article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
an experiment using x - ray single crystal diffraction at the swiss - norwegian beam lines ( snbl ) of european synchrotron radiation facility ( esrf ) on a single crystal of cubic perovskite bmt ( @xmath0 ) has shown an interesting diffuse scattering pattern , consisting of cube - shaped domains of nearly constant intensity , with the centers located on the ( 1/2,1/2,1/2 ) points of the reciprocal lattice and with an edge of @xmath6 of a reciprocal lattice unit ( rlu ) .
all the cubes have the same intensity level , apart from the usual slow modulations due to atomic and thermal factors .
bmt belongs to perovskites of formula ab@xmath1b@xmath2o@xmath3 ( a = ba , b@xmath4ta , b@xmath5 mg ) where different species on the b - site can be ordered to various degrees .
the scattering depends on the site occupancy and when the ordering is directly related to the pair correlation function of the different atomic sites @xcite .
in fact , bmt in its cubic phase is necessarily intrinsically disordered in the b - site . in this paper
we shall mainly explore the chemical order at the b - site and show that the cube - like diffuse scattering can be fully explained in this way .
lattice deformations and displacements , due to different atomic radii of mg and ta ( see ref .
@xcite ) , are expected .
so there is additional diffuse scattering originating from these correlated displacements .
in fact streak - like diffuse scattering - relatively weaker - has been found at the foot of the cubic bragg peaks , see fig .
we shall address also this point in a future paper .
the simplest possibility for b - site ordering is the formation of subregions enriched in either mg or ta cations .
this possibility , however , has been thoroughly ruled out as energetically unfavourable in ab@xmath1b@xmath2o@xmath3 perovskites @xcite because of the large charges in the subregions .
there are then different homogeneous ways of ordering the b@xmath7-b@xmath8 cations .
these are : * homogeneous random distribution of b - cations ; * correlated distribution with occupation numbers being a well - defined function of distance between two lattice nodes , but without periodicity ; * periodic long range ordering of the mg / ta concentrations in different sublattices . situation i ) would give an almost wavevector - independent diffuse scattering , but this is not observed .
b - site ordering of type iii ) has been observed in bmt , with simultaneous lowering of the cubic symmetry .
in fact , bmt may exist in an ordered trigonal structure ( @xmath9 ) modification @xcite and it has been shown that this modification @xcite mainly entails perfect periodic ordering of the b - site .
the trigonal lattice is essentially a threefold superlattice of the cubic one , with extremely small geometric distortions , realized by stacking periodically a sequence of one mg plane and two ta planes orthogonally to one of the cubic @xmath10 directions .
the atomic sites are also very close to those of the underlying cubic sublattice , with only small atomic displacements consistent with the reduced symmetry .
it is noteworthy that the trigonal bmt superstructure - usually cast unto four equally populated domain orientations with the threefold axis along either of the cubic @xmath11 directions , so that the average symmetry is conserved - can be considered as a commensurate distortion of cubic bmt . in this framework
, the modulation contributes eight additional ( @xmath12 ) satellite bragg peaks to the diffraction pattern , and these coincide with the vertices of the diffuse scattering cubes .
another partial b - site ordering type of class iii ) has been observed @xcite in other ab@xmath1b@xmath2o@xmath3 cubic perovskites ( pmt , with a = pb , b@xmath4ta , b@xmath5 mg ) . these have ( 1/2,1/2,1/2 ) satellites and correspond to a partial b - site ordering with a 2@xmath132@xmath132 @xmath14 superstructure unit cell .
the b - sites are of two types , one of pure b@xmath8 and the other containing the rest of b@xmath8 and all of b@xmath7 cations , and the two b - site types arrange in a nacl - type structure that consists of two inter - penetrating fcc lattices .
these satellites coincide now with the center of the diffuse scattering cubes observed for bmt . diffuse scattering cubes straddle the reciprocal space location of the ( 1/2,1/2,1/2 ) satellites and the ( @xmath12 ) satellites identifying the two kinds of periodically ordered type iii ) b - site structures . although this diffuse scattering is not bragg - like , in a sense that will be specified in sec . [
sec : smooth ] , this suggests that the diffuse cubes result from b - site ordering as well , possibly from a non - periodic order of type ii ) .
we stress here that the observed cubes of diffuse intensity are not ideal , as careful inspection ( _ cf .
[ fig1 ] ) shows that `` cubes '' have a minimum in the center about 10% lower than at the periphery and the corresponding intensity distribution is not a flat - top .
we see in particular that intensity tends to increase near the vertices ( the ( @xmath12 ) satellites of the trigonal superstructure ) and decrease in the center - the ( 1/2,1/2,1/2 ) satellite of the hypothetical partly ordered nacl - type superstructure , indicating an embryo of transition to the trigonal phase .
we must therefore consider the phase with idealized cubes as a limiting situation that may not be perfectly realized in our sample . as we argue in the following that the phase with ideal diffuse cubes is not the ground state at room temperature but only a metastable phase , kinetically hindered from transforming to the trigonal ground state , this is perfectly possible , as a germ of transition may have occurred due to finite quenching speed .
however , in this framework , the limiting phase with ideal diffuse cubes is a key to understanding the b - site ordering phenomena in bmt , and therefore we will deal mainly with this phase . as we are discussing the multiple phases of bmt
, we should comment on an important point about phase stability . in the thermodynamic sense ,
a stable structure observed in an experiment corresponds to a minimum of the free energy at the conditions of crystal formation . at ambient conditions
the mg - ta distribution is quenched and experimentally observed configurations may not necessarily correspond to a true ground state .
a disordered or incompletely ordered state can be stabilized at high temperature by a strong entropy term and could be a metastable state at lower t. in fact , at room temperature the diffusion of mg and ta is practically nonexistent , so that the non - ergodic effects in the scattering experiments can be neglected .
if the first born approximation is valid and the scattering can be considered as elastic , the scattered intensity is the fourier transform of the scattering density s pair ( two - body ) correlation function @xcite .
the scattering density is the atomic structure , and for x - rays it is the sum of the single atom electron densities and the only degrees of freedom come from the spatial distribution ( structure ) of the atoms themselves .
the pair correlation function ( crystallographers name it the `` patterson function '' ) is the information contained in the scattering pattern .
a natural ambiguity arises in the relationship between the atomic structure and the pair correlation function , as there may be many atomic structures that have the same pair correlation function @xcite . in some cases ,
this is not an accidental degeneracy .
when the internal energy @xmath15 depends only on the pairwise interactions , different structures with identical pair correlation functions are isoenergetic . at any finite temperature @xmath16 ,
the multiplicity of configurations increases the entropy @xmath17 and this in turn decreases the free energy , @xmath18 , to the advantage of the stability of the ( statistically defined ) structure .
we intend to show that the b - site order in bmt , that gives rise to the cubes of the diffuse intensity belongs to this case , based on the fact that pair ( electrostatic ) interactions between the b cations are sufficient to explain the phase diagram of the different bmt phases which derive from the ordering of b - sites .
the pair ( two - body ) and the set of the higher order ( many - body ) correlation functions can fully describe any structure in an unambiguous way , including stochastically defined ones @xcite . considering higher order correlation functions is necessary , however , only when they make important contributions to the internal energy .
this is sometimes the case , for instance , for structures that show displacive diffuse scattering @xcite . in any case , higher correlation functions are also a way of _ describing _ a structure in the crystal - chemical sense .
this is especially convenient when the structure itself can not be easily represented because it is only statistically defined . however , simpler ways of presenting the crystal - chemical informations are also possible , see sec .
[ mcsol ] .
our approach is as follows .
the pair correlation function that is related to the diffuse cubes can be simply analyzed in terms of the b - site chemical order as a site concentration pair correlation function @xcite .
this turns out to have a simple analytical form that shows a rather peculiar power - law decrease with distance that is observed in only a few cases for rather exotic systems .
monte - carlo simulations were performed for a large scale in order to find the b - site configurations that give the same pair correlation as the diffuse intensity .
successfully determined configurations were then analyzed in terms of a long - range order parameter showing that these partially ordered configurations that are compatible with the observed scattering phenomena are not distinguishable from completely disordered systems .
next , we consider that bmt is an ionic compound and that the electrostatic energy is the main term in the crystal hamiltonian .
the electrostatic energy of a rigid system depends only on pair interactions and the electrostatic energies and the entropy terms for the different bmt phases were calculated .
these gave a good correspondence with the observed phase diagram , confirming that the cubic phase must be a high - temperature ground state of bmt that is metastable at room temperature .
the single crystals of bmt were synthesized according to the procedure described by galasso and pinto @xcite .
the powder of bmt was prepared according to two - stage synthesis described in ref .
@xcite and is essentially the same as used in the neutron diffraction study @xcite .
the experiments were performed on x - ray scattering instruments at the esrf and sls . at the esrf the diffuse scattering from bmt was measured at the swiss - norwegian beam line ( snbl ) and the xmas beamline , and at the sls the measurements were made at the beamline x06sa .
more details of the measurements and the experimental results for bmt are given in ref .
@xcite . in this paper
we concentrate on the unusual diffuse scattering observed from bmt and typical results are shown in fig .
similar results were obtained from the sls diffractometer x06sa , but from the xmas data the scattering contained sharp peaks at the ( h+1/3 , k+1/3,l+1/3 ) positions showing the structure was trigonal .
a bmt powder sample was also studied at the sls on the beamline x04sa .
it was revealed that the trigonal satellites were significantly broader than the intense peaks of the cubic phase showing that the domain size was about 20 nm for the trigonal structure .
these results support the suggestion that our powder and the surfaces of relatively large bmt crystals consist to some extent of trigonal - structured domains while the bulk of crystals is cubic and shows the diffuse scattering of fig .
[ fig1 ] . to @xmath19 .
, title="fig:",scaledwidth=49.0% ] to @xmath19 .
, title="fig:",scaledwidth=49.0% ] to @xmath19 .
, title="fig:",scaledwidth=49.0% ] this paper concentrates on the discussion and modeling of the unusual cube - shaped distribution of scattering and on how the different structures of bmt can exist .
it is suggested that the structures are unable to relax when rapidly cooled from high temperature .
bamg@xmath20ta@xmath21o@xmath3 is cubic with lattice parameter @xmath22 nm .
the a cation ( ba ) is located within the unit cell in ( 1/2,1/2,1/2 ) , the mixed b - site is in ( 0,0,0 ) and the three oxygens occupy the ( 0,1/2,1/2 ) , ( 1/2,0,1/2 ) , ( 1/2,1/2,0 ) positions .
in this section a simple mathematical model of the diffuse intensity is described .
the model is simple because each atom is assumed to be a point scatterer , with element xx having a constant scattering length @xmath23 .
this means neglecting the atomic form factors and debye - waller factors , whose effect is just a smooth decay of the intensity at large transferred momentum values .
the scattering from bmt has diffuse scattering cubes centered at @xmath24 with half - side 1/3 ( all in rlu ) .
a simple mathematical description of their shape can be given as follows .
in one dimension , define a box function @xmath25 then a periodic sequence of such boxes centered on the half - integers is @xmath26 multiplying three such terms ( one for each coordinate ) and adding a constant background ( that we will show to be necessary ) we obtain a model of the x - ray diffuse intensity : @xmath27 here @xmath28 is the halfwidth of the observed cubes in reciprocal lattice units . taking the fourier transform of the periodic array of cubes of x - ray diffuse intensity @xmath29 ( eqs .
( [ icube1],[icube2 ] ) ) , we evaluate the pair correlation function as @xmath30 note that , as @xmath31 is periodic and its fourier transform is necessarily a weighted set of dirac deltas on the direct lattice @xmath32 where @xmath33 is a positive constant related to the actual scattering lengths ( see sec . [
sec : bord ] , eq .
( [ etaval ] ) ) and @xmath34 are the correlation coefficients .
in fact , substituting @xmath31 from eqs .
( [ icube1],[icube2 ] ) , we obtain @xmath35 \label{eq : corr } \ ] ] whereas the correlation coefficients are given by @xmath36,\qquad { { { \ensuremath{{\bm{m}}}}}}\neq(0,0,0 ) ; \label{eq : corr1}\\ s_{(0,0,0)}&= & { { \ensuremath{\displaystyle{\frac{{{\ensuremath{\displaystyle{27i_b+i_0}}}}}{{{\ensuremath{\displaystyle{27\eta^2 } } } } } } } } } \label{eq : corr2 } \ ] ] here @xmath37 are direct lattice nodes , where the density ( and the pair correlation function ) are different from zero .
the distance @xmath38^{1/2}$ ] , so that @xmath39 , with @xmath40 , depending on the direction .
this is a very rare case of power - law correlation decay of structural origin . in quite exotic systems , including spin glasses and nematic crystals , such correlations have been described @xcite and named as quasi - long range order ( qlro ) .
we shall name qlrco ( as quasi - long - range cation order ) the kind of order exhibited by the bmt sample .
two parameters are often used to characterize the degree and extension of the structural chemical order .
the first is the long - range order parameter of refs .
@xcite , see appendix , sec .
the value of the long - range order parameter @xmath41 characterizes periodically ordered structures , as the type iii ) above , while zero is the value for randomly ordered structures , as type i ) .
conversely , the short - range order parameters of ref .
@xcite characterize the local degrees of order - supposedly when long - range order is absent .
we obtain ( see sec . [
corr : lro ] , sec .
[ chem : sro ] ) that both long- and short - range parameters indicate a substantial degree of order , which is an unusual occurrence .
the model of the diffuse intensity described above can be improved in one significant aspect .
the cubes as observed are not perfect , as there are intensity fluctuations within them .
their edges , especially , are not very sharply defined .
this effect can be included by convoluting @xmath42 with a point - spread function ( psf ) @xmath43 .
accordingly , the pair correlation function @xmath44 is multiplied by the fourier transform of the psf , @xmath45 as shown in fig . [ fig3 ] .
the latter will be named here the _ correlation decay function _ ( cdf ) .
this will give a stronger decay law for the pair correlation function .
this is just what happens with every known diffraction phenomenon - as far as infinite crystals do not exist , even the most perfect standard crystal will show bragg peaks convoluted by a suitable psf and pair correlations damped by the corresponding cdf .
we will not , in the following , burden the notation by including a psf , but we will calculate some of its effects in sec .
[ eesec ] .
different possible psf are discussed in sec .
[ app ] . here , as an example , we consider a lorentzian psf along the cubic axes ( fig . [ fig3 ] )
@xmath46}}}^{-1 } \label{exampsf}\ ] ] its fourier transform is @xmath47 the correlation decay length is @xmath48 . from our data , taking @xmath49 as the rise of the half - width of the cube edges , we can coarsely estimate that @xmath50 is about 10 unit cells .
this is well below the coherent domain size , however it remains a considerable distance when dealing with diffuse scattering .
as the psf @xmath43 is convoluted with the scattered intensity , the cdf @xmath45 is multiplied by the pair correlation function , that is the fourier transform of the intensity .
its effect is then to limit the spatial extension of the pair correlation function .
this is a familiar issue in crystallography .
crystal size can be estimated from the broadening of the bragg peaks , due to the limited extension of the crystal - and , in turn , of the pair correlation function .
we wish to remark here that the cube scattering is quite different from bragg scattering with respect to its response to a limitation of correlation length .
in fact , convoluting the psf with a bragg peak gives the psf itself .
this is the reason for the familiar behaviour of bragg peaks - namely , that their width is inversely proportional to the size of the correlated domain ( a good estimate of this is @xmath50 ) .
the diffuse cubes , however , show a negligible change in width , second - order or more in @xmath51 .
this scattering form is therefore fundamentally different from bragg scattering .
to explain such diffuse scattering in terms of b - site cation order we split the scattering density @xmath52 as @xmath53 with @xmath54 the periodic average and @xmath55 a non - periodic zero - average fluctuation . if the b site is the only one that varies , all of ba and o partial densities are included in @xmath54 . as already mentioned , we simplify the picture by considering point - like atoms .
then @xmath56 is zero everywhere except on the b sites , and for b = ta @xmath57 ; for b = mg @xmath58 .
if @xmath59 and @xmath60 are the scattering lengths of mg and ta , respectively , then @xmath61 as @xmath55 assumes only two values ( @xmath62 on ta sites and @xmath63 on mg sites ) we can consider ( see @xcite ) the site concentration of one of the two species , say mg , in each lattice node .
clearly this can be only 0 or 1 .
we indicate it with @xmath64 , that is 1 if site @xmath65 is occupied by mg and 0 otherwise .
similarly we take @xmath66 as the ta local concentration , that is 1 if site @xmath65 occupied by ta and 0 otherwise . as we have no vacancies , @xmath67 .
averaging over all sites , we have @xmath68 furthermore , as @xmath69 and @xmath70 can be only 0 or 1 , @xmath71 so the r.m.s
. fluctuation of the concentration is given by @xmath72 the fluctuation of the density is given by @xmath73 in order to study the diffuse scattering from bmt , we need to evaluate the pair correlation function of @xmath55 : @xmath74 here @xmath75 is the crystal volume , @xmath76 the number of lattice nodes within it , @xmath77 the unit cell volume ( taken as 1 hereafter , hence the last equality ) .
we shall equate this to the pair correlation function derived from the data , see eq .
( [ eq : corr ] ) .
the rightmost form is because @xmath55 is a weighted dirac lattice , and this property is invariant upon self - convolution .
this is the same form as eq .
( [ eq : corr ] ) , so clearly now we have to calculate the @xmath34 pair correlation coefficients for a given fluctuation density ( eq . ( [ flucden ] ) ) and to compare them with those obtained from the fourier transform of the data ( eqs .
( [ eq : corr1],[eq : corr2 ] ) ) . substituting eq .
( [ flucden ] ) into eq .
( [ eq : gdelta ] ) gives @xmath78 where , comparing with eq .
( [ eq : gdelta ] ) , we obtain @xmath79 now , the probability of finding a pair of mg atoms separated by a lattice vector @xmath80 is given by @xmath81 similarly we can calculate the probabilities of all the other pairs ( mg - ta / ta - mg and ta - ta ) for any spacing vector @xmath80 , and the result depends only on the @xmath34 correlation coefficients : @xmath82 where the subscript mix refers to mg - ta pairs , in any order .
therefore , if we know the correlation coefficients we can calculate all the properties that depend on the pair interactions . comparing eq .
( [ eq : corr ] ) with eqs .
( [ greq],[bongo ] ) , considering @xmath83 , we obtain @xmath84 \qquad\leftrightarrow\qquad i_0={54}{\eta^2}-27i_b \label{theo000 } \]]we define a parameter @xmath85 , so that if @xmath86 , then @xmath87 this result describes how much of the total diffuse intensity in one brillouin zone is a constant background and how much comes to form the cubes of diffuse scattering . as described by eq .
( [ pmgmg ] ) , @xmath88 is a probability , hence it must range between 0 and 1 . in order for it to be positive for all @xmath89 , @xmath90 for @xmath91 , this constraint - using eq .
( [ eq : corr1 ] ) - implies @xmath92 which tells us that a consistent fraction - at least about 40% - of the diffuse intensity must be spread as a constant background while the rest forms the diffuse cubes .
this is also a lower bound only , so possibly the background fraction is even larger . in eqs .
( [ pmgmg],[pother ] ) we have calculated the probability of finding pairs of atoms of equal or different species ( limited to the b sublattice ) for each separation vector @xmath80 . that is all we need to calculate the electrostatic energy .
in fact , the structure consists of the fixed part ( the a sublattice and the oxygen sublattices ) and the variable part ( the b sublattice ) .
a result of ref .
@xcite is that the electrostatic energy does not depend on the interaction terms between the different sublattices ; the fixed part then has no influence except for a constant term and one needs to calculate only the interactions between atoms of the b sublattice .
similarly to the scattering density ( eq . ( [ eq : flu ] ) ) , we can separate the charge density @xmath93 into the the average part and fluctuating part @xmath94 with @xmath95 the average component and @xmath96 a non - periodic zero - average fluctuation . again , on ta - occupied sites we will have @xmath97 and on mg sites ,
@xmath98 with @xmath99 if we neglect covalent bonding and assume the formal charges @xmath100 , @xmath101 , then we have @xmath102 .
it can be shown @xcite that only the fluctuation density @xmath96 - that is the only term that varies between the different possible phases of bmt - needs be included in the calculations .
therefore , the electrostatic energy is simply expressed - using eqs .
( [ pmgmg],[pother ] ) - as @xmath103 the prime meaning that @xmath104 is excluded from the summation .
an additional factor of 1/2 was included in order not to count twice each bond . at this point we know that a distribution of b - site cations whose correlation coefficients are given by eqs .
( [ eq : corr1],[eq : corr2 ] ) would produce a diffuse intensity as described by eqs .
( [ icube1],[icube1 ] ) , that well matches the observed diffuse scattering cubes .
we can not , however , prove analytically that a distribution of two different cations on a cubic lattice with respective abundances of 1/3 and 2/3 exists that can satisfy eqs .
( [ eq : corr1],[eq : corr2 ] ) ( or alternatively , produce a diffuse intensity as in eqs .
( [ icube1],[icube2 ] ) ) . in order to prove that such a distribution exists , we decided to try to fabricate one or more by a monte - carlo simulation . in this way
we could also get some structural insight by analyzing the configurations so obtained . a large cube ( @xmath105 unit cells with the total number of atomic sites @xmath106 ) was placed on the lattice nodes with a mg atom - represented by a scattering length @xmath107 - or by a ta atom - represented by a scattering length @xmath108 .
these values are chosen so that the average density ( _ cf . _ sec .
[ sec : bord ] , eq .
( [ eq : flu ] ) ) @xmath54 is zero and only the fluctuation density @xmath55 is not zero ; moreover , we set @xmath109 for simplicity .
the fft power spectrum was evaluated from the intensity @xmath29 in one brillouin zone .
the atoms were moved around until a good agreement with eqs .
( [ icube1],[icube2 ] ) was achieved .
the simulation was very successful .
[ qbint ] a ) a section of the calculated diffuse intensity is shown for one of the optimal configurations found .
the result of the calculations compares well with the experimental data .
( [ icube1],[icube1 ] ) ) .
other sections ( not plotted ) are consistent with the one shown . + plane .
+ b ) : red bars : correlations @xmath34 - for @xmath110 - from the monte carlo calculation . as allowed by the cubic symmetry , values have been averaged over @xmath111 , and plotted _ vs. _ @xmath112 . blue line : correlations @xmath34
calculated ( on the same @xmath80 points ) using eqs .
( [ eq : corr1],[eq : corr2 ] ) .
this is plotted as a double envelope curve ( because of the factor @xmath113 ) to guide the eye .
green line : same as the former , multiplied by a cdf @xmath114 as from eq .
( [ examcdf ] ) with @xmath115 . ,
title="fig:",scaledwidth=49.0% ] + + plane .
+ b ) : red bars : correlations @xmath34 - for @xmath110 - from the monte carlo calculation . as allowed by the cubic symmetry , values have been averaged over @xmath111 , and plotted _ vs. _ @xmath112 .
blue line : correlations @xmath34 calculated ( on the same @xmath80 points ) using eqs .
( [ eq : corr1],[eq : corr2 ] ) .
this is plotted as a double envelope curve ( because of the factor @xmath113 ) to guide the eye .
green line : same as the former , multiplied by a cdf @xmath114 as from eq .
( [ examcdf ] ) with @xmath115 .
, title="fig:",scaledwidth=49.0% ] several equivalent configurations were constructed as above and used to evaluate the pair correlations .
firstly , the ratio @xmath116 for the solutions was found , and it resulted that @xmath117 .
this gives a value of the intensity partition parameter @xmath118 - actually much lower than its upper bound of @xmath119 .
from that , exploiting eq .
( [ xoft ] ) in the appendix , the value of the long - range order parameter was extracted as @xmath120 .
we can then evaluate @xmath121 , @xmath122 .
[ qbint]b ) shows the correlations @xmath34 from the monte carlo compared with those evaluated above , for @xmath89 along the family of @xmath123 directions ) .
the good agreement with values calculated from eqs .
( [ eq : corr1],[eq : corr2 ] ) , with and/or without an additional psf with @xmath124 , shows more quantitatively that the monte carlo converged simulation converged towards the observed pattern . in order to display some crystal - chemical information in the montecarlo - obtained qlrco configurations
, we directly evaluated the probabilities that a mg ( respectively , ta ) atom has a mg ( respectively , ta ) neighbor at a distance @xmath112 along any of the equivalent @xmath125 directions .
the results are shown in fig .
[ xxcorr ] and have been calculated as follows .
each site in a cluster is treated as the central atom .
the fraction of atoms of a given species over the neighbors having distance @xmath112 along the family of @xmath123 directions was calculated .
periodic boundary conditions were used .
the fractions for all four chemical pairs where the atom in the origin is mg or ta and the neighbor is mg or ta , respectively . in fig . [ xxcorr ] , we plot the ta - ta , ta - mg , mg - ta , mg - mg @xmath126 chemical pair fractions for the qlrco cluster . for comparison we add the values for a cluster with randomly assigned mg or ta at each site , fulfilling a composition mg@xmath20ta@xmath21 . these of course do not depend on the choice of the atom at the origin , and amount to 1/3 when the neighbor is mg and 2/3 when it is ta . from the values of the ta - mg ( or mg - ta ) chemical pair fractions
we immediately obtain the value of the short - range order parameter for the first coordination shell @xcitefor the qlrco phase . the parameter @xmath127 is found to be @xmath128 . for this composition it has a maximum ( in absolute value ) at @xmath129 , for the fully ordered state , while it is zero for the random state . combined with similar results obtained for the long - range order parameter in sec .
[ chem : sro ] this gives a further justification for the name quasi - long range order. separation vectors .
these correspond to @xmath130 ( _ cf .
( [ pmgmg],[pother ] ) , the factor 3 weighs the probability of the central atom being mg ) . for the same cluster : red : mg neighbor fractions for a central ta atom , equal to @xmath131 ( here factor 3/4 is obtained as product of 1/2 - because @xmath132 includes both bond directions , and 3/2 - inverse of the probability of the central atom being ta ) .
green : ta neighbor fractions for a central mg atom , @xmath133 .
brown : ta neighbor fractions for a central ta atom , @xmath134 . for a cluster with randomly assigned mg or ta at each site , but with the exact composition mg@xmath20ta@xmath21 , we plot in black the mg neighbor concentrations of any central atom ; in violet , the ta neighbor concentrations of any central atom .
these values can be calculated as above , simply substituting @xmath135 .
, title="fig:",scaledwidth=49.0% ]
in this section and the next we shall demonstrate how the electrostatic energy and the configuration entropy , as calculated from the structure and pair correlations for the different possible bmt phases is sufficient to explain the phase diagram of bmt .
in particular , the fact that the qlrco phase is metastable at room temperature , and the fact that the nacl phase , although common in similar perovskites , has never been found for bmt .
the electrostatic energy , is not the only component of the crystal energy , but has been previously identified as the driving force of the cation ordering in complex perovskites . at least it has been shown @xcite that electrostatic energy - based predictions describe quite effectively the ordering behaviour of a large class of perovskites .
the electrostatic energy was evaluated in a similar fashion as in ref .
@xcite . as already mentioned in sec .
[ ee - cor ] , only the fluctuation term relevant to the b - cation sublattice need to be calculated , because the other contributions can be considered either as fixed or as zero .
we have used a relative permittivity @xmath136 and a cubic cell side @xmath137 and the formal charges ta@xmath138 , mg@xmath139 , ba@xmath139 , o@xmath140 which gives @xmath102 .
it is easy to scale the results to more realistic values of the parameters later on .
the electrostatic energy was calculated using different ways . in all cases ,
the zero of the energy was fixed to the energy of a reference perovskite - batio@xmath141 - that has a + 4-charged cation ( equal to the average charge of mg@xmath20ta@xmath21 ) on the b site .
electrostatic energies were calculated - apart from the reference batio@xmath142 perovskite , labeled uniform , for all the different b - site ordering discussed in this paper . the bmt phase ( we label it qlrco for conciseness ) , the structure with nacl - type partial b - site order ( label nacl ) , the trigonal phase with full b - site order ( label @xmath143 ) , and the phase having fully random b - site occupancies ( label random ) have been calculated . as a first method of calculation
, we created large clusters of atoms and computed the interactions directly , summing up all atom pair contributions .
calculations of this kind were performed for clusters of increasing size up to @xmath144 cubic cells , at which limit the results were well stabilized and appeared to have reached their asymptotic value ( with residual oscillations smaller than @xmath145 ev / formula unit ) .
fractional o- and ba- site occupancies at the boundary have been used to keep the neutrality of the clusters .
this fact @xcite , together with the cubic cluster symmetry , explains the very rapid convergence of the electrostatic energy sum . as a second approach ,
the madelung method @xcite was used .
this gave consistent results within @xmath146 ev / formula unit .
thirdly , we use the knowledge of the correlation coefficients @xmath34 ( cf .
[ sec : corr ] ) . in this case , the electrostatic energy is simply expressed as in eq .
( [ calcule ] ) .
such sums were evaluated in all cases from the analytical @xmath34 .
again , we evaluated them on cubes of increasing size ( up to @xmath147 unit cells ) in order to verify the convergence - which was always excellent .
calculation times in this case were also extremely fast - less than one minute on a single pc for the largest cubes with @xmath148 nodes . with this latter method for the qlrco case
was also evaluated - the effect of a psf spreading the diffuse scattering cubes ( see fig .
[ cdffig ] ) .
it turns out that even with small correlation lengths ( @xmath149 unit cells ) the energy was within few % from the value as @xmath150 . ) - for the qlrco structure , when correlations are reduced by various decay functions ( cf .
[ herecdf ] , sec .
[ app ] ) . , title="fig:",scaledwidth=49.0% ] for qlrco , nacl and random cases ( which are not deterministic ) we have repeated the calculation for many configurations , evaluating the average values and their standard deviations . in all cases
the number of trials was chosen high enough for the standard deviations to be below @xmath145 ev / formula unit .
permittivity was taken as @xmath136 , and the true energy levels can be obtained by dividing the results by the permittivity of the material .
the results for the uniform structure could be compared with those evaluated using the madelung constants ( see @xcite ) and turn out to be quite accurate . also the electrostatic energy of the @xmath143 structure - calculated by the madelung method by assuming that the site geometry remains cubic is found in ref .
@xcite and compares well with our results .
furthermore we are not interested in absolute values but in comparative ones , and these are numerically much easier to compute than the absolute ones . for each structure
we evaluate @xmath151 .
the results show that , with respect to the uniform structure , in order of increasing stability : * the random structure has the same energy within 0.1% , @xmath152 ev / formula unit ; * the qlrco structure has 1.1% lower energy or about @xmath153 ev / formula unit ; * the nacl structure has 1.8% lower energy or about @xmath154 ev / formula unit ; * the @xmath143 structure has 3.0% lower energy or about @xmath155 ev / formula unit .
noteworthy is the level spacing ratio @xmath156 which is smaller than @xmath157 there are several factors which will influence the absolute values of these energy levels , and the relative values of @xmath158 and @xmath159 as well .
these are : * the energies have to be divided by the material s relative ( static ) permittivity @xmath160 , that can be large in perovskites .
for bamg@xmath20ta@xmath21o@xmath142 it is found that @xmath161 @xcite .
this factor would not change the ratio . *
the electrostatic energy is evaluated using the full formal charges of the ions .
covalency corrections may play a role .
a value of @xmath162 can be extrapolated from the valence sum corrections evaluated for bmt in ref .
this entails an absolute change of the energy levels , not affecting the ratio @xmath158 . *
the electrostatic energy is inversely proportional to the cubic lattice parameter @xmath163 .
appropriate scaling is needed .
however , for the known bamg@xmath20ta@xmath21o@xmath142 phases , @xmath143 and qlrco , @xmath163 ( evaluated from the cubic main reflections ) is the same within one part in @xmath164 @xcite .
this effect can not change the ratios . *
geometric distortions are present in the ordered nacl and @xmath143 structures , but they affect the distances on few parts in @xmath165 @xcite , and as their effect on the energies is of comparable order , we can neglect them altogether . *
a psf is observed , leading to smoothed cube edges .
this term is possibly accidental , depending on the detailed crystallization kinetics , so it will not be considered quantitatively in the following .
its effect is also small enough ( see fig . [ cdffig ] ) that it can not qualitatively affect our results .
the crystal structure can be stabilized by the entropy . for disordered structures configurational
entropy plays an important role .
a mixed - occupancy atomic site with occupancy @xmath166 for one species and @xmath167 for another gives an entropy contribution of @xmath168\ ] ] in the @xmath143 structure no mixed sites exist , therefore they have no entropic contribution to the gibbs free energy .
there will be an entropy term related to the domain distribution , but that is smaller by many orders of magnitude .
the maximal configurational entropy per formula unit is expected in the random structure and it amounts to @xmath169=54.85\ \mu\text{ev / k}\ ] ] for the qlrco structure , as discussed , the entropy value is slightly smaller , but the difference can be evaluated numerically ( in a simple but tedious way ) from the correlation coefficients and the result is that the correction is negligible , so the entropy is @xmath170 . for the nacl structure ,
only 50% of the sites are mixed so the entropy is @xmath171 .
one more important contribution is the phonon entropy
. a different b - site degree of order can in principle affect the phonon entropy .
however , this effect is much smaller than the configurational term .
this is supported by the scarce literature , see _ e.g. _
the specific heat @xmath172 for bmt has been measured @xcite , however the phase was not fully characterized at the time ( being , however , either qlrco or @xmath143 )
. we can only compare the entropy evaluated from the experimental specific heat with data for pb containing relaxor perovskite ( pbmg@xmath20ta@xmath21o@xmath142-pmt ) @xcite .
the phonon entropy of pmt extrapolated to t@xmath173 exceeds that of bmt by 59 @xmath174ev / k .
this large difference can be explained either by the polar nano - regions which are only present in pmt @xcite or by the difference in the masses of pb and ba .
entropy is important at high temperature for stabilizing different structures . for instance , considering the various free energies as calculated in the previous section , we can construct a phase diagram of bmt ( fig . [ phased ] ) .
we have taken , as previously stated , realistic values of @xmath175 ( assumed constant with @xmath16 ) , @xmath162 , and a relative linear expansion coefficient of 9@xmath176 .
we have not included any contribution from imperfect ordering ( namely , the effect of a cdf limiting the spatial extent of correlations ) because it is not easy to evaluate quantitatively and it is not an intrinsic effect but rather a kinetic effect while its contribution is small ( fig .
[ cdffig ] ) and the leading approximation would only compress the energy range .
the entropy term is canceled by the energy difference between qlrco and @xmath143 phases at about 1600 @xmath177c , that is about the melting point . above this temperature
the qlrco state would be stable while the @xmath143 state would be stable below .
we can also see from fig .
[ phased ] that there is no region that is stable for the nacl - ordered phase , and in fact , this phase has never been observed in bmt , at variance with pmt @xcite . due to the slow kinetics associated with cation ordering , chemical ordering
is usually achieved only after annealing for a long time at very high @xmath16 , where the equilibrium state may be dictated by the entropy .
cooling - due to the slow kinetics - can not change the high-@xmath16 equilibrium state so that it would remain preserved ( metastable ) at room temperature .
the random structure is always less stable than the qlrco structure , because of the tiny entropy difference and the large energy difference that favours the latter structure .
therefore the random structure is a disordered metastable state .
the ground state - or the thermodynamic equilibrium state at 0 k - is the @xmath143 structure .
the true high - temperature thermodynamic equilibrium state is qlrco , because the electrostatic energy difference with the @xmath143 structure is overcome by the entropy term at high @xmath16 , even if the dielectric constant is of moderate size .
the nacl structure is intermediate between these two . given that its gibbs free energy decreases with increasing temperature half as fast as the free energy of the qlrco state
, the ratio @xmath158 becomes crucial to determine if the nacl order can be the thermodynamic equilibrium state in some temperature range .
in fact for @xmath178 the gibbs free energy of qlrco is always lower than that of nacl structure , while for @xmath179 there exists a temperature interval where nacl structure is the equilibrium state .
the qlrco structure is favoured above that temperature and the @xmath143 structure below .
we are grateful for the opportunity to carry out the measurements using the spectrometers : snbl and xmas at the esrf , and the beamlines x04sa ( ms - powder ) and x06sa at the sls .
we thank dr .
laurence bouchenoire for invaluable assistance in the xmas experiment .
firstly , as in eqs . ( [ exampsf],[examcdf ] ) , we consider the symmetry of the psf to be cubic .
a lorentzian psf and the corresponding exponential cdf are presented in eqs .
( [ exampsf],[examcdf ] ) . a simpler psf would be a sinc function , corresponding to a box function cdf .
the negative parts of the sinc tails are , however , a problem .
therefore we consider a sinc squared function : @xmath180 correspondingly , if the symmetry is spherical , the psf / cdf are given by the radial lorentzian or exponential : @xmath181}}}^{-2 } ; \qquad\text{cdf:}\quad \widetilde{d}({{\ensuremath{{\bm{r}}}}})={{\ensuremath{\mathrm{e}}}}^{-2\pi r k_0},\ ] ] and the analog of the squared sinc / box pair : @xmath182}}}^2 ; \quad\text{cdf:}\ \widetilde{d}({{\ensuremath{{\bm{r}}}}})=(1+\pi r k_0/2)(1-\pi r k_0)^2\mathrm{box}{{\ensuremath{\left({\pi r k_0}\right)}}}\ ] ] the trigonal superstructure has as the 3-fold axis one of the equivalent @xmath10 cubic axes .
this means that it exists in four different orientations , where the 3-fold axis @xmath188 is @xmath189 , @xmath190 , @xmath191 , @xmath192 , respectively .
the b - site superstructure is obtained by positioning in a lattice node @xmath65 on a mg atom if @xmath193 .
therefore , we will have pure mg planes alternating with pairs of ta planes , stacked orthogonally to @xmath188 . in any crystallite
, domains of the four different orientations will be present in equal volumes , therefore conserving the global cubic symmetry , in agreement with landau s theory .
the correlations can be calculated for each orientation and then averaged .
however , one can also calculate them from the scattering pattern , as an average over domains .
the superstructure satellites - considering equipopulated domains - are eight equal bragg peaks ( each with intensity @xmath194 ) at @xmath195 .
then @xmath196 where @xmath112 is the number of components of @xmath89 that are equal to zero , or @xmath197 . from the condition @xmath184
, we obtain @xmath198 , so that @xmath199 the hypothetical nacl - type superstructure is simpler .
there is only one bragg peak at ( 1/2,1/2,1/2 ) with intensity @xmath200 plus a small constant background @xmath201 due to residual randomness on half of the sites .
so one obtains @xmath202 as half the sites contribute to the background and the occupancy of the random sites is always divided as 1/3 - 2/3 , @xmath203 and again using the condition @xmath184 , we have @xmath204 in this section we show how the former results can be related to the formalism of the long - range order parameter . a simple way of describing systems where there is a fully ordered state is by the long - range order parameter @xcite . here
we will not go to all the detail as it is a well - known theory .
for bmt the ordered state is the trigonal modification . here
the cubic b - site lattice is divided into 3 sublattices , denoted ( a),(b),(c ) for convenience of which one is occupied by mg and two by ta .
there are four ways of choosing the three - fold axis of the trigonal structure and for each three permutations of the mg sublattice between ( a),(b),(c ) .
suppose that we chose one axis and that the mg sublattice is ( a ) . in the ordered state
all mg atoms are actually on the ( a ) sublattice ; if we have @xmath76 total atoms , we will have @xmath205 mg atoms in the ( a ) sublattice . in the random state ,
the mg concentration on any site will be the same , so we will find only @xmath206 mg atoms in the ( a ) sublattice .
we can parametrize this situation if we say that the number of mg atoms on the ( a ) sublattice is given by @xmath207 where @xmath208 is a parameter that is 1 in the ordered state , 0 in the random one and assumes intermediate values for all intermediate states . using the same @xmath208 , we can parametrize the number of ta atoms on the ( b)+(c ) double sublattice , and all cross - frequencies .
more involved calculations ( see @xcite for details of the method ) lead to the expressions for the nearest - neighbor pairs frequencies as a function of @xmath208 . if the interaction is fundamentally short - ranged - or if the interactions beyond the first neighbors are not very important - one can then use different interaction hypotheses and calculate the internal energy . finally , from the stochastic occupancy of each site , it is possible to evaluate the configurational entropy .
the nearest - neighbor bond fractions turn out to be @xmath209@xmath210@xmath211 as nearest - neighbor bond vectors are just the @xmath212 family , it is easy to see that @xmath213 similar relations for the other @xmath214 s can be given using the @xmath215 s . since @xmath216 , and using eqs .
( [ eq : corr1],[defit ] ) , a relationship is found between @xmath208 and the background parameter @xmath85 ( see sec . [ bkg ] ) @xmath217}}}^{1/2 } \label{xoft}\ ] ] the correlation coefficients are evaluated as in sec .
[ pha ] , while @xmath218 for the trigonal phase but also for the partially ordered hypothetical nacl - type phase .
obviously @xmath219 for the random b - site phase , while for the qlrco phase , using the @xmath118 value estimated from the montecarlo runs , we have @xmath220 , which is well above the value of zero for the completely random structure .
this - combined with similar results obtained for the short - range order parameter of @xcite in sec .
[ chem : sro ] - is a further justification for the name quasi - long range order. a. cervellino , s. n. gvasaliya , b. roessli , g. m. rotaru , o. zaharko , r. a. cowley , s. g. lushnikov , t. a. shaplygina , m. t. fernandez - diaz , d. chernyshov , a. bossak , and l. bouchenoire . . ,
submit/0458884 , 2011 . | a quite unusual diffuse scattering phenomenology was observed in the single - crystal x - ray diffraction pattern of cubic perovskite bmt ( @xmath0 ) .
the intensity of the scattering is parametrized as a set of cube - like objects located at the centers of reciprocal space unit cells , resembling very broad and cubic - shaped ( 1/2,1/2,1/2)-satellites .
bmt belongs to perovskites of formula ab@xmath1b@xmath2o@xmath3 ( a = mg , b@xmath4ta , b@xmath5 mg ) .
the cubes of the intensity can be attributed to the partial correlations of the occupancies of the b site .
the pair correlation function is the fourier transform of the diffuse scattering intensity and the latter s idealized form yields the unusual property of a power - law correlation decay with distance . up to now
this is observed only in a few exotic instances of magnetic order or nematic crystals .
therefore it can not be classified as a short - range order phenomenon , as in most situations originating diffuse scattering . a monte - carlo search in configuration space yielded solutions that reproduce faithfully the observed diffuse scattering .
analysis of the results in terms of the electrostatic energy and the entropy point to this phase of bmt as a metastable state , kinetically locked , which could be the equilibrium state just below the melting point . |
graphene , one monolayer of carbon atoms tightly packed into a two - dimensional honeycomb lattice , is actively being pursued as a material for next - generation electronics due to its promising electronic properties , such as high carrier mobility @xcite , long phase coherence lengths @xcite . on the other side ,
the unique two - dimensional atomic structure of graphene implies unique confinement on electron system and offers a perfect platform to explore the amazing physics phenomenons , such as quantum hall effect @xcite and massless dirac fermions @xcite .
the first task for experimentalists to study graphene electronics is to fabricate high quality single layer graphene . until now
, several different experimental methods have been proposed and realized to prepare single layer ( or few layers ) graphene , including mechanical exfoliation of highly oriented pyrolytic graphite @xcite , patterned epitaxially grown graphene on silicon carbide or transition metal ( e.g. ru , ni ) substrates @xcite , liquid - phase exfoliation of graphite @xcite , substrate - free gas - phase synthesis @xcite , and chemical vapor deposition @xcite .
the success in fabricating single layer graphene has stimulated the extensive research efforts ( both theoretical and experimental ) in graphene related research area .
the ultimate goal of the use of graphene in next - generation electronics is to realize all - graphene circuit with functional devices built from graphene layers or graphene nanoribbons ( gnrs ) @xcite . as the basic building blocks of such circuit , the concept of electronic devices based on graphene
have been proposed theoretically and realized by experiments recently , such as field effect transistors @xcite , _
junctions @xcite , gas molecule sensor @xcite , and so on . in this article
, we will focus our discussion on the basic electronic and transport properties of gnrs and their application to electronic devices .
in particular , the theoretical investigations of gnrs physics and the technical aspects of gnr based electronic devices will be reviewed in detail .
for other topics on the recent experimental and theoretical research efforts on graphene , please refer to the reviews by katsnelson @xcite , geim _ et al . _
@xcite , beenakker @xcite , and castro neto _
the realization of graphene electronics relies on the ability to modify the electronic properties of finite - size graphenes ( for example , from semiconducting to metallic ) by varying their size , shape , and edge orientation . such unique property compared to traditional semiconductor materials , such as silicon , would ultimately enable the design and miniaturization of future electronic circuit by patterned graphene .
one of the most important issues in patterned graphene fabrication is the control of the nanoribbon width . in order to take advantage of quantum confinement effects in graphene
, the ribbon width should go down to nanometer scale . to realize the patterning of graphene with nano - scale width ,
several different techniques have been proposed including standard e - beam lithography ( fig .
1a ) @xcite , microscope lithography ( fig .
1b ) @xcite , chemical method ( fig .
1c ) @xcite , metallic nanoparticle etching @xcite , and e - beam irradiation of ultrathin poly(methylmethacrylate ) ( pmma ) @xcite . as shown in fig .
1a , the scanning electron microscopy ( sem ) image reveals the graphene can be patterned by traditional e - beam lithography technique into nanoribbons with various widths ranging from 20 to 500 nm @xcite .
figure 1b shows 10-nm - wide nanoribbon etched via scanning tunnelling microscope ( stm ) lithography . by setting the optimal lithographic parameters , it is possible to cut gnrs with suitably regular edges , which constitutes a great advance towards the reproducibility of gnr - based devices @xcite .
figure 1c shows atomic force microscopy images of chemically derived gnrs with various widths ranging from 50 nm to sub-10 nm .
these gnrs have atomic - scale ultrasmooth edges @xcite .
[ tbp ] ( color online ) various gnrs got from different experimental methods : ( a ) the sem image of gnrs patterned by e - beam lithography .
reprinted with permission from ref .
@xcite , z. chen et al .
, physica e * 40 * , 228 ( 2007 ) .
2007 , elsevier .
( b ) an 8-nm - wide 30@xmath0 gnr bent junction connecting an armchair and a zigzag nanoribbon etched by stm lithography . reprinted with permission from ref .
@xcite , l. tapaszt et al . , nat
. nanotechnol . * 3 * , 397 ( 2008 ) . 2008 , nature publishing group .
( c ) gnrs are got by using simple chemical methods . reprinted with permission from ref .
@xcite , x. li et al . , science * 319 * , 1229 ( 2008 )
. 2008 , american association for the advancement of science.,title="fig:",scaledwidth=48.0% ] the electronic properties of gnrs exhibit a strong dependence on the orientation of their edges .
as two typical types , armchair gnrs ( agnrs ) and zigzag gnrs ( zgnrs ) can be obtained by lithography technology along the specific orientation on graphene ( fig .
2b ) @xcite .
actually , the detailed edge structures ( both armchair and zigzag ) have already been clearly observed in recent experiments @xcite .
one of the most serious obstacle to graphene electronic application is the reliable control of the edge structure of gnrs .
theoretical studies predict that edge states ( in a manner similar to the well - known concept of surface states of a 3d crystal ) in graphene are strongly dependent on the edge termination and affect the physical properties of gnrs @xcite .
however , until now there is no reliable experimental method which is able to exactly control the edge structures and reduce their roughness .
an interesting experimental observation is that the band gaps of gnrs show little orientation dependence @xcite and all fabricated gnrs show semiconducting behavior @xcite , which seems inconsistent with theoretical results @xcite .
one of the reason for such inconsistency comes from the roughness of gnr edges and our explanation is also given in the following part of the article .
another issue related to gnr edges is the edge passivation .
since the dangling bonds from the edge carbon atoms have relatively high chemical activity , there is the possibility that other chemical elements present in the material fabrication process ( such as c , o , n , h and other chemical groups formed by these atoms ) would interact with the edge atoms and modify the electronic properties of gnrs . to the best of our knowledge , this issue has not been properly solved experimentally .
[ tbp ] ( color online ) the structures of h - passivated 11-agnr ( a ) and 6-zgnr ( b ) , where big green balls and small blue balls represent carbon atoms and hydrogen atoms , respectively .
integer @xmath1 is their width index.,title="fig:",scaledwidth=48.0% ]
next we will review some basic electronic and transport properties of gnrs from the theoretical viewpoint . figs .
2a and 2b show two typical models of armchair and zigzag gnrs in first - principles or other atomic - level electronic structure calculations , noting as 11-agnr and 6-zgnr , respectively . here
the numbers 11 and 6 are defined as the width index , @xmath1 . in order to remove the effect of dangling bonds ,
the edges of gnrs are saturated by hydrogen atoms .
as geometrically terminated graphene , the electronic structure of gnrs can be modelled by imposing appropriate boundary conditions on schrdinger s equation with simple tight - binding ( tb ) approximations based on @xmath2-states of carbon @xcite . another way to get
the band structure is to solve two - dimensional dirac s equation of massless free particles with an effective speed of light to model gnr system @xcite . within these models
, it is predicted that gnrs with armchair - shaped edges can be either metallic or semiconducting depending on their widths , as shown in fig . 3a . on the other side ,
the gnrs with zigzag - shaped edges are metallic with peculiar edge states on both sides of ribbons regardless of their widths , as shown in fig . 4a .
@xcite [ tbp ] ( color online ) the variation of band gaps of n@xmath3-agnrs as a function of width ( w@xmath3 ) obtained ( a ) from tb calculations and ( b ) from first - principles calculations ( symbols ) .
( c ) first - principles band structures of n@xmath3-agnrs with n@xmath3= 12 , 13 , and 14 , respectively . reprinted with permission from ref .
@xcite , y. -w .
son et al .
lett . * 97 * , 216803 ( 2006 ) .
2006 , american physical society.,title="fig:",scaledwidth=48.0% ] further detailed _ ab initio _ and gw quasiparticle calculations show that all of the agnrs exhibit semiconducting behavior and the energy gaps decrease as a function of increasing ribbon widths .
the variation in energy gaps can be separated into three distinct family behaviors @xcite , as shown in fig .
3b . as mentioned above , such dependence of band gap on the geometrical structure of gnr offers unique possibility to modify the electronic properties of gnrs simply by controlling the width and edge orientation in order to realize all - graphene functional devices .
[ tbp ] ( color online ) electronic structures of graphene nanoribbons . in all figures ,
the fermi energy ( e@xmath4 ) is set to zero .
a , the spin - unpolarized band structure of a 16-zgnr .
b , the spatial distribution of the charge difference between @xmath5-spin and @xmath6-spin for the ground state when there is no external field .
the magnetization per edge atom for each spin on each sublattice is 0.43@xmath7 with opposite orientation , where @xmath7 is the bohr magneton .
the graph is the electron density integrated in the z direction , and the scale bar is in units of 10@xmath8e@xmath8 .
c , from left to right , the spin - resolved band structures of a 16-zgnr with the external field of 0.0 , 0.05 and 0.1v , respectively .
the red and blue lines denote bands of @xmath5-spin and @xmath6-spin states , respectively . reprinted with permission from ref .
@xcite , y. -w .
son et al .
, nature * 444 * , 347 ( 2006 ) .
2006 , nature publishing group.,title="fig:",scaledwidth=48.0% ] upon inclusion of the spin degrees of freedom within density functional theory ( dft ) calculations , zgnrs are predicted to have a magnetic insulating ground state with ferromagnetic ordering at each zigzag edge and antiparallel spin orientation between the two edges @xcite , as shown in fig .
the spin polarization originates from the edge states that introduce a high density of state ( dos ) at the fermi energy .
it can be qualitatively understood in terms of the stoner magnetism of @xmath9 electrons ( in analogy to conventional @xmath10 electrons ) , which occupy a very narrow edge band and render instability of spin - band splitting @xcite .
what is more interesting , the zigzag gnrs show half - metallic behavior when external transverse electric field is applied across the zgnrs along the lateral direction @xcite , as shown in fig .
however , such spin related half - metallic phenomenon becomes weak with increasing ribbon width ( since the total energy difference per edge atom between spin - unpolarized and spin - polarized edge states is only about 20 mev in their simulation system and decreases with increasing width ) and is not energetically stable if the width of gnr is significantly larger than the decay length of the spin - polarized edge states @xcite . on the other hand
, it is predicted that the half - metallicity can be also achieved in edge - modified or doped zgnrs @xcite another important issue regarding the basic electronic structures of gnrs relies on the edge states . due to the presence of the edge states , the @xmath2 and @xmath11 subbands of metallic zgnrs ( in the spin - unpolarized state )
do not cross with each other at the fermi level to span the whole energy range like metallic armchair carbon nanotubes ( cnts ) ( the left panel of fig .
this leads to the fact that the transport property of zgnrs under a low bias voltage ( or a small potential step ) is only determined by the transmission between @xmath2 and @xmath11 subbands ( as shown in fig .
5a ) . with the presence of such unique band structures ,
zgnrs exhibit two distinct transport behaviors depending on the existence of @xmath12 mirror symmetry with respect to the midplane between two edges @xcite , although all the zgnrs have similar metallic energy band structures .
since the @xmath2 and @xmath11 subbands of symmetric zgnrs ( i.e. , width index @xmath1 is an even number ) have opposite definite @xmath12 parities , the transmission between them is forbidden ( the left panel of fig .
5b ) . for asymmetric zgnrs ( i.e.
, width index @xmath1 is an odd number ) , however , their @xmath2 and @xmath11 subbands do not have definite @xmath12 parities , so the coupling between them can contribute to about one conductance quantum ( the right panel of fig .
this transport difference can be clearly reflected in the current - bias - voltage ( @xmath13-@xmath14 ) characteristics of zgnrs by using the first - principle transport simulation , as shown in fig . 5c .
although metallic armchair carbon nanotubes also have @xmath2 and @xmath11 subbands with definite parities , such symmetry - depending ( or band - selective ) @xmath13-@xmath14 characteristics can not be observed in them because of the crossover of their subbands , i.e. , the absence of edge states .
recently , theoretical work predicts a very large magnetoresistance in a graphene nanoribbon device due to the existence of edge states @xcite .
[ tbp ] ( color online ) ( a ) schematic band structure around the fermi level of a zgnr under a positive ( @xmath15 ) and a negative ( @xmath16 ) potential step . ( b ) conductance of 8-zgnr and 7-zgnr under two potential steps shown in ( a ) .
( c ) @xmath13-@xmath14 curves of the two - probe system ( see the inset ) made of zgnrs with different widths @xmath1 .
reprinted with permission from ref .
@xcite , z. li et al .
lett * 100 * , 206802 ( 2008 ) . 2008 , american physical society.,title="fig:",scaledwidth=47.0% ] besides the fabrication and theoretical study of monolayer graphene and gnrs , recent experimental @xcite and theoretical @xcite studies are also carried out on bilayer graphene and gnrs .
theoretically , it is shown that the bilayer gnrs and monolayer gnrs have some similar electronic properties such as edge states localized at the zigzag edges and semiconducting behavior of armchair bilayer gnrs @xcite .
experimentally , it is found that the bilayer graphene has unique features such as anomalous integer quantum hall effects @xcite , which is absent in single layer graphene . and the size of energy gap of such bilayer structures can be controlled by adjusting carrier concentration @xcite as well as by an external electric field @xcite .
these unique properties open an opportunity to implement bilayer graphene or gnrs in various electronic applications .
as mentioned above , current experimental techniques ( such as lithography ) are not able to realize exact control of the edge structures of gnrs and the edges are always very rough due to the limitation of the fabrication technology @xcite .
there are theoretical evidences that such edge disorders can significantly change the electronic properties of gnrs @xcite , and lead to some unexpected physics effect , such as the anderson localization @xcite and coulomb blockade effect @xcite .
these effects have already been observed in lithographically obtained graphene nanoribbons @xcite .
the edge roughness is also crucial for the spin polarized properties of gnrs . as we know , the magnetic properties of gnrs depend on the highly degenerate edge states . in principle
a perfect edge structure is necessary for stabilizing magnetic properties of gnrs as theoretically predicted .
an important question is , how robust the spin - polarized state is in the presence of edge defects and impurities ?
the answer to this question is not only scientifically interesting to better understand the physical mechanism of spin polarization in gnrs but also has important technological implications in the reliability of gnrs as a new class of spintronic materials .
first - principles theoretical studies reveals the effect of edge vacancies and substitutionally doped boron atoms , as typical examples of structural edge defects and impurities , on the spin - polarization of zgnrs @xcite . the calculated energy difference between the magnetic state [ both antiferromagnetic ( af ) and ferromagnetic ( fm ) ] and the nonmagnetic state
is found to rapidly decrease with increasing defect concentration and eventually decrease to zero ( nonmagnetic ) , as shown in fig .
the critical defect ( impurity ) concentration is found to be @xmath17 0.10/ when the ribbon width is larger than 2 nm .
[ tbp ] ( color online ) the energy difference per edge atom between the magnetic ( af or fm ) and paramagnetic ( pa ) state as a function of vacancy concentration in the edge .
two different ribbon widths of @xmath1=6 and @xmath1=4 are shown .
the inset shows the critical concentration as a function of the ribbon width up to 5 nm .
reprinted with permission from ref .
@xcite , b. huang et al .
b * 77 * , 153411 ( 2008 ) . 2008 , american physical society.,title="fig:",scaledwidth=40.0% ] evidently , the magnetism in gnrs depends on a high density of state ( dos ) around the fermi energy coming from the highly degenerate edge states in a perfect ribbon edge ( @xmath18 ) that renders instability of spin polarization @xcite .
the presence of edge vacancies and impurities would decrease the dos at @xmath18 since they do not contribute to the same edge state . from stoner model ,
such decrease of dos will suppress the spin polarization of gnr systems .
therefore , the practical realization of the spin polarization in gnrs for spintronics applications could be rather challenging @xcite .
recently , an interesting theoretical work systematically studied the spin current in rough gnrs and predicted that only gnrs with imperfect edges exhibit a nonzero spin conductance while there is no spin current in perfect gnrs @xcite .
it confirms that the edge effect is of great importance to spin related properties of gnrs .
furthermore , the problem of edge passivation has not yet clearly resolved by experiment until now . from the theoretical viewpoint
, the edge passivation can be well modeled by the modifications of the hopping energies in the tight - binding approach @xcite or via additional phases in the boundary conditions @xcite .
recent theoretical modeling and calculations have indicated that the edge passivation has a strong effect on the electronic and spin - polarized properties of gnrs @xcite .
the possible passivation species include hydrogen , carbon , oxygen , nitrogen , and other chemical groups .
further experimental works are needed to explore the realistic edge structures of gnrs at atomic scale and determine which types of edge passivation are favorable .
the interesting and unique electronic properties of gnrs , such as orientation and width dependence of transport behavior , offer great possibilities for their electronic device applications . compared with other electronic materials ,
one of the most promising advantage of gnrs is that gnr - based devices and even integrated circuits can be fabricated by a single process of patterning a graphene sheet @xcite .
figures 7a-7c illustrate three basic device building blocks : ( i ) metal - semiconductor junction , ( ii ) _ p - n _ junction , and ( iii ) hetero - junction , which can be , respectively , made by patterned gnrs ( i ) along different direction , ( ii ) with different edge doping , and ( iii ) with different widths . it was proposed that a variety of devices can be constructed from these building blocks . for example , a field effect transistor ( fet ) can be made simply by two metal - semiconductor junctions , as shown in fig .
7d . there are some potential key advantages in designing and constructing device architectures based on gnrs .
the first advantage is the perfect atomic interface , a feature that is difficult to achieve for the interconnection between nanotubes of different diameter and chirality .
second , it is generally difficult to find a robust method to make contact with the molecular device unit , because there exists usually a large contact resistance between the metal electrodes and molecules ( _ e.g. _ , single - walled cnt ) due to a very small contact area .
this difficulty may be circumvented by using gnrs , because the gnr - based devices can be connected to the outside circuits exclusively via metallic gnrs ( or graphene ) , as illustrated in fig .
7d , which serve as extensions of metal electrodes to make contact with the semiconducting gnrs so that an atomically smooth metal - semiconductor interface is maintained with minimum contact resistance .
last but not the least , the edges of gnrs may serve as effective sites for doping . in principle , by introducing different types of dopants at different sections of gnr edges , one can realize a _
p - n _ junction by selective doping , as shown in fig . 7b .
[ tbp ] ( color online ) schematics of three device building blocks : ( a ) a metal - semiconductor junction between a zigzag and an armchair gnr , ( b ) a _ p - n _ junction between two armchair gnrs with different edge doping , and ( c ) a heterojunction between two armchair gnrs of different widths ( band gaps ) .
( d ) schematics of a gnr - fet , made from one semiconducting 10-agnr channel and two metallic 7-zgnr leads connected to two external metal electrodes . reprinted with permission from ref .
@xcite , q. yan et al .
, nano lett .
* 7 * , 1469 ( 2007 ) . 2007 , american chemical society.,title="fig:",scaledwidth=38.0% ] one of the most important electronic applications based on gnrs is field effect transistors .
recently , experimental studies @xcite have indicated the possibility of fabricating gnr - based transistors .
the advantage of gnrs as an alternative material for transistors is that it could bypass the chirality challenge of cnts while retaining the excellent electronic properties of graphene sheets , such as the high @xmath19/@xmath20 ratio and excellent electron / hole mobilities .
the performance of one sub-10-nm gnr - fet in the latest work from dai s group is shown in figs . 8a and 8b ( the transfer and output characteristics , respectively , for the gnr device with the width of @xmath17 2 @xmath21 0.5 nm and the channel length of @xmath17 236 nm)@xcite .
this device delivered @xmath19 @xmath17 4 @xmath22a at @xmath23 = 1 v , @xmath19/@xmath20 ratio @xmath24 10@xmath25 at @xmath23 = 0.5 v , subthreshold slope @xmath26=d@xmath27/@xmath28 @xmath17 210 mv / decade and transconductance @xmath17 1.8 @xmath22s ( @xmath17900 @xmath22s/@xmath22 m ) .
the device performance is comparable with the best cnt - based transistors .
however , the dirac point was not observed around zero gate bias in this measurement , indicating _ p_-doping effects at the edges or by physisorbed species during the chemical treatment steps .
[ tbp ] ( color online ) ( a ) and ( b ) transistor performance of gnr - fets with width of @xmath17 2 nm and channel length of @xmath17 236 nm [ ( c ) and ( d ) , width of @xmath17 60 nm , and channel length of @xmath17 190 nm ] .
( a ) transfer characteristics ( current vs gate voltage @xmath29-@xmath30 ) under various @xmath23 . @xmath19/@xmath20 ratio of @xmath24 10@xmath25
is achieved at room temperature .
( b ) output characteristics ( @xmath29-@xmath23 ) under various @xmath30 . on current density is @xmath17 2000 @xmath22a/@xmath22 m in this device .
( c ) transfer and ( d ) output characteristics of the 60 nm width gnr - fet device . reprinted with permission from ref .
@xcite , x. wang et al .
lett . * 100 * , 206803 ( 2008 ) .
2008 , american physical society.,title="fig:",scaledwidth=48.0% ] together with experimental progress on gnr - based transistors , theoretical studies using semiclassical and quantum transport models show that gnr - based fets could have a similar performance as cnt - based fets and might outperform traditional si - based fets @xcite .
figure 9 shows a first - principles study on the performance of a typical gnr - based fet made with a 5.91 nm long intrinsic semiconducting 10-agnr channel connected to two metallic 7-zgnr leads ( source and drain ) @xcite . in fig .
9a , the near - symmetric @xmath31 curve shows an excellent ambipolar transistor with on / off ratio @xmath32/@xmath20 @xmath17 2000 and subthreshold swing of @xmath26 @xmath17 60 mv / decade , which are comparable to those of high performance cnt - fets . such the field effect can be clearly reflected in the change of @xmath33 characteristics under different gate voltages ( fig .
figure 9c shows the @xmath31 curves of the gnr - fets made from the same 10-agnr channel with its length ranging from 1.69 to 6.76 nm , from which the values of @xmath26 are derived as a function of @xmath34 as shown in figure 9d .
clearly , @xmath26 decreases with increasing @xmath34 , and gradually approaches @xmath1760 mv / decade when @xmath34 becomes longer than 6 nm .
meanwhile , the on - current stays the same , independent of @xmath34 , but the off - state leakage current increases rapidly with decreasing @xmath34 , which gives rise to a large @xmath26 .
the performance of the ambipolar gnr - fets made of intrinsic semiconductor channels can be understood in terms of metal - semiconductor tunneling junctionh within the semiclassical band - bending model .
[ tbp ] ( color online ) ( a ) @xmath31 curve for a 5.91 nm long intrinsic 10-agnr channel ( @xmath14=20 mv ) .
( b ) @xmath33 curves under different gate voltage ( @xmath35 ) .
( c ) @xmath31 curves for different channel lengths ( @xmath14=20 mv ) .
( d ) the subthreshold swing ( @xmath26 ) as a function of channel length @xmath34 .
( e),(f ) @xmath31 curves for a 5.91 nm long 10-agnr channel with selective n and b doping , respectively ( @xmath14=20 mv ) . reprinted with permission from ref .
@xcite , q. yan et al .
, nano lett .
* 7 * , 1469 ( 2007 ) .
2007 , american chemical society.,title="fig:",scaledwidth=48.0% ] compared with the basic ambipolar fets , it is well known that _
n_-type ( or _ p_-type ) fets serve as critical transistor devices for digital electronics applications @xcite .
to realize such device design based on gnrs , a method was proposed using n ( or b ) atoms as selective dopants at the channel region of perfect gnr - fets ( the positions of b or n are indicated by arrows in fig
figure 9e ( 9f ) shows the calculated @xmath31 curves under @xmath14 = 20 mv , exhibiting the typical behavior of a _ n_-type ( _ p_-type ) fet@xcite .
it is suggested that all of the functional transistor devices that work in traditional si - based circuits could be realized by gnrs and gnr - based junctions in principle .
[ tbp ] ( color online ) ( a ) the schematic structure of the field effect transistor ( fet ) made from a single 5-zgnr .
the semiconducting channel is obtained by edge doping of n in a finite - length region ( the center region ) .
( b ) simulated @xmath13-@xmath27 curves of n - doped gnr - fets under @xmath14 = 0.01 v. the channel length is 8.54 nm and the linear doping concentration is 0.1365 @xmath36 .
( c ) the dependence of the subthreshold swing @xmath26 ( blue line ) and the on / off current ratio ( red line ) on the channel length @xmath34 .
reprinted with permission from ref .
@xcite , b. huang et al .
. lett . * 91 * , 253122 ( 2007 ) .
2007 , american institute of physics.,title="fig:",scaledwidth=48.0% ] noting the current experimental difficulty to get an accurate z - shape junction ( i.e. , fet shown in fig .
7d ) due to the limitation of lithography technique , a new type of field effect transistor has also been proposed taking advantage of the metal - semiconductor transition in zgnrs induced by substitutional doping of nitrogen or boron atoms at their edges @xcite , as shown in fig .
10a . besides simplifying the fabrication process
, such a linear configuration can also increase the device density in electronic circuits .
figure 10b shows a typical @xmath37 curve for the n - doped gnr - fet ( with the channel length of 8.54 nm ) under the bias voltage @xmath14 = 0.01 v. clearly , the doped fet exhibits ambipolar characteristics , similar to the z - shape fets .
the relationship between the device performance and the channel length is demonstrated by calculating @xmath31 curve of n - doped gnr - fets as a function of the doped channel length while keeping the bias voltage @xmath14 at 0.01 v. as shown in fig .
10c , the subthreshold swing @xmath26 of these doped gnr - fets decreases and the on / off current ratio increases exponentially .
it can be seen that for good device performance with small @xmath26 value ( e.g. , below 100 mv / decade ) and high on / off current ratio ( e.g. , above 100 ) , the doped channel length should be longer than 5 nm .
the minimum leakage current of those fets with the doped channels shorter than this critical length will be greatly enhanced by direct tunneling , which lowers the device performance . besides ideal case
, some more practice issues concerning gnr - based fets are discussed in recent theoretical works .
for example , the effects of the various contact types and shapes on the performance of schottky - barrier - type gnr - fets have been investigated theoretically @xcite , which indicates that the semi - infinite normal metal can potentially provide promising performance .
in addition , the effect of edge roughness and carrier scattering on gnr - fets have been studied @xcite .
the presence of edge disorder significantly reduces on - state currents and increases off - state currents ( the on / off ratio decreases ) , and introduces wide variability across devices .
these effects become weaker for gnrs with larger width and smoother edges
. however , the band gap decreases with increasing width , thereby increasing the band - to - band tunneling mediated subthreshold leakage current even with perfect gnrs .
obviously , without atomically precise edge control during fabrication , it is hard to get reliable and stable performance of gnr - fets .
due to their unusual basic properties , gnrs as well as graphene are promising for a large number of applications @xcite , from spin filters @xcite , valley filters @xcite , to chemical sensors @xcite .
gnrs can be chemically and/or structurally modified in order to change its functionality and hence its potential applications .
in summary , we review the basic electronic and transport properties of graphene nanoribbons , and discuss recent theoretical and experimental progress on gnr - based field effect transistors from the viewpoint of device application . due to the interesting electronic and magnetic properties , gnrs have been demonstrated as a promising candidate material for future post - silicon electronics such as transport materials , field effect transistors , and spin injection or filter .
more experimental efforts will focus on fabricating high quality nanoribbon samples with accurate control of the edge structures .
d. martoccia , p. r. willmott , t. brugger , m. bjorck , s. gunther , c. m. schleputz , a. cervellino , s. a. pauli , b. d. patterson , s. marchini , j. wintterlin , w. moritz , and t. greber , phys .
* 101 * , 126102 ( 2008 ) .
y. hernandez , v. nicolosi , m. lotya , f .
m. blighe , z. sun , s. de , i. t. mcgovern , b. holland , m. byrne , y. k. gunko , j. j. boland , p. niraj , g. duesberg , s. krishnamurthy , r. goodhue , j. hutchison , v. scardaci , a. c. ferrari , and j. n. coleman , nat . nanotechnol . * 3 * , 563 ( 2008 ) . | the successful fabrication of single layer graphene has greatly stimulated the progress of the research on graphene . in this article , focusing on the basic electronic and transport properties of graphene nanoribbons ( gnrs ) , we review the recent progress of experimental fabrication of gnrs , and the theoretical and experimental investigations of physical properties and device applications of gnrs . we also briefly discuss the research efforts on the spin polarization of gnrs in relation to the edge states . |
universal conductance fluctuations ( ucf s ) are one of the most meaningful manifestations of the quantum - interference electron transport in mesoscopic and nanoscale systems .
@xcite in weakly disordered miniature metals and at low temperatures , the aperiodic " ucf patterns are highly reproducible .
those fluctuation patterns are determined by the specific impurity configuration that is frozen " in a given sample at a given cooldown .
@xcite under such conditions , one may sweep a magnetic field @xcite or gate voltage @xcite sufficiently widely to realize statistically distinct subsystems ( independent members ) of the specific ensemble which embraces the given sample under study .
the sample may then be thermally cycled up to room temperatures to induce possible rearrangement of the impurity configuration .
if an impurity / disorder reconfiguration should occur , the ucf patterns would significantly or completely alter after the sample is remeasured at low @xmath6 . in this context , the impurities and defects are essentially _ static _ both in space and with time at liquid - helium temperatures .
apart from the mesoscopic metal and semiconductor structures that are fabricated by the top - down electron - beam lithography , @xcite the ucf s have recently been studied in bottom - up artificially synthesized nanowires ( nw s ) .
@xcite besides , the ucf s have been searched in newly developed materials , such as epitaxial ferromagnets , @xcite carbon nanotubes , @xcite graphene , @xcite and topological insulators .
@xcite tin - doped indium oxide ( in@xmath7sn@xmath8o@xmath9 , or so - called ito ) is a metal oxide which exhibits low electrical resistivities @xmath10 ( @xmath11 10@xmath12 @xmath13 cm ) @xcite and , in particular , free - carrier - like electronic conduction properties , @xcite such as a linear diffusive thermopower in the wide @xmath6 interval 5300 k. @xcite single - crystalline ito nw s possess similar metallic characteristics to those of ito films .
@xcite at not too low temperatures , the @xmath10@xmath6 behavior can be described by the standard boltzmann transport equation .
below a few tens degree of k , the quantum - interference weak - localization ( wl ) and electron - electron interaction effects @xcite ( among other possible effects @xcite ) cause notable corrections to @xmath10 [ see the inset of fig .
[ fig2](b ) ] . furthermore ,
in _ sufficiently short _ ito nw s , pronounced ucf s can arise at cryogenic temperatures , owing to the absence of classical self - averaging in the sample conductance @xmath14 .
@xcite in this work , we report our experimental results for three ito nw s that reveal marked ucf s in sweeping , perpendicular magnetic fields .
two of our nw s had been intentionally thermally cycled to 300 k and then remeasured at liquid - helium temperatures .
the shape of the ucf patterns completely altered due to the impurity / disorder reconfigurations induced by the room - temperature thermal energy @xmath15 , where @xmath16 is the boltzmann constant . in sharp contrast , the ucf magnitudes remained basically unchanged .
our measured @xmath6 dependence of the root - mean - square ( rms ) ucf magnitudes , however , can not be described by the existing theory .
the reason why seems to be associated with the concept of thermal averaging " that is brought forward in the current mesoscopic theory .
@xcite under the condition of @xmath17 ( which is pertinent to the present study , where @xmath18 is the normalized planck s constant , and @xmath19 is the electron dephasing time ) , the canonical ucf theory predicts that the effect of the thermal averaging would lead to an extra @xmath20 temperature dependence of the conductance fluctuation magnitudes , in addition to that to result from the @xmath6 dependence of @xmath19 .
unexpectedly , such a strong @xmath20 temperature dependence is not seen in this work .
further experimental investigations focusing on the temperature characteristics of the ucf magnitudes would be useful to improve our understanding of this issue . in this
regard , self - assembled nw s can provide valuable platforms due to their marked ucf phenomena , as compared with those in conventional lithographic metal structures .
empirically , the ucf effect can often be seen in nw s up to above 10 k , @xcite while it is seen only below 1 k in top - down lithographic metal structures .
@xcite this paper is organized as follows .
section ii contains our experimental method .
section iii includes our experimental results and theoretical analysis .
our conclusion is given in sec .
iv . appendix a contains a discussion of the magnetoresistance in the wl effect in one dimension .
ito nanowires were fabricated by the implantation of sn ions into in@xmath21o@xmath9 nw s .
the in@xmath21o@xmath9 nw s were grown by the vapor - solid - liquid ( vls ) method , as described previously .
@xcite the morphology and the cubic bixbyite structure ( the prototype structure being mn@xmath21o@xmath22 and the space group : ia3 ) of the single - crystalline sn - doped in@xmath21o@xmath9 nw s were studied by the scanning electron microscopy ( sem ) and the transmission electron microscopy ( tem ) .
the nominal composition of our nw s has previously been determined to be in@xmath23sn@xmath24o@xmath9 ( ref . ) .
four - probe single nw devices were fabricated by the electron - beam lithography .
[ figure [ fig2](a ) shows an sem image of the nw14 nanowire device . ]
the magnetoresistance ( mr ) measurements were performed on an oxford heliox @xmath25he cryostat equipped with a 4-t superconducting magnet . a linear research lr-700 ac resistance bridge operating at a frequency of 16 hz was employed for the mr measurements .
to avoid electron heating , an excitation current of @xmath26 10 na ( so that the voltage drop along the nw was @xmath27 , where @xmath28 is the electron charge ) was applied . in all cases ,
the magnetic field @xmath29 was applied _
perpendicular _ to the nw axis .
.sample parameters for ito nanowires .
@xmath30 is the diameter , @xmath31 is the voltage probe distance in a four - probe geometry , @xmath32 is the electron diffusion constant , @xmath33 is the electron elastic mean free path , and @xmath34 is the fermi wave number .
the uncertainty in @xmath30 is @xmath35 nm . @xmath32 , @xmath36 , and @xmath37 are for 10 k. the samples are labeled according to their voltage probe distance . the nw28 nanowire was taken from ref . .
[ cols="<,^,^,^,^,^,^,^ " , ] for 1d quasielastic nyquist @xmath28-@xmath28 scattering , the theory @xcite predicts a coupling strength @xmath38^{2/3}$ ] , where @xmath39 is the resistance of the nw , and @xmath31 is the nw segment between the two voltage probes .
we have compared our experimental @xmath5 ( but not @xmath40 ) with eq .
( [ tauphi ] ) , and our least - squares fitted values of the adjustable parameters @xmath41 and @xmath42 are listed in table ii .
it can be seen that our experimental value of @xmath42 agrees with the theoretical value @xmath43 to within a factor of @xmath44 ( @xmath45 ) for the nw12 ( nw14 ) nanowire .
these results provide meaningful self - consistency check of our experimental method and data analyses .
we have observed universal conductance fluctuations with varying magnetic field in indium tin oxide nanowires from 0.26 k up to @xmath0 k. the ucf s originate from the inherent quantum - interference nature of the electron transport in weakly disordered nanoscale structures .
we found that the shape of the ucf patterns is very sensitive to the specific impurity configuration , and it alters completely after thermal cycling the sample to room temperatures .
the root - mean - square ucf magnitudes increase with reducing temperature , reaching a fraction of @xmath1 at @xmath2 k. however , the temperature dependence of our measured ucf magnitudes can not be explained by the existing theory .
the discrepancy between the experiment and theory seems to arise from the absence of the thermal averaging effect in our measurements . in our experiment , we are always in the regime of @xmath17 , corresponding to @xmath4 ( @xmath40 ) . under such conditions
, one would expect the responsible phase - coherence region in our nanowires to be cut off by @xmath3 , instead of by @xmath5 ( @xmath40 ) .
the reason why this does not happen such that @xmath46 does not lead to a notable temperature dependence of the measured ucf magnitudes is not understood . in this work ,
the ucf magnitudes in every nanowire are deduced from averaging over about 20 to 40 correlation field @xmath47 periods for our applied magnetic field of 4 t. whether such averaging over a somewhat limited range of measuring magnetic field is fully equivalent to the theoretically concerned averaging over a complete change of the impurity configurations , i.e. , whether the ergodic hypothesis assumed in the original ucf theory @xcite is faithfully met in our measurements , deserves further investigations .
finally , the approximate values of the electron dephasing length @xmath40 has been evaluated and found to be in reasonable semiquantitative agreement with the dephasing length @xmath5 extracted from the weak - localization magnetoresistance studies .
this work demonstrates that the ucf effect is particularly pronounced in self - assembled conducting nanowires .
studies in this direction using metallic nanowires may thus provide insightful information on the ucf mechanism in miniature conductors .
the authors are grateful to s. p. chiu and y. h. lin for valuable experimental assistance and discussion , c. s. chu for useful discussion , and f. r. chen and j. j. kai for providing us with the ito nanowires used in this study .
this work was supported by taiwan national science council through grant nos .
nsc 99 - 2120-m-009 - 001 and nsc 100 - 2120-m-009 - 008 , and by the moe atu program .
the magnetoresistance in the weak - localization effect has been well established over the years , both theoretically @xcite and experimentally .
@xcite our measured low - field mr s in every nw have been least - squares fitted to the 1d wl theoretical prediction as given below ( @xmath29 is applied perpendicular to the nw axis ) : @xcite where @xmath49 , @xmath39 is the resistance of a nanowire of width @xmath50 and length @xmath31 , @xmath51 is the magnetic length , @xmath52 is the electron dephasing length , @xmath53 is the spin - orbit scattering length ( @xmath54 being the spin - orbit scattering time ) , and the electron diffusion constant @xmath55 ( @xmath56 being the fermi velocity , and @xmath57 being the electron elastic mean free time ) . notice that our nw s are 1d with regard to the wl effect ( i.e. , @xmath58 ) , while 3d with regard to the usual boltzmann transport ( i.e. , @xmath59 ) .
the spin - orbit scattering length ( time ) is a temperature independent quantity whose size , relative to the inelastic electron scattering strength , determines the sign of the weak-(anti)localization effects in the low - field mr .
@xcite our method for estimating the @xmath56 , @xmath57 , and @xmath32 values in the ito nw s through our measured fermi energy @xmath60 values in the ito material has recently been described in ref . .
normalized magnetoresistance as a function of perpendicular magnetic field of the nw14 nanowire at several temperatures , as indicated .
the symbols are the experimental data and the solid curves are the theoretical predictions of eq .
( [ a1 ] ) .
note that the theory can well describe the experimental data .
note also that signatures of the ucf s in @xmath61 t can be seen especially at 0.26 and 3.0 k.,width=377,height=302 ] figure [ 1dmr ] shows the normalized magnetoresistance @xmath62 as a function of magnetic field at several temperatures of the nw14 nanowire at first cooldown .
the symbols are the experimental data and the solid curves are the theoretical predictions of eq .
( [ a1 ] ) .
this figure clearly demonstrates that our measured low - field mr s can be well described by the 1d wl theory in the wide temperature interval of 0.2640 k. therefore , the characteristic electron dephasing length @xmath5 can be very reliably extracted .
the @xmath5 values thus obtained are plotted in fig .
[ fig8](a ) . similarly , our measured low - field mr s in the nw12 nanowire can also be well described by eq .
( [ a1 ] ) ( not shown ) , and the extracted @xmath5 values are plotted in fig .
[ fig8](b ) . besides , our measured low - field mr s of the nw28 nanowire , along with the inferred @xmath5 values ,
have previously been reported in ref . . finally , our extracted spin - orbit scattering length is @xmath63 nm in the nw28 nanowire . on the other hand ,
the spin - orbit scattering is comparatively weak in the relatively cleaner nw12 and nw14 nanowires , and hence only a lower bound can be estimated , i.e. , @xmath64 @xmath65 m ( corresponding to @xmath66 ps ) in these two nw s .
we would like to note that a weak spin - orbit scattering strength has recently also been found in a series of homogeneous and inhomogeneous ito thin films .
@xcite s. estvez hernndez , m. akabori , k. sladek , ch .
volk , s. alagha , h. hardtdegen , m. g. pala , n. demarina , d. grtzmacher , and th .
schpers , phys .
b * 82 * , 235303 ( 2010 ) ; c. blmers , m. i. lepsa , m. luysberg , d. grtzmacher , h. lth , and th .
schpers , nano lett . *
11 * , 3550 ( 2011 ) .
s. p. chiu , y. h. lin , and j. j. lin , nanotechnology * 20 * , 015203 ( 2009 ) ; l. t. tsai , s. p. chiu , j. g. lu , and j. j. lin , _ ibid .
_ * 21 * , 145202 ( 2010 ) . in the case of semiconductor nw s
, the point defects often act as unintentional donors or acceptors . in our recent studies of the temporal ucf s in ruo@xmath21 nw s due to mobile defects ( ref . ) , we had compared our measured rms ucf magnitudes with the theoretical predictions in the absence of the thermal averaging effect [ i.e. , a factor of @xmath67 is not multiplied to the eqs .
( 1 ) and ( 2 ) of ref . ] .
a good agreement was obtained . in that case , it is also not understood why the thermal averaging effect played no significant role .
it should be of fundamental importance to reexamine whether any ingredients in the theoretical concepts for thermal averaging in mesoscopic physics might have been overlooked ( overestimated ) .
around 10 k , the experimental magnitudes of @xmath68 are relatively small , rendering the evaluated @xmath47 and @xmath40 values less accurate compared with those extracted for @xmath69 k. moreover , the uncertainty in @xmath30 could introduce some error in @xmath40 . during the course of our mr measurements on the nw12 nanowire , unintentional electrical drifts and electrical shocks
had delivered into the nw , which made the subtraction of the background noise subject to greater uncertainties than in other nw s .
[ the electrical shocks had caused slight rearrangement of the scatterers , p. y. yang , master thesis , national chiao tung university , 2011 ( unpublished ) .
] therefore , the extracted @xmath40 values scattered to a greater extent in this particular nw than those in other nw s . since the nw28 nanowire is not strictly in the 1d regime at @xmath70 k , the @xmath71 values in fig .
[ fig8](b ) are plotted only for reference .
j. j. lin and n. giordano , phys . rev .
b * 35 * , 1071 ( 1987 ) ; p. mohanty , e. m. q. jariwala , and r. a. webb , phys . rev
* 78 * , 3366 ( 1997 ) ; s. m. huang , t. c. lee , h. akimoto , k. kono , and j. j. lin , phys .
lett . * 99 * , 046601 ( 2007 ) .
the electron - phonon relaxation rate , which often dominates in normal metals at a few degrees k and higher , is relatively weak in the low - carrier - concentration ito material , and thus can be ignored , see ref . .
y. niimi , y. baines , t. capron , d. mailly , f .- y .
lo , a. d. wieck , t. meunier , l. saminadayar , and c. bauerle , phys .
b * 81 * , 245306 ( 2010 ) ; f. pierre , a. b. gougam , a. anthore , h. pothier , d. esteve , and n. o. birge , phys . rev .
b * 68 * , 085413 ( 2003 ) ; y. j. zeng , l. m. c. pereira , m. menghini , k. temst , a. vantomme , j .- p . locquet , and c. van haesendonck , nano lett .
, doi : 10.1021/nl2034656 ( 2012 ) . | magnetic field dependent universal conductance fluctuations ( ucf s ) are observed in weakly disordered indium tin oxide nanowires from 0.26 k up to @xmath0 k. the fluctuation magnitudes increase with decreasing temperature , reaching a fraction of @xmath1 at @xmath2 k. the shape of the ucf patterns is found to be very sensitive to thermal cycling of the sample to room temperatures , which induces irreversible impurity reconfigurations . on the other hand , the ucf magnitudes are insensitive to thermal cycling . our measured temperature dependence of the root - mean - square ucf magnitudes are compared with the existing theory [ c. w. j. beenakker and h. van houten , phys . rev .
b * 37 * , 6544 ( 1988 ) ] . a notable discrepancy is found , which seems to imply that the experimental ucf s are not cut off by the thermal diffusion length @xmath3 , as would be expected by the theoretical prediction when @xmath4 , where @xmath5 is the electron dephasing length .
the approximate electron dephasing length is inferred from the ucf magnitudes and compared with that extracted from the weak - localization magnetoresistance studies . a reasonable semiquantitative agreement is observed . |
pairs of high frequencies , known as kilohertz quasi - periodic oscillations ( khz qpos ) , have been detected in the x - ray emission of low - mass x - ray binaries ( lmxbs ) for more than 20 neutron stars and several black holes .
it has been suggested that a non - linear resonance within an accretion disc in a general - relativistic space - time metric plays a role in the excitation of the two oscillations ( abramowicz & kluniak 2001 , kluniak & abramowicz 2001 ) .
it is now recognized that the frequencies of the two peaks in the power spectrum of x - ray variability in black holes are in rational ratios , @xmath1 , with @xmath2 for two sources , and @xmath3 in a third source ( abramowicz & kluniak 2001 ; kluniak & abramowicz 2002 ; remillard et al .
2002 , abramowicz et al .
these black hole qpos are thought to have stable frequencies , and until the discovery of the puzzling rational ratios they have been thought to correspond to a trapped g - mode or c - mode of disc oscillation in the kerr metric ( okazaki et al .
1987 , wagoner 1999 , wagoner et al .
2001 , kato 2001 ) . in neutron star sources ,
the twin khz qpo frequencies are known to vary considerably and to be mutually correlated . in the context of disc oscillations , two frequencies varying in this fashion
can be explained as the fundamental and the first harmonic of the non - axisymmetric ( _ m_=1 ) g - mode ( kato 2002 , 2003 ) . in sco x-1 , a prototypical z - source ,
the slope of the correlation line is somewhat steeper than 2/3 , but it is not uncommon to find the source in a state when @xmath4 hz , and @xmath5 hz , i.e. , at about @xmath0 ratio ( abramowicz et al .
2002b ) .
it would appear that the @xmath0 frequency ratio is common to neutron star and black hole systems .
kluniak & abramowicz ( 2002 ) , and abramowicz & kluniak ( 2003 ) suggest that the @xmath0 ratio follows from the properties of parametric resonance between two eigenfrequencies , of which one is always lower than the other . specifically , two modes of disc oscillation are thought possible with frequencies close to the two epicyclic frequencies of free orbital motion . within general relativity ,
the radial epicyclic frequency is generically lower than the meridional one , their ratio varying from one to zero as the radius of circular orbits decreases from infinity down to that of the marginally stable orbit .
the observed rational ratios of frequencies may follow from fundamental properties of both strong gravity and a non - linear resonance between radial and vertical oscillations in accretion discs with fluid lines that are nearly geodesic , nearly circular , and nearly planar . in this paper
we investigate a simple mathematical model of such motions . to this aim
, we discuss the properties of parametric resonance in the paczyski - wiita ( 1980 ) model of the schwarzschild metric .
let us consider a fluid in which the flow lines are only slightly non - circular , and slightly off the @xmath6 symmetry plane .
in spherical coordinates , @xmath7 with accuracy to the third order in @xmath8 and @xmath9 , equations of fluid motion read : @xmath10\delta \theta\ ] ] @xmath11 @xmath12 @xmath13 @xmath14 here the time derivative is denoted by a dot , the radial derivative by a prime , and @xmath15 are components of a small force of non - gravitational origin ( pressure , viscous , magnetic , or other ) . the epicyclic eigenfrequencies @xmath16 and @xmath17 are defined , in terms of the effective potential @xmath18 and the specific angular momentum @xmath19 , by @xmath20 derivation of these equations assumes equatorial plane symmetry for the gravitational potential @xmath21 . here
we adopt a spherically symmetric potential , and @xmath22 at the marginally stable orbit . ]
@xmath23 from eqs .
( [ 5 ] ) and ( [ 6 ] ) it follows @xmath24 where @xmath25 is the marginally stable orbit .
( 0.48,48)obr3as.eps ( 0.48,48)obr3b.eps
a parametric resonance instability occurs near @xmath26 for @xmath27 , in an oscillator that obeys the mathieu type equation of motion ( landau & lifshitz 1976 ) , @xmath28\delta \theta + \lambda\,\delta { \dot \theta}=0 . \label{8}\ ] ] when the coupling is weak ( @xmath29 small ) , the strongest instability occurs for the lowest value of @xmath30 possible ( see figure [ fig2 ] ) . with this textbook mathematics in mind ,
let us consider a simplified version of eqs .
( [ 2.6a])([2.6c ] ) , with @xmath31 , and some of the higher order terms neglected . with accuracy up to linear terms in ( [ 2.6b ] )
one obtains , obviously , a solution @xmath32 . substituting this solution in ( [ 2.6a ] ) brings this equation close to the standard form ( [ 8 ] ) .
we have solved these equations for @xmath33 , with @xmath30 a positive real parameter .
the instability regions for a lower - order version of equations ( [ 2.6a])([2.6b ] ) resemble the well - known tongues of instability of the mathieu equation . since @xmath34 in general relativity , the lowest value for which the resonance can occur is @xmath35 , and the first two tongues of instability in fig .
[ fig2 ] will be absent ; i.e. , the ratio of the two eigenfrequencies in ( [ 8 ] ) is @xmath0 for the strongest resonance ( kluniak & abramowicz 2002 ) .
however , the behaviour of the resonance can only be properly studied if the third - order terms in ( [ 2.6a ] ) and the second - order terms in ( [ 2.6b ] ) are retained .
these terms provide the non - linearity necessary to saturate the amplitude at a finite value and the damping which affects the frequency of oscillations . using the standard analytic method of successive approximations ( landau & lifshitz 1976 )
, we verified that , as expected , no parametric resonance occurs for strictly geodesic motion , i.e. , for @xmath36 in eqs .
( [ 2.6a])([2.6c ] ) .
we also checked numerically that there is no resonance in the geodesic case .
however , the resonance does occur ( as anticipated in kluniak & abramowicz 2002 ) for even very slightly non - geodesic motion , when the higher order terms are influenced by non - geodesic effects of a rather general form and right amplitude .
in general it is not possible to specify the form of the very small force @xmath15 exactly , because the nature of non - geodesic forces in accretion discs is not yet very accurately known . here
, we only explore the basic mathematical form of the solutions , not the physical origin of the @xmath15 force .
we assume _ ad hoc _ that the possible non - geodesic effects have the form of a non - linear isotropic coupling between the two components of geodesic deviation : @xmath37 \delta \theta , \label{9}\ ] ] @xmath38 r { \left(\dot{\delta\theta}\right)^2}. \label{10}\ ] ] here , @xmath39 and @xmath40 measure the strength of non - geodesic forces in comparison with the geodesic terms of corresponding form and order .
we first assumed @xmath41 for the sake of simplicity . after choosing a particular value of @xmath42 , we solved eqs .
( [ 2.6a])([2.6b ] ) numerically by using the runge - kutta adaptive step - size routine .
the integration was performed over a wide range of starting radii , @xmath43 . for the initial conditions
, we took @xmath44 , @xmath45 , @xmath46 , and @xmath47 .
we then repeated the same procedure for another value of @xmath42 , and in this way we constructed a sequence of solutions that is parameterized by @xmath42 and @xmath48 .
one can locate radii @xmath49 for which @xmath50 grows in time up to a maximum amplitude .
the saturation amplitude , @xmath51 , is shown in fig .
[ fig3 ] ( right panel ) as a function of radius . clearly
, the largest amplitudes correspond to parametric resonance .
close to those radii the test - particle epicyclic frequencies are in rational ratios @xmath52 ; @xmath53 , @xmath54 , and @xmath55 .
note that the growth rate of the amplitudes @xmath56 , @xmath57 decreases with decreasing @xmath42 .
the @xmath0 resonance is the strongest one , as expected . the exact slope of the @xmath0 resonance trace @xmath49 , starting from the point @xmath58 , depends on the choice of coupling parameters in eqs .
( [ 9])([10 ] ) , so the above assumption of identical values for @xmath59 is not universal . for example
, assuming @xmath60 and setting @xmath61 , one finds that the end of the trace is moved to the point @xmath62 .
it is thus shifted in the @xmath63-plane with respect to the case @xmath64 , for which the trace ends at the point @xmath65 , as shown in the left panel of fig .
also the initial conditions for @xmath66 and @xmath67 have an influence on the trace slope .
therefore , the corresponding frequencies that are present in the solutions for @xmath68 , @xmath69 depend on the details of the solution .
nevertheless , the ratio of the frequencies stays at @xmath0 at the maximum amplitude .
this fact is in agreement with the intuition about the properties of eqs .
( [ 2.6a])([2.6b ] ) , and we checked it by fourier - analyzing their solution .
in this paper we assumed the gravitational potential of eq .
( [ 6 ] ) for the sake of simplicity .
when a more accurate , full formulation in general relativity is developed together with a physical mechanism responsible for the non - geodesic coupling term ( [ 9])([10 ] ) , one will be able to constrain the absolute value of the mass of the accreting body .
as noted in section 1 , the frequency ratio of the qpos in some black hole sources is accurately @xmath0 .
specifically , in gro j1655 - 40 the frequency ratio is that of 300 hz and 450 hz , while in xte j1550 - 564 it is that of 184 hz and 276 hz ( abramowicz & kluniak 2001 , remillard et al .
the properties of parametric resonance , discussed above , provide a natural explanation for this ratio .
( 0.48,0.5)obr4.eps can the same mechanism be responsible for the observed frequencies of the khz qpos in neutron stars ?
in neutron star sources the two khz qpos vary considerably in frequency , and their ratio is not always @xmath0 .
a case in point is sco x-1 , whose two khz frequencies are linearly correlated ( fig . [ fig4 ] ) , but not directly proportional to each other ( van der klis et al . 1997 ) .
however , the distribution of points along the line of correlation is not uniform , and the resulting distribution of frequency ratios has a prominent peak at about @xmath0 value , strongly suggesting the presence of a non - linear resonance ( abramowicz et al .
2002b ) .
we note that qpos are not coherent oscillations , and that the typical integration time ( minutes ) greatly exceeds the coherence time of the signal ( fraction of a second ) .
we find that the slope of sco x-1 frequency correlation can be reproduced if we assume that the signal observed at a given time corresponds to the frequencies originating at a particular radius , for example the radius @xmath49 discussed in the previous section .
if this is _ not _ strictly equal to the radius where the maximum amplitude is attained , the ratio of the frequencies departs from 2/3 .
specifically , taking an arbitrary initial amplitude @xmath70 we constructed fourier power spectra of @xmath71 , and @xmath72 .
these power spectra peak at frequencies @xmath73 and @xmath74 , which we interpret here as the observed pair of oscillations .
figure [ fig4 ] shows the correlation between @xmath75 and @xmath76 along the @xmath42-trace from fig .
we assumed @xmath64 as before .
the actual value of @xmath70 has been selected to match the slope of sco x-1 data ( van der klis 1997 ) . because @xmath42 measures the deviation from geodesicity , its value is likely to be a function of the accretion rate in the disc , @xmath77 .
this notion agrees with the reported correlation between the frequencies and the position of the source along the z curve in the color - color diagram .
[ fig4 ] illustrates that the simple scheme that we introduced in the present paper may , indeed , explain the slope of @xmath78 relation .
we have expanded , through the third order , relativistic equations of test motion about a circular geodesic , and have shown that the lowest order expansion corresponds to mathieu s equation with non - standard damping . as expected , the solutions of these ( unperturbed ) equations do not show any resonant phenomenon they simply describe geodesic motion .
when a sufficiently large non - geodesic perturbation is imposed , the deviations grow rapidly at certain resonant radii and the motion departs from circular geodesics .
this is caused by parametric resonance between the meridional and radial epicyclic motions , and the strongest resonance occurs when the two dominant frequencies are near ratio @xmath0 .
we thank michiel van der klis for providing the sco x-1 data and for a very helpful discussion .
most of this work was done during maa s and vk s visit to the supercomputer centre ukaff at leicester university , and maa s , wk s and whl s visits to iap ( paris ) and sissa ( trieste ) .
wk held a post rouge at cesr , funded by cnrs .
vk acknowledges support from gacr 205/03/0902 and gauk 188/2001 , and whl from conacyt ( 36632e ) .
the center for particle physics is supported by the czech ministry of education project ln00a006 . | we have explored the ideas that parametric resonance affects nearly geodesic motion around a black hole or a neutron star , and that it may be relevant to the high frequency ( twin ) quasi - periodic oscillations occurring in some low - mass x - ray binaries .
we have assumed the particles or fluid elements of an accretion disc to be subject to an isotropic perturbation of a hypothetical but rather general form .
we find that the parametric resonance is indeed excited close to the radius where epicyclic frequencies of radial and meridional oscillations are in a @xmath0 ratio .
the location and frequencies of the highest amplitude excitation vary with the strength of the perturbation .
these results agree with actual frequency ratios of twin khz qpos that have been reported in some black hole candidates , and they may be consistent also with correlation of the twin peaks in sco x-1 . |
we thank s. v. kalinin , t. mikolajick , j. m@xmath1ller , t. schenk , d. g. schlom , u. schr@xmath26der , and d. vanderbilt for valuable discussions . s.e.r .-
l . would like to thank s. trolier - mckinstry and c. randall for their hospitality at mri - penn state .
this work was supported by the office of naval research grant no .
n00014 - 12 - 1 - 1040 .
l . would also like to thank the support of conicyt and the sponsor of fulbright foundation .
99 m. t. bohr , r. s. chau , t. ghani and k. mistry , spectrum , ieee * 44 * ( 10 ) , 29 ( 2007 ) .
j. a. kittl _
et al . _ ,
* 86 * ( 7 - 9 ) , 1789 ( 2009 ) .
s. k. kim , s. w. lee , j. h. han , b. lee , s. han and c. s. hwang , adv .
mater . * 20 * 2989 ( 2010 ) .
d. panda and t. -y .
tseng , thin solid films * 531 * , 1 ( 2013 ) .
j. h. choi , y. mao , j. p. chang , mater .
r * 72 * ( 6 ) , 97 ( 2011 ) .
r. ruh , h. j. garrett , r. f. domagala , n. m. tallan , j. am .
* 51 * ( 1 ) , 23 ( 1968 ) .
a. p. mirgorodsky , m. b. smirnov , p. e. quintard and t. merle - mejean , phys .
b * 52 * , 9111 ( 1995 ) .
j. m@xmath1ller , t. s. b@xmath26scke , u. schr@xmath26der , s. mueller , d. br@xmath27uhaus , u. b@xmath26ttger , l. frey , and t. mikolajick , nano .
* 12 * , 4318 ( 2012 ) .
k. s. keun and c. s. hwang , electrochem . and solid - state lett .
* 11 * , g9 ( 2008 ) .
h. kim , p. c. mcintyre , k. c. saraswat , j. mater .
* 19 * ( 2 ) , 643 ( 2004 ) .
r. c. garvie , j. phys .
, 82 , 218 ( 1978 ) .
s. mueller , j. mueller , a. singh , s. riedel , j. sundqvist , u. schroeder , and t. mikolajick , adv . funct . mater . * 22 * ( 11 ) , 2412 ( 2012 ) .
j. m@xmath1ller , u. schr@xmath26der , t. s. b@xmath26scke , i. m@xmath1ller , u. b@xmath26ttger , l. wilde , j. sundqvist , m. lemberger , p. k@xmath1cher , t. mikolajick , and l. frey , j. appl . phys . * 110 * , 114113 ( 2011 ) . s. mueller , c. adelmann , a. singh , s. van elshocht , u. schroeder and t. mikolajick , ecs j. solid state sci . technol . * 1 * ( 6 ) , n123-n126 ( 2012 ) .
t. s. b@xmath26scke , st .
teichert , d. br@xmath27uhaus , j. m@xmath1ller , u. schr@xmath26der , u. b@xmath26ttger , and t. mikolajick , appl .
. lett . * 99 * , 112904 ( 2011 ) .
t. s. b@xmath26scke , j. m@xmath1ller , d. br@xmath27uhaus , u. schr@xmath26der and u. b@xmath26ttger , appl .
lett . * 99 * , 102903 ( 2011 ) . t. schenk , s. mueller , u. schroeder , r. materlik , a. kersch , m. popovici , c. adelmann , s. van elshocht , and t. mikolajick , essderc ( 2013 ) .
j. m@xmath1ller , t. s. b@xmath26scke , d. br@xmath27uhaus , u. schr@xmath26der , u. b@xmath26ttger , j. sundqvist , p. k@xmath1cher , t. mikolajick , and l. frey , appl .
. lett . * 99 * , 112901 ( 2011 ) . m. h. park , h. j. kim , y. j. kim , w. lee , t. moon , and c. s. hwang , appl .
. lett . * 102 * , 242905 ( 2013 ) .
e. h. kisi , c. j. howard and r. j. hill , j. am .
. soc . * 72 * , 1757 ( 1989 ) .
k. m. rabe , in functional metal oxides : new science and novel applications , ed . by satish ogale and v. venkateshan , wiley ( 2013 ) .
d. berlincourt , ieee trans .
sonics ultrason . ,
* 13 * , 116 ( 1966 ) .
zhang , a. b. kounga , w. jo , c. jamin , k. seifert , t. granzow , j. r@xmath26del , and d. damjanovic , adv .
* 21 * , 1 ( 2009 ) .
b. jaffe , proc .
ire , vol * 49 * , 1264 ( 1961 ) .
g. r. love , j. am .
, vol * 73 * , 323 ( 1990 ) .
a. s. mischenko , q. zhang , j. f. scott , r. w. whatmore , n. d. mathur , science , vol * 311 * , 1270 ( 2006 ) .
x. tan , c. ma , j. frederick , s. beckman , and k. g. webber , j. am .
* 94*[2 ] , 4091 ( 2011 ) .
j. w. bennett , k. f. garrity , k. m. rabe , and d. vanderbilt , phys . rev . lett . * 110 * , 017603 ( 2013 ) .
d. j. singh , phys .
b * 52 * , 12559 ( 1995 ) .
m. d. johannes and d. j. singh , phys . rev .
b * 71 * , 212101 ( 2005 ) .
s. e. reyes - lillo and k. m. rabe , phys .
b * 88 * , 180102 ( 2013 ) .
j. lasave , s. koval , n. s. dalal , and r. l. migoni , phys .
* 98 * , 267601 ( 2007 ) .
x. gonze , b. amadon , p. anglade , j. m. beuken , f. bottin , p. boulanger , f. bruneval , d. caliste , r. caracas , m. cote _ et al .
_ , comput .
phys . commun . *
180 * , 2582 ( 2009 ) . h. j. monkhorst and j. d. pack , phys . rev .
b * 13 * , 5188 ( 1976 ) .
r. d. king - smith and d. vanderbilt , phys .
b * 47 * , r1651 ( 1993 ) . j. w. bennett , phys .
procedia * 34 * , 14 ( 2012 ) .
h. j. f. jansen , phys .
b * 43 * , 7267 ( 1991 ) g. -m .
rignanese , f. detraux , x. gonze , and a. pasquarello , phys . rev .
b * 64 * , 134301 ( 2001 ) . x. zhao and d. vanderbilt , phys .
b * 65 * , 075105 ( 2002 ) .
j. e. lowther , j. k. dewhurst , j. m. leger , and j. haines , phys .
b * 60 * , 14485 ( 1999 ) .
h. ding , a. v. virkar and l. feng , solid state ionics * 215 * , 16 ( 2012 ) .
j. jin , y. taek - yung , k. y. woon , p. h. min .
w. fan - xin , z. jin , h. taeghwan , j. of the amer .
soc . * 125 * , 6553 ( 2003 ) . c. jovalekic , m. zdujic , d. poleti , lj .
karanovic , m. mitric , j. sol .
stat . chem . *
181 * , 1321 ( 2008 ) .
k. parlinski , z- .
q. li , and y. kawazoe , phys .
* 78 * , 4063 ( 1997 ) .
note that each of these modes transform as a single partner function of a multidimensional irrep of _
we note the fact that freezing @xmath28 or @xmath29 alone in the tetragonal structure gives rise to a _
pca2@xmath5 _ structure distinct from the orthorhombic phase of zro@xmath0 considered here .
see supplemental material at [ url will be inserted by publisher ] for a description of the structural modes and the construction of the invariants using group theory .
n. a. pertsev , a. g. zembilgotov , and a. k. tagantsev , phys .
lett . * 80 * , 1988 ( 1998 ) .
o. dieguez , k. m . rabe and d. vanderbilt ,
b * 72 * , 144101 ( 2005 ) .
d. fischer , and a. kersch , appl .
lett . * 92 * , 012908 ( 2008 ) .
d. fischer , and a. kersch , j. appl .
phys . * 104 * , 084104 ( 2008 ) .
x. zhao and d. vanderbilt , phys .
b * 65 * , 233106 ( 2002 ) .
s. clima , d. j. wouters , c. adelman , t. schenk , u. schroeder , m. jurczak , and g. pourtois , appl .
104 * , 092906 ( 2014 ) . | density functional calculations are performed to investigate the experimentally - reported field - induced phase transition in thin - film zro@xmath0 ( j. m@xmath1ller _ et al .
_ , nano .
lett .
12 , 4318 ) .
we find a small energy difference of @xmath2 1 mev / f.u . between the nonpolar tetragonal and polar orthorhombic structures , characteristic of antiferroelectricity .
the requisite first - order transition between the two phases , which atypically for antiferroelectrics have a group - subgroup relation , results from coupling to other zone - boundary modes , as we show with a landau - devonshire model .
tetragonal zro@xmath0 is thus established as a previously unrecognized lead - free antiferroelectric with excellent dielectric properties and compatibility with silicon .
in addition , we demonstrate that a ferroelectric phase of zro@xmath0 can be stabilized through epitaxial strain , and suggest an alternative stabilization mechanism through continuous substitution of zr by hf .
2.5 in 0.5 in # 1 # 1 # 1 # 1 # 1 zirconia ( zro@xmath0 ) is a high - k dielectric @xcite , chemically and structurally similar to hfo@xmath0 , and likewise is a candidate for dynamic random access memory ( dram ) applications @xcite and complementary metal - oxide - semiconductor ( cmos ) devices @xcite .
bulk zro@xmath0 has a high - symmetry cubic ( _ fm@xmath3 m _ ) structure ( fig .
[ fig : zro2-phases](a ) ) above 2400 k , and a tetragonal ( _ p4@xmath0/nmc _ ) structure ( fig .
[ fig : zro2-phases](b ) ) between 2400 k and 1200 k @xcite .
the tetragonal structure is related to the cubic structure by freezing in an unstable @xmath4 mode @xcite and is nonpolar . below 1200 k , zro@xmath0 is monoclinic ( _ p2@xmath5/c _ ) ( fig .
[ fig : zro2-phases](c ) ) .
the first - order transition from the tetragonal phase to the monoclinic phase changes the coordination number of zr from 8 to 7 and increases the volume by @xmath2 5 % . in light of the extensive research which has been conducted over the past fifty years on this relatively simple dielectric
, the recent report of antiferroelectric - like double - hysteresis loops in thin film zro@xmath0 @xcite at first seems rather surprising . in thin film zro@xmath0
, the tetragonal - monoclinic transition temperature is suppressed and the structure is tetragonal at room temperature @xcite ; in contrast , thin film hfo@xmath0 is monoclinic at room temperature and exhibits simple dielectric behavior . the field - induced polar phase in zro@xmath0 , which appears above a critical field on the order of 2 mv / cm , is isostructural with the ferroelectric phases that have been observed in thin films of hfo@xmath0 doped with al @xcite , y @xcite , gd @xcite , si @xcite and sr @xcite , as well as in ( hf@xmath6zr@xmath6)o@xmath0 thin films @xcite .
the structure of the polar phase is orthorhombic ( _ pca2@xmath5 _ ) @xcite and corresponds to a distortion of the high - symmetry cubic structure , as depicted in fig .
[ fig : zro2-phases ] .
antiferroelectrics have recently been the subject of increasing interest @xcite .
the characteristic electric - field - induced transition from a nonpolar to a strongly polar phase is the source of functional properties and promising technological applications
. non - linear strain and dielectric responses due to the phase switching are useful for transducers and electro - optic applications @xcite .
the shape of the double hysteresis loop suggests applications in high - energy storage capacitors @xcite .
in addition , an electro - caloric effect can be observed in systems with a large entropy change between the two phases @xcite . while most attention has focused on pbzro@xmath7 and related perovskites @xcite ,
a recent theoretical materials design search @xcite suggested that there are many more antiferroelectric compounds to be discovered .
experimentally reported phases of zro@xmath0 : ( a ) cubic ( _ fm@xmath3 m _ ) , ( b ) tetragonal ( _ p4@xmath0/nmc _ ) , ( c ) monoclinic ( _ p2@xmath5/c _ ) and ( d ) orthorhombic ( _ pca2@xmath5 _ ) . ] in this work , we use first - principles calculations to provide clear evidence that the tetragonal phase of zro@xmath0 is a previously unrecognized antiferroelectric material , and that the behavior observed in thin films is intrinsic .
we find a remarkably small energy difference of @xmath2 1 mev per formula unit ( mev / f.u . ) between the nonpolar tetragonal and polar orthorhombic structures , which is a key characteristic of antiferroelectricity @xcite .
we show with a polynomial expansion of the energy that the requisite first - order transition between nonpolar and polar phases , which in the present case , atypically for antiferroelectrics , have a group - subgroup relation , results from coupling to other zone - boundary modes .
this novel mechanism for antiferroelectricity provides a basis for broader searches for antiferroelectric materials with optimal functional properties .
in addition , we demonstrate that the polar phase of zro@xmath0 can be stabilized through epitaxial strain , and we discuss the possibility of ferroelectricity in the solid solution ( zr@xmath8hf@xmath9)o@xmath0 .
density - functional theory ( dft ) calculations are performed using version 7.4.1 of ` abinit ` @xcite and the local - density approximation ( lda ) .
we use a plane - wave energy cutoff of 680 ev , and a @xmath10 monkhorst - pack sampling of the brillouin zone @xcite for all structural optimizations .
polarization was calculated on a @xmath11 grid using the modern theory of polarization @xcite as implemented in ` abinit ` .
we used norm - conserving pseudopotentials from the bennett - rappe library @xcite with reference configurations : _ zr _ ( [ _ kr_]4d@xmath125s@xmath12 ) , _ hf _ ( [ _ xe_]4f@xmath135d@xmath146s@xmath12 ) and _ o _ ( [ _ he_]2s@xmath152@xmath16 ) , generated by the ` opium ` code @xcite .
table [ table : polar - phases - zro2 ] reports relaxed bulk energies and lattice constants for the experimentally observed bulk phases of zro@xmath0 .
our results are in good agreement with previous first principles results @xcite .
the monoclinic phase is the lowest energy structure , consistent with experiments .
experimental lattice constants @xcite are underestimated by 1 % , typical for lda .
the spontaneous polarization of the orthorhombic structure was found to be p@xmath17 @xmath2 58 @xmath18c / cm@xmath15 . lcccccc phase & @xmath20 & @xmath21 & latt . & this work & prev .
work@xmath22 & exp.@xmath23 + .energy @xmath20 ( mev / f.u . ) , volume expansion @xmath21 ( @xmath24 ) and lattice parameters ( latt . ) in ( @xmath25 : monoclinic angle ) of the experimental phases of zro@xmath0 .
[ cols= " < " , ] + [ table : polar - phases - hfo2 ] in summary , the experimentally observed field - induced ferroelectric transition corresponds to an intrinsic behavior of thin - film zro@xmath0 .
the tetragonal - to - orthorhombic antiferroelectric transition is explained as a field - induced first - order transition where the polar mode is stabilized from the cubic phase by a coupling with two zone - boundary modes .
our results suggest that a ferroelectric phase could be favored over the antiferroelectric phase by appropriate epitaxial strain or isovalent substitution of zr for hf .
the absence of toxic elements and the compatibility of zro@xmath0 with silicon make these results especially relevant for technological applications . |
this article is concerned with a topological quantity , the so - called _ fixed - point index _ of a homeomorphism of jordan curves , which has proven useful in the study of various areas of complex analysis .
we begin with its definition : a _ jordan curve _ is a homeomorphic image of a topological circle @xmath0 in the complex plane @xmath1 .
jordan domain _ is a bounded open set in @xmath1 with jordan curve boundary .
we use the term _ closed jordan domain _ or _ compact jordan domain _ to refer to the closure of a jordan domain .
we define the _ positive orientation _ on a jordan curve as usual .
that is , if @xmath2 is a closed jordan domain , then as we traverse @xmath3 in what we call the _ positive _ direction , the interior of @xmath2 stays to the left .
let @xmath2 and @xmath4 be closed jordan domains .
let @xmath5 be a homeomorphism of jordan curves which is fixed - point - free and orientation - preserving .
we call such a homeomorphism _ indexable_. then @xmath6 is a closed curve in the plane which misses the origin .
it has a natural orientation induced by traversing @xmath3 positively .
then we define the _ fixed - point index _ of @xmath7 , denoted @xmath8 , to be the winding number of @xmath6 around the origin .
two examples are shown in figures [ fig : ex disjoint then winding 0 ] and [ fig : ex neg winding forced ] .
we remark that the fixed - point index depends crucially on the choice of homeomorphism , and also on the way that the sets @xmath2 and @xmath4 sit on top of each other .
it is a worthwhile exercise to construct an indexable homeomorphism @xmath9 , for @xmath2 and @xmath4 as in figure [ fig : ex neg winding forced ] , having fixed - point index unequal to @xmath10 .
( 0,-1.02)(10.129063,1.015 ) ( 0.28453124,-0.015)@xmath2 ( 7.004531,-0.295)@xmath11 ( 2.5345314,-0.675)@xmath12 ( 4.6945314,0.105)@xmath13 ( 1.28,-0.43)(1.5,-0.615 ) ( 8.67,0.585)(8.57,0.685 ) ( 2.29,-0.425)(7.23,0.055 ) ( 2.29,-0.425 ) ( 7.23,0.055 ) ( 9.234531,0.165)@xmath4 ( 0,-3.0367188)(6.9990625,3.0567188 ) ( 6.174531,-0.46671876)@xmath4 ( 3.0245314,2.7932813)@xmath2 ( 5.82,1.3232813)(0.0,-1.6767187 ) ( 4.74,2.5832813)(1.12,-3.0167189 ) ( 5.82,1.3232813)(4.74,2.5832813 ) ( 4.3,1.3232813)(4.1,2.5832813 ) ( 3.26,1.3232813)(3.24,2.5832813 ) ( 1.6,1.3232813)(2.16,2.5832813 ) ( 0.0,1.3232813)(1.12,2.5832813 ) ( 0.0,0.66328126)(1.12,0.98328125 ) ( 0.0,0.02328125)(1.12,-0.05671875 ) ( 0.0,-0.75671875)(1.12,-1.2967187 ) ( 0.0,-1.6767187)(1.12,-3.0167189 ) ( 1.52,-1.6767187)(1.78,-3.0167189 ) ( 2.64,-1.6767187)(2.62,-3.0167189 ) ( 3.92,-1.6767187)(3.52,-3.0167189 ) ( 5.82,-1.6767187)(4.74,-3.0167189 ) ( 5.82,-1.0967188)(4.74,-1.4167187 ) ( 5.82,-0.49671876)(4.74,-0.5967187 ) ( 5.82,0.44328126)(4.74,0.88328123 ) ( 4.74,0.00328125)(4.74,0.30328125 ) ( 5.82,-0.03671875)(5.82,0.16328125 ) ( 6.164531,2.0132813)@xmath13 fixed - point index has found applications for example in the theories of circle packing @xcite , koebe uniformization @xcite , and sierpinski carpets @xcite*section 12 . in all of these settings
, it has been applied to prove powerful existence , rigidity , and uniformization statements .
most recently , the current author has used fixed - point index , including torus parametrization , to prove rigidity statements for collections of possibly - overlapping round disks , see .
the fixed - point index measures the following topological quantity : suppose that @xmath2 and @xmath4 are closed jordan domains , and @xmath14 is a homeomorphism , having finitely many fixed points , which restricts to an indexable homeomorphism @xmath15 .
there is a well - understood notion of the _ multiplicity _ of a fixed point of @xmath16 .
then the fixed - point index @xmath17 counts the number of fixed - points of @xmath16 , with multiplicity . for more discussion along these lines , see @xcite*section 2 . in this article , we describe a new technique for working with fixed - point index , which we call the _ torus parametrization _ of a pair of jordan curves , defined in section [ chap : torus ] .
we apply torus parametrization to give a new , elementary proof of the following fundamental lemma : [ threep ] [ 3p ] [ lem:3p ] [ lem : three points ] [ tppl ] let @xmath2 and @xmath4 be compact jordan domains in transverse position , with boundaries oriented positively .
let @xmath18 appear in counterclockwise order , similarly @xmath19 .
then there is an indexable homeomorphism @xmath5 sending @xmath20 for @xmath21 , so that @xmath22 .
two jordan domains are in _ transverse position _ if their boundary jordan curves cross wherever they meet , c.f . definition [ transverse position ] in section [ sec background ] . the example given in figure [ fig : ex neg winding forced ] shows that if we prescribe the images of four points , then a negative fixed - point index may be forced .
a version of the three point prescription lemma [ threep ] is stated in @xcite*lemma 8.14 , but we have not been able to fill in the details of the argument .
the idea of the approach is as follows : first , any riemann mapping @xmath23 between open jordan domains extends to a homeomorphism @xmath24 of their boundaries , and we may prescribe the images of three points of @xmath25 in @xmath26 by post - composing with self - biholomorphisms of @xmath27 .
next , it is known that any isolated fixed point of a holomorphic map has non - negative multiplicity , see @xcite*section 2 .
thus the map @xmath28 , if it is indexable , has non - negative fixed - point index , because the fixed - point index of @xmath28 counts the fixed points of @xmath16 with multiplicity , completing the argument .
however , it is not clear how to deal with possible fixed points in the induced boundary map @xmath28 .
our proof of lemma [ 3p ] uses only induction and plane topology arguments and is given in section [ sec:3p ] .
we also state and prove a new fundamental lemma on fixed - point index , generalizing the well - known circle index lemma [ cil ] which states that the fixed - point index of an indexable homeomorphism between circles is always non - negative .
the circle index lemma was a crucial ingredient in all of the applications of fixed - point index described above .
in our generalization , round disks are replaced by closed jordan domains which do not disconnect each other . in particular
, the closed jordan domains @xmath2 and @xmath4 are said to _ cut each other _ if @xmath29 or @xmath30 is disconnected . then : [ no cut index ] let @xmath2 and @xmath4 be closed jordan domains in transverse position , which do not cut each other , having boundaries oriented positively .
let @xmath31 be an indexable homeomorphism
. then @xmath22 .
the proof appears at the end of section [ incompat section ] . as an example of the power of fixed - point index ,
we apply the three point prescription lemma [ threep ] and lemma [ no cut index ] to prove a version of the incompatibility theorem of schramm @xcite*theorem 3.1 , as described in section [ incompat section ] .
the incompatibility theorem is then easily applied in section [ cp proofs ] to prove some well - known rigidity theorems for circle packings .
the ideas for these proofs are borrowed from .
* acknowledgments . *
thanks to jordan watkins for many fruitful discussions , especially for pointing us strongly in the direction of what we call _ torus parametrization_. thanks also to mario bonk for many fruitful discussions , especially for helpful suggestions that greatly simplified the proof of lemma [ no cut index ] .
in the upcoming discussion it will be useful to have access to two well - known lemmas on fixed - point index .
this section is devoted to introducing these two lemmas .
our first background lemma says essentially that `` the fixed - point index between two circles is always non - negative '' : [ cil ] [ lem : circle index lemma ] let @xmath2 and @xmath4 be closed jordan domains in @xmath1 , with boundaries oriented positively , and let @xmath5 be an indexable homeomorphism .
then the following hold . 1
. we have @xmath32 .
2 . if @xmath33 or @xmath34 , then @xmath35 .
3 . if @xmath2 and @xmath4 have disjoint interiors , then @xmath36 .
4 . if @xmath3 and @xmath37 intersect in exactly two points , then @xmath38 . as a consequence of the above , if @xmath2 and @xmath4 are closed disks in the plane , then @xmath22 .
lemma [ cil ] can be found in @xcite*lemma 2.2 , with a clear and complete proof .
there it is indicated that a version of the lemma appeared earlier in @xcite .
the moral of our second background lemma is that fixed - point indices `` add nicely '' : [ lem : indices add ] [ additivity ] [ ial ] suppose that @xmath2 and @xmath39 are interiorwise disjoint closed jordan domains which meet along a single positive - length jordan arc @xmath40 , similarly for @xmath4 and @xmath41 .
then @xmath42 and @xmath43 are closed jordan domains .
let @xmath44 and @xmath45 be indexable homeomorphisms .
suppose that @xmath7 and @xmath46 agree on @xmath40 .
let @xmath47 be induced via restriction to @xmath7 or @xmath46 as necessary .
then @xmath48 is an indexable homeomorphism and @xmath49 .
( 0,-2.49)(7.2690625,2.51 ) ( 3.69,-0.5)(3.69,0.4 ) ( 3.09,-0.5)(3.09,0.4 ) ( 3.33,1.67 ) ( 3.57,-1.29 ) ( 3.7545313,1.6)@xmath50 ( 3.954531,-1.18)@xmath51 ( -0.035,0.275)(0.09,-0.52 ) ( 6.7,0.035)(6.84,-0.55 ) ( 0.42453125,2.14)@xmath2 ( 6.9045315,1.0)@xmath39 [ ex : indices add ] the situation is as depicted in figure [ fig : lem indices add 1 ] . we may consider @xmath8 to be @xmath52 times the change in argument of the vector @xmath13 , as @xmath12 traverses @xmath3 once in the positive direction .
then as @xmath12 varies positively in @xmath3 and in @xmath53 the contributions to the sum @xmath54 along @xmath40 cancel .
next , we wish to make precise our notion of _ transverse position _ : [ transverse position ] two jordan curves @xmath55 and @xmath56 are in _ transverse position _
if for any point @xmath57 where they meet , they _ cross transversely _ , which here means that there is an open neighborhood @xmath58 of @xmath12 and a homeomorphism @xmath59 between @xmath60 and the open unit disk @xmath61 , so that @xmath62 and @xmath63 .
two ( open or closed ) jordan domains are said to be in _ transverse position _ if their boundary jordan curves are in transverse position .
note that if two jordan curves are in transverse position , then they meet finitely many times , by a compactness argument .
finally , it will be helpful to have available to us the terminology of the following definition : suppose that @xmath64 and @xmath65 are all subsets of @xmath1 .
then we say that the collections @xmath66 and @xmath67 are in the same _ topological configuration _ if there is an orientation - preserving homeomorphism @xmath68 so that @xmath69 for all @xmath70 . in practice
the collections of objects under consideration will not be labeled @xmath71 and @xmath72 , but there will be some natural bijection between them .
then our requirement is that @xmath73 respects this natural bijection .
we will say that certain conditions on some objects _ uniquely determine _ their topological configuration if any two collections of objects satisfying the given conditions are in the same topological configuration . for example , considering a point @xmath74 and an open jordan domain @xmath75 , the condition that @xmath76 uniquely determines the topological configuration of @xmath77 , but the condition that @xmath78 does not uniquely determine the topological configuration of @xmath77 .
( we may have @xmath79 , or @xmath80 , and these situations are topologically distinct . )
the following lemma says that when working with fixed - point index , we need to consider our jordan domains only `` up to topological configuration . ''
[ prop : index invariance ] suppose @xmath2 and @xmath4 are closed jordan domains .
let @xmath81 be an indexable homeomorphism .
suppose that @xmath82 and @xmath83 are also closed jordan domains , so that @xmath84 and @xmath85 are in the same topological configuration , via the homeomorphism @xmath86 .
let @xmath87 be induced in the natural way , explicitly as @xmath88 .
then @xmath89 is indexable with respect to the usual orientation on @xmath90 and @xmath91 , and @xmath92 .
the following is well - known . for a reference ,
see chapters 1 and 2 of @xcite .
every orientation - preserving homeomorphism @xmath93 is homotopic to the identity map via homeomorphisms .
thus let @xmath94\to { \mathbb{c}}$ ] be such a homotopy from the identity to @xmath73 .
explicitly , for fixed @xmath95 we have that @xmath96 is an orientation - preserving homeomorphism @xmath93 , with @xmath97 equal to the identity on @xmath1 and @xmath98 .
let @xmath99 and @xmath100 .
then @xmath101 and @xmath102 are closed jordan domains , because @xmath96 is a homeomorphism .
let @xmath103 be induced in the natural way , explicitly as @xmath104 .
let @xmath105 .
then tautologically @xmath106 is the winding number of @xmath107 around the origin , and @xmath108 is the winding number of @xmath109 around the origin .
every @xmath110 is a closed curve because @xmath111 is a closed curve and @xmath112 is continuous .
once we establish that no @xmath110 passes through the origin lemma [ prop : index invariance ] will be proved because we have an induced homotopy from @xmath107 to @xmath109 , and two curves homotopic in @xmath113 have the same winding number around the origin .
suppose for contradiction that @xmath114 . then there is a @xmath115 so that @xmath116 .
thus @xmath117 , and so @xmath118 , contradicting the fixed - point - free condition on @xmath119 .
in this section , we state and prove the incompatibility theorem of oded schramm , appearing in @xcite*theorem 3.1 . before doing so , we need some preliminary definitions : first , a _ topological rectangle _ is a closed jordan domain @xmath120 with four marked points on its boundary @xmath121 , which we naturally call its _ corners_. a _ side _ of @xmath120 is a closed sub - arc of @xmath121 having two corners of @xmath120 as its endpoints , and containing no other corner of @xmath120 .
we abuse notation slightly and use the same symbol , in this case @xmath120 , to refer both to a topological rectangle and to its constituent closed jordan domain .
we define _ topological triangles _ , their _ corners _ , and their _ sides _ analogously , and will employ the same abuse of notation .
a _ packing _ of a topological rectangle @xmath120 consists of a finite collection @xmath122 of closed jordan domains so that the following hold : * every @xmath123 is contained in @xmath120 . *
the @xmath123 are pairwise interiorwise disjoint , and any two of them meet at at most one point . *
each of the @xmath123 meets @xmath121 at at most one point , and no @xmath123 meets a corner of @xmath120 . * for every connected component @xmath60 of @xmath124 , we have that the closure of @xmath60 is a topological triangle @xmath125 each of whose sides is contained in one of the @xmath126 , or in a side of @xmath120 .
then every corner of @xmath125 is either a corner of @xmath120 , or an intersection point of some pair of the @xmath123 .
see figure [ clown1 ] for two examples of packings of topological rectangles .
let @xmath127 denote the sides of @xmath120 .
the _ contact graph _ of the packing of @xmath120 by @xmath128 is the graph having vertices @xmath129 corresponding to the @xmath128 and @xmath130 corresponding to the sides of @xmath120 , both in the natural way , so that two distinct vertices share an edge if and only if the corresponding sets meet .
note that for example , the contact graph of a packing of a rectangle is always a triangulation of a square , that is , a triangulation of a topological closed disk having four boundary edges .
( 0,-2.52)(11.74,2.54 ) ( 5.54,1.22)(0.46,-1.36 ) ( 1.7,-0.12)(1.4,-0.12)(0.45218492,0.28114378)(0.5,0.58)(0.5478151,0.8788562)(1.4001769,1.1611928)(1.8,1.18)(2.1998231,1.1988072)(3.0059357,0.9609888)(3.0,0.68)(2.9940643,0.3990112)(2.6,0.28)(2.5,0.18)(2.4,0.08)(2.0,-0.12)(1.7,-0.12 ) ( 2.5,0.18)(2.7,0.28)(3.123653,0.2064788)(3.3,0.08)(3.476347,-0.0464788)(3.2,-0.42)(2.9,-0.42)(2.6,-0.42)(2.3,0.08)(2.5,0.18 ) ( 3.3,0.08)(3.2,0.18)(3.0,0.38)(3.0,0.68)(3.0,0.98)(3.4026165,1.1522919)(4.1,1.18)(4.7973833,1.2077081)(5.5714183,0.5651983)(5.5,0.28)(5.4285817,-0.0051983013)(4.698314,0.03803818)(4.2,-0.02)(3.7016857,-0.07803818)(3.4,-0.02)(3.3,0.08 ) ( 4.2,-0.02)(4.8,-0.02)(5.5,-0.02)(5.5,-0.52)(5.5,-1.02)(4.0,-1.32)(3.0,-1.32)(2.0,-1.32)(0.5,-1.22)(0.5,-0.52)(0.5,0.18)(0.9,-0.12)(1.7,-0.12)(2.5,-0.12)(1.9003581,-0.44675693)(2.9,-0.42)(3.899642,-0.39324304)(3.6,-0.02)(4.2,-0.02 ) ( 10.62,2.4)(6.54,-2.48 ) ( 6.58,1.16)(6.6309495,1.3284905)(6.68,1.76)(6.88,1.96)(7.08,2.16)(7.38,2.36)(7.78,2.36)(8.18,2.36)(8.58,2.06)(8.58,1.66)(8.58,1.26)(8.375797,0.76556337)(8.18,0.66)(7.984203,0.5544366)(7.48,0.56)(7.28,0.56)(7.08,0.56)(6.5290504,0.9915095)(6.58,1.16 ) ( 8.88,0.86)(8.76,0.94)(8.380138,0.82594526)(8.18,0.66)(7.9798613,0.49405473)(8.18,0.36)(8.28,0.26)(8.38,0.16)(8.48,-0.14)(8.68,-0.14)(8.88,-0.14)(8.86,0.12)(8.98,0.26)(9.1,0.4)(9.1,0.36)(9.16,0.5)(9.22,0.64)(9.0,0.78)(8.88,0.86 ) ( 9.18,2.36)(8.98,2.36)(8.602183,1.9525076)(8.58,1.66)(8.5578165,1.3674924)(8.68,1.46)(8.78,1.26)(8.88,1.06)(8.754272,0.94004184)(8.88,0.86)(9.005728,0.7799582)(9.15612,0.70268345)(9.28,0.66)(9.403879,0.61731654)(9.34,0.54)(9.58,0.56)(9.82,0.58)(9.98,0.66)(10.18,0.76)(10.38,0.86)(10.48,0.96)(10.58,1.16)(10.68,1.36)(10.25945,1.8262779)(10.18,1.96)(10.100551,2.093722)(9.812722,2.1994033)(9.68,2.26)(9.547278,2.3205967)(9.38,2.36)(9.18,2.36 ) ( 6.58,-1.04)(6.574384,-0.74032813)(7.0397286,0.6195747)(7.28,0.56)(7.520272,0.5004253)(8.18,-0.14)(8.68,-0.14)(9.18,-0.14)(9.352913,0.5601822)(9.58,0.56)(9.807088,0.55981773)(10.600395,-0.640208)(10.58,-0.94)(10.559605,-1.239792)(9.999562,-1.8670098)(9.78,-2.04)(9.560438,-2.2129903)(8.98,-2.44)(8.38,-2.44)(7.78,-2.44)(7.24302,-2.2863748)(7.08,-2.04)(6.91698,-1.7936251)(6.5856156,-1.3396719)(6.58,-1.04 ) ( 5.06,-1.615)@xmath120 ( 10.98,-2.095)@xmath131 next , suppose that we have packings of topological rectangles @xmath120 and @xmath131 by collections of closed jordan domains @xmath128 and @xmath132 , respectively .
the two packings are said to be in _ transverse position _ if : * for every pair of integers @xmath133 and @xmath134 , both between @xmath135 and @xmath136 , with possibly @xmath137 , we have that @xmath123 and @xmath138 are in transverse position as closed jordan domains . * for every @xmath70 , we have that @xmath123 and @xmath131 are in transverse position as closed jordan domains , as are @xmath139 and @xmath120 . *
no intersection point of a pair of the sets @xmath140 lies on any @xmath141 nor on @xmath142 .
similarly , no intersection point of a pair of the sets @xmath143 lies on any @xmath126 nor on @xmath121 . for two examples of packings in transverse position , see figure [ clown2 ] .
finally , two closed jordan domains @xmath2 and @xmath4 which are in transverse position are said to _ cut _ one another if at least one of the sets @xmath29 and @xmath144 is disconnected .
though it is not important for us , it holds for such @xmath2 and @xmath4 that @xmath29 is connected if and only if @xmath30 is .
we are now ready to state our version of the incompatibility theorem originally due to schramm .
the original appears in @xcite*theorem 3.1 , where it is used to give the first proof of the rigidity of circle packings filling the complex plane . when processing our statement of the incompatibility theorem [ incompat ] , it may be helpful to keep the following in mind : in schramm s original formulation , our notion of shapes _ cutting one another _ is referred to as
incompatibility_. then the incompatibility theorem [ incompat ] may be remembered as , `` if two incompatible rectangles are packed in the same combinatorial way , then at least one pair of corresponding shapes of the two packings will be incompatible . ''
the precise statement given in @xcite is somewhat different from the one we give here .
our statement communicates the main idea of the theorem , and suffices for our applications .
our proof uses fixed - point index , with the main new tool being lemma [ no cut index ] .
the original proof by schramm is via different methods , but is also elementary .
( 0,-1.7292187)(6.6690626,1.7292187 ) ( 5.81,0.5757812)(0.81,-0.62421876 ) ( 5.21,1.1757812)(1.41,-1.2242187 ) ( 1.41,1.1757812 ) ( 5.21,1.1757812 ) ( 5.81,0.5757812 ) ( 5.81,-0.62421876 ) ( 0.81,-0.62421876 ) ( 0.81,0.5757812 ) ( 1.41,-1.2242187 ) ( 5.21,-1.2242187 ) ( 6.1045313,-0.03421875)@xmath145 ( 2.8545313,0.30578125)@xmath146 ( 0.45453125,0.02578125)@xmath147 ( 3.7145312,-0.37421876)@xmath148 ( 4.8545313,0.00578125)@xmath149 ( 3.2845314,1.4657812)@xmath150 ( 1.7645313,-0.03421875)@xmath151 ( 3.2045312,-1.4942187)@xmath152 ( 6.054531,0.74578124)@xmath120 ( 1.2645313,1.5257813)@xmath131 [ incompat ] let @xmath120 and @xmath131 be topological rectangles having sides @xmath153 and @xmath154 respectively , in the topological configuration depicted in figure [ topo conf incompat rect ]
. suppose that we are given packings of @xmath120 and @xmath131 by collections of closed jordan domains @xmath128 and @xmath132 respectively , in transverse position , so that the packings are combinatorially equivalent in the following precise sense : denoting the contact graphs of the packings by @xmath155 and @xmath156 respectively , we insist that the following holds : letting @xmath157 and @xmath158 denote the vertex sets of @xmath155 and @xmath156 in the natural way , we have that @xmath155 and @xmath156 are isomorphic via the identification @xmath159 for @xmath160 . then , there is an @xmath70 so that @xmath123 and @xmath139 cut each other .
for example , the packings shown in figure [ clown1 ] share a contact graph . in figure [ clown2 ]
we have overlaid them in two different ways , in both cases ensuring that the hypotheses of theorem [ incompat ] are satisfied .
we see that in each case , a pair of corresponding closed jordan domains cut one another .
the rest of this section consists of a proof of the incompatibility theorem [ incompat ] .
our proof relies on the following simple lemma , which appeared in the introduction but is restated here for the convenience of the reader : let @xmath2 and @xmath4 be closed jordan domains in transverse position , which do not cut each other , having boundaries oriented positively .
let @xmath31 be an indexable homeomorphism .
then @xmath22 .
we now give the proof of theorem [ incompat ] assuming lemma [ no cut index ] : first , note that there is a natural bijection between the @xmath161 and the @xmath162 , where we write @xmath163 to denote the connected components of @xmath164 , similarly @xmath165 .
moreover , for fixed @xmath119 , we have that @xmath161 and @xmath162 are topological triangles , and that the corners of @xmath161 correspond in a natural way to those of its partner @xmath162 .
( to see this , one may consider the graphs @xmath155 and @xmath156 .
each is the 1-skeleton of a triangulation of a topological closed disk , and because the graphs are isomorphic , we get that the triangulations are combinatorially equivalent
. then the @xmath161 are in natural bijection with the faces @xmath166 of this combinatorial triangulation , as are the @xmath162 . )
given such a pair @xmath161 and @xmath162 , we have that @xmath161 and @xmath162 are in transverse position as closed jordan domains , because the packings we began with are in transverse position .
for every such pair @xmath161 and @xmath162 , orient @xmath167 and @xmath168 positively , and let @xmath169 be an indexable homeomorphism identifying corresponding corners , satisfying @xmath170 .
we may do so by the three point prescription lemma [ tppl ] .
next , for every @xmath70 , orienting @xmath126 and @xmath141 positively , we obtain an indexable homeomorphism @xmath171 by restriction to the @xmath172 as necessary .
also , via the same procedure , orienting @xmath121 and @xmath142 positively , we obtain an indexable homeomorphism @xmath173 .
then , by the index additivity lemma [ ial ] , we have : @xmath174 now , as we saw in figure [ fig : ex neg winding forced ] , we have that @xmath175 . on the other hand ,
every @xmath176 is non - negative by construction , and the @xmath177 are non - negative by lemma [ no cut index ] , which gives us a contradiction . to complete the proof of the incompatibility theorem [ incompat ]
, we require the proofs of several lemmas .
the proof of the three point prescription lemma [ tppl ] is given in section [ sec:3p ] using torus parametrization .
our proof of lemma [ no cut index ] is inspired by an argument given by schramm in @xcite*proof of lemma 2.2 in support of the circle index lemma [ cil ] .
we first need to prove a topological lemma on jordan domains which do not cut each other : [ uniqa ] suppose @xmath2 and @xmath4 are closed jordan domains in transverse position which do not cut each other , whose boundaries meet at least twice .
then the topological configuration of @xmath84 is determined by how many times @xmath3 and @xmath37 meet .
first , suppose without loss of generality that @xmath2 is the closed unit disk @xmath178 .
let @xmath179 be the number of intersection points of @xmath3 with @xmath37 .
note also that we may without loss of generality pick these arbitrarily along @xmath180 .
label these @xmath181 in clockwise order around @xmath180 .
this brings us to the situation of figure [ jon1 ] .
orient @xmath3 and @xmath37 as usual .
we now follow what happens as we traverse @xmath37 . by relabeling
we may suppose that @xmath37 enters @xmath2 at @xmath182 . because the interior of @xmath4 stays to the left of @xmath37 , and because @xmath37 crosses @xmath3 at every point where the two curves meet
, it follows that @xmath37 exits @xmath2 at @xmath183 .
the same reasoning allows us to conclude that @xmath37 enters @xmath2 at @xmath184 , etc .
, so we get that @xmath37 enters @xmath2 at @xmath185 for all odd @xmath186 , and exits @xmath2 at @xmath185 for all even @xmath186 .
this brings us to the situation of figure [ jon2 ] .
we now consider where @xmath37 goes after it crosses @xmath182 .
denote by @xmath185 the point of @xmath187 at which it arrives immediately after crossing @xmath182 , noting that then @xmath133 is even .
we wish to establish that then @xmath188 , so suppose for contradiction that @xmath189 .
this brings us to the situation of figure [ jon3 ] .
then @xmath190_{\partial \tilde k}$ ] disconnects @xmath191 into two components , call them @xmath192 and @xmath193 .
we have that every connected component of @xmath194 must then be completely contained in one of @xmath192 and @xmath193 .
but @xmath29 is connected by hypothesis , so one of @xmath192 and @xmath193 must be disjoint from @xmath29 .
we then get a contradiction , because , keeping careful track of the orientation of @xmath37 , we see that there are points of @xmath194 immediately counterclockwise from @xmath182 along @xmath3 , and points of @xmath194 immediately clockwise from @xmath183 along @xmath3 , and these lie in different components of @xmath195_{\partial \tilde k}$ ] unless @xmath188 .
the same reasoning allows us to conclude that , for any odd @xmath196 , after entering @xmath2 at @xmath197 , the curve @xmath37 exits @xmath2 at @xmath198 , bringing us to the situation of figure [ jon4 ] . by the same reasoning as in the previous paragraph
, we get that , for any even @xmath196 , after exiting @xmath2 at @xmath197 , the curve @xmath37 enters @xmath2 at @xmath198 , adopting the convention that @xmath199 . however , in this case , there are two ways to connect @xmath197 to @xmath198 : roughly speaking , we may either travel clockwise around the complement of @xmath200 from @xmath197 to @xmath198 , or counterclockwise .
if we connect every such pair @xmath197 and @xmath198 with `` clockwise '' arcs , then the resulting orientation on @xmath37 is not positive with respect to the jordan domain it bounds , see figure [ jon5 ] .
thus suppose without loss of generality , by relabeling if necessary , that @xmath201 and @xmath182 are connected with a `` counterclockwise '' arc .
now for the remaining pairs @xmath202 , with @xmath203 even , there are no choices ( up to topological equivalence ) about how to draw the connecting sub - arcs of @xmath37 between them , and we arrive at the situation of figure [ jon6 ] .
which _ a priori _ topological configurations can occur for two jordan curves in transverse position is a poorly understood question , and is known as the study of _
meanders_. we are fortunate that our setting is nice enough that a statement like that of lemma [ uniqa ] is possible .
thanks to thomas lam for informing us of the topic of meander theory .
[ no cut index proof ] in light of lemmas [ prop : index invariance ] and [ uniqa ] , we may suppose that @xmath2 and @xmath4 are as in figure [ no cut index figure ] , of course drawn with the correct number of meeting points between @xmath3 and @xmath37 .
recalling that @xmath106 counts the winding number of the vector @xmath204 around the origin , we consider when it is possible for @xmath204 to be a positive real number .
if @xmath12 does not lie on the intersection of the left side of the rectangle @xmath3 with the interior of @xmath4 , then @xmath204 certainly has either a negative real component , or a non - zero imaginary component .
thus pick a @xmath12 lying in this intersection . then
only way for @xmath204 to be real and positive is for @xmath205 to lie on the semicircular sub - arc of @xmath37 to the right of @xmath12 , as in the figure .
but then , considering the orientations on @xmath3 and @xmath37 , we get that the vector @xmath204 must always be turning counterclockwise at such a @xmath12 , as we traverse @xmath3 in the positive direction .
we conclude that whenever the curve @xmath206 crosses the positive real axis , it does so in the positive direction , thus the winding number of this curve around the origin is non - negative .
( 0,-3.49)(7.3028126,3.497 ) ( 6.6609373,2.91)(3.6609375,-3.49 ) ( 3.6609375,2.11)0.6 - 90.090.0 ( 3.6609375,-0.29)0.6 - 90.090.0 ( 3.6609375,-2.69)0.6 - 90.090.0 ( 3.6609375,0.91)0.690.0270.0 ( 3.6609375,-1.49)0.690.0270.0 ( 3.6609375,-0.29)3.090.0270.0 ( 3.6609375,2.11)(4.2609377,2.11 ) ( 3.6609375,2.11 ) ( 4.2609377,2.11 ) ( 4.782344,2.135)@xmath205 ( 3.3223438,2.115)@xmath12 ( 6.952344,1.235)@xmath2 ( 0.67234373,1.535)@xmath4 ( 3.9209375,2.19)(4.0209374,3.11)(4.0209374,3.49)(3.4009376,3.37)(2.7809374,3.25)(2.4409375,3.11)(1.9409375,3.09 ) ( 1.1623437,3.075)@xmath204 the same figure [ no cut index figure ] can be used to show that under the hypotheses of lemma [ no cut index ] , we get that @xmath207 .
thus the only fixed - point indices which can be achieved in this setting are @xmath208 , and all three of these occur .
we do not use these facts , so working out the details is left as an exercise for the interested reader .
[ cp proofs ] as an example of the power of fixed - point index , in this section we prove three rigidity and uniformization theorems on circle packings .
theorem [ rigid in sphere ] is usually credited to koebe @xcite , andreev @xcite , and thurston .
all of their proofs were via methods different from ours .
theorems [ rigid in plane ] and [ thm : no luv ] are originally due to schramm @xcite*rigidity theorems 1.1 , 5.1 , who proved them essentially via the incompatibility theorem [ incompat ] , although his proof of the incompatibility theorem is not via fixed - point index techniques .
the first ones to study circle packings by fixed - point index techniques were he and schramm in @xcite , although they did not proceed via the incompatibility theorem , instead applying other normalizations to the packings in question , to similarly argue by contradiction . for further references on circle packing ,
see for example the articles and their bibliographies .
we begin with some basic definitions .
a _ circle packing _ is a collection @xmath209 of pairwise interiorwise disjoint round closed disks in the riemann sphere @xmath210 .
the _ contact graph _ of a circle packing is the graph having a vertex for every disk of the packing , so that two vertices share an edge if and only if the corresponding disks meet . then the following hold : [ rigid in sphere ]
suppose that @xmath211 and @xmath212 are circle packings in @xmath210 , sharing a contact graph that triangulates the 2-sphere @xmath213 .
then @xmath211 and @xmath212 differ by a mbius or anti - mbius transformation .
[ rigid in plane ] suppose that @xmath211 and @xmath212 are circle packings which are locally finite in @xmath1 , sharing a contact graph that triangulates a topological open disk .
then @xmath211 and @xmath212 differ by a euclidean similarity .
[ thm : no luv ] there can not be two circle packings @xmath211 and @xmath214 sharing a contact graph triangulating a topological open disk , so that one is locally finite in @xmath1 and the other is locally finite in the hyperbolic plane @xmath215 , equivalently the open unit disk @xmath61 .
the rest of this section consists of the proofs of these three theorems . before moving on ,
we make one note : the statement for locally finite packings in the hyperbolic plane @xmath216 analogous to theorem [ rigid in plane ] also holds
. however , to apply our techniques to prove it , one would need to show that two combinatorially equivalent packings , both locally finite in @xmath215 , having contact graphs triangulating @xmath215 , induce a homeomorphism , or at least some appropriately behaved identification , on the boundary @xmath217 .
this turns out to be true ( in fact , it follows from the rigidity theorem we discuss in this remark ) , but non - trivial .
the argument is by contradiction .
the main idea is to superimpose the packings @xmath211 and @xmath212 on the riemann sphere in a convenient way . in particular
, we will isolate two topological quadrilaterals @xmath218 and @xmath219 so that they are packed in combinatorially equivalent ways by disks of @xmath211 and @xmath214 , but so that @xmath218 and @xmath219 cut each other , as in figure [ clown2 ] .
then the simple observation that two round disks can not cut each other will give us our desired contradiction , via the incompatibility theorem [ incompat ] .
let @xmath220 be the common triangulation of @xmath213 which @xmath211 and @xmath212 realize .
suppose for contradiction that @xmath211 and @xmath214 are not equivalent under any mbius or anti - mbius transformation . for the first part of the proof
, we apply a sequence of normalizations to @xmath211 and to @xmath212 .
let @xmath221 be a face of @xmath222 .
we first normalize via a mbius transformation so that @xmath223 for @xmath21 . here
@xmath224 is the disk corresponding to the vertex @xmath225 of @xmath222 , similarly @xmath226 . in particular , the correct mbius transformation is the one that sends the intersection points of the @xmath227 to the corresponding ones of the @xmath228 .
our next normalization is in our initial choice of @xmath229 and our labeling of the @xmath230 , as per the following observation : [ obs : old prf ] let @xmath231 denote the vertex of @xmath222 other than @xmath232 so that @xmath233 is a face of @xmath222 .
then there is some choice of @xmath234 so that the disks @xmath235 and @xmath236 are not equal after our normalization identifying @xmath223 for @xmath21 .
if there were no such choice of @xmath229 then in fact every pair of corresponding disks @xmath237 and @xmath238 would coincide after our first normalization , and so @xmath211 and @xmath214 coincide .
an _ interstice _ of the packing @xmath211 is a connected component of @xmath239 .
every interstice is necessarily a curvilinear triangle , because the packing s contact graph triangulates @xmath210 . for our next normalization , we insist that @xmath240 lies in the interstice formed by @xmath241 which contains no other disks of either packing . finally , we insist that @xmath242 all have euclidean radius 1 , and that @xmath243 and @xmath244 are tangent at a point lying on the horizontal axis , so that @xmath245 lies to their left .
the situation for @xmath211 is depicted in figure [ fig : calp ] . from now on we work in the plane @xmath1 , in the sense that @xmath246 will not move again for the remainder of the proof .
note that every face @xmath119 of @xmath166 corresponds to some interstices @xmath247 and @xmath248 of @xmath211 and @xmath214 respectively , except for @xmath229 , for which the interstices @xmath249 contain @xmath240 .
let @xmath218 be the topological quadrilateral shown in figure [ fig : qq ] .
more precisely , let @xmath250 , and let @xmath251 .
then we define @xmath252 .
define the analogous objects for @xmath212 in the obvious way .
we now apply one final transformation to @xmath211 . first
, suppose without loss of generality that the euclidean radius of @xmath235 is larger than that of @xmath236 .
then translate every disk of @xmath211 to the right by a small amount @xmath253 , leaving the disks of @xmath212 unchanged .
denote this transformation by @xmath254 .
we will discuss more precise requirements on @xmath255 later .
the situation is depicted in figure [ fig : qqmain ] .
the essential point is that there is an open interval of values that @xmath253 may take so that after all of our transformations , the topological quadrilaterals @xmath218 and @xmath219 are arranged qualitatively as in figure [ fig : qq3 ] . in particular , we may choose @xmath255 so that the packings of @xmath218 and @xmath219 by the remaining disks of @xmath211 and @xmath212 are in transverse position , because there are only finitely many values of @xmath253 for which this fails .
then our desired contradiction follows immediately from the incompatibility theorem [ incompat ] , because round closed disks can not cut one another .
the proof of theorem [ rigid in plane ] proceeds along the same lines , except that after our first round of normalizations identifying @xmath237 and @xmath238 for @xmath21 , and sending a point of their common interstice to @xmath240 , the remaining disks of @xmath211 accumulate around a point @xmath256 , as do those of @xmath214 around a point @xmath257 .
the points @xmath258 and @xmath259 may coincide or may be different .
we define and apply @xmath254 as before , this time making sure that @xmath258 and @xmath259 differ after applying @xmath254 .
next , pick small disjoint neighborhoods @xmath260 and @xmath261 of @xmath258 and @xmath259 respectively , and contained in @xmath218 and @xmath219 respectively . then , let @xmath262 be the set of vertices @xmath263 so that both @xmath264 and @xmath265 . remove vertices from @xmath262 until the sub - triangulation of @xmath222 having vertices @xmath266 is a triangulation of a topological closed disk . let @xmath267 be the set of faces of @xmath222 corresponding to interstices formed by disks whose vertices are in @xmath262 .
let @xmath39 ( which stands for leftovers ) be the union @xmath268 , and define @xmath41 similarly .
then @xmath269 together with @xmath39 form a packing of the topological quadrilateral @xmath218 by closed jordan domains , as do @xmath270 in @xmath219 . furthermore ,
because @xmath39 and @xmath41 are disjoint by construction , these two domains do not cut each other .
then we get our desired contradiction by the incompatibility theorem [ incompat ] as before .
the adaptation here is along similar lines as the adaptation to prove theorem [ rigid in plane ] .
suppose for contradiction that @xmath211 is locally finite in @xmath1 , and @xmath212 is locally finite in @xmath216 .
this time , after our normalizations , the disks of @xmath211 accumulate around a single point @xmath258 , and the disks of @xmath212 accumulate around some round circle @xmath271 contained in the bounded region in the plane formed between @xmath272 .
this time , ensure that we chose @xmath255 so that @xmath258 does not lie on the circle @xmath271 .
we define @xmath39 and @xmath41 by throwing away disks of our circle packings , as before , but this time , either @xmath39 and @xmath41 are disjoint , or one contains the other . in either case
, the two do not cut each other , and the conclusion of the proof proceeds as before .
before defining torus parametrization , it will be helpful to have access to the following simple lemma : [ lem : gen pos lemma ] suppose @xmath2 and @xmath4 are closed jordan domains in transverse position .
suppose that @xmath273 .
orient @xmath3 and @xmath37 positively as usual .
then one of the following two mutually exclusive possibilities holds at the point @xmath12 . 1 .
[ case : lem : orientations:1 ] the curve @xmath37 is entering @xmath2 , and the curve @xmath3 is exiting @xmath4 .
[ case : lem : orientations:2 ] the curve @xmath3 is entering @xmath4 , and the curve @xmath37 is exiting @xmath2 .
thus as we traverse @xmath3 , we alternate arriving at points of @xmath187 where ( [ case : lem : orientations:1 ] ) occurs and those where ( [ case : lem : orientations:2 ] ) occurs , and the same holds as we traverse @xmath37 .
let @xmath273 .
we may assume , by applying a homeomorphism , that locally near @xmath12 the picture looks like figure [ fig : in out in out ] , with @xmath3 oriented down - to - up as shown .
then @xmath2 lies to the left of @xmath3 .
now , certainly @xmath37 is either entering or exiting @xmath2 at @xmath12 .
suppose @xmath37 is entering @xmath2 at @xmath12 .
then @xmath37 is oriented right - to - left , and so @xmath4 is below @xmath37 .
thus @xmath3 is exiting @xmath4 , and case ( 1 ) occurs .
similarly , if @xmath37 is exiting @xmath2 at @xmath12 then @xmath3 is entering @xmath4 at @xmath12 , so case ( 2 ) occurs .
( 0,-1.7067188)(5.8790627,1.7267188 ) ( 1.925,1.0632813)(2.1,0.5132812 ) ( 2.3945312,1.5232812)@xmath3 ( 4.304531,-0.37671876)@xmath37 ( 2.24,-0.24671875 ) ( 2.4845312,0.02328125)@xmath12 we are now ready to define _
torus parametrization_. throughout the definition , refer to figure [ tpex1 ] for an example .
let @xmath2 and @xmath4 be closed jordan domains in transverse position , so that @xmath3 and @xmath37 meet at @xmath274 points , with boundaries oriented as usual .
let @xmath275 , where @xmath276 and @xmath277 are labeled so that at every @xmath276 we have that @xmath3 is entering @xmath4 , and at every @xmath277 we have that @xmath37 is entering @xmath2 .
imbue @xmath0 with an orientation and let @xmath278 and @xmath279 be orientation - preserving homeomorphisms .
we refer to this as fixing a _ torus parametrization _ for @xmath2 and @xmath4 .
we consider a point @xmath280 on the 2-torus @xmath281 to be parametrizing simultaneously a point @xmath282 and a point @xmath283 .
we denote by @xmath284 be the unique point @xmath285 satisfying @xmath286 , similarly @xmath287 . note that by the transverse position hypothesis no pair of points in @xmath288 share a first coordinate , nor a second coordinate .
suppose we pick @xmath289 .
then we may draw an image of @xmath281 by letting @xmath290 be the vertical axis and letting @xmath291 be the horizontal axis .
then we call @xmath292 a _ base point _ for the drawing .
suppose that @xmath5 is an orientation - preserving homeomorphism .
then @xmath7 determines an oriented curve @xmath55 in @xmath293 for us , namely its graph @xmath294 , with orientation obtained by traversing @xmath3 positively .
note that @xmath7 is fixed - point - free if and only if its associated curve @xmath55 misses all of the @xmath295 and @xmath296 .
pick @xmath297 and denote @xmath298 .
then if we draw the torus parametrization for @xmath2 and @xmath4 using the base point @xmath299 , the curve @xmath55 associated to @xmath7 `` looks like the graph of a strictly increasing function . ''
the converse is also true : given any such @xmath55 , it determines for us an orientation - preserving homeomorphism @xmath9 sending @xmath51 to @xmath300 , which is fixed - point - free if and only if @xmath55 misses all of the @xmath295 and @xmath296 .
suppose that @xmath301 , equivalently that @xmath302 .
the curve @xmath55 and the horizontal and vertical axes @xmath303 and @xmath304 divide @xmath293 into two simply connected open sets @xmath305 and @xmath306 as shown in figure [ fig : proving amazing index ] .
we suppress the dependence on @xmath300 in the notation because @xmath298 . if neither @xmath307 nor @xmath308 then every @xmath295 and every @xmath296 lies in either @xmath306 or @xmath305 . in this case
we write @xmath309 to denote @xmath310 the number of points @xmath295 which lie in @xmath306 , and we define @xmath311 , @xmath312 , and @xmath313 in the analogous way .
denote by @xmath314 the winding number of the closed curve @xmath315 around the point @xmath316 .
the following central lemma says that given an indexable homeomorphism @xmath317 , we may read off its fixed - point index @xmath8 simply by examining the curve @xmath55 associated to @xmath7 in the way we just described : [ prop : computing index from torus ] let @xmath2 and @xmath4 be closed jordan domains . fix a torus parametrization of @xmath2 and @xmath4 via @xmath318 and @xmath319 .
let @xmath5 be an indexable homeomorphism , with graph @xmath55 in @xmath293 .
suppose that @xmath301 , where @xmath320 and @xmath321 .
then : @xmath322 the remainder of the section is spent proving lemma [ prop : computing index from torus ] .
suppose @xmath107 is any oriented closed curve in @xmath323 .
then the closed curve @xmath324 misses the origin , and has a natural orientation obtained by traversing @xmath107 positively .
we denote by @xmath325 the winding number around the origin of @xmath324 .
[ lem : homotopy classes ] if @xmath109 and @xmath326 are homotopic in @xmath323 then @xmath327 .
this is because the homotopy between @xmath109 and @xmath326 in @xmath328 induces a homotopy between the closed curves @xmath329 and @xmath330 in the punctured plane @xmath113 .
suppose that @xmath44 is a fixed - point - free orientation - preserving homeomorphism .
let @xmath55 be its graph in @xmath293 .
if @xmath55 has orientation induced by traversing @xmath3 and @xmath37 positively , then the following is a tautology .
[ ex : never gon na reference this lol ] @xmath331 ( 0,-3.2275)(9.279062,3.1875 ) ( 5.82,3.1875)(0.0,-2.6325 ) ( 0.0,-2.6325 ) ( 0.0,-3.0025)@xmath299 ( 4.0445313,2.1175)@xmath305 ( 4.7945313,1.2975)@xmath306 ( 1.3845313,-0.3825)@xmath55 ( 5.7,0.8475)(5.0,0.0675 ) ( 2.68,-0.1525)(2.86,-0.0625 ) ( 0.0,-2.6325)(5.82,-2.6325 ) ( 0.0,-2.6325)(1.82,-2.6325 ) ( 5.82,-2.6325)(5.82,3.1875 ) ( 5.82,-2.6325)(5.82,0.1875 ) ( 0.84,0.9475 ) ( 2.18,2.0075 ) ( 2.64,-1.8325 ) ( 4.84,-0.4925 ) ( 3.2245312,-3.0025)@xmath332 ( 7.044531,0.5775)@xmath333 ( 0.8045313,1.3175)@xmath334 ( 2.1645312,2.3975)@xmath335 ( 2.9245312,-1.8025)@xmath336 ( 4.5645313,-0.4825)@xmath337 ( 5.02,-0.0525)(4.26,-0.8125 ) ( 5.02,-0.0525)(4.46,-0.0525 ) ( 3.18,-1.3925)(2.38,-2.1925 ) ( 2.98,-2.1925)(2.38,-2.1925 ) ( 3.18,-1.3925)(4.26,-0.8125 ) ( 3.38,-1.3925)(4.16,-0.9625 ) ( 3.28,-1.2425)(4.06,-0.8125 ) ( 0.8,-2.3325)(2.08,-2.1125 ) ( 2.66,-0.5925)(2.98,-1.1925 ) ( 3.74,0.6675)(4.22,0.0475 ) ( 5.58,-1.1525)(5.16,-0.8525 ) ( 4.26,-2.4125)(3.72,-1.4125 ) ( 3.34,-2.5125)(3.24,-2.1925 ) ( 1.54,-1.3325)(2.3,-1.4325 ) ( 5.58,-2.4525)(4.38,-1.1325 ) ( 3.16,0.1275)(3.84,-0.8725 ) orient @xmath338 as shown in figure [ fig : proving amazing index ] .
then @xmath338 is the concatenation of the curve @xmath55 traversed backwards with @xmath332 and @xmath333 , where the two latter curves are oriented according to the positive orientation on @xmath0 . if @xmath332 and @xmath333 are oriented according to the positive orientation on @xmath0 , then @xmath339 and @xmath340 .
it is also easy to see that if we concatenate two closed curves @xmath109 and @xmath326 that meet at a point , we get @xmath341 .
thus in light of the orientations on @xmath338 and all other curves concerned we get : @xmath342 for every @xmath133 let @xmath343 and @xmath344 be small squares around @xmath295 and @xmath296 respectively in @xmath293 , oriented as shown in figure [ fig : proving amazing index ] . by _
we mean a simple closed curve which decomposes into four `` sides , '' so that on a given side one of the two coordinates of @xmath345 is constant .
pick the rectangles small enough so that the closed boxes they bound are pairwise disjoint and do not meet @xmath338 .
let @xmath346 be the closed curve in @xmath306 obtained in the following way .
first , start with every loop @xmath343 and @xmath344 for those @xmath295 and @xmath296 lying in @xmath306 .
let @xmath347 be an arc contained in the interior of @xmath306 which meets each @xmath343 and @xmath344 contained in @xmath306 at exactly one point .
it is easy to prove inductively that such an arc exists .
let @xmath348 be the closed curve obtained by traversing @xmath347 first in one direction , then in the other .
then let @xmath346 be obtained by concatenating @xmath348 with every @xmath343 and @xmath344 contained in @xmath306 .
the curves @xmath346 and @xmath338 are homotopic in @xmath349 .
also @xmath350 .
it follows that : @xmath351 see figure [ fig : proving amazing index ] for an example . on the other hand ,
the following holds .
( 0,-3.953125)(9.065,3.953125 ) ( 5.1,0.7596875)(5.31,0.9296875 ) ( 2.97,1.0796875)(3.17,0.6896875 ) ( 5.5845313,3.7496874)@xmath2 ( 5.7145314,-3.7503126)@xmath4 ( 3.77,0.0196875 ) ( 3.1945312,0.0496875)@xmath276 ( 2.55,1.7796875 ) ( 5.25,-1.4603125 ) ( 2.31,-1.1403126 ) ( 6.17,1.5996875 ) ( 1.8545312,1.6096874)@xmath352 ( 11.174531,-1.2303125)@xmath353 ( 5.9045315,2.0296875)@xmath354 ( 1.7045312,-0.7903125)@xmath355 ( 5.58,-1.2303125)(5.41,-1.3603125 ) [ lem : local windings for pi qi ] @xmath356 , @xmath357 to see why , suppose that @xmath358_{{\mathbb{s}}^1 } \times [ y_0\to y_1]_{{\mathbb{s}}^1})$ ] .
then up to orientation - preserving homeomorphism the picture near @xmath276 is as in figure [ fig : proving local winding pi ] .
we let @xmath280 traverse @xmath343 positively starting at @xmath292 , keeping track of the vector @xmath359 as we do so .
the vector @xmath360 points to the right .
as @xmath361 varies from @xmath362 to @xmath363 , the vector @xmath359 rotates in the positive direction , that is , counter - clockwise , until it arrives at @xmath364 , which points upward .
continuing in this fashion , we see that @xmath359 makes one full counter - clockwise rotation as we traverse @xmath343 .
the proof that @xmath357 is similar . combining all of our observations establishes equation [ eq 1:prop : computing index from torus eq ] .
the proof that equation [ eq 2:prop : computing index from torus eq ] holds is similar .
we restate the lemma for the convenience of the reader .
let @xmath2 and @xmath4 be compact jordan domains in transverse position .
let @xmath18 appear in counterclockwise order , similarly @xmath19 .
an indexable homeomorphism @xmath44 is called _ faithful _ if it sends @xmath365 for @xmath21 .
we wish to find a faithful @xmath7 with @xmath38 .
we refer to the @xmath185 and the @xmath366 as our _ constraint points_. we proceed by induction on the number of intersection points @xmath187 , noting that this number is always even .
the circle index lemma [ cil ] takes care of the cases where @xmath3 and @xmath37 meet @xmath367 or @xmath368 times .
thus suppose that @xmath3 and @xmath37 meet at least @xmath369 times .
( 0,-1.123711)(13.18291,1.1221094 ) ( 1.4998766,0.9437109)(1.9012852,0.45693192)(2.8851926,0.06627866)(3.7512546,-0.029847104)(4.6173167,-0.12597287)(5.0427427,-0.045549653)(5.7059393,0.33131152 ) ( 1.0810156,-0.9562975)(1.7442123,-0.061252203)(2.2557118,0.5768708)(2.9309852,0.6767676)(3.6062586,0.7766644)(3.9183872,0.21804778)(4.396999,0.31560898)(4.8756104,0.41317016)(5.3219833,-0.07695475)(5.7408447,-0.9248924 ) ( 6.1410155,0.12371094)(7.5410156,0.12371094 ) ( 7.905242,1.0837109)(8.2910185,0.7153742)(9.2366085,0.419774)(10.068943,0.3470374)(10.9012785,0.27430078)(11.310137,0.33515555)(11.947506,0.6203195 ) ( 7.5010157,-0.95667714)(8.138643,-0.36444452)(8.630423,0.05778803)(9.279662,0.12388763)(9.928902,0.18998723)(10.228996,-0.17963761)(10.689156,-0.115083426)(11.149315,-0.050529245)(11.578479,-0.37483457)(11.981192,-0.93589705 ) ( 1.0324707,0.92871094)@xmath2 ( 0.6924707,-0.95128906)@xmath4 ( 12.292471,0.6887109)@xmath82 ( 12.412471,-0.89128906)@xmath83 the main idea of the proof is to find sub - arcs of @xmath3 and @xmath37 which `` cross minimally , '' and `` pull them apart , '' see figure [ fig : pull me ] .
when `` pulling these sub - arcs apart , '' we leave the rest of @xmath3 and @xmath37 fixed , as in figure [ fig : pull me 2 ] .
we then apply the induction hypothesis to the resulting jordan domains @xmath82 and @xmath83 , obtaining some indexable homeomorphism @xmath370 with @xmath371 .
finally , we use @xmath372 to construct @xmath7 , arguing that the fixed - point index is preserved or increased in this last step .
the details of the proof will be worked out in a torus parametrization .
this is because the torus parametrization allows us to work systematically and somewhat combinatorially through many cases , and unfortunately much of the proof is case analysis . also unfortunately , it is often difficult to discuss the plane topological pictures under consideration using prose , so much of the proof involves chasing the definitions of various terms and mathematical expressions around .
( 0,-1.8308277)(9.78291,1.8308275 ) ( 4.301016,-0.10917247)(5.7010155,-0.10917247 ) ( 3.9524708,1.0158275)@xmath2 ( 0.6924707,1.0758275)@xmath4 ( 9.43247,0.37582752)@xmath82 ( 6.332471,1.4958276)@xmath83 ( 2.4610157,-0.089172475)(2.7010157,0.25082752)(2.189661,0.5789255)(2.4210157,0.83082753)(2.6523702,1.0827296)(3.2210157,0.83082753)(3.4210157,0.9908275)(3.6210155,1.1508275)(4.2010155,-0.049172472)(4.1210155,-0.30917248)(4.0410156,-0.5691725)(3.5410156,-1.0691725)(3.2210157,-1.3891724)(2.9010155,-1.7091725)(2.1010156,-1.7691724)(2.0610156,-1.3691725)(2.0210156,-0.9691725)(2.2210157,-0.4291725)(2.4610157,-0.089172475 ) ( 0.64101565,-1.3291725)(0.8010156,-1.5491725)(1.484355,-1.8108275)(1.7610157,-1.8091725)(2.0376763,-1.8075174)(2.985388,-1.2667643)(2.5210156,-0.92917246)(2.0566432,-0.5915806)(1.9610156,0.11082753)(2.5810156,0.43082753)(3.2010157,0.75082755)(2.8410156,1.7508276)(2.3610156,1.7708275)(1.8810157,1.7908275)(1.1210157,0.9308275)(1.0010157,0.7108275)(0.8810156,0.49082753)(0.48101562,-1.1091725)(0.64101565,-1.3291725 ) ( 7.6010156,-0.06917247)(7.841016,0.27082753)(7.789661,0.3189255)(8.021015,0.5308275)(8.25237,0.74272954)(8.361015,0.8508275)(8.561016,1.0108275)(8.761016,1.1708275)(9.341016,-0.029172473)(9.261016,-0.28917247)(9.181016,-0.54917246)(8.681016,-1.0491725)(8.361015,-1.3691725)(8.041016,-1.6891725)(7.2410154,-1.7491724)(7.2010155,-1.3491725)(7.1610155,-0.9491725)(7.361016,-0.40917248)(7.6010156,-0.06917247 ) ( 5.7810154,-1.3091725)(5.9410157,-1.5291724)(6.624355,-1.7908275)(6.9010158,-1.7891725)(7.177676,-1.7875174)(8.125388,-1.2467643)(7.6610155,-0.9091725)(7.1966434,-0.57158065)(7.1010156,0.13082753)(7.7210155,0.45082754)(8.341016,0.77082753)(7.9810157,1.7708275)(7.5010157,1.7908275)(7.0210156,1.8108275)(6.2610154,0.95082754)(6.1410155,0.7308275)(6.0210156,0.51082754)(5.6210155,-1.0891725)(5.7810154,-1.3091725 ) two intersection points @xmath373 are called _ adjacent in @xmath3 _ if one of the two arcs @xmath374_{\partial k}$ ] and @xmath375_{\partial k}$ ] does not contain any other intersection points @xmath376 .
note that two intersection points of @xmath187 may be adjacent only if @xmath3 enters @xmath4 at one and leaves @xmath4 at the other , so our notation is in keeping with our conventions from section [ chap : torus ] .
we define _ adjacency in @xmath37 _ similarly .
two intersection points @xmath377 are called _ doubly adjacent _ if they are adjacent both in @xmath3 and in @xmath37 .
then the notion of sub - arcs of @xmath3 and @xmath37 `` crossing minimally '' which we informally described before is exactly captured by the property of double adjacency . if @xmath187 meet , then there is always at least one doubly adjacent pair of points in @xmath187 . from now on
, we fix a torus parametrization of @xmath3 and @xmath37 in @xmath281 via @xmath378 and @xmath379 .
let @xmath380 be doubly adjacent .
then we denote @xmath381 and @xmath382 as usual , so @xmath383 parametrizes the point @xmath384 , similarly @xmath385 for @xmath386 .
we can now see how we plan to apply induction : [ obs : construct induction ] let the torus parametrization @xmath387 be obtained from @xmath293 by deleting @xmath388 and @xmath385 .
then @xmath387 parametrizes a pair @xmath90 and @xmath91 , where @xmath82 and @xmath83 are compact jordan domains in transverse position , meeting two times fewer than do @xmath2 and @xmath4 .
thus our strategy will be as follows : first , apply the induction hypothesis to @xmath387 , to get a faithful indexable @xmath389 so that @xmath390 ; then , `` reinsert '' the points @xmath388 and @xmath385 , in such a way that the fixed - point index is preserved or increased
. we will refer to the collection of the sets @xmath391 and @xmath392 for @xmath21 as our _ grid lines _ , and to their pairwise intersection points as our _ lattice points_. furthermore we refer to the points @xmath393 , for @xmath21 , as our _ constraint lattice points_. if we draw our torus parametrization with a lattice point chosen to be the base point , then the grid lines actually divide the drawing into a grid , with nine disjoint rectangular cells in total
. we will refer to these as our _ lattice cells_. within any lattice cell there are well - defined left , right , up , and down directions .
from now on , whenever we draw the torus parametrization @xmath293 , we will always pick one of the lattice points for the base point .
suppose for the rest of this paragraph that we have fixed such a drawing
. then we may refer to the lattice cells according to their position in this drawing in the natural way , for example we may refer to the bottom - left lattice cell .
we denote the lattice cells by @xmath394 in the natural way , and write @xmath395 to denote the central lattice cell .
then the union of the closed upper - left , middle - left , and lower - left lattice cells is called the _ left lattice column_. we define the _ central _ and _ right lattice columns _ analogously , as well as the _ bottom _ , _ central _ , and _ top lattice rows_. we again emphasize that these descriptors depend crucially on the choice of base point for the drawing of the torus parametrization .
as usual , we denote @xmath396 and @xmath397 , where @xmath398 , with the same notational conventions as in section [ chap : torus ]
. we will apply lemma [ prop : computing index from torus ] frequently . before moving on to the next part of the proof
, we make an observation that will simplify things later : ( 0,-1.868)(3.728,1.868 ) ( 3.46,-0.8)(0.26,-1.6 ) ( 3.66,1.8)(0.06,-1.8 ) ( 0.06,0.6)(3.66,0.6 ) ( 0.06,-0.6)(3.66,-0.6 ) ( 1.26,1.8)(1.26,-1.8 ) ( 2.46,1.8)(2.46,-1.8 ) ( 0.06,-1.8)(0.2,-1.18)(0.13289322,-0.78710675)(0.22,-0.72)(0.3071068,-0.65289325)(0.99612045,-0.76268345)(1.26,-0.6)(1.5238795,-0.43731657)(2.14,0.26)(2.46,0.6)(2.78,0.94)(3.3,1.42)(3.66,1.8 ) ( 2.861455,1.345)@xmath55 ( 1.8614551,-1.195)@xmath399 ( 0.06,-1.8 ) ( 1.26,-0.6 ) ( 2.46,0.6 ) ( 3.66,1.8 ) to see why , we walk through figure [ fig : first steps ] .
we suppose that all of the @xmath276 and @xmath277 lie in a single lattice row , call it @xmath120 .
the same argument will work if they all lie in a single lattice column .
without loss of generality , we have drawn the torus parametrization so that @xmath120 is the bottom lattice row , and so that the base point is a constraint lattice point .
we may do so by picking the correct constraint lattice point to serve as our base point .
then the other two constraint lattice points are the ones along the diagonal , drawn in the figure as dots . as usual
, a faithful indexable homeomorphism @xmath5 can be parametrized by a `` strictly increasing '' curve * from the bottom - left corner of the bottom - left cell , * to the top - right corner of the top - right cell , * passing through the two other constraint lattice points , and * missing all of the @xmath276 and @xmath277 .
thus the @xmath55 we have drawn in figure [ fig : first steps ] parametrizes such a homeomorphism .
furthermore , because there are finitely many @xmath276 and @xmath277 , and all of them lie in the bottom lattice row @xmath120 , in fact they lie in an open rectangle @xmath399 as depicted in figure [ fig : first steps ] .
thus @xmath22 by lemma [ prop : computing index from torus ] .
let @xmath400 and @xmath386 be adjacent in @xmath3 , and let @xmath401 denote the arc connecting @xmath400 and @xmath386 along @xmath3 which contains no other intersection points @xmath187 .
thus @xmath401 is equal to one of @xmath374_{\partial k}$ ] and @xmath375_{\partial k}$ ] .
there is a unique such @xmath401 so long as @xmath3 and @xmath37 meet at least four times , as is our running assumption .
we call this @xmath401 the _ short adjacency arc of @xmath380 in @xmath3 _ , and define the _ short adjacency arc @xmath402 of @xmath380 in @xmath37 _ similarly . continuing with the notation of the last paragraph , let @xmath403 be obtained from @xmath401 by extending it slightly on both ends , so that @xmath404 is a closed jordan arc containing @xmath400 and @xmath386 , but no other points of @xmath376 , and no constraint point @xmath185 which was not already contained in @xmath401 .
there is a topologically unique such @xmath403 .
then we say that @xmath403 is the _ adjacency arc of @xmath380 in @xmath3 _ of @xmath380 .
we define the _ adjacency arc @xmath405 of @xmath380 in @xmath37 _ similarly . again continuing with the notation of before ,
let @xmath406 be the closed rectangle so that @xmath407 if and only if @xmath408 and @xmath409 . because @xmath403 and @xmath405 are essentially unique , we get also that there is a topologically unique such @xmath410
. then @xmath410 is called the _ adjacency box _ of @xmath380 .
pick doubly adjacent @xmath380 , having adjacency arcs @xmath411 and let @xmath412 be their adjacency box .
let @xmath413 and @xmath414 , and set @xmath415 .
then the parametrization in @xmath293 of every point in @xmath416 lies @xmath417 .
however , we have basically no _ a priori _ information about how these points are arranged in @xmath417 .
however : suppose that @xmath418 .
then @xmath419 is contained in a lattice column , regardless of which lattice point is chosen for the base point of our drawing of the torus parametrization .
thus lemma [ lem:3p ] holds by observation [ obs : wrap ] . to see why ,
let @xmath420 be so that @xmath421 .
then traverse @xmath3 starting from @xmath185 , positively , until arriving at the initial endpoint of @xmath401 .
let @xmath197 be the last one of @xmath422 crossed during this traversal .
then we may draw the torus parametrization using @xmath423 as the base point .
after relabeling if necessary , we suppose from now on that our torus parametrization is drawn with @xmath424 as our base point . in particular , this means that @xmath425 , so our adjacency box @xmath410 does not meet the left , nor the right grid line @xmath426 of our torus parametrization . because @xmath410 is a topological rectangle in the natural way , it has well - defined left , right , up , and down directions .
more precisely , we suppose that the positive orientation on the first coordinate of @xmath281 goes left - to - right , and on the second coordinate goes down - to - up .
then : we will denote the cases for the topological location of @xmath410 in the following way : let @xmath429 equal one of @xmath430 , and let @xmath431 equal one of @xmath432 .
then we say that case @xmath433 occurs if and only if @xmath427 lies in lattice cell @xmath434 , and @xmath428 lies in lattice cell @xmath435 .
_ a priori _ this gives us @xmath436 possibilities for @xmath410 .
fortunately we will see that many of the cases are handled in more or less the same way . for reference we have depicted the 27 cases in figure [ alcohol ] .
for the remainder of the proof , let @xmath387 be the torus parametrization obtained from @xmath293 by simply deleting @xmath388 and @xmath385 , and let @xmath389 be the curve in @xmath387 obtained inductively as described in observation [ obs : construct induction ] , so that in particular @xmath437 .
we wish to construct @xmath55 in @xmath293 from @xmath389 , so that @xmath438 . in comparing @xmath439 with @xmath440 ,
the germane issue is the effect of the presence of @xmath441 in @xmath293 compared to their absence from @xmath387 . in light of lemma
[ prop : computing index from torus ] , there are precisely two ways that reinserting these points can change the fixed - point indices under consideration : * first , they may affect the winding numbers @xmath442 and @xmath443 as compared to @xmath444 and @xmath445 . here
@xmath446 are the closed jordan domains parametrized by the torus parametrization @xmath387 .
* second , they may affect the sum @xmath447 , equivalently the sum @xmath448 , as compared to the respective sums in @xmath387 .
this is because @xmath449 is determined completely by whether we arrive first at a @xmath450 or a @xmath451 as we traverse the first coordinate @xmath0 of @xmath293 positively starting at @xmath452 , and this is invariant under reinserting @xmath441 , as the points @xmath400 and @xmath386 occur immediately after one another , in some order , as we traverse @xmath3 starting at @xmath182 . in this case , similarly to observation [ obs : no left ] , we get that @xmath456 . also , because @xmath453 does not meet @xmath410 , we have that @xmath388 and @xmath385 either both lie above @xmath55 in @xmath293 , or both lie below it .
thus @xmath457 , completing the proof of the claim by lemma [ prop : computing index from torus ] . in this case
, the bottom / top grid line @xmath458 cuts @xmath410 into two connected components @xmath459 and @xmath460 , one of which contains @xmath388 and the other of which contains @xmath385 .
the `` upper half '' of @xmath410 , which we have denoted @xmath460 , actually lies in the bottom lattice row of @xmath293 , and furthermore lies beneath @xmath55 because @xmath55 does not meet @xmath410 .
similarly @xmath459 lies in the upper lattice row of @xmath293 and above @xmath55 .
there are two sub - cases : in this case @xmath463 in @xmath293 is @xmath135 less than @xmath464 in @xmath387 . on the other hand @xmath465 and @xmath466 .
thus the total change between indices in the two settings cancels out and we get @xmath467 by lemma [ prop : computing index from torus ] .
we continue with the notation from before . _
a priori _ there are exactly four topologically distinct ways that @xmath388 and @xmath385 may be arranged in the adjacency box @xmath410 of @xmath380 , as depicted in figure [ fig : four brothers ] .
[ obs : move p ] we may move @xmath388 and @xmath385 as we please within @xmath410 , without changing the topological configuration of @xmath2 , @xmath4 , and the constraint points , so long as the following two requirements are satisfied : during the eventual induction step , it will often be desirable to move the points @xmath388 and @xmath385 . by observation [ obs : move p ] we may do so , within limits , without affecting any relevant aspects of the situation .
it will also be useful to note the following : [ obs : replace a ] we may replace @xmath410 by any topological rectangle contained in @xmath410 , having sides `` parallel to '' those of @xmath410 in the natural sense , so long as the interior of this new rectangle continues to contain @xmath388 and @xmath385 .
this replacement preserves the germane features of @xmath410 in that every argument we have made in this proof so far continues to work under this replacement .
[ claim : monster1 ] suppose that @xmath55 meets only a single connected component of the set @xmath472 , and furthermore that this component contains a corner of @xmath410 .
then we may reduce to claim [ obs : no meetingz ] , completing the proof of lemma [ lem:3p ] . first note that @xmath475 has only a single connected component , because @xmath55 is `` strictly increasing '' in @xmath293 , thus also in the single connected component of @xmath472 which @xmath55 meets .
this also implies that if the component of @xmath472 which meets @xmath55 contains a corner of @xmath410 , then it must contain either the upper - left corner @xmath476 of @xmath410 , or the lower - right corner @xmath477 of @xmath410 , or both of these .
suppose it contains the upper - left corner .
the argument will be the same if it contains the lower - right corner .
a potential drawing of the situation is depicted in figure [ fig : shells ] .
( 0,-1.8991992)(6.72291,1.9191992 ) ( 2.0810156,1.3208008)(3.7810156,1.3208008 ) ( 2.0810156,1.3208008)(2.0810156,-0.3791992 ) ( 3.8810155,1.3208008)(4.5810156,1.3208008 ) ( 4.5810156,1.3208008)(5.9810157,1.3208008 ) ( 5.9810157,1.3208008)(5.9810157,-0.3791992 ) ( 5.9810157,-0.3791992)(5.9810157,-1.0791992 ) ( 2.0810156,-0.3791992)(2.0810156,-0.97919923 ) ( 3.8810155,-1.8791993)(4.5810156,-1.8791993 ) ( 2.0810156,-1.0791992)(2.0810156,-1.8791993 ) ( 2.0810156,-1.8791993)(3.7810156,-1.8791993 ) ( 4.5810156,-1.8791993)(5.9810157,-1.8791993 ) ( 5.9810157,-1.8791993)(5.9810157,-1.0791992 ) ( 6.272471,0.9258008)@xmath410 ( 2.0810156,-0.07919922)(3.7810156,-0.07919922 ) ( 3.5810156,1.3208008)(3.5810156,-0.3791992 ) ( 3.5810156,-1.0791992)(3.5810156,-1.8791993 ) ( 4.5810156,-0.07919922)(5.9810157,-0.07919922 ) ( 3.5810156,-0.3791992)(3.5810156,-1.0791992 ) ( 3.8810155,-0.07919922)(4.5810156,-0.07919922 ) ( 2.0810156,0.42080078)(2.2810156,0.52080077)(2.2810156,0.6208008)(2.3810155,0.72080076)(2.4810157,0.8208008)(2.7461746,0.87030977)(2.8810155,0.9208008)(3.0158567,0.9712918)(3.0810156,1.1208007)(3.1810157,1.3208008 ) ( 1.2724707,0.42580077)@xmath478 ( 3.1724708,1.7258008)@xmath479 ( 3.2810156,0.22080079)(3.7810156,0.22080079 ) ( 3.2810156,0.22080079)(3.2810156,-0.3791992 ) ( 4.5810156,0.22080079)(5.9810157,0.22080079 ) ( 5.9810157,0.22080079)(5.9810157,-0.3791992 ) ( 5.9810157,-1.0791992)(5.9810157,-1.8791993 ) ( 5.9810157,-1.8791993)(4.5810156,-1.8791993 ) ( 3.7810156,-1.8791993)(3.2810156,-1.8791993 ) ( 3.2810156,-1.8791993)(3.2810156,-1.0791992 ) ( 3.2810156,-0.3791992)(3.2810156,-1.0791992 ) ( 3.8810155,0.22080079)(4.4810157,0.22080079 ) ( 5.1524706,0.42580077)@xmath480 ( 2.8224707,0.6258008)@xmath55 then it is clear that within the constraints described in observation [ obs : move p ] it is possible to move @xmath388 and @xmath385 if necessary so that both lie to the right of @xmath479 and below @xmath478 , and then it is clear how to choose @xmath480 our replacement rectangle for @xmath410 .
this completes the proof of claim [ claim : monster1 ] by observation [ obs : replace a ] .
let @xmath380 be doubly adjacent , with short adjacency arcs @xmath401 and @xmath402 .
the unoriented union @xmath481 is an unoriented jordan curve .
the orientations on @xmath401 and @xmath402 may or may not agree , and may or may not induce the positive orientation on @xmath481 .
let @xmath60 be the open jordan domain bounded by @xmath481 .
then @xmath60 is called the _ adjacency domain _ of @xmath380 .
we make the following observation : [ obs : nice direction ] there exist at least two distinct pairs of doubly adjacent points in @xmath376 , so that if @xmath380 is such a pair , then their short adjacency arc in @xmath3 is @xmath482_{\partial k}$ ] .
[ obs : nice sides ] there exist at least two distinct pairs of doubly adjacent points in @xmath187 , so that if @xmath483 parametrize one of these pairs , then @xmath388 lies to the left of @xmath385 in their adjacency box @xmath410 .
[ [ completing - the - proof - if - a - constraint - lattice - point - lies - in - the - adjacency - box ] ] completing the proof if a constraint lattice point lies in the adjacency box ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [ claim : monster2 ] suppose that @xmath410 contains a constraint lattice point
. then we may move @xmath388 and @xmath385 around in such a way that we get @xmath484 , completing the proof of lemma [ lem:3p ] .
it is not hard to see that then @xmath388 and @xmath385 are on different sides of @xmath55 in our drawing of @xmath293 , and furthermore that @xmath388 lies above @xmath55 and @xmath385 below it .
then , arguing straightforwardly as in the proof of claim [ obs : no meetingz ] , case [ mermaid1 ] , we get @xmath485 .
[ obs : why am i doing this ] suppose that @xmath388 lies in a connected component of @xmath486 through which @xmath55 passes
. then we may decide whether @xmath388 lies above or below @xmath55 in our drawing of @xmath293 , in the sense that we may move @xmath388 within the constraints given in observation [ obs : move p ] to arrange either situation . the same is true for @xmath385 .
we argue similarly to the proof of claim [ obs : no meetingz ] , case [ mermaid2 ] . in this case
it can always be arranged that @xmath388 lies below @xmath55 and that @xmath385 lies above it .
there are three cases to check , and in each it is clear how to proceed .
the cases are shown in figure [ fig : keyboardme0 ] .
again we get @xmath485 . as per our usual arguments
, we have @xmath456 and @xmath487 .
thus we will be done if we can argue that we may arrange so that @xmath388 and @xmath385 lie on the same side of @xmath55 in our drawing of @xmath293 .
there are five cases to check , and in each it is clear how to proceed .
figure [ fig : keyboardme1 ] shows each case . in this case
we have that @xmath487 as per observation [ obs : no left ] , but unfortunately @xmath488 .
thus we will be done if we can argue that we may arrange so that @xmath388 lies above @xmath55 , and @xmath385 below it , in our drawing of @xmath293 .
there are only three cases to check .
figure [ fig : keyboardme2 ] shows each .
this completes the proof of claim [ claim : monster2 ] .
we summarize the situation so far .
there are @xmath436 possibilities for the location of @xmath410 in @xmath293 .
most of the cases fall into one of the following three categories . for a given type @xmath489 of @xmath410 , it is quick and easy to check the conditions that define membership in the categories .
we also refer the reader back to figure [ alcohol ] , where each of the 27 cases is drawn .
case @xmath494 is easily handled in a manner similar to the proof of claim [ claim : monster1 ] .
if case @xmath495 occurs , then our running assumption that @xmath388 lies to the left of @xmath385 in @xmath410 implies that @xmath485 by lemma [ prop : computing index from torus ] as usual .
we have no way to deal with case @xmath496 directly .
however , recall from observation [ obs : nice sides ] that we had two choices for @xmath441 for which every step of the proof until this point go through .
because the adjacency arc @xmath402 contains two constraint points @xmath366 if case @xmath497 occurs , it can occur for only one of these two choices of @xmath441 , so we do not need to handle this case directly .
this completes the proof of lemma [ lem:3p ] . | the fixed - point index of a homeomorphism of jordan curves measures the number of fixed - points , with multiplicity , of the extension of the homeomorphism to the full jordan domains in question .
the now - classical circle index lemma says that the fixed - point index of a positive - orientation - preserving homeomorphism of round circles is always non - negative .
we begin by proving a generalization of this lemma , to accommodate jordan curves bounding domains which do not disconnect each other .
we then apply this generalization to give a new proof of schramm s incompatibility theorem , which was used by schramm to give the first proof of the rigidity of circle packings filling the complex and hyperbolic planes .
as an example application , we include outlines of proofs of these circle packing theorems .
we then introduce a new tool , the so - called torus parametrization , for working with fixed - point index , which allows some problems concerning this quantity to be approached combinatorially .
we apply torus parametrization to give the first purely topological proof of the following lemma : given two positively oriented jordan curves , one may essentially prescribe the images of three points of one of the curves in the other , and obtain an orientation - preserving homeomorphism between the curves , having non - negative fixed - point index , which respects this prescription .
this lemma is essential to our proof of the incompatibility theorem . |
the nature of the galactic bulge morphology has been discussed using different interpretations over the last decades .
its non - axisymmetry was quickly recognized with the arrival of the first near - infrared surveys ( e.g. , weiland et al .
1994 ; lpez - corredoira et al .
1997 ; babusiaux & gilmore 2005 ) , but its shape has not been free from debate .
a peanut shape was evident in the projection of the bulge , especially in infrared surveys covering the full sky , like cobe - dirbe , and extinction was considered to be responsible for the presence of the pronounced `` peanut shape '' waist located near @xmath0 , @xmath1 in the uncorrected dirbe maps of the bulge at 1.25 and 2.2@xmath2 m ( weiland et al .
1994 ) . from another perspective
, astronomers have seen in this peanut shape some imprint of a boxy bulge ( e.g. , kent et al .
1991 ; dwek et al .
1995 ; lpez - corredoira et al .
2005 ) as observed in other galaxies ( e.g. , shaw 1987 ) or predicted by some theories that argue that the boxiness of the bulge was a composite effect expected to appear when one considers stable orbits belonging to several families of periodic orbits ( e.g. , combes & sanders 1981 ; stiavelli et al .
1991 ; merrifield 1996 ; patsis et al .
2003 ) . using similar images of the milky way in the infrared , some observers ( nataf et al . 2015
; ness & lang 2016 ) now see an x - shaped bulge behind the projected peanut shape , whereas other authors have seen elliptical or boxy bulges .
these , however , are not direct interpretations of the image but depend on the image processing with the subtraction of some particular disk model , or these may be artifacts from subtracting the bulge as an ellipsoid instead of as a boxy bulge ( lee & jang 2016 ) .
this follows more recent theories of bulge formation ( e.g. , patsis et al .
2002 , quillen et al .
2014 , li & shen 2015 ) .
one may wonder why astronomers have not seen an x - shaped bulge in the 1990s in the cobe - dirbe images .
the structure along the line of sight has also been claimed to show an x - shapeusing metal rich red clumps as standard candles ( nataf et al .
2010 , 2015 ; mcwilliam & zoccali 2010 ; saito et al . 2011
; wegg & gerhard 2013 ) , but some doubts have been cast on whether the second peak along the line of sight is a real density structure or an artifact in the luminosity function of red clumps ( rattenbury et al . 2007 ; lee et al . 2015 ; joo et al .
2016 ; lee & jang 2016 ; lpez - corredoira 2016 ) .
there are also signs of a non - x - shape bulge in other populations : very old and metal poor stars like rr - lyrae in the bulge are not x - shaped according to some authors ( dkny et al . 2013 ; pietrukowicz et al .
2015 ) , and young ( @xmath3 gyr ) populations like f0-f5v stars are not x - shaped according to different interpretations ( lpez - corredoira 2016 ) , interpretations that might also have doubt cast upon .
for instance , the argument for non - x - shape by lpez - corredoira ( 2016 ) might also raise doubts based on the possible dispersion of absolute magnitudes , which might be much larger than assumed and thus may convolve the two peaks along the line of sight into one .
if we assume that all of these results are correct , we would have the surprising result of a galaxy with three bulges , which raises suspicions something is wrong with some of the previous analyses . here
we want to throw further light on the question by analyzing a new population , mira variables , which includes ages of both young and old stars .
catchpole et al . ( 2016 ) report periods and jhkl observations for 643 oxygen - rich mira variables found in two outer bulge fields centered at @xmath4 and @xmath5 , each approximately 25 square degrees .
they also use the 6528 mira stars from the ogle - iii catalog of variable stars ( soszyski et al .
2013 ) in regions with lower galactic latitudes , providing periods and photometry in v and i , which are dominated by oxygen - rich sources ( matsunaga et al .
here we use both sets of data and we cross - correlate ogle data with 2mass ( skrutskie et al .
2006 ) within 1 arcsec to also obtain the @xmath6-magnitude ( in a single epoch ) , finding 6363 common stars . neither catchpole et al . ( 2016 ) nor soszyski et al .
( 2013 ) have carried out a proper analysis to test the x - shape hypothesis , so these data can be further explored with that purpose .
12 of catchpole et al .
( 2016 ) , with the intermediate - age population , is misleading since the clumpiness of the distribution is mostly due to noise : there are not enough stars to produce a map with bins of 0.15x0.15 kpc@xmath7 ( 630 stars in @xmath8 kpc@xmath7 gives an average of 3.5 star / bin , which is insufficient for performing a statistical analysis ) , and there are some incomplete counts because the entire selected region -6@xmath9 , @xmath10 was not totally covered . in any case ,
catchpole et al . (
2016 ) have not tested the x - shaped bulge hypothesis in this intermediate - age population . instead , they just plotted fig .
12 in this range of absolute magnitudes with the projected counts and put their results together with those of mcwilliam & zoccali ( 2010 ) , which correspond to @xmath11 , and observed that some stars are where they should be .
the rough comparison between fig .
12 and mcwilliam & zoccali ( 2010 ) is invalid because they correspond to different latitudes . using the period luminosity relationship ( whitelock et al .
2008 ; catchpole et al .
2016 , equation ( 2 ) ) @xmath12@xmath13 - 2.38 , \ ] ] and the @xmath6-band extinction @xmath14 ( fitzpatrick 1999 ) .
the extinction @xmath15 is calculated by catchpole et al .
( 2016 ) for their data and by us using the schlegel et al .
( 1998 ) maps for ogle data ; we can use this extinction to obtain the distance of each mira star .
for the ogle data , the @xmath6-magnitude is derived for a single epoch from 2mass ; it is not the average @xmath6-magnitude .
this can generate a large dispersion in distances , due to the variability of the sources ( matsunaga et al .
we wonder why catchpole et al .
( 2016 ) did not make any attempt to correct this dispersion of magnitudes , which introduces noise in the density maps . here
, in order to avoid this error in distance ( although with average zero , which should be compensated for when large numbers of stars are used ) due to the difference between the average magnitude of a mira stars and the magnitude in a single epoch , we calibrate the average relationship between the color @xmath16(ogle)-@xmath6(2mass ) ( corrected for extinction ) and the period in the ogle sample of 1235 stars with @xmath17 ( to avoid high - extinction regions ) , and we use this relationship to calculate the `` average '' @xmath6-band magnitude in the ogle stars used here .
this relationship is @xmath18 the dispersion of 2mass single - epoch @xmath6-magnitudes with respect to this law gives an average rms=0.85 mag ( see fig .
[ fig : imk ] ) , which is on the order of magnitude of the expected value due to variability ( matsunaga et al . 2005 ) , with higher dispersion toward higher periods .
given that we have used 1235 stars , the error of @xmath19 given in the above expression is 0.024 mag .
if we choose stars with @xmath20<2.6 $ ] , the rms is reduced to 0.62 mag .
we will use this constraint in ogle stars to avoid the highest deviations from the color period relationship . then , for the ogle data , rather than using the single - epoch 2mass magnitude , we use the @xmath16-magnitude for each of the stars and this color to calculate the average @xmath6-magnitude , i.e. @xmath21 , where @xmath22 is calculated from the period in eq .
( [ imk ] ) .
( ogle)-@xmath6(2mass ) ( corrected for extinction ) in the sample of ogle - iii data with @xmath17 , and the best fit of a cubic polynomial given by equation ( [ imk]).,width=302 ] in fig .
[ fig : mirasdens ] , we see the distribution of cathpole et al .
( 2016 ) along different lines of sight and for some interesting regions of ogle data . for our analysis , since x - shape features are observable mainly at @xmath23 ( nataf et al .
2010 , wegg & gerhard 2013 ) , we study the region with the lowest negative longitude and the highest abs(latitude ) : @xmath24 , @xmath25 . the density and star counts are related through @xmath26=\frac{5}{\ln 10}\frac{1}{\omega \,r(m_k)^3}a(m_k)$ ] , where @xmath27/5}$ ] , @xmath28 is the extinction - corrected @xmath6-band magnitude and the differential star counts are @xmath29 .
there may be some incompleteness of sources due to the patchiness of extinction , especially in the regions with lower @xmath30 , but this does not affect the shape of the density along the line of sight , although its amplitude is affected .
in the same plot , we do a linear fit in the regions with 4.5 kpc@xmath31 kpc and 11 kpc@xmath32 kpc , which represents approximately the disk counts along those lines of sight , and the excess over this linear fit is the bulge density .
the analysis of the disk counts is not the matter here , but note that this may be composed of both thick and thin disks , and the inner flare of the disk ( lpez - corredoira et al .
2004 ) leads to a sizeable amount of disk stars even at @xmath33 kpc from the plane . for the bulge density
, we can observe that there is only one peak in all lines of sight .
a prediction of the bulge densities of two models , x - shaped and boxy , is given in fig .
[ fig : pred_miras ] .
they were derived from the analytical expressions given in lpez - corredoira ( 2016 , 4.1 ) for the boxy and x - shaped bulges observed by lpez - corredoira et al .
( 2005 ) and wegg & gerhard ( 2013 ) respectively : @xmath34 @xmath35@xmath36 , \]]@xmath37 , \]]@xmath38@xmath39 for the positive longitude @xmath40 there is not much distinction in the functional shape of the density between the both models .
however , for the negative longitude @xmath41 line of sight and for the three selected ogle regions , the two predictions are very different , with a double peak for the x - shaped bulge and a single peak for the boxy bulge . fig .
[ fig : wegg ] shows a plot that helps us to understand the origin of this double peak along @xmath41 , @xmath4 .
this line of sight crosses two maxima in a constant @xmath42 slice , first in the left - hand bump , at around @xmath43 kpc , followed by a constant or slight decrease of the density in the hollow created by the peanut shape ( the line remains in the region between the blue and violet colors ) , and when we take into account the increase of @xmath44 at higher values of @xmath45 , this becomes a decrease of density .
then there is a second big peak at around 9.5 kpc associated with the right - hand bump ; this second peak is indeed not as high in amplitude in fig .
[ fig : pred_miras ] because we must add the effect of the increase of @xmath44 with distance , which makes the amplitude of the second peak similar to the first peak .
one may wonder why fig .
6 of wegg & gerhard ( 2013 ) and fig . 3 of saito et al .
( 2011 ) do not explictly show the double peaks .
indeed , these images show double bumps , but the lowest contours do not reach the longitude @xmath46 .
it is a question of constrast , as it is not enough in these figures to see the lowest contours .
however , in fig .
3/top - right panel of mcwilliam & zoccali ( 2010 ) for ( @xmath47 , @xmath11 ) , a double peak is clearly observed ; see also their fig .
6 , which shows the effect of double peaks between @xmath48 and @xmath49 . for lower values of @xmath50
, it is even more evident that the x - shape should provide double peaks at negative longitudes . in any case , if there is no double peak for the red clumps , it would remove the question regarding the x - shape topic at least for this region , although not for the other regions . clearly , the boxy bulge resembles the results of fig .
[ fig : mirasdens ] , whereas the x - shaped bulge does not . for the ogle data , the covered areas are lower , and that is the reason it produces a noisier density . a @xmath51 analysis in the comparison of the models with the data in the range 5
kpc@xmath52 kpc ( once subtracted the disk ) gives 1 .
for the line of sight @xmath53 , @xmath4 : @xmath54 ( @xmath55 values of @xmath45 ) equal to 2.81 and 1.09 , respectively , for the x - shaped and boxy bulges , corresponding to probabilities of @xmath56 and 0.37 , i.e. the x - bulge is excluded within 99.36% c.l .
( @xmath57 ) whereas the boxy bulge matches the data ; and 2 . for the line of sight @xmath58 , @xmath59 : @xmath60 equal to 2.30 and 1.40 , respectively , for the x - shaped and boxy bulges , corresponding to probabilities of @xmath61 and 0.20 , i.e. the x - bulge is excluded within 97.6% c.l . ( @xmath62 ) and the boxy bulge fit is not good but it is only excluded within 80% c.l .
( 1.3@xmath63 ) , which is much likely than the x - shaped bulge .
the combination of the two lines of sight gives , respectively , for the x - shaped and boxy bulges a probability of @xmath64 and 0.23 , i.e. the x - bulge is excluded within 99.89% c.l .
( @xmath65 ) whereas the boxy bulge is only excluded within 77% c.l .
( @xmath66 ) , which can be considered as a likely fit .
the average relative error of distance of these mira stars using eq . ( [ mkp ] ) is 3.6% ( catchpole et al .
2016 ) , which is too low to produce a smoothing of the two peaks into one .
the error in eq .
( [ imk ] ) is also low , as mentioned , around 0.024 mag .
the error in the extinction is around 10% ( schlegel et al . 1998 ) , with the average extinction @xmath67 mag for ogle regions with @xmath17 .
this produces another error of @xmath68 mag and @xmath69 mag , which is again too low to produce any confusion of the two peaks .
the effect of metallicity dispersion is also negligible : first because the variations of the metallicity in the bulge along a line of sight are very low ( lpez - corredoira 2016 ) , and second because even large variations of metallicity would produce variations of the magnitude of less than 0.1 mag ( matsunaga 2012 ) , which would mean an error lower than @xmath70 mag in each bin with @xmath71 the number of stars in that bin .
this makes this standard candle much more precise in the distance determination than the f0-f5v stars used by lpez - corredoira ( 2016 ) , but the result is still the same : a single peak along the line of sight . the @xmath72 is related to the age of the star . roughly , the ages of 1.0 , 1.8 , 3.0 , 5.0 , 7.0 gyr correspond to @xmath73 ( feast 2009 ; catchpole et al .
2016 ) so an approximate relationship can be given by a second - order polynomial fit of these numbers : @xmath74 the average @xmath75 of the mira variables in fig .
[ fig : mirasdens ] is 2.46 at the ( @xmath53 , @xmath4 ) line of sight and 2.45 for the combination of the three ogle lines of sight , which represents , according to eq .
( [ age ] ) , average ages of 8.8 and 9.1 gyr respectively . in fig .
[ fig : mirasdensage ] , we represent the density along ( @xmath53 , @xmath4 ) for three bins with approximately the same number of stars and ranges of @xmath75 ranges of @xmath76 , 2.39 - 2.53 and @xmath77 , which correspond , using equation ( [ age ] ) , to ages of @xmath78 gyr , 6.4 - 11.6 gyr , and @xmath79 gyr .
the c.l.s at which x - shapes are excluded in a @xmath80 analysis are 2.1@xmath63 , 1.2@xmath63 , and 1.1@xmath63 , respectively , for these ages ; for the boxy bulge , it is excluded at @xmath81 , 0.8@xmath63 , and 0.5@xmath63 , respectively .
the only significant exclusion here is for the oldest miras to be x - shaped . for the ogle line of sight ,
the star counts are much lower , so these data are even more inconclusive .
we would need a higher number of mira stars to be able to distinguish the shapes of the bulge at different ages .
, @xmath4 for different ranges of ages.,width=321 ]
the bulge with an x - shaped morphology traced by the average population of mira variables is excluded at the 3.3@xmath63 level , whereas a boxy bulge density distribution matches the data .
the significance is reduced to 2.3@xmath63 if we exclude the region of @xmath53 , @xmath4 .
apparently , there is no agreement regarding the structure of red clumps for this line of sight of @xmath53 , @xmath4 : some authors obtained a double peak there ( e.g , mcwilliam & zoccali 2010 ) , others do not see it there ( e.g. , saito et al . 2011 ) , others see it but relatively noisy ( e.g. , nataf et al .
the fiducial model of the x - shaped bulge used here ( equation ( [ densx ] ) ) reproduces the wegg & gerhard ( 2013 ) maps derived from red clumps . in any case , in our analysis using mira stars , we have not seen double peaks along the line of sight , not in @xmath53 , @xmath4 , nor in @xmath58 , @xmath59 .
this population has an average age of 9 gyr . at present , using mira variable stars we can not distinguish the morphology of stars with different ages .
this casts some doubt on the validity of red clump analyses claiming the existence of an x - shaped structure among stars with similar ages to those analyzed here , although an analysis of the metallicity of these stars is still pending .
bulge dynamics , formation , and evolution do not distinguish among the different types of stars with the same origin ( same age and metallicity ) .
stars with the same origin should belong to the same structure .
rr lyrae are thought to be part of a different bulge compared to red clumps because rr lyrae are older . also , the f0 - 5v stars are younger than red clumps on average .
but now we have a population of stars with ages similar to the red clumps with high metallicity , it is not expected that all stellar populations are part of a boxy bulge except for the red clumps , which are furthermore suspected to be contaminated by the second peak , unless we demonstrate a different origin for these red clumps .
even if we can not be 100% sure , at the very least , doubts regarding the very existence of the x - shaped bulge should be established from the present work .
thanks are given to the anonymous referee for the critical comments that motivated new calculations that reinforced the robustness of the analysis to be carried out .
this work has been supported by the grant aya2015 - 66506-p from the spanish ministry of economy and competitiveness ( mineco ) . | the stellar density distribution of the bulge is analyzed through one of its tracers .
we use oxygen - rich miras variables from the catchpole et al .
( 2016 ) survey and ogle - iii survey as standard candles .
the average age of these stars is around 9 gyr .
the population traced by mira variables matches a boxy bulge prediction , not an x - shaped one , because only one peak is observed in the density along the analyzed lines of sight , whereas the prediction of an x - shape gives two clear peaks . |
cavity polaritons are half - light half - matter quasiparticles resulting from the strong coupling of the photon mode of a microcavity with an exciton resonance of the embedded semiconductor structure @xcite
. being mixed exciton - photon quasiparticles , the cavity polaritons have integer spins and can reveal bosonic properties @xcite responsible for a number of interesting effects both predicted and observed , namely , stimulated scattering @xcite , polariton lasing @xcite , bose - einstein condensation @xcite , superfluidity @xcite etc .
many important properties of the polaritonic systems are connected with their spin degree of freedom @xcite .
it was shown that an external magnetic field strongly affects non - linear dynamics of the polaritons and optical properties of polariton condensates .
recently , the magnetic field induced suppression of polariton superfluidity and the so - called `` spin meissner effect '' have been predicted @xcite . for understanding of the rich variety of non - linear phenomena induced by magnetic field in microcavities a thorough knowledge of the linear properties of magneto - polaritons ( i.e. polaritons subject to a magnetic field ) is needed . though the problem on the surface seems simple it becomes really complex if the polarization of light is included into consideration .
in particular , to the best of our knowledge , till now there was no accurate analysis of the dependence of eigenenergies of exciton - polariton modes and their polarization on the magnetic field .
the present paper is aimed at filling the apparent gap in the theory of cavity polaritons .
we expand the non - local dielectric response theory to describe oblique reflection and transmission of light through quantum wells ( qws ) in the presence of the magnetic field normal to the qw plane .
we present also the transfer - matrix calculation of the eigenenergies and polarization of exciton - polaritons accounting for the te - tm splitting of cavity modes .
finally , faraday rotation of the polarization plane of light passing through the microcavity is analyzed .
the polarization of cavity polaritons depends on many factors . in the absence of the external magnetic field the exciton - polariton states having a non - zero in - plane wave vector ( @xmath0
are linearly polarized due to the long range electron - hole exchange interaction @xcite and te - tm splitting of the photonic mode of the cavity @xcite .
application of a magnetic field normal to the qw plane ( faraday geometry ) results in the zeeman splitting of the exciton doublet and thus changes the polarization of the polariton eigenstates from linear to elliptical or circular . as the bare photonic modes are unaffected by the magnetic field , the degree of circular polarization of exciton - polariton states is strongly dependent on the relative weights of photonic and excitonic components within the given polariton state . in strong magnetic fields these weights are field - dependent due to the shrinking of the wavefunction of the relative motion of the electron and hole in the real space resulting in an increase of the exciton - photon coupling strength .
the interplay between these phenomena has not been addressed in its complexity to the best of our knowledge .
experimentally , the magnetic field effect on the spectrum of exciton - polaritons in microcavities has been studied in 1990s by several groups @xcite .
the magnetic field induced weak - strong coupling transition has been observed , and the enhancement of the rabi - splitting of the exciton - polariton modes with the magnetic field increase has been measured by means of _ cw _ and time - resolved optical spectroscopies .
the shape and polarization of the polariton dispersion curves have not been studied , however .
later on , mixing of the bright - polariton and dark exciton states due to the magnetic field applied in the cavity plane has been studied using the polarized photoluminescence @xcite and time - resolved kerr rotation @xcite techniques .
theoretically , the dispersion of exciton - polaritons in microcavities in the absence of the magnetic field has been described in 1990s @xcite , the magnetic field effect on the exciton - photon coupling strength has been calculated in @xcite , and the exciton - polariton longitudinal - transverse splitting has been studied in detail in @xcite . the present paper is organized as follows . in the second section
we calculate the reflectivity of a qw structure in the presence of the external magnetic field applied perpendicular to the well plane . in the section iii
we calculate the dispersion of magneto - exciton - polaritons in a microcavity and analyze their polarization . in section
iv kerr and faraday effects in microcavities are discussed .
for a heavy - hole exciton usually being the lowest energy exciton state in a qw , the allowed spin projection @xmath1 to the structure growth axis is either @xmath2 or @xmath3 depending on the mutual orientation of the electron and hole spins .
the states with @xmath4 are decoupled from the cavity modes while the coupling of the @xmath3 states with the right or left circularly polarized cavity photons gives rise to the polariton doublets . in the following
we will bound ourselves to consider only the bright states @xmath5 .
let us consider the reflection of light incident at oblique angle on a qw which is sandwiched between the semi - infinite barriers and subject to the magnetic field @xmath6 normal to its plane .
we shall take into account the zeeman splitting of the exciton resonance and neglect for a while the magnetic field effect on the orbital motion of electron and hole in the exciton .
then the resonance frequencies of the @xmath7 and @xmath8 excitons can be presented respectively by @xmath9 , where @xmath10 is the bohr magneton , @xmath11 is the exciton @xmath11-factor , @xmath12 is the exciton resonance frequency given by @xmath13 @xmath14 is the effective band gap of the qw calculated with allowance for size - quantization energy , @xmath15 is the exciton binding energy , @xmath16 is the exciton translational mass and @xmath17 is its in - plane wave vector . the electric field of the electromagnetic wave is presented in the standard form @xmath18 , where the light frequency @xmath19 lies in the vicinity of the qw exciton resonance .
the wave equation for the vector @xmath20 can be written as @xmath21 where @xmath22 , @xmath23 is the background dielectric constant of the material surrounding the qw , @xmath24 is the exciton polarization induced by the electromagnetic wave . in our further consideration
we denote the structure growth axis as @xmath25 , and suppose that the light propagates in the @xmath26 plane . the electric field of te - polarized light is parallel to the @xmath27-axis while the electric field of tm - polarized light lies in the @xmath26 plane .
let @xmath28 be the incidence angle , i.e. the angle between @xmath25 axis and light wavevector .
representing the solution in the form @xmath29 , taking into account that @xmath30 and using one - dimensional green s function of the wave equation @xmath31 , one can reduce eq .
( [ eq1 ] ) to the following integral equation @xmath32 this equation should be completed by the material relation linking the electric field and polarization .
the latter can be written assuming the nonlocal dielectric response of the qw in the exciton resonance frequency region @xcite . in order to do so
, we remark that the amplitudes of the right and left circularly polarized components , @xmath33 or @xmath34 , are related to those of linearly polarized components by @xmath35 note that we consider only the polarization induced by heavy - hole exciton and do not take into account the @xmath25-component of the excitonic polarization which may be induced due to a light - hole exciton @xcite , so that @xmath36 . in the basis of circular polarized components the material equation has the form @xmath37 here
@xmath38 is the exciton wave - function taken with equal electron and hole coordinates , @xmath39 @xmath40 is the homogeneous broadening of the exciton resonance caused by the acoustic phonon scattering and the qw imperfections , @xmath41 with @xmath42 and @xmath43 being the longitudinal - transverse splitting and bohr radius of bulk excitons , respectively .
equations represent the generalization of the expression for the dielectric polarization within the framework of the nonlocal model of the dielectric response for qws with a spin - degenerate exciton resonance .
the condition @xmath44 allows one to decouple the equation for the @xmath25-component of the electromagnetic field from those for the @xmath45- and @xmath27-components which read @xmath46 @xmath47 where @xmath48 multiplying both parts of eq . by @xmath49 and integrating over @xmath25 from @xmath50 to @xmath51 one can obtain a closed set of equations for @xmath52 @xmath53 @xmath54 here @xmath55 is a renormalization shift of the exciton resonance frequency due to the light - matter coupling effect at normal incidence in the absence of the external magnetic field , @xmath56 is the exciton radiative broadening in the same conditions , and @xmath57 in general , both the exciton resonance frequency and oscillator strength depend on the magnetic field due to the orbital motion of electron and hole .
the magnetic field presses an electron and a hole to each other thus increases its binding energy and oscillator strength @xcite .
these effects are neglected in the following calculation but they can be readily taken into account by renormalizing the values of @xmath58 and @xmath59 .
ev , @xmath60 mev , @xmath61 mev , zeeman splitting is 0.1 mev .
inset shows faraday rotation angle calculated as a function of energy for the same quantum well at normal light incidence . ]
the solution of linear equations ( [ eq9 ] ) can be presented in the form @xmath62 @xmath63 where @xmath64 and @xmath65 . in order to find the reflection and transmission coefficients of the qw we should find the asymptotic values of the electric field far beyond the qw . coming back to eq . ( [ eq6a ] ) and
putting @xmath66 one easily obtains @xmath67 the transmission and reflection coefficients are polarization - dependent and can be represented in the form of @xmath68 matrices @xmath69 and @xmath70 defined by @xmath71 using eqs .
( [ eq13])([eq24 ] ) one obtains after simple algebra the reflection matrix elements , @xmath72 the transmission and reflection coefficients are interrelated by @xmath73 where @xmath74 is the kronecker delta .
note that , due to the zeeman splitting of the exciton resonance into a circularly polarized doublet , the coupling of light with an exciton leads to the mixing of te and tm polarizations both in reflection and transmission .
this results in the resonant faraday and kerr rotation and dichroism @xcite .
figure [ fig : faraday ] shows the angle of rotation of the polarization plane of transmitted light with respect to the polarization plane of incident light ( which is assumed to be te - polarized ) versus the angle of incidence for different energies close of the exciton resonance for the magnetic field of 2 t ( the induced zeeman splitting is 0.1 mev ) .
we have considered a 10 nm gaas qw with @xmath75 mev .
the homogeneous broadening is 0.5 mev .
the effect includes both linear - circular dichroism and faraday rotation .
the angle of kerr rotation can be calculated by using the following equation , see , e.g. , @xcite , @xmath76 here @xmath77 .
the faraday rotation angle @xmath78 is obtained by the replacement @xmath79 . for small rotations
a simplified equation @xmath80 can be applied .
one can check that the effect is strongest at the resonance and decreases with the increasing angle of incidence .
let us now consider a qw embedded in the center of a planar microcavity in the external magnetic field parallel to the structure growth axis , fig .
[ fig : cavity ] .
we shall calculate the dispersion of the exciton - polaritons in such a system using the generalized transfer matrix formalism .
then we compare this exact approach with a simple model of coupled oscillators where exciton and photon states are represented by the effective classical oscillators .
the eigenstates of the cavity can be found from the condition that a non - zero field exists inside the cavity without any external field .
we choose the qw center as an origin @xmath81 and present the electric field inside the cavity ( but outside the qw ) in the form @xmath82 respectively , on the left and right - hand side from the qw .
the fields @xmath83 are interrelated by the transfer matrix through the qw as follows @xmath84 here @xmath85 is a @xmath86 matrix , its explicit form is given later , and @xmath87 are two - component columns with components @xmath88 etc . according to fig . 3 @xmath89 and @xmath90 can be expressed via @xmath91 and @xmath92 by @xmath93 where @xmath94 is the width of the active layer and @xmath95 is a @xmath96 diagonal matrix of reflection from the bragg mirror .
note that , at oblique incidence ( @xmath97 ) , its components @xmath98 and @xmath99 are different . equations ( [ inter+- ] ) can be interpreted as follows .
a qw exciton creates the outgoing electromagnetic waves @xmath100 and @xmath101 propagating to the left and to the right , respectively .
the waves reaching the bragg mirror acquire the phase @xmath102 , the field reflected from the mirror returns to the qw with the phase @xmath103 and the additional factors @xmath104 .
whence , we obtain equation ( [ inter+- ] ) for the incoming waves . from eqs .
( [ gendisp ] ) , ( [ inter+- ] ) we arrive at the matrix dispersion equation @xmath105 the considered structure is symmetric , therefore the solutions of eq .
can be classified as _ even _ [ @xmath106 in eq . ] and _ odd _ ( @xmath107 ) with respect to the mirror reflection in the qw plane .
the electric field in odd solutions is uncoupled with the even lowest - exciton state , thus these modes are purely photonic . on the contrary ,
the even - parity cavity mode is coupled with the qw exciton to form mixed exciton - polariton modes .
the matrix equation ( [ eigenmodes ] ) is equivalent to four scalar equations . among them , for the waves of certain parity , only two are linearly independent .
if we present the transfer matrix in the block form @xmath108 where @xmath109 are @xmath96 matrices , then , for the even modes , the dispersion equation relating the frequency @xmath19 with the in - plane wave vector @xmath17 can be reduced to @xmath110 using the definition of reflection and transmission coefficients , eq . , one can present the blocks @xmath109 in the form @xmath111 @xmath112 and @xmath113 where @xmath114 . in general , eq .
( [ dispersion ] ) has four solutions for each incidence angle which corresponds to four exciton - polariton dispersion branches .
the polarization of polariton eigenmodes is determined by the electric - field complex amplitudes @xmath115 and @xmath116 .
the degree of polariton circular polarization is given by the standard expression @xcite @xmath117 in the absence of an external magnetic field the cross - polarized reflection and transmission coefficients vanish , @xmath118 , and eq .
( [ dispersion ] ) reduces to a couple of independent equations for te- and tm - polarized polariton modes satisfying the dispersion equation @xcite : @xmath119 where @xmath120 ( tm ) or @xmath121 ( te ) . in this case
exciton - polariton modes have definite linear polarizations .
in order to demonstrate underlying physics of the light - matter coupling we also put forward a simplified model of four coupled oscillators representing the te- and tm- polarized optical modes of empty cavity and two excitonic modes split by the zeeman interaction .
the effective dispersion equation can be written as @xmath122 here @xmath123 is the rabi splitting determined by the system parameters , @xmath124 , @xmath125 are dispersions of the te and tm bare - cavity photons .
we neglected the longitudinal - transverse splitting of the exciton .
the difference between this approach and the rigorous formalism presented above consists in ( i ) neglecting of the coupling of four polariton modes with all other light modes in the cavity and ( ii ) disregarding of the polarization dependence of the light - matter coupling constant @xmath126 @xcite .
t ( @xmath127 mev ) calculated for the exact eq . .
red and black mark lower and upper polariton branches , correspondingly , and solid and dashed lines correspond to different polarizations . cavity width is @xmath128 nm .
other parameters the same as in fig . 1 . ]
figure [ fig : dispersion ] shows the exact dispersion of cavity polaritons for an artificially high value of the magnetic field 11.5 t allowing to make visible all peculiarities of the dispersion .
this field corresponds to the zeeman splitting of 0.6 mev .
we consider a @xmath129-cavity with 10 nm gaas qw ( parameters are the same as in fig . [
fig : faraday ] ) .
the distributed bragg reflectors ( dbrs ) are typical 20-period gaas / al@xmath130ga@xmath131as mirrors .
approximate dispersion curves calculated from eq . are not shown , because they do not differ from the exact one in the range of angles considered .
four branches of exciton - polaritons having a characteristic non - parabolic dispersion are clearly seen .
the anticrossings take place between branches having the same circular polarization .
in such a strong field , the polariton eigenstates are almost completely circularly polarized , so we do not show their polarization degree in this case .
t ( @xmath132 mev ) ( solid line exact model , dashed line approximate model ; red color shows exciton - like branch and black color shows photon - like branch ) . only branches with positive polarization degree are shown . ]
however , we do show the the circular polarization degree of the cavity eigenstates in the exact and approximate models for a small magnetic field of 0.04 t ( zeeman splitting is 2 @xmath133ev ) in figure [ fig : dispersionpolarization ] .
the results match well except for the exciton - like branch ( red ) , the difference coming from the fact that in the exact model the exciton eigenstates are circularly polarized , whereas in the approximate model they are linearly polarized ( their te - tm splitting is taken into account ) .
the circular polarization degree has been calculated using eq . in both models , the difference being that in one case the complex amplitudes are the eigenvectors of exact system eq . and in the other case they are the eigenvectors of approximate system eq . .
for such a small field , the splittings that appear in the polaritonic branches , are invisible relative to the rabi splitting , therefore we do not show the corresponding dispersions .
in this section we consider kerr and faraday effects in quantum microcavities .
it is assumed that @xmath134 or @xmath135 polarized light is incident on a microcavity from the vacuum and the rotation of polarization plane of reflected ( kerr effect ) and transmitted ( faraday effect ) wave is monitored as a function of incidence angle .
the kerr and faraday rotation angles can be determined once the matrices of amplitude reflection ( transmission ) coefficients @xmath136 ( @xmath137 ) of the whole microcavity are known .
the latter can be found from the microcavity transfer matrix @xmath138 which , in accordance with its definition , can be recast as a product @xmath139 where @xmath140 is the transfer matrix through the bragg mirror , @xmath141 is the transfer matrix through a half of the cavity and @xmath142 is the quantum well transfer matrix , determined in eqs .
, . the transfer matrix through the homogeneous layer of the thickness @xmath143 is diagonal and given by @xmath144 where @xmath145 is the @xmath68 unit matrix .
polarized light .
the qw and cavity parameters are the same as before .
the zeeman splitting is 0.1 mev .
inset shows energy dependence of faraday rotation angle at the same parameters at normal light incidence . ] for example , the transmission and reflection coefficients in the @xmath134 polarization can be determined from the following system of equations @xmath146 figure [ fig : cavityfaraday ] shows calculated faraday rotation angle at oblique incidence as a function of incidence angle .
an inset in the figure presents the faraday rotation angle as a function of the energy .
the function shows a non - monotonous behaviour reflecting the resonant character of light reflection and transmission in microcavities .
the figure clearly demonstrates that faraday rotation angles are almost by an order of magnitude larger than those for the faraday rotation in the case of a single quantum well .
this enhancement can be interpreted as a result of multiple passage of a photon inside a cavity @xcite . namely , consider a cavity mode as a beam of light travelling back and forth inside the cavity .
let @xmath147 be an average number of photon round trips between the bragg mirrors made during its life - time ( @xmath148 , where @xmath149 is the cavity quality factor ) .
at each trip the polarization plane of the photon exhibits a rotation by @xmath150 and the faraday effect is strongly enhanced .
the observable faraday rotation angle is smaller than @xmath151 because there is a finite probability for the photon to escape cavity after any number of passages @xcite .
the absolute value of the faraday rotation angle decreases with the increase of the incidence angle , due to the decrease of the polariton lifetime in the cavity . in average , the polaritons make less rount - trips inside the cavity at oblique angles than at normal angle .
we note that the kerr rotation angles ( observed in reflection geometry ) are smaller than the faraday rotation angles because the light reflection from the microcavity is dominated by a surface reflection of the bragg mirror .
in conclusion , we have analyzed the dispersion of the exciton - polaritons in a microcavity subjected into the external magnetic field taking in account both zeeman splitting of the exciton and te - tm splitting of the photonic modes .
we have shown that the polarization of the polariton eigenstates is neither linear nor circular , but elliptical , in general .
it is very sensitive to the polariton in - plane wave vector as well as to the detuning between the exciton and photon modes .
we have also studied theoretically the faraday rotation in the planar microcavities .
we found that the giant faraday rotation due to the multiple passages of light across the quantum well takes place at normal incidence , while this effect is reduced at oblique incidence angles .
see , e.g. , papers in the special issue on microcavities , semicond .
technol . * 18 * , n 10 ( 2003 ) or a.v . kavokin and g. malpuech , _ cavity polaritons _ , elsevier , amsterdam , 2003 .
m. richard , j. kasprzak , r. andr , r. romestain , le si dang , g. malpuech , and a. kavokin , phys .
b * 72 * , 201301 ( 2005 ) . p.g .
savvidis , j.j .
baumberg , r. m. stevenson , m.s .
skolnick , d.m .
whittaker , and j.s .
roberts , phys .
84 * , 1547 , ( 2000 ) .
a. imamoglu and j.r .
ram , phys .
a * 214 * , 193 ( 1996 ) .
j. kasprzak , m. richard , s. kundermann , a. baas , p. jeambrun , j.m.j .
keeling , f.m .
marchetti , m.h .
szymaska , r. andr , j.l .
staehli , v. savona , p.b .
littlewood , b. deveaud , and le si dang , nature * 443 * , 409 ( 2006 ) .
s. christopoulos _
et al _ , phys .
98 , 126405 ( 2007 ) .
shelykh , yu.g .
rubo , g. malpuech , d.d .
solnyshkov , and a.v .
kavokin , phys .
lett . * 97 * , 66402 ( 2006 ) .
i. carusotto and c. ciuti , phys .
* 93 * , 166401 ( 2004 ) .
shelykh , a.v .
kavokin , and g. malpuech , phys .
( b ) * 242 * , 2271 ( 2005 ) .
rubo , a.v .
kavokin , and i.a .
shelykh , phys .
lett . a * 358 * , 227 ( 2006 ) .
maialle , d.e . de andrada e silva , and l.j .
sham , phys .
b * 47 * , 15776 ( 1993 ) .
g. panzarini , l.c .
andreani , a. armitage , d. baxter , m.s .
skolnick , v.n .
astratov , j.s .
roberts , a.v .
kavokin , m.r .
vladimirova , and m.a .
kaliteevski , phys .
b * 59 * , 5082 ( 1999 ) .
j. tignon , r. ferreira , j. wainstain , c. delalande , p. voisin , m. voos , r. houdr , u. oesterle , and r.p .
stanley , phys .
b * 56 * , 4068 ( 1997 )
berger , o. lyngnes , h.m .
gibbs , g. khitrova , t.r .
nelson , e.k .
lindmark , a.v .
kavokin , m.a .
kaliteevski , and v.v .
zapasskii , phys .
b * 54 * , 1975 ( 1996 ) .
a. armitage , t.a .
fisher , m.s .
skolnick , d.m .
whittaker , p. kinsler , and j.s .
roberts , phys .
b * 55 * , 16395 ( 1997 ) .
p. renucci , t. amand , and x. marie , physica e * 17 * , 329 ( 2003 ) .
a. brunetti , m. vladimirova , d. scalbert , r. andr , d. solnyshkov , g. malpuech , i.a .
shelykh , and a.v .
kavokin , phys . rev .
b * 73 * , 205337 ( 2006 ) .
a. brunetti , m. vladimirova , d. scalbert , m. nawrocki , a.v .
kavokin , i.a .
shelykh , and j. bloch , _ phys .
rev . b _ * 74 * , 241101 ( 2006 )
landau and e.m .
lifshits , _ the classical theory of fields _ , volume 2 ( course of theoretical physics series ) elsevier , 1980
. v. savona , l.c .
andreani , p. schwendimann , and a. quattropani , solid state commun . * 93 * , 733 ( 1995 ) .
a.v . kavokin and m.a .
kaliteevski , solid state commun . *
95 * , 859 ( 1995 ) .
andreani , f. tassone , and f. bassani , solid state commun . * 77 * , 641 ( 1991 ) .
ivchenko , sov .
phys . solid state * 33 * , 1344 ( 1992 ) .
f. tassone , f. bassani , and l.c .
andreani , phys .
b * 45 * , 6023 ( 1992 ) .
kavokin , m.r .
vladimirova , m. a. kaliteevski , o. lyngnes , j.d .
berger , h.m .
gibbs , and g. khitrova , phys . rev .
b * 56 * , 1087 ( 1997 ) .
kavokin , a.i .
nesvizhskii , and r.p .
seisyan , sov .
. semicond .
* 27 * , 530 ( 1993 ) .
ivchenko , optical spectroscopy of semiconductor nanostructures ( alpha science , harrow uk , 2005 ) .
s.v goupalov , e.l .
ivchenko , and a.v .
kavokin , jetp * 86 * , 388 ( 1998 ) . | the non - local dielectric response theory is extended to describe oblique reflection of light from quantum wells subjected to the magnetic field .
this allows us to calculate the dispersion and polarization of the exciton - polariton modes in semiconductor microcavities in the presence of a magnetic field normal to the plane of the structure .
we show that due to the interplay between the exciton zeeman splitting and te - tm splitting of the photon modes , four polariton dispersion branches are formed with a polarization gradually changing from circular in the exciton - like part to linear in the photon - like part of each branch .
faraday rotation in quantum microcavities is shown to be strongly enhanced as compared with the rotation in quantum wells . |
images obtained through a turbulent medium ( like the atmosphere ) suffer from serious quality degradation .
there are several methods developed to alleviate this problem .
one of them is the adaptive optics @xcite , which is a method commonly used in astronomy , microscopy and surveillance . in this technique ,
a high - speed camera monitors continuously the shape of the wave front and the closed - loop system compensates for the wave distortions by deforming a special mirror in the optical path @xcite .
however , this approach is very complex , expensive and requires sophisticated equipment ( e.g. the deformable mirror or the wavefront sensing devices ) .
moreover , the required wavefront diagnostics is available only for the objects much brighter than the background .
if there is not enough light to perform the wavefront analysis , only the tip - tilt correction is applied @xcite .
it is accomplished by oscillating an additional mirror in optical path , which results in real - time image shifting @xcite .
another , much simplified yet efficient approach , is the lucky imaging @xcite , where consecutive short exposures are analyzed , shifted to the brightest speckle and stacked together , as depicted in fig .
[ fig : f1 ] . only the images with the highest strehl ratio
are included , hence the final image is the average of the frames captured under best atmospheric conditions .
the probability of obtaining such good quality and unblurred frames is a function of the aperture size of @xmath0 and the fried parameter @xmath1 @xcite : [ ] @xmath2 however , even those good quality frames are shifted relative to each other .
the cancellation of tip - tilt component recovers about 50% of spatial distribution in most cases @xcite . that kind of recovery is possible only if the source is intensive enough , so that a single frame contains clearly visible brightest speckles .
moreover , the brightest speckle varies very quickly and it is assumed to be stable for a very short , so - called coherence time ( about 10ms ) @xcite . to the best of our knowledge , there were no successful attempts to recover the high quality spatial distribution of the objects using shift - and - add method , if neither the brightest speckle nor the signal centroid is visible in a single exposure .
in this paper we present a novel technique based on the genetic algorithm ( ga ) , which allows for individual frames centring , even if the brightest speckle is not distinguishable , ( see example of such frame in fig . [
fig : f2]c ) .
our method does not require any additional information about observed object , so it may be employed in a wide range of applications .
( russell ) at ( 0,0 ) ; on ( -4.05,1.75 ) in node at ( -8,1 ) ; ( -4.2,2.4 ) node [ above ] a ) ; ( 0,2.4 ) node [ above ] b ) ; ( 4.7,2.4 ) node [ above ] c ) ; ga is an iterative technique employed for the optimization purposes in highly nonlinear spaces @xcite .
it searches a solution space in order to achieve a given goal , so that in each evolution of the algorithm , a part of individuals is randomly modified in hope to produce new , better solutions .
the best individuals ( elite ) are transferred directly between the evolutions .
another group in the population is created by the mutation of the elite .
finally , in each iteration , a part of individuals is created randomly to prevent getting stuck at a local minimum .
the use of ga on a large number of frames allows to evolve the set of frames shifts toward the highest energy of averaged frame , i.e. to maximize the value of @xmath3 defined as : @xmath4 where @xmath5 is the number of counts in a pixel at position ( @xmath6 , @xmath7 ) and _ n _ , _ m _ denotes the frame s width and height .
we used such a fitness function , because only the perfect alignment of all images provides the maximum sum of squared counts .
as the errors occur in shifts estimations , the counts from the object are spread between many pixels in the averaged stack , thus our metric decreases . before the energy calculation
, we perform background estimation utilizing a median over all pixels in stacked images .
then , the background level is subtracted , hence the influence of background counts on the energy measure @xmath3 is significantly reduced . in our simulations , first we used airy discs as sources @xcite .
it is the best description of spot light that a perfect lens with circular aperture can produce .
the disc size is governed by the diffraction limit , what may be observed in spaceborne telescope missions due to a lack of turbulent medium between the imager and the object ( the interested reader is referred to e.g. @xcite ) . to simulate real count results in each pixel ,
we utilized poisson distribution , where the mean value was determined from the sum of background and object counts @xcite .
we present an exemplary result of simulated airy discs in fig .
[ fig : f2]a , where the objects produce 61 photons and the entire background counts equals 677 .
finally , we combined the counts of the background with the ones from the object and added gaussian readout noise of imager amplifier .
we adjusted the level of noise and source , so that the object can not be distinguished from the background noise ( see fig .
[ fig : f2]c ) .
in ga we optimized two variables per each frame the shifts in _ x _ and _ y _ directions .
such an approach produces a huge number of variables , thus a whole set of combinations can not be tested in reasonable time even with the most powerful computers .
assuming that the object tip - tilt effect is limited to @xmath8__p _ _ pixels in each direction , we have the following number of possible shift combinations @xmath9 for @xmath10 images in a series : @xmath11 in contrast , with the use of ga , this optimization process can be significantly reduced , since the algorithm promotes and modifies only viable shifts combinations while the others are discarded .
for example , in our experiments , the recovery of 70 frames of size 100@xmath12100 , took about 30 minutes ( 1500 generations ) in the case of simple point - like sources .
according to eq .
( [ eq : eq3 ] ) , 3.75e+479 years would be required on the same computer to search a whole combinations space ( @xmath9=1.97e+489 combinations for @xmath10=70 images in a series , maximum shift @xmath13=7 ; the computational time of a single shift - and - add procedure is 6ms ) .
we used 700 individuals in our ga population . by an individual
we understand the vector of shifts in @xmath14 and @xmath15 directions for all images in a series ( e.g. for 100 images the individual is 200-elements vector ) . in each iteration
, the best solution was transferred directly from the previous population and 500 individuals were created by the crossover procedure from the shifts estimations of previous 140 individuals ( elite ) .
we employed scattered crossover , thus each shift of newly created solution was selected from a random individual from the elite .
next 100 individuals were created by mutation , where we changed the shifts of previous best individual using the gaussian distribution with a standard deviation equal to 1/3 of maximum available shift .
the ga final outcomes were the shifts estimations obtained after 1500 iterations . such value was chosen due to the observed satisfactory results in this initial part of the convergence process and due to the reasonable computational time ( 30 minutes ) .
it should be noted , that better outcomes may be obtained after much longer ga run .
however it would require randomly varying calculation time , often reaching tens of hours .
it is difficult to present strict numeric properties of the method , since each individual run depends on its randomly generated starting point . nevertheless , from multiple tests we observed , that the method allows for the recovery of the spatial distribution of objects as dim , as 2% of their mean background .
for such intensity of light , about 70 exposures were required to obtain satisfactory results using our ga - based algorithm .
[ every node/.style = anchor = south west , inner sep=0pt , font= * * * * , x=1 mm , y=1 mm ] ( fig1 ) at ( 0,30 ) ; 5 ( 3.3,215 ) node [ above ] a ) ; ( 39,215 ) node [ above ] b ) ; ( 74.5,215 ) node [ above ] c ) ; ( 110,215 ) node [ above ] d ) ; ( 145,215 ) node [ above ] e ) ; ( 47,192 ) node [ above]blur : 0.000 ; ( 83,192 ) node [ above ] blur : 1.000 ; ( 118.5,192 ) node [ above ] blur : 0.028 ; ( 83.5,188 ) node [ above ] rmse : 22.94 ; ( 119,188 ) node [ above ] rmse : 14.10 ; ( 3,149 ) node [ above ] f ) ; ( 39,149 ) node [ above ] g ) ; ( 74.5,149 ) node [ above ] h ) ; ( 110,149 ) node [ above ] i ) ; ( 145,149 ) node [ above ] j ) ; ( 47,126 ) node [ above]blur : 0.000 ; ( 83,126 ) node [ above ] blur : 1.000 ; ( 118.5,126 ) node [ above ] blur : 0.040 ; ( 83.5,122 ) node [ above ] rmse : 22.16 ; ( 119,122 ) node [ above ] rmse : 13.53 ; ( 3,85 ) node [ above ] k ) ; ( 39,85 ) node [ above ] l ) ; ( 75.3,85 ) node [ above ] m ) ; ( 110,85 ) node [ above ] n ) ; ( 145.2,85 ) node [ above ] o ) ; ( 47,61 ) node [ above]blur : 0.000 ; ( 83,61 ) node [ above ] blur : 1.000 ; ( 118.5,61 ) node [ above ] blur : 0.041 ; ( 83.5,57 ) node [ above ] rmse : 23.40 ; ( 119,57 ) node [ above ] rmse : 13.76 ; in fig .
[ fig : f3 ] we present the outcomes of our algorithm for exemplary images .
the first one depicts the aforementioned point - like objects . to obtain the remaining two complex images
, we cropped 100@xmath12100 pixels from the image of a car licence plate and 200@xmath12200 pixels of human retina .
we simulated the required short exposures by reducing significantly the pixels intensities .
then , we shifted the frames randomly .
we decided to select uniform distribution , so that the estimation was more challenging , than it would be for gaussian distribution , in which a majority of small shift deviations is observed .
we also accounted for the poisson and read - out noise of imager amplifier .
then , we ran our ga procedures on such a set of frames .
the presented experiment reflects the imaging process through the turbulent medium , where the fried parameter @xmath16 is equal to the lens size .
in such a case , the most prominent resolution degradation originates in a global tip - tilt effect within the image plane . in our experiments we obtained visually satisfactory results , even if the light flux was about 2 - 3% stronger than the background .
the image enhancement is significant and the error maps contain only the noise without any visible object signature .
the calculations of the signal to background level were based on the photon count , measured within green ( background only ) and blue ( background and signal ) regions ( see fig [ fig : f3]b , g and l ) .
for the point - like object , the signal - to - background ratio was 3.7% .
for the complex images it was , respectively 1.78% and 1.92% for the license plate and human retina . in fig .
[ fig : f3 ] we also included the results of frequently used image quality measures : rmse ( root - mean - square error , in counts ) and blur ( see @xcite for description ) .
for better readability , blur outcome was normalized , so that for the reference its value is 0 , and for the stack with simulated turbulence , it is 1 . to reduce the impact of background noise on results of quality measures ,
all images were smoothed using a 3@xmath123 moving average .
the reference for the comparisons was obtained by stacking images without simulated turbulence .
[ ] . , title="fig:",width=453 ] in fig .
[ fig : f4 ] we present a comparison between calculated and simulated _ x _ and _ y _ shifts in an exemplary experiment run .
the calculated mean shift error was , respectively , 0.13 and -0.2 for _ x _ and _ y _ direction .
the standard deviation was 1.43 ( _ x _ ) and 1.08 ( _ y _ ) .
as predicted , for the most of the frames , the shifts estimations provided by our method were the same or very close to the reference ones .
in this paper we demonstrated the first results of spatial distribution recovery in shift - and - add method when no brightest speckle nor the signal centroid is visible in a frame .
the spatial distribution , to the best of our knowledge , not possible to resolve by any other approach , was successfully reconstructed .
the proposed ga method may be employed in a wide range of applications including astronomy , medicine or surveillance .
moreover , it is an automatic method and no knowledge about spatial distribution , number of sources or their position is required .
it is also very easy to implement , as ga solvers are presently available in numerous software packages .
the proposed technique is suitable for real and simulated data in both optical and infrared bands .
our method may be also employed as a digital image stabilizer in extremely low light level conditions in both professional and consumer applications .
adam popowicz was supported by polish national science center , grant no .
2013/11/n / st6/03051 : novel methods of impulsive noise reduction in astronomical images .
the research was performed using the infrastructure supported by poig.02.03.01 - 24 - 099/13 grant : geconii upper silesian center for scientific computation .
this research was supported by the dean of the faculty of physics , astronomy and applied computer sciences of the jagiellonian university , decision no .
this research was also supported by the national science center grant umo-2012/07/b / st9/04425 .
j. m. backers , `` adaptive optics for astronomy principles , performance , and applications '' , annu .
* 31 , * 13 - 62 ( 1993 ) http://adsabs.harvard.edu/abs/1993ara%26a..31...13b [ online version ] n. baba , s. isobe , et al . ,
stellar speckle image reconstruction by the shift - and - add method " , appl .
optics .. * 24 , * 1403 - 1405 ( 1985 ) http://ukads.nottingham.ac.uk/cgi-bin/nph-bib_query?bibcode=1985apopt..24.1403b&db_key=ast [ online version ] a. sivaramakrishnan , r. weymann , et al . ,
measurements of the angular correlation of stellar centroid motion " , astrophys .
j. * 110 , * 430 - 438 ( 1995 ) http://adsabs.harvard.edu/abs/1995aj....110..430s [ online version ] j. christou , image quality , tip - tilt correction , and shift - and - add infrared imaging " , publ .
* 103 , * 1040 - 1048 ( 1991 ) ttp://adsabs.harvard.edu / abs/1991pasp .. 103.1040c [ online version ] j. hecquet , g. coupinot , a gain in resolution by the superposition of selected recentered short exposures " , j. opt .
* 16 , * 21 - 26 ( 1985 ) http://adsabs.harvard.edu/abs/1985jopt...16...21h [ online version ] r. haupt , s. haupt , `` practical genetic algorithms '' , john wiley & sons ( 1998 ) http://library.alibris.com/practical-genetic-algorithms-randy-l-haupt-ph-d/book/5278379?qsort=c&matches=33 [ online version ] j. harvey , c. ftaclas , diffraction effects of telescope secondary mirror spiders on various image - quality criteria " , appl
* 34 , * 6337 ( 1995 ) http://adsabs.harvard.edu/abs/1995apopt..34.6337h [ online version ] j. holberg , s. casewell , hubble space telescope imaging and spectroscopy of the sirius - like triple star system hd 217411 " , mon . not .
r. astron .
* 444 , * 2022 - 2030 ( 2014 ) http://adsabs.harvard.edu/abs/2014mnras.444.2022h [ online version ] f. crete , t. dolmiere , et al .
, the blur effect : perception and estimation with a new no - reference perceptual blur metric " , soc .
photo .- opt .
instru . * 6492 , * ei 6492 - 16 ( 2007 ) http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=1298489 [ online version ] | shift - and - add is an approach employed to mitigate the phenomenon of resolution degradation in images acquired through a turbulent medium . using this technique ,
a large number of consecutive short exposures is registered below the coherence time of the atmosphere or other blurring medium .
the acquired images are shifted to the position of the brightest speckle and stacked together to obtain high - resolution and high signal - to - noise frame .
+ in this paper we present a highly efficient method for determination of frames shifts , even if in a single frame the object can not be distinguished from the background noise .
the technique utilizes our custom genetic algorithm , which iteratively evolves a set of image shifts .
we used the maximal energy of stacked images as an objective function for shifts estimation and validate the efficiency of the method on simulated and real images of simple and complex sources .
obtained results confirmed , that our proposed method allows for the recovery of spatial distribution of objects even only 2% brighter than their background .
the presented approach extends significantly current limits of image reconstruction with the use of shift - and - add method .
the applications of our algorithm include both the optical and the infrared imaging .
our method may be also employed as a digital image stabilizer in extremely low light level conditions in professional and consumer applications .
+ https://www.osapublishing.org/ol/abstract.cfm?uri=ol-40-10-2181 [ original @ optics letters website ] + + |
the factorization calculation@xcite has successfully reproduced the decay rate for many two - body channels of @xmath0 decay .
it has been argued@xcite that the factorization should become exact in the large @xmath7-quark mass limit and that the only deviation from the factorization is short - distance corrections of @xmath8 . for some channels of @xmath0 decay
, however , the factorization calculation appears to be in clear disagreement with current measurement@xcite .
theoretically the factorization and the final - state interaction ( fsi ) are closely tied together@xcite .
if the factorization is a good approximation , the fsi phase should be small in general . if the fsi phase turns out to be large in experiment , it is a warning sign against the factorization .
some _ initial state _ interaction can also affect the factorization@xcite .
the observed smallness of the upper bounds on the fsi phases of the two - body @xmath9 processes such as @xmath10 is consistent with absence of long - distance fsi@xcite . on the other hand ,
purely phenomenological suggestions were recently made about possibility of large fsi phases in @xmath11 , @xmath12@xcite . at present ,
little is known in experiment about fsi phases of decays other than the charm producing channels .
the factorization clearly fails in the @xmath13 decay in spite of the relatively small value of @xmath14 .
the observed relative fsi phase in @xmath15 is close to @xmath16@xcite though one would naively expect the factorization to work reasonably well if @xmath17 is really the controlling parameter of its corrections . in most of the two - body decay channels of @xmath18
so far analyzed@xcite . ] in two - body @xmath0 decays the _ c.m .
_ energy between a fast quark and a spectator grows only like @xmath19 with @xmath20 , not linearly in @xmath20 .
therefore , @xmath19 must be large in order for perturbative qcd to be applicable . even if one accepts the asymptotic validity of the factorization in theory@xcite , one should test in experiment whether or not the @xmath0 mass of 5.3 mev is high enough to be in the asymptopia of its final states . in this paper ,
we extend the basic premise of the factorization and explore its consequences in three - meson decay channels of @xmath0 decay in which one of mesons is soft .
we propose simple selection rules which will provide a further test of the fractorization at the @xmath0 mass scale .
our approach is purely phenomenological and we do not attempt here a rigorous justification of the selection rules by perturbative qcd .
when we compare the selection rules with the soft - pion theorem of chiral symmetry , it appears as if they were incompatible .
the origin of this superficial incompatibility is traced to difference in kinematics : the selection rules hold when the _ c.m . _
energy @xmath19 between a fast quark and a soft quark is sufficiently large , while the soft - pion theorem holds when the same _ c.m .
_ energy is equal to a light quark mass , _
i.e. , _ negligibly small .
the purpose of our selection rules is to probe if @xmath19 in @xmath0 decay is in the qcd asymptopia or not .
we state the factorization of two - body decays as follows : when a @xmath0 meson decays with a current - current interaction through the quark decay process , @xmath21 either @xmath22 or @xmath23 forms one meson and the remaining @xmath24 or @xmath25 forms the other meson with a soft quark @xmath26 from the light spectator cloud around @xmath27 .
( see fig.1a . )
the formation of a meson by two hard quarks through a color - singlet current is given by a decay constant while the formation of the other meson is described by a transition form factor of the other current .
it is important that once combinations of @xmath28 have been chosen for meson formation , no flavor exchange occurs thereafter , not even a charge exchange .
any flavor - changing process is supposedly a higher order process of perturbative qcd .
therefore the final mesons retain their flavor contents , including charges , determined by the short - distance quark decay process .
this leads us to various constraints of the factorization which relate different isospin amplitudes of two - body decays for a given decay operator . what happens if three mesons are produced ?
in the case that all final mesons are energetic , each meson should contain one of the three final quarks of eq.([qqq ] ) picking up one soft quark from the spectator cloud .
( see fig.1b . ) in the case that two of the mesons are energetic and one is very soft , the energetic mesons are formed just as in the two - body decay and the third one emerges from the spectator cloud with energy no larger than @xmath29 . in this case
the energetic mesons retain the history of the weak interaction while the soft meson is made of whatever left in the spectator cloud .
this is the natural extension of the factorization from two - body decays to three - body decays .
we call this picture as the factorization for three - body decays and explore its consequences here
. the _ c.m .
_ energy of the fsi between an energetic meson and the soft cloud is @xmath30 , which is the same as the _ c.m . _
energy of @xmath25 and @xmath26 of the two - body decay in fig.1a . , but our selection rules do not depend on a soft interaction between them . ]
if @xmath19 is large and short - distance qcd dominates in fsi , flavor exchange between any pair of final mesons would be a higher order correction . in this case , the three quarks produced by the @xmath27-quark can be tracked down into the flavor contents of the final mesons .
we illustrate derivation of our selection rule in the decay , @xmath32 which proceeds through the tree - operator @xmath33 and its gluon corrections . the boldface letter @xmath34 in eq.([dpipi ] ) represents a cloud of light quark - antiquarks and gluons around @xmath27 which carries the net flavor of the @xmath35-quark ; @xmath36 we refer to @xmath34 as the spectator @xmath35-quark cloud .
when the decay products @xmath37 of @xmath27 enter three mesons one each , all mesons are energetic . in this case
we can not find a simple flavor selection rule of the factorization .
the reason is as follows : even if no flavor exchange occurs between final mesons , the process @xmath38 occurs perpetually inside the cloud prior to meson formation .
this annihilation process has the same effect as a flavor exchange between final mesons since each meson picks up one soft quark from the cloud .
therefore , no simple testable selection rule results in this case . on the other hand ,
if one of the pions is soft , it must be generated by the spectator cloud without involving a hard quark .
the three quarks from @xmath39 must go into @xmath40 and the energetic @xmath41 . and @xmath42 are negligible . ]
the hard @xmath43 can form either @xmath6 , @xmath44 , or @xmath40 while the hard @xmath42 can form either @xmath6 or @xmath44 . even though @xmath42 or @xmath45 picks one soft quark from the cloud
, the resulting meson is energetic .
after one quark ( @xmath43 or @xmath35 ) is removed from the spectator cloud * d * of total charge @xmath46 , the remainder of the cloud is either negatively charged or neutral .
that is , the soft pion can be either @xmath5 or @xmath44 , but not @xmath6 .
we thus conclude that @xmath6 can not be soft in @xmath47 .
this selection rule is depicted in fig.2 .
in fact , the selection rule holds irrespective of the decay interaction .
it depends only on the flavor content of the spectator * d * : _ when one quark is taken away from a cloud of charge @xmath46 , the remainder can not be positively charged . _
the key is that mesons carry quantum numbers of quark - antiquark no matter what their fock - space compositions are .
the only assumption we have made here is that the energetic mesons do not exchange a flavor with the soft meson by short - distance qcd dominance in fsi
. our prediction should be tested in the form , @xmath48 where @xmath49 for the soft pion .
the bands of the @xmath12 and @xmath50 resonances should be excluded in testing eq.([sel ] ) . if , contrary to the factorization , elastic or inelastic rescattering is important at @xmath51 ( @xmath52 gev ) , @xmath53 can be fed through the unsuppressed decay modes , for instance , through @xmath54 followed by the forward charge exchange rescattering @xmath55 or through @xmath56 followed by the backward elastic rescattering @xmath57 , _
i.e. , _ the forward rescattering of two units of charge exchange .
thus the selection rule is sensitive to fsi .
the fsi of our concern is the soft - hard process at _ c.m .
_ energy of @xmath30 that involves a hard quark , since the soft - soft fsi is no more than a different pick of soft quarks from the cloud .
selection rules similar to eq.([sel ] ) can be derived for many other three - body decay modes .
considering their relative simplicity , it is worthwhile exploring such selection rules for better understanding of strong interaction dynamics in @xmath0 decay .
the @xmath40 meson can be replaced by any meson or meson resonance in eq.([dpipi ] ) ; @xmath58 , where @xmath59 denotes a neutral meson or meson resonance such as @xmath60 , @xmath44 , or @xmath61 .
the selection rule reads : @xmath62 we can apply the same reasoning to @xmath63 by replacing the spectator @xmath34 by @xmath64 ; @xmath65 where @xmath66 is a positively charged meson or meson resonance such as @xmath67 , @xmath6 , and @xmath68 . for @xmath69 , we can write the selection rule @xmath70 if we include @xmath71 in the spectator cloud , we obtain corresponding selection rules , @xmath72 furthermore , @xmath73 since the rest energy of a kaon exceeds @xmath74 , formation of any kaon from the spectator cloud is kinematically suppressed .
eqs.(8 ) to ( [ kaon4 ] ) mean that the factorization - forbidden modes are even further suppressed than the modes in the denominators .
the lightest meson resonance that the spectator cloud can produce is the @xmath75-meson .
since its rest mass far exceeds @xmath74 , @xmath76 is suppressed no matter what charge state the @xmath75-meson is in .
the cloud simply does not have enough energy to produce @xmath75 even at rest . to be concrete ,
consider @xmath77 .
then the lack of energy in producing a resonance from the cloud leads us to @xmath78 where @xmath75 can be in any charge state . in eq.([selrho ] ) @xmath75 may be replaced by @xmath79 or some other resonance .
similarly , @xmath80 though we have listed only the selection rules that are relatively simple and clean , one can write down many more selection rules by the same reasoning . since meson or meson resonance production of energy higher than @xmath29 from the cloud
is highly suppressed kinematically , we may raise the softness limit considerably from @xmath81 to , for instance , @xmath82 in actual testing of the selection rules , provided that one can avoid resonance bands of quasi - two - body decays .
if we raise it too high , however , the tail of the energy spectrum of a decay product quark enters the `` soft meson '' . to find a compromise , we need some numerical study .
our selection rules are based on the postulate that a soft meson must be produced entirely from the spectator cloud .
therefore , they would fail if a soft meson contains a significant contribution from the low - energy tail of a decay product of @xmath83 . for this reason we would like to know how much the low - energy tail contributes when we define the soft meson by @xmath84 .
if the tail is indeed negligible , then we should ask how much we can raise the upper limit of softness in actual test without violating the selection rules substantially . to be concrete , consider @xmath85 again .
suppose that we define the soft pion by its momentum @xmath86 .
estimate of the matrix element involves uncertainties in soft meson formation .
if we use the quark - hadron duality for the low - energy tail region and replace the meson wavefunctions by their average values , we can obtain a crude estimate of the decay rate ratio as @xmath87 where @xmath88 . the factor @xmath89 , which arises from difference in meson formation probability , is given by the ratio of the pion wavefunction @xmath90 on the light - cone as @xmath91 where @xmath92 .
the value of @xmath89 is sensitive to @xmath93 near the end point .
let us extrapolate @xmath94 smoothly to @xmath95 .
then for @xmath96 @xmath97 the right - hand side is a fraction of 1% for @xmath98 mev .
it is indeed small due to the phase space suppression of the soft @xmath43-quark and a smaller meson formation probability for a quark pair of highly uneven energies . from this estimate
we expect that accuracy of our selection rules should be very high .
when @xmath99 is raised to @xmath100 , the ratio in eq.([q1 ] ) goes up to 8% . in carrying out experimental test , therefore , we may be able to raise the softness upper limit from @xmath74 to @xmath100 .
the main background in testing our this selection rule will be a spillover from the resonance bands of @xmath61 and @xmath101 due to errors in pion measurement .
in the zero four - momentum limit of the soft pion , we can relate three - body decay matrix elements to two - body decay matrix elements by the soft - pion theorem of chiral symmetry .
it can be expressed with @xmath102 mev as @xmath103|b\rangle + k_{\mu}m^{\mu},\ ] ] where the @xmath104 term gives the correction to the soft - pion limit@xcite . for soft @xmath6 emission ,
@xmath105 should be replaced by @xmath106 . upon extrapolation to the physical pion at rest @xmath107 ,
the correction term @xmath104 is of @xmath108 where @xmath109 is a typical excitation energy of the @xmath0 system . according to the traditional wisdom ,
the born diagram of @xmath110 is likely the most important correction ( see fig.3a ) since it is nearly degenerate with @xmath0 ( @xmath111 mev ) .
this correction can be of @xmath112 or even dominate over the zero - four - momentum limit for a physical soft pion .
it is easy to see that the @xmath113 contribution to @xmath104 satisfies the selection rule eq.([sel ] ) .
in fact , this is the case for the contributions from all born diagrams .
the reason is that the quark diagrams for the born diagrams of fig.3a are topologically equivalent to those of the three - body factorization process .
the only difference between them is whether a soft pion is emitted from the cloud before or after weak interaction takes place .
if we keep only the soft - pion limit , we obtain @xmath114 eq.([s2 ] ) results from the vanishing commutator , @xmath115=0 $ ] .
in contrast , the zero four - momentum limit of the factorization - forbidden decay @xmath116 is related to @xmath117 . while @xmath118 is @xmath119 for @xmath120 , it is the leading term for @xmath121 .
eqs.([s1 ] ) and ( [ s2 ] ) look puzzling at the first sight .
if we work out the soft - pion limit for @xmath122 , our three - body factorization predicts @xmath123 while the soft - pion theorem gives for the tree - operator @xmath124 and @xmath125 @xmath126 the situation looks equally bizzare in @xmath127 through the penguin operators . since the penguin operators of @xmath128 are all singlets of chiral su(2)@xmath129su(2 ) , their commutators with the axial isospin charges are zero .
that is , the soft - pion limit should disappear for all matrix elements of @xmath130 as @xmath131 for the penguin operators .
it appears as if the selection rules contradicted the soft - pion theorem of chiral symmetry .
we can trace the origin of this apparent discrepancy to the _ c.m . _
energy between a fast quark ( @xmath132 ) from @xmath27-decay and a soft quark ( @xmath133 ) in the cloud . in the _ physical _ soft pion limit where our selection rules should hold ,
@xmath134 while in the zero four - momentum limit @xmath135 in order for the factorization to work , the right - hand side of eq.([cm1 ] ) must be large enough to be in the asymptopia of qcd .
the same variable is equal to the light quark mass in the soft - pion theorem .
their predictions apply to different kinematical regions .
the chiral lagrangian has been successful for which the soft - pion theorem gives the leading terms .
however , extrapolation to the physical region of @xmath0-decays is a long way because of large @xmath0 mass ( cf eqs.([cm1 ] ) and ( [ cm2 ] ) ) and involves large incalculable corrections .
we can understand what physical process the soft - pion theorem represents@xcite .
the commutator of the axial charge @xmath136 with weak decay operator @xmath137 arises from its commutator with the currents in @xmath137 . in the case of @xmath138 at @xmath139 with @xmath140 , for instance ,
the commutator with @xmath141 survives to give @xmath142 = ( \overline{b}_{l\alpha}\gamma_{\mu}c_{l\beta } ) ( \overline{u}_{l\beta}\gamma^{\mu}u_{l\alpha } -\overline{d}_{l\beta}\gamma^{\mu}d_{l\alpha}),\ ] ] where the right - hand side is the @xmath143 partner of @xmath121 .
this commutator describes production of states of charge @xmath144 through @xmath141 of @xmath121 and subsequent fragmentation of the soft @xmath6 from them .
emission of a zero - four - momentum pion from the energetic quark pair @xmath141 is not suppressed by asymptotic freedom since no large momentum transfer occurs .
after the @xmath6 emission , these states ( @xmath145 ) and the operator @xmath146 together produce the final state @xmath147 .
( see fig.3b . ) for soft @xmath5 emission , the commutator vanishes so that there is no diagram corresponding to fig.3b .
thus physics in the zero - four - momentum limit is quite different from what we have introduced as an extension of the factorization in this paper .
an important question is which picture gives a better description of the three - body @xmath0 decay in which one pion is soft .
is the approximation of @xmath148 better than the zero four - momentum approximation ?
the issue should be settled by experimental test of the selection rules proposed here .
the dynamics underlying the factorization can be tested with the selection rules in three - body decays in which one of mesons is soft . the selection rules test whether the soft - hard fsi at _ c.m .
_ energy of @xmath30 is in the qcd asymptopia or not .
the rules are intuitive and do not depend on decay operators nor on form factors .
if the selection rules are verified in experiment , we would obtain better understanding of physics underlying the factorization and strong interactions in @xmath0 decay in general .
this work was supported in part by the director , office of science , division of high energy and nuclear physics , of the u.s .
department of energy under contract de ac0376sf00098 and in part by the national science foundation under grant phy9514797 .
d. fakirov and b. stech , nucl . phys .
* b 232 * , 315 ( 1978 ) ; n. cabibbo and l. maiani , phys .
73b * , 418 ( 1978 ) ; m. wirbel , b. stech , and m. bauer , z. phys .
c * 29 * , 637 ( 1985 ) .
bjorken , nucl .
b(proc . suppl . )
* 11 * , 325 ( 1991 ) ; m.j .
dugan and b. grinstein , phys .
255b * , 583 ( 1991 ) ; h.d .
politzer and m.b .
wise , phys .
257b * , 399 ( 1991 ) ; c.h .
chang and h .- n .
li , phys .
d*55 * , 5577 ( 1997 ) ; t.w .
yeh and h .- n .
li , phys .
rev . d*56 * , 1615 ( 1997 ) ; y .- y . keum , h .-
li , and a.i .
sanda , hep - ph/0004004 ; hep - ph/0004173 .
m. beneke , g. buchalla , m. neubert , and c.t .
sachrajda , phys .
lett . * 83 * , 1914 ( 1999 ) and hep - ph/0006124 .
m. beneke , g. buchalla , m. neubert , c.t .
sachrajda , hep - ph/0007256 ; d. cronin - hennessy _ et al . _
( cleo collaboration ) , hep - ex/0001010 . for a viewpoint different from perturbative qcd , m. suzuki , in _ proceedings of the third international conference on b physics and cp violation _ ,
taipei , december , 1999 ( world scientific , singapore , 2000 ) , to be published .
a. deandrea , r. gatto , m. ladisa , g. nardulli , and p.santorelli , hep - ph/0002038 . m. neubert and b. stech , hep - ph/9705292 ; m. suzuki , phys . rev .
d*58 * , 111504 ( 1998 ) .
hou and k .- c .
yang , phys .
lett * 84 * , 4906 ( 2000 ) ; y.f .
zhou , y.l .
ng , and c.q .
geng , hep - ph/0006225 .
m. bishai _
et al._(cleo collaboration ) , phys .
lett . * 78 * , 3261 ( 1997 ) .
j. jousset _
et al . _ ,
d * 41 * , 1389 ( 1990 ) : see m. suzuki , hep - ph/0006296 for recent references .
s.l . adler and r.f .
dashen , _ current algebra and applications to particle physics _ , benjamin , new york , 1968 . | extending the dynamics underlying the factorization calculation of two - body decays , we propose simple selection rules for nonresonant three - body @xmath0 decays .
we predict , for instance , that in the dalitz plot of @xmath1 , practically no events should be found in the corner region of @xmath2 as compared with the corner of @xmath3 .
we also predict that there should be very few three - body decay events containing one soft meson resonance and two energetic mesons or meson resonances .
the selection rules are quite different from the soft - pion theorem , since they apply to different kinematical regions . for @xmath4 ,
the latter predicts that the decay matrix element vanishes in the zero four - momentum limit of @xmath5 instead of @xmath6 . since this marked difference from the soft - pion theorem
is directly related to the issue of short - distance qcd dominance in two - body b decays , experimental test of the selection rules will shed light on strong interaction dynamics of @xmath0 decay . |
[ sec : defi ] the issue of this work is to introduce and to study the computational model of quantum finite state transducers .
these can be understood as finite automata with the addition of an output tape which compute a relation between strings , instead of a decision ( which we read as a binary valued function ) .
after the necessary definitions , the relation to quantum finite automata is clarified ( section [ sec : qfa : qfst ] ) , then decidability questions are addressed ( section [ sec : empty ] ) : it is shown that emptiness of the computed relation is undecidable both for quantum and probabilistic transducers .
however , the membership problem for a specific output is decidable .
next , the relation between deterministic and probabilistic transducers is explored ( section [ sec : det : prob ] ) , and in section [ sec : qfst ] quantum and probabilistic transducers are compared .
we feel our extension of quantum automata studies to this new model justified by the following quote from d. scott @xcite : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ` the author ( along with many other people ) has come recently to the conclusion that the functions computed by the various machines are more important or at least more basic than the sets accepted by these devices .
( ... ) in fact by putting the functions first , the relationship between various classes of sets becomes much clearer ' .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we start be reviewing the concept of probabilistic finite state transducer . for a finite set @xmath0
we denote by @xmath1 the set of all finite strings formed from @xmath0 , the empty string is denoted @xmath2 .
[ defi : pfst ] a _ probabilistic finite state transducer ( pfst ) _ is a tuple @xmath3 , where @xmath4 is a finite set of states , @xmath5 is the input / output alphabet , @xmath6 is the initial state , and @xmath7 are ( disjoint ) sets of accepting and rejecting states , respectively .
( the other states , forming set @xmath8 , are called non halting ) . the transition function @xmath9
is such that for all @xmath10 the matrix @xmath11 is stochastic , and @xmath12 is the output function .
if all matrix entries are either @xmath13 or @xmath14 the machine is called a _ deterministic finite state transducer (
dfst)_. the meaning of this definition is that , being in state @xmath15 , and reading input symbol @xmath16 , the transducer prints @xmath17 on the output tape , and changes to state @xmath18 with probability @xmath11 , moving input and output head to the right .
after each such step , if the machine is found in a halting state , the computation stops , accepting or rejecting the input , respectively . to capture this formally ,
we introduce the _ total state _ of the machine , which is an element @xmath19 with the natural norm @xmath20 at the beginning , the total state is @xmath21 ( where we identify an element of @xmath22 with its characteristic function ) .
the computation is represented by the ( linear extensions of the ) transformations @xmath23 of the total state , for @xmath10 , with @xmath24 and @xmath25 . for a string
@xmath26 the map @xmath27 is just the concatenation of the @xmath28 . observe that all the @xmath29 conserve the probability .
implicitely , we add initial and end marker symbols ( @xmath30 ) at the input , with additional stochastic matrices @xmath31 and @xmath32 , executed only at the very beginning , and at the very end .
we assume that @xmath32 puts no probability outside @xmath33 .
by virtue of the computation , to each input string @xmath34 there corresponds a probability distribution @xmath35 on the set @xmath36 : @xmath37\ ] ] is the probability to reject the input @xmath38 , whereas @xmath39\ ] ] is the probability to accept , after having produced the output @xmath40 .
[ defi : compute ] let @xmath41 . for @xmath42
we say that @xmath43 _ computes the relation @xmath44 with probability @xmath45 _ if for all @xmath38 , whenever @xmath46 , then @xmath47 , and whenever @xmath48 , then @xmath49 for @xmath50 we say that @xmath43 _ computes the relation @xmath44 with isolated cutpoint @xmath45 _ if there exists @xmath51 such that for all @xmath38 , whenever @xmath46 , then @xmath52 , but whenever @xmath48 , then @xmath53 .
the following definition is modelled after the ones for pfst for quantum finite state automata @xcite : [ defi : qfst ] a _ quantum finite state transducer ( qfst ) _ is a tuple @xmath54 , where @xmath4 is a finite set of states , @xmath5 is the input / output alphabet , @xmath6 is the initial state , and @xmath7 are ( disjoint ) sets of accepting and rejecting states , respectively .
the transition function @xmath9 is such that for all @xmath10 the matrix @xmath11 is unitary , and @xmath12 is the output function .
like before , implicitely matrices @xmath31 and @xmath32 are assumed , @xmath32 carrying no amplitude from @xmath8 to outside @xmath33 . the computation proceeds as follows : being in state @xmath15 , and reading @xmath16 , the machine prints @xmath17 on the output tape , and moves to the superposition @xmath55 of internal states .
then a measurement of the orthogonal decomposition @xmath56 ( with the subspaces @xmath57 , which we identify with their respective projections ) is performed , stopping the computation with accepting the input on the second outcome ( while observing the output ) , with rejecting it on the third . here ,
too , we define total states : these are elements @xmath58 with norm @xmath59 at the beginning the total state is @xmath60 , the total state transformations , for @xmath61 are ( for @xmath10 ) @xmath62 where @xmath63 , and @xmath64 observe that the @xmath29 do not exactly preserve the norm , but that there is a constant @xmath65 such that @xmath66 for any total state @xmath0 . quite straightforwardly , the distributions @xmath35 are defined , and so are the concepts of computation with probability @xmath45 or with isolated cutpoint @xmath45 .
observe also that we defined our model in closest possible analogy to quantum finite automata @xcite .
this is of course to be able to compare qfst to the latter . in principle
however other definitions are conceivable , e.g. a mixed state computation where the @xmath29 are any completely positive , trace preserving , linear maps ( the same of course applies to quantum finite automata ! ) .
we defer the study of such a model to another occasion .
notice the physical benefits of having the output tape : whereas for finite automata a superposition of states means that the amplitudes of the various transitions are to be added , this is no longer true for transducers if we face a superposition of states _ with different output tape content_. i.e. the entanglement of the internal state with the output may prohibit certain interferences .
this will be a crucial feature in some of our later constructions .
the definition of qfst is taylored in such a way that by excluding the output tape and the output function , we get a quantum finite automaton .
one , however , with distinct acceptance and rejection properties , as compared to the qfst .
nevertheless , the decision capabilities of qfst equal those of quantum finite automata : [ thm : qfa - c - qfst ] a language @xmath67 is accepted by a 1way quantum finite automaton with probability bounded away from 1/2 if and only if the relation @xmath68 is computed with isolated cutpoint .
_ proof _ : first observe that for finite automata ( probabilistic and quantum ) , recognizability with an isolated cutpoint is equivalent to recognizability with probability bounded away from @xmath69 ( by `` shifting the cutpoint '' : just add in the @xmath70step possibilities to accept or reject right away with certain probabilities ) .
we have to exhibit two constructions : let there be given a quantum finite automaton .
we may assume that it is such that @xmath32 is a permutation on @xmath4 .
this can be forced by duplicating each @xmath71 by a new state @xmath72 , and modifying the transition function as follows : denote by @xmath73 the map interchanging @xmath15 with @xmath72 for @xmath74 , and being the identity on @xmath75 for @xmath76 .
define a unitary @xmath77 such that for @xmath76 @xmath78 and @xmath79 for @xmath71 .
now let @xmath80 it is easily checked that this automaton behaves exactly like the initial one .
construct a qfst as follows : its states are @xmath81 , with @xmath82 being the accepting states , and no rejecting states .
let the transition function be @xmath83 with @xmath84 since @xmath32 is the permutation @xmath73 on @xmath4 , we may define @xmath85 finally , let the output function be ( for @xmath86 ) @xmath87 and @xmath2 in all other cases .
it can be checked that it behaves in the desired way .
given a qfst , construct a quantum finite automaton as follows : its states are @xmath88 , where the second component represents the tape content up to @xmath89 many symbols .
initial state is @xmath90 . observe that by definition of the @xmath29 amplitude that once is shifted onto output tapes of length larger than @xmath14 is never recovered for smaller lengths .
hence we may as well cut such branches by immediate rejection : the states in @xmath91 are all rejecting , and so are @xmath92 .
the accepting states are @xmath93 . the transition function is partially defined by @xmath94 ( for @xmath95 this is followed by mapping @xmath96 to a rejecting state , while leaving the other halting states alone ) , i.e. the automaton performs like the qfst on the elements of @xmath4 , and uses the second component to simulate the output tape .
we think of @xmath97 being extended in an arbitary way to a unitary map .
one can check that this construction behaves in the desired way .
as is well known , the emptiness problem for the language accepted by a deterministic ( or nondeterministic ) finite automaton is decidable .
since the languages accepted by probabilistic and quantum finite automata with bounded error are regular @xcite , these problems are decidable , too . for finite state transducers
the situation is more complicated : in @xcite it is shown that the emptiness problem for deterministic and nondeterministic fst is decidable .
in contrast we have [ thm : empty1 ] the emptiness problem for pfst computing a relation with probability @xmath98 is undecidable .
likewise , the emptiness problem for qfst computing a relation with probability @xmath98 is undecidable .
_ proof _ : by reduction to the post correspondence problem : let an instance @xmath99 , @xmath100 of pcp be given ( i.e. @xmath101 ) .
it is to be decided whether there exists a sequence @xmath102 ( @xmath103 ) such that @xmath104 construct the following qfst with input alphabet @xmath105 : it has states @xmath106 , and @xmath107 .
the initial transformation produces a superposition of @xmath108 , each with amplitude @xmath109 .
the unitaries @xmath110 are all identity , but the output function is defined as @xmath111 , for @xmath112 .
the endmarker maps @xmath113 to accepting states .
it is clear that @xmath102 is a solution iff @xmath114 is in the relation computed with probability @xmath98 ( the automaton is easily modified so that it rejects when the input was the empty word , in this way we force @xmath103 ) . by replacing the unitaries by stochastic matrices ( with entries the squared moduli of the corresponding amplitudes )
the same applies to pfst . since it is well known that pcp is undecidable
, it follows that there can be no decision procedure for emptiness of the relation computed by the constructed pfst , or qfst , respectively .
undecidable questions for quantum finite automata were noted first for `` @xmath115way '' automata , i.e. ones which move only to the right on their input , but may also keep their position on the tape . in
@xcite it is shown that the equivalence problem for these is undecidable .
the same was proved for 1way2tape quantum finite automata in @xcite .
[ conj : decidable ] the emptiness problem for probabilistic and quantum fst computing a relation with probability @xmath116 is decidable .
the emptiness problem for probabilistic and quantum fst computing a relation with a single letter input alphabet , with probability @xmath117 is decidable . to prove this
, we would like to apply a packing argument in the space of all total states , equipped with the above metric .
however , this fails because of the infinite volume of this space ( for finite automata it is finite , see @xcite and @xcite ) . in any case
, a proof must involve the size of the gap between the upper and the lower probability point , as the above theorem shows that it can not possibly work with gap @xmath118 .
still , we can prove : if the relation @xmath44 is computed by a pfst or a qfst with an isolated cutpoint , then @xmath119 is a recursive set ( so , for each specific output , it is decidable if it is ever produced above the threshold probability ) .
_ proof _ : let the cutpoint be @xmath45 , with isolation radius @xmath120 , and let @xmath121 .
define @xmath122 , the set of prefixes of @xmath123 .
consider the _ output@xmath123truncated total state _
, which is an element @xmath124 it is obtained from @xmath125 with @xmath126 by defining @xmath127 let us denote this transformation by @xmath128
. now observe that in the total state evolution of the qfst probability once put outside @xmath129 never returns , and likewise , amplitude once put outside @xmath130 never returns ( compare proof of theorem [ thm : qfa - c - qfst ] ) .
formally , this is reflected in the relation @xmath131 hence , if we want to know if @xmath132 for some @xmath133 , we may concentrate on the space of output@xmath123truncated total states , which is finite dimensional , and its transformation functions @xmath134
. it is easily seen that there is a constant @xmath65 such that for all truncated total states @xmath135 and all @xmath136 @xmath137 hence , for @xmath138 , if @xmath139 then @xmath140 because of the cutpoint isolation we find that either both or none of @xmath141 , @xmath142 is in @xmath44 .
now , because of compactness of the set of truncated total states reachable from the starting state , it follows that there is a constant @xmath143 such that for all @xmath144 of length @xmath145 one can write @xmath146 , with @xmath147 , such that @xmath148 hence @xmath149 and thus , if @xmath133 had produced @xmath123 with probability at least @xmath150 , so had the shorter string @xmath151 .
this means that we only have to consider input strings of length up to @xmath152 to decide whether @xmath153 .
obviously , this reasoning applies to pfst , too .
unlike the situation for finite automata , pfst are strictly more powerful than their deterministic counterparts : [ thm : mmm ] for arbitrary @xmath51 the relation @xmath154 can be computed by a pfst with probability @xmath155 .
it can not be computed by a dfst .
_ proof _ : the idea is essentially from @xcite : for a natural number @xmath156 choose initially an alternative @xmath157 , uniformly .
then do the following : repeatedly read @xmath156 @xmath13 s , and output @xmath158 @xmath159 s , until the @xmath14 s start ( remember the remainder modulo @xmath156 ) , then repeatedly read @xmath156 @xmath14 s , and output @xmath160 @xmath159 s .
compare the remainder modulo @xmath156 with what you remembered : if the two are equal , output this number of @xmath159 s and accept , otherwise reject .
it is immediate that on input @xmath161 this machine outputs @xmath162 with certainty .
however , on input @xmath163 each @xmath164 receives probability at most @xmath165 . that this can not be done deterministically is straightforward : assume that a dfst has produced @xmath166 @xmath159 s after having read @xmath167 @xmath13 s . because of finiteness there are @xmath168 such that after reading @xmath156 @xmath14 s ( while @xmath169 @xmath159 s were output ) the internal state is the same as after reading @xmath170 further @xmath14 s ( while @xmath171 @xmath159 s are output ) .
so , the output for input @xmath172 is @xmath173 , and these pairs are either all accepted or all rejected . hence they are all rejected , contradicting acceptance for @xmath174 . by observing that the random choice at the beginning can be mimicked quantumly , and that all intermediate computations are in fact reversible , we immediately get [ thm : mmm : q ] for arbitrary @xmath51 the relation @xmath175 can be computed by a qfst with probability @xmath155 .
note that this puts qfst in contrast to quantum finite automata : in @xcite it was shown that if a language is recognized with probability strictly exceeding @xmath176 then it is possible to accept it with probability @xmath14 , i.e. reversibly deterministically .
the relation @xmath177 can be computed by a pfst and by a qfst with probability @xmath98 .
_ proof _ : we do this only for qfst ( the pfst is obtained by replacing the unitaries involved by the stochastic matrices obtained by computing the squared moduli of the entries ) : let the input be @xmath178 ( other forms are rejected ) .
with amplitude @xmath109 each go to one of three ` subprograms ' : either copy @xmath133 to the output , or @xmath123 ( and accept ) , or reject without output .
this works by the same reasoning as the proof of theorem [ thm : empty1 ]
after seeing a few examples one might wonder if everything that can be done by a qfst can be done by a pfst .
that this is not so is shown as follows : [ thm : mnk : q ] the relation @xmath179 can be computed by a qfst with probability @xmath180 , for arbitrary @xmath51 .
[ thm : mnk ] the relation @xmath181 can not be computed by a pfst with probability bounded away from @xmath69 . in fact , not even with an isolated cutpoint . _ proof ( of theorem [ thm : mnk : q ] ) _ : for a natural number @xmath170 construct the following transducer : from @xmath182 go to one of the states @xmath183 , @xmath184 ( @xmath185 , @xmath186 ) , with amplitude @xmath187 for @xmath183 and with amplitude @xmath188 each , for the others .
then proceed as follows ( we assume the form of the input to be @xmath189 , others are rejected ) : for @xmath183 output one @xmath190 for each @xmath13 , and finally accept .
for @xmath184 repeatedly read @xmath170 @xmath13 s and output @xmath158 @xmath190 s ( remember the remainder @xmath191 ) .
then repeatedly read @xmath170 @xmath192 s and output @xmath193 @xmath190 s ( output nothing on the @xmath194 s ) .
compare the remainder with the one remembered , and reject if they are unequal , otherwise output this number of @xmath190 s .
reading @xmath195 perform the following unitary on the subspace spanned by the @xmath184 and duplicte states @xmath196 : @xmath197 accepting are all @xmath198 , rejecting are all @xmath199 .
now assume that the input does not occur as the left member in the relation : this means either @xmath200 and @xmath201 , or @xmath202 . in the first case all the outputs in each of the @xmath192branches of the program are of different length , so
get amplitude @xmath188 .
the final step combines at most two of them , so any output is accepted with probability at most @xmath203 .
the second case is more interesting : in all branches the amplitude is concentrated on the output @xmath204 .
the rotation @xmath32 however is made such that the amplitude on @xmath198 cancels out , so we end up in a rejecting state @xmath199 . in total , any output is accepted with probability at most @xmath205 . on the other hand , if the input occurs as the left member in the relation , exactly one of the two @xmath192branches of the program concentrates all amplitude on output @xmath204 , whereas the other spreads it to @xmath170 different lengths .
this means that the output @xmath204 is accepted with probability at least @xmath206 , and others are accepted with probability at most @xmath207 each . in total , the output @xmath204 is accepted with probability at least @xmath180 , all others are accepted with probability at most @xmath205 .
_ proof ( of theorem [ thm : mnk ] ) _ : by contradiction .
suppose @xmath181 is computed by a pfst @xmath43 with isolated cutpoint @xmath45 .
the following construction computes it with probability bounded away from @xmath69 : assuming @xmath208 ( the other case is similar ) , let @xmath209 .
run one of the following subprograms probabilistically : with probability @xmath18 output one @xmath190 for each @xmath13 , and ignore the other symbols ( we may assume that the input has the form @xmath189 ) , with probability @xmath210 run @xmath43 on the input .
it is easily seen that this new pfst computes the same relation with probability bounded away from @xmath69 .
hence , we may assume that @xmath43 computes @xmath44 with probability @xmath211 , from this we shall derive a contradiction .
the state set @xmath4 together with any of the stochastic matrices @xmath212 is a markov chain .
we shall use the classification of states for finite markov chains ( see @xcite ) : for @xmath213 @xmath4 is partitioned into the set @xmath214 of _ transient _ states ( i.e. the probability to find the process in @xmath214 tends to @xmath13 ) and a number of sets @xmath215 of _ ergodic _ states ( i.e. once in @xmath215 the process does not leave this set , and all states inside can be reached from each other , though maybe only by a number of steps ) .
each @xmath215 is divided further into its _
classes @xmath216 ( @xmath217 ) , @xmath213 mapping @xmath216 into @xmath218 .
by considering sufficiently high powers @xmath219 ( e.g. product of all the periods @xmath220 ) as transition matrices , all these cyclic sets become ergodic , in fact , @xmath219 restricted to each is _
regular_. using only these powers amounts to concentrating on input of the form @xmath221 , with @xmath222 , which we will do from now on .
relabelling , the ergodic sets of @xmath223 will be denoted @xmath215 .
each has its unique equilibrium distribution , to which every initial one converges : denote it by @xmath224 .
furthermore , there are limit probabilities @xmath225 to find the process @xmath226 in @xmath227 after long time , starting from @xmath182 . likewise , there are limit probabilities @xmath228 to find the process @xmath229 in @xmath230 after long time , starting from @xmath231 , and similarly @xmath232 .
so , by the law of large numbers , for large enough @xmath233 the probability that @xmath226 has passed into @xmath227 after @xmath234 steps , after which @xmath229 has passed into @xmath230 after @xmath235 steps , after which @xmath236 has passed into @xmath237 after @xmath238 steps , is arbitrarily close to @xmath239 .
( note that these probabilities sum to one ) . as a consequence of the ergodic theorem ( or law of large numbers ) , see @xcite , ch .
4.2 , in each of these events @xmath240 the probable number of @xmath190 s written after the final @xmath195 , is linear in @xmath233 : @xmath241 } |{\bf 0}^m{\bf 1}^n{\bf 2}^k , j)\rightarrow 1,\ ] ] as @xmath242 , with @xmath243 and non negative constants @xmath244 .
since we require that for @xmath245 @xmath246 it is necessary that for a set @xmath247 of events @xmath240 @xmath248 in fact , as for @xmath249 @xmath250 for certain sequences @xmath251 , we even have @xmath252 for @xmath253 it is obvious that the transducer outputs _ no more _ @xmath190 s , once in @xmath237 . but
this implies that for @xmath254 large enough , @xmath255 is arbitrarily close to @xmath256 , hence @xmath257 which implies that @xmath258 contradicting @xmath259 . in general however , computing with isolated cutpoint
is strictly weaker than with probability bounded away from @xmath69 ( observe that for finite automata , probabilistic and quantum , recognizability with an isolated cutpoint is equivalent to recognizability with probability bounded away from @xmath69 , see theorem [ thm : qfa - c - qfst ] ) : [ thm : cutpoint ] the relation @xmath260 can be computed by a pfst and by a qfst with an isolated cutpoint ( in fact , one arbitrarily close to @xmath69 ) , but not with a probability bounded away from @xmath69 .
_ proof _ : first the construction ( again , only for qfst ) : initially branch into two possibilities @xmath261 , each with amplitude @xmath262 .
assume that the input is of the correct form ( otherwise reject ) , and in state @xmath263 output one @xmath264 for each @xmath265 , ignoring the @xmath266 s .
then , if @xmath267 , accept , if @xmath268 , reject .
it is easily seen that @xmath269 is accepted with probability @xmath69 if @xmath270 , and with probability @xmath13 otherwise . that this can not be done with probability above @xmath69
is clear intuitively : the machine has to produce some output ( because of memory limitations ) , but whether to output @xmath271 or @xmath272 it can not decide until seeing the last symbol .
formally , assume that @xmath273 , with @xmath274 .
if @xmath275= t(4^m|0^m1^n2)\geq 1/2+\delta,\ ] ] necessarily @xmath276 + t_{\ddag 0^m1^n}((q_0,\epsilon),{\bf 0},0)[q_{\rm non}\times 4^{[m-2t , m+2 t ] } ] \geq 1/2+\delta.\ ] ] but this implies @xmath277 + t_{\ddag 0^m1^n}((q_0,\epsilon),{\bf 0},0)[q_{\rm non}\times 4^{[n-2t , n+2 t ] } ] \leq 1/2-\delta,\ ] ] hence @xmath278= t(4^n|0^m1^n3)\leq 1/2-\delta,\ ] ] contradicting @xmath279 . to conclude from these examples , however , that quantum is even better than probabilistic ,
would be premature : [ thm : lastsymbol ] the relation @xmath280 can not be computed by a qfst with an isolated cutpoint .
( obviously it is computed by a pfst with probability @xmath14 , i.e. a dfst ) .
_ proof _ : the construction of a dfst computing the relation is straightforward . to show that no qfst doing this job exists , we recall from @xcite that @xmath281 is not recognized by a @xmath14way quantum finite automaton with probability bounded away from @xmath69 , and use theorem [ thm : qfa - c - qfst ] for this language .
we introduced quantum finite state transducers , and showed some of their unique properties : undecidability of the emptiness problem , as opposed to deterministic finite state transducers and finite automata , and incomparability of their power to that of probabilistic and deterministic finite state transducers .
as open questions we would like to point out primarily our conjecture [ conj : decidable ] .
another interesting question is whether a relation computed by a qfst with probability _ sufficiently close to @xmath14 _ can be computed by a pfst .
this would be the closest possible analog to the `` @xmath176theorem '' from @xcite .
research of rf supported by contract ist199911234 ( qaip ) from the european commission , and grant no .
96.0282 from the latvian council of science .
this work was carried out during rf s stay at bielefeld university in october 2000 . at this time aw
was at fakultt fr mathematik , universitt bielefeld , germany , and was supported by sfb 343 `` diskrete strukturen in der mathematik '' of the deutsche forschungsgemeinschaft .
r. bonner , r. freivalds , r. gailis , `` undecidability of 2tape quantum finite automata '' , in proceedings of _ quantum computation and learning .
sundbyholms slott , sweden , 2729 may , 2000 _
, r. bonner and r. freivalds ( eds . ) , malardalen university , 2000 , pp . 93100 . | we introduce _ quantum finite state transducers ( qfst ) _ , and study the class of relations which they compute .
it turns out that they share many features with probabilistic finite state transducers , especially regarding undecidability of emptiness ( at least for low probability of success ) .
however , like their ` little brothers ' , the quantum finite automata , the power of qfst is incomparable to that of their probabilistic counterpart .
this we show by discussing a number of characteristic examples . |
the search for dark matter is currently one of the most active fields of research in experimental astroparticle physics .
common candidates are weakly interacting massive particles ( wimps ) , with only weak and gravitational interactions with normal matter , and which encompass a variety of particle types in the mass range from a few tens of gev to a few hundred tev , where the upper limit is based on theoretical arguments to preserve unitarity @xcite . among wimp candidates
are the lightest neutralino arising in the minimal supersymmetric extension of the standard model ( mssm ) , or the lightest kaluza - klein mode in models of universal extra dimensions ( ued ) @xcite .
these candidates are thermal relics from the big bang , and since they are stable , assumed to be able to contribute to the dark matter content in the halos of galaxies . in a different scenario
, it has been shown that super - massive particles can be produced in the early universe and account for dark matter , independently of their interaction strength with normal matter . in order to avoid the mass limit imposed by unitarity constraints
, one can model non thermal production of super - massive particles @xcite .
the large mass will prevent the particle to ever get into thermal equilibrium with the primordial plasma .
super - massive dark matter candidates were coined wimpzillas @xcite if assumed to interact weakly with matter and simpzillas @xcite if the interaction is strong . the production mechanism discussed in @xcite favours particles with masses of the order of the inflaton mass ( @xmath310@xmath4 gev ) , but particles of masses as low as a few hundred gev
can also be produced , normally denoted as strongly interacting massive particles ( simps ) .
although simps are in general taken to be in the mass range @xmath310@xmath5 m@xmath610@xmath7 gev , and simpzillas , with masses m@xmath810@xmath7 gev , we will generically call them either simpzillas . searches for simpzillas have been carried out in a variety of ways in the past .
figure [ fig : pexc ] shows the current exclusion region in the mass versus simpzilla nucleon cross section parameter space .
the `` mine / space '' region ( blue square hatched ) was ruled out by many different experiments underground or space - borne @xcite .
more recently , a search for simpzillas using results from direct detection experiments excluded most of the remaining parameter space for high masses ( solid yellow , labeled `` direct '' ) @xcite . a recent analysis further constrained the allowed region by using the earth heat flow to set constrains on the annihilation of captured dark matter in the earth s center ( green striped , labelled `` earth heat '' ) @xcite .
the blank triangular regions in figure [ fig : pexc ] remained unprobed until this work . whether thermally produced or not , stable dark matter candidates can gravitationally accumulate as a halo in our galaxy , becoming bound in orbits in the solar system .
further energy losses can occur in elastic interactions with matter in large celestial bodies , like the sun .
if they lose enough energy they can fall below the escape velocity of the object , accumulating in its center and annihilating therein @xcite .
neutrinos of energies _ o_(100 gev ) can be produced from the decays of the annihilation products .
this allows for an indirect search for dark matter with neutrino telescopes , based on a search for an excess neutrino flux from the direction of the sun over the known atmospheric neutrino background .
predictions of simpzilla event rates and neutrino telescopes sensitivity have been anticipated in @xcite . in this paper
we explore the capabilities of icecube to detect high - energy neutrinos from simpzilla annihilations in the sun .
the baseline icecube geometry consists of 80 strings with 60 digital optical modules each , instrumenting 1 km@xmath9 of ice at depths between 1450 m2450 m near the geographic south pole @xcite .
recently , an addition of six strings forming a denser core in the middle of the icecube array has been proposed in order to lower the neutrino energy threshold of the detector to about 10 gev , the so called `` deepcore '' array @xcite .
although the icecube geometry has been optimized to detect ultra - high energy ( @xmath10tev ) neutrinos from potential cosmic sources with sub - degree angular resolution , its angular response is still adequate at _ o_(100 gev ) to perform directional searches for dark matter .
icecube has recently published results on the searches for wimps arising in the mssm @xcite or the simplest ued model @xcite using 22 strings deployed in the 2006/2007 antarctic season . in this work
we use this 22-string detector ( hereafter referred to as icecube-22 ) which took data during the 2007 austral winter .
we set a limit in the simpzilla mass , m@xmath11 , versus simpzilla - nucleon cross section , @xmath12 , phase space using the fact that icecube-22 has not detected any neutrino excess from the direction of the sun .
we also estimate the sensitivity of the full icecube detector in its expected final configuration with 86 strings ( hereafter referred to as icecube-86 ) .
in order to estimate the neutrino rate from simpzilla annihilations in the sun , we use the capture rate , as well as the full - flavour neutrino flux and energy spectrum at the core of the sun as determined in @xcite .
we then simulate the neutrino propagation to the earth , including energy losses and oscillation effects .
the capture rate , @xmath13 , depends on the mass of the simpzilla , m@xmath11 , and the strength of the interaction of simpzillas with nucleons , @xmath12 .
the capture efficiency can be described by the parameter @xmath14 , where m@xmath15 is the average nucleon mass , n@xmath15 is the number density of nucleons in the sun and r@xmath16 is the radius of the sun . for large cross sections , q
@xmath17 1 , and simpzillas lose enough energy in their passage through the sun to be efficiently captured .
the capture rate ( in s@xmath18 ) is then given by @xmath19 on the other hand , if q@xmath101 , capture is determined by the relative velocity , and only low velocity simpzillas are trapped .
in such case , the capture rate is @xmath20\left(\frac{10^{12}\ , \text{gev}}{m_\text{\tiny x}}\right)\,\times \\ & \times \left ( \frac{\text{u}_{th}}{240\ , \text{km}\ , \text{s}^{-1}}\right ) \,\left ( \frac{r_\odot}{7\times 10^{10 } \text{cm}}\right)^2 \\ \end{split } \label{eq : gamma2_c}\ ] ] where in the above expressions @xmath21 and u@xmath22 is the speed of the sun in the galaxy and @xmath23 is the sun escape velocity . note that if @xmath12 goes to zero , then the capture rate goes to zero through its dependency on @xmath24 , as it is expected .
the simpzilla capture rate is large enough to reach equilibrium with the annihilation rate , @xmath25 , in the lifetime of the solar system , and therefore we can assume @xmath26 . + the neutrino spectrum from simpzilla annihilations is modeled as producing a pair of quarks or gluons , which will fragment into high multiplicity hadronic jets .
long - lived light and charmed hadrons will lose energy through interactions in the dense solar interior before decaying , producing low energy neutrinos .
on the other hand , due to its short lifetime , @xmath27 s , the top quark will not have time to hadronize or lose energy through interaction in the medium before decaying into w@xmath28 .
the w then decays into @xmath29 , with similar probability into each flavour @xmath30 .
annihilations into top quarks is therefore a promising channel to produce high energetic neutrinos detectable by neutrino telescopes .
bottom quarks are also a potential source of high energy neutrinos @xcite .
however in this work we only consider neutrinos produced from the top decay chain , which makes our results conservative .
note that although the initially available energy for each top is very high ( the mass m@xmath31 of the simpzilla ) , the available phase space is shared among many produced particles ( 2.8@xmath3210@xmath33 tops are produced per annihilation @xcite , where @xmath34 is the mass of the simpzilla in units of 10@xmath4 gev ) , so the average energy per produced particle is of _ o _ ( tev ) .
the main consequence of this is that initial - state gluon radiation can be neglected , not playing an important role in further degrading the energy of the tops before they decay .
this is illustrated in figure [ fig : e_t ] , which shows the energy distribution of the tops at the production point ( dashed line ) and just before decaying ( full line ) , obtained with a full ` pythia ` @xcite simulation . taking into account the above considerations
, the energy distribution of the neutrinos produced from top decays at the center of the sun can be parametrized as @xmath35[(e_{\nu}+\text{m}_{\mbox{\tiny w}})^2-\text{m}_{\mbox{\tiny w}}^2]}}\\ \label{eq : parametrization}\ ] ] where m@xmath36 and m@xmath37 are the w boson and top quark masses respectively , and we have omitted a constant normalization factor .
the energy dependence of the neutrino flux at the detector ( earth ) will be modified by neutrino energy losses in their way out of the sun and by oscillations . in order to include these effects we use the publicly available wimpsim code @xcite , which we have modified so it takes the neutrino spectrum from equation ( [ eq : parametrization ] ) .
we use as input the three flavours with their correct relative intensity from the top decay chain . figure [ fig : nuflux ] shows the injection spectrum of muon neutrinos at the sun from simpzilla annihilations according to equation ( [ eq : parametrization ] ) ( dashed line ) , and the resulting spectrum at the earth ( full line ) .
the injection spectrum has been truncated at 10 tev for display purposes .
as mentioned earlier , the initial energy available in the annihilation is shared among an enormous amount of annihilation products and , furthermore , the energy of the resulting neutrinos is further reduced by energy losses in the sun .
the spectrum at the detector is therefore not very different from those of lighter dark matter candidates , like mssm neutralinos or kaluza - klein modes .
this allows us to use icecube-22 background determination found in both wimp and kaluza - klein dark matter analysis @xcite .
since icecube is mainly sensitive to muon neutrinos through the detection of muon tracks , we only use the muon - neutrino flux at the detector for our calculations in what follows .
the number of signal events , n@xmath38 , from simpzilla annihilations in the sun predicted in a neutrino telescope of effective area a@xmath39 during an exposure time t , is given by @xmath40 where @xmath41 is the muon neutrino flux at the detector , @xmath42 is the number of tops per annihilation , @xmath43 is the branching ratio of w decaying into neutrinos ( 0.326 ) and @xmath25 is the annihilation rate in the center of the sun . the effective area is a measure of the efficiency of the detector and includes the neutrino - nucleon interaction probability , the energy loss of the produced muon from the interaction point to the detector and the detector trigger and analysis efficiency . in order to compare our expected signal with icecube-22 observations
, we use the neutrino effective area of icecube-22 published in @xcite , and shown in figure [ fig : icecube_aeff ] . in 104.3 days of live time icecube-22
has detected 13 events from the direction of the sun , while the expected number of atmospheric neutrinos is 18.5 @xcite .
the expected signal strength for the given live time is shown in figure [ fig : evr ] . we have assumed a 15% uncertainty in the expectation of the atmospheric neutrino flux at the relevant energies ( e@xmath4450 gev ) in order to calculate the 90% cl upper limit on a possible signal , @xmath45 , following the prescription in @xcite .
we obtain @xmath45 = 3.1 . a scan in the ( m@xmath11 , @xmath12 ) parameter space , selecting models that predict less events than @xmath45 leads to the 90% cl exclusion region shown in figure [ fig : exclusion ] ( read cross hatched area ) .
the results disfavour a large region of the parameter space , and we discuss their significance in section [ sec : conclusions ] .
the sensitivity of the completed icecube detector can be estimated from the expected number of background events ( due to atmospheric neutrinos ) and the detector effective area .
an estimation of the expected effective area of icecube-86 at trigger level has been given in @xcite .
previous analyses of icecube and amanda show that the effective area is reduced between a factor of 5 to 10 between trigger level and the final analysis level , due to the cuts that are applied to clean the data sets and select high quality tracks .
we have taken a conservative approach here and we have reduced the trigger effective area given in @xcite by a factor between 15 for low energies down to 5 at the higher energies , and used that as our estimate of where the effective area of icecube-86 at a final analysis level might be .
the curve used is shown as the upper curve in figure [ fig : icecube_aeff ] .
we have calculated the number of atmospheric neutrino events in icecube using the measured atmospheric neutrino flux by amanda @xcite .
we parametrized the curve in figure 10 in @xcite as dn@xmath46/ded@xmath47 = a e@xmath48 gev@xmath18 @xmath49 s@xmath18 sr@xmath18 , with a=0.11407 and @xmath50 = 3.142 .
the expected number of atmospheric neutrino events in icecube-86 from the direction of the sun in a exposure time t is then given by @xmath51 , where the angular integration is performed over the 0.5@xmath52 that the sun subtends in the sky .
we have assumed no contamination of misreconstructed atmospheric muons in the final sample , and therefore we use only the 3.4 atmospheric neutrino events from the direction of the sun predicted by the integral above in 365 days live time . as in the previous section ,
we have used a 15% uncertainty on the overall normalization of the atmospheric neutrino flux , as also indicated by the amanda measurement , and calculated the sensitivity including this uncertainty .
the 90% cl exclusion region from icecube-86 is shown in figure [ fig : exclusion ] as the black line .
it agrees well with the predictions made in @xcite .
we have used recently published results from the 22-string icecube detector on wimp dark matter searches @xcite to extend the search to strongly interacting dark matter candidates in the mass range @xmath310@xmath5 m@xmath610@xmath2 gev .
the fact that the icecube results are compatible with the expected atmospheric neutrino background allows us to set restrictive limits on the mass versus nucleon cross section parameter space .
a note on the complementarity of direct and indirect searches is due at this point .
generally , the total cross section with nucleons , @xmath12 , can have at least two contributions , a spin - independent ( or scalar ) component and a spin - dependent ( or axial - vector ) component , the relative strength of each depending on the exact assumptions made on the structure and interaction of the simpzilla .
there is no detailed modelling in the literature of the way that simpzillas interact with matter , other than the interaction is assumed to be `` strong '' , as opposed to weak as in the case of wimps .
however for experimental searches such distinction is important because different experiments probe different components of the cross section .
the results from the experiments shown in figure [ fig : pexc ] refer mainly to the spin - independent component of the simpzilla - nucleon cross section , due to the nature of the targets involved or , in the case of the earth heat flow analysis , due to the fact that the capture of dark matter candidates in the earth proceeds mainly through scattering on spinless nuclei ( fe , si , o , mg ) . on the other hand
the icecube results are sensitive to the spin - dependent cross section since the sun is the target for capture , which is essentially a proton target . in view of the previous considerations ,
the current results from icecube provide the most restrictive limits so far on the spin - dependent cross section of simpzillas .
the fact that they overlap in the ( m@xmath11 , @xmath12 ) parameter space with previous exclusion regions from experiments which probe mainly the spin - independent cross section , practically rules out simpzillas as dark matter .
it would require an unreasonably narrow fine tuning of the mass and the structure of the cross section to propose a viable candidate that would evade all the current limits . in conclusion , strongly interacting heavy relics from the early universe can only account for the dark matter if they have masses above @xmath53 gev .
as these ultra heavy masses are not currently favored by any model , the window for simpzillas as the only component of dark matter can be considered closed .
+ we thank t. sjstrand for his kind help with ` pythia ` .
ia was partially funded by the u.s .
department of energy under contract number de - ac02 - 07ch11359 and the brazilian national counsel for scientific research ( cnpq ) 100 k. griest and m. kamionkowski .
lett . * 64 * , 615 ( 1990 ) .
d. hooper and s. profumo .
phys . rep . *
453 * , 29 ( 2007 ) .
d. j. h chung , e. w. kolb and a. riotto .
phys . rev . * d59 * , 023501 ( 1998 ) .
d. j. h chung , e. w. kolb and a. riotto .
phys . rev . * d60 * , 063504 ( 1999 ) .
d. j. h chung , e. w. kolb and a. riotto .
lett . * 81 * , 4048 ( 1998 ) . i. f. m. albuquerque , l. hui and e. w. kolb .
phys . rev . * d64 * , 083504 ( 2001 ) .
p. c. mcguire and p. j. steinhardt .
proceedings of the 27th international cosmic ray conferences ( icrc 2001 ) , hamburg , germany , aug 2001 . `
astro - ph/0105567 ` .
g. d. starkman , a. gould , r. esmailzadeh and s. dimopoulos , phys .
d * 41 * , 3594 ( 1990 ) .
i. f. m. albuquerque and l. baudis , phys .
lett . * 90 * , 221301 ( 2003 ) .
g. d. mack , j. f. beacom and g. bertone , phys . rev . *
d76 * , 043523 ( 2007 ) .
w. h. press and d. n. spergel .
j. * 296 * , 769 ( 1985 ) ; t. k. gaisser , g. steigman and s. tilav .
phys . rev .
* d34 * , 2206 ( 1986 ) .
j. * 328 * , 919 ( 1988 ) .
i. f. m. albuquerque , j. lamoureux and g. f. smoot , phys .
* d66 * , 125006 ( 2002 ) .
j. ahrens _
et al_. astropart .
phys.*20 * , 507 , ( 2004 ) c. wiebusch _
et al_. proceedings of the 31st international cosmic ray conferences ( icrc 2009 ) , lodz , poland , july 2009 . `
arxiv:0907.2263 ` .
r. abassi _
et al_. phys .
lett . * 102 * , 201302 ( 2009 ) .
r. abassi _
et al_. arxiv:0910.4480 .
t. sjstrand , s. mrenna and p. skands , jhep05 , 025 , ( 2006 ) .
j. edsj , wimpsim neutrino monte carlo , w. a. rolke , a m. lpez and j. conrad .
a551 * , 493 - 503 ( 2005 ) .
_ et al_. phys . rev .
* d79*,102005 ( 2009 ) . | we use the recent results on dark matter searches of the 22-string icecube detector to probe the remaining allowed window for strongly interacting dark matter in the mass range 10@xmath0m@xmath110@xmath2 gev .
we calculate the expected signal in the 22-string icecube detector from the annihilation of such particles captured in the sun and compare it to the detected background . as a result
, the remaining allowed region in the mass versus cross section parameter space is ruled out .
we also show the expected sensitivity of the complete icecube detector with 86 strings . |
strontium ruthenate is an intriguing low t@xmath2 superconductor @xcite .
it is widely believed to have spin triplet order parameter .
the orbital symmetry of it , however , is still unknown .
knight shift @xcite and spin susceptibility @xcite being constant below @xmath4 point to the chiral state with d - vector @xmath5 .
this state ( called ( a ) in the following ) is also consistent with the @xmath6-sr experiments which show spontaneous time reversal symmetry breaking at @xmath4 @xcite .
it was recently confirmed in phase sensitive experiments @xcite .
the original knight shift measurements were performed for magnetic fields aligned in the ruthenate a - b plane .
recently the knight shift @xcite has been measured in magnetic field parallel to the c - axis . in contradiction to expectations it was found that the susceptibility is _ also unchanged _ from the normal state value below @xmath4 .
it was noted@xcite that this result would be consistent with the assumption that the @xmath7 vector rotates away from the @xmath8 direction under the influence of the external field .
however a simple rotated state , such as @xmath9 , is not allowed in tetragonal symmetry . instead , the symmetry allowed states @xmath10 and @xmath11 ( which below we refer to as @xmath12 and @xmath13 respectively ) would also explain the data as the spin susceptibility for either of these states is of the form @xcite @xmath14 corresponding to a constant spin susceptibility for c - axis fields consistent with the results of murakawa _ et al .
_ @xcite .
it is the purpose of this work to study in some detail the scenario of d - vector rotation in the c - axis magnetic field .
specifically , we shall study the possible effect of the weak external ( zeeman only ) magnetic field on the symmetry of the lowest energy pairing state . to this end
we use a recently proposed @xcite phenomenological three - band , three dimensional model with a realistic band structure @xcite .
it takes into account the three ruthenium @xmath15 orbitals @xmath16 , @xmath17 and @xmath18 .
the hopping integrals @xmath19 for orbitals @xmath20 and @xmath21 and ru sites @xmath22 and @xmath23 as well as the site energies @xmath24 are fitted to reproduce the experimentally determined fermi surface @xcite .
the effective spin and orbital dependent attractive hubbard parameters @xmath25 are chosen to give correct transition temperature . without spin - orbit coupling the model with only two values of interaction parameters : an in - plane @xmath26 and out - of - plane @xmath27 results in the chiral state @xmath28 , as the ground state of the system .
for spin - independent interactions the alternative pairing states @xmath12 and @xmath13 are the ground state for any value of @xmath29 . to get the state ( a ) as the ground state in the presence of non - zero spin - orbit interaction @xmath30 one has to take into account that the spin - orbit coupling leads to a small spin dependence of the effective pairing interaction @xcite .
choosing @xmath31 about 1% larger than @xmath32 is sufficient to stabilize the chiral state even for large spin orbit coupling . in fig .
[ fig1 ] we show the free energies of the @xmath28 @xmath12 and @xmath13 symmetry pairing states as a function of external magnetic field @xmath33 .
the model parameters were chosen to make the chiral state @xmath28 stable at zero field .
one can see that the chiral phase increases its free energy as the field is increased , until at a certain critical field the @xmath12 or @xmath13 solutions become more stable .
therefore we expect a `` spin flop '' type phase transition from a d - vector oriented along the @xmath34-axis to one where the d - vector lies in the @xmath35 plane .
a similar rotation is observed in the bulk superfluid state of @xmath36-a where the d - vector rotates continuously to remain perpendicular to the applied field @xcite . in the case of sr@xmath0ruo@xmath1
the transition is not simply a rotation of the chiral d - vector but is also a transition from a chiral to non - chiral pairing state of different symmetry .
the two distinct solutions shown in fig . [ fig1 ] ( note that @xmath12 and @xmath13 are essentially degenerate ) have different entropies , as shown in the figure inset , and hence this freedericksz - like @xcite spin - flop is a first order thermodynamic phase transition . it has not yet been observed , because the specific heat jump at this transition is very small in comparison with the main jump at 1.5k , as our ( unpublished ) calculations indicate .
00 y. maeno , t.m .
rice and m. sigrist , physics today * 54 * , 42 ( 2001 ) ; a.p .
mackenzie , y. maeno , rev .
* 75 * , 657 ( 2003 ) ; i. eremin , d. manske , s.g .
ovchinnikov , j. f. annett , ann .
( leipzig)*13 * , 149 ( 2004 ) . | the superconductor sr@xmath0ruo@xmath1 is widely believed to be a spin triplet system with a chiral order parameter analogous to the a phase of superfluid helium-3 .
the best evidence for this pairing state is that the knight shift or spin susceptibility measured in neutron scattering is constant below t@xmath2 , unlike in a spin - singlet superconductor .
the original knight shift and neutron scattering measurements were performed for magnetic fields aligned in the ruthenate a - b plane .
these would be consistent with a triplet d - vector @xmath3 aligned along the c - axis .
however recently the knight shift for fields along c was also found to be constant below t@xmath2 , which is not expected for this symmetry state . in this paper
we show that while spin - orbit interaction stabilises the c - axis oriented d - vector , it is possible that only a very small external b field may be sufficient to rotate the d - vector into the a - b plane . in this case
the triplet pairing model remains valid .
we discuss characteristics of the transition and the prospects to detect it in thermodynamic quantities .
ruthenate superconductors , pairing symmetry , d - vector rotation 74.20.-z , 74.20.rp , 74.70.pq , 74.25.bt |
this paper is a follow - up of @xcite and establishes wegner - type ( more precisely , wegmer stollmann - type ) bounds for random continuous schrdinger operators .
we focus here on a two - particle interactive anderson model in a continuous space , subject to a random external field with a continuous argument .
the infinite - volume hamiltonian of the model is a schrdinger operator @xmath2 acting on functions @xmath3 : @xmath4 here @xmath5 is the kinetic energy operator : @xmath6 where @xmath7 is the laplacian in variable @xmath8 corresponding to the @xmath9th particle : @xmath10 the randomness in hamiltonian @xmath11 is concentrated in the potential energy function @xmath12 which is written in the form @xmath13 the term @xmath14 represents the ( nonrandom ) interaction potential while the sum @xmath15 $ ] describes the action of an external potential field on the two - particle system .
more precisely , the external potential field is represented by a family of real random variables ( rvs ) @xmath16 , or , in probabilistic terminology , by a real - valued random field with a continuous argument ( or simply ) .
a sample of such a random field is a ( measurable ) function @xmath17 , and its its graph is a ` hypersurface ' in @xmath18 , with a unique point of intersection with any straight line orthogonal to the ` argument space ' @xmath0 .
the argument @xmath19 in the notation @xmath20 stresses randomness of the external potential .
the distribution of the @xmath21 is denoted by @xmath22 : it is a probability measure on a ( suitably chosen ) sigma - algebra @xmath23 in the space of sample functions @xmath24 .
formal definitions are given in section [ sectextpot ] ; cf .
a classical example is a gaussian distribution on @xmath23 ; see , e.g. , @xcite .
the randomness introduced in hamiltonian @xmath25 perplexes its spectral properties .
a physically - motivated result expected here is that the spectrum of @xmath25 near its ` lower edge ' is pure point with probability one , and the corresponding eigenfunctions decay exponentially in space .
( this is called lifshits - tail - based exponential localisation . )
similar results have been established for a variety of single - particle models where the hamiltonian acts on a function @xmath26 as @xmath27 , @xmath28 .
see @xcite , @xcite and references therein . as far as localisation is concerned
, the main difference between a single - particle model and its two - particle counterpart is that replacing term @xmath20 by the sum @xmath29 leads to ` strong ' dependencies between rvs @xmath12 and @xmath30 , no matter how far points @xmath31 are positioned from each other .
the ensuing difficulty can not be relieved no matter how quickly correlations between rvs @xmath20 and @xmath32 decay when @xmath33 are far from each other .
@xcite where this discussion has been conducted for the so - called tight - binding two - particle anderson model on a lattice
. an important ingredient of the existing localisation proofs for single - particle anderson models or their two - particle counterparts is the so - called wegner - type ( or wegner stollmann type ) estimates for the corresponding hamiltonian .
[ exceptions are one - dimensional models where the whole physics of localisation is rather special . ]
this explains the title of this paper .
the wegner - type estimates are produced for the eigen - values of a finite - volume approximation of the hamiltonian ; cf .
@xcite . in our situation , these are operators @xmath34 acting on functions @xmath35 : @xmath36 with @xmath37 and @xmath12 defined , as before , by equation ( 1.2.2 ) . here
@xmath38 is a bounded domain in @xmath39 which we choose to be the cartesian product of two @xmath40-dimensional cubes : for @xmath41 and @xmath42 , @xmath43 where , for @xmath44 and @xmath45 , @xmath46 .
\eqno ( 1.8)\ ] ] the shorthand notation @xmath38 for @xmath47 ( and @xmath48 for @xmath49 ) will be systematically used in this paper .
further , @xmath50 and @xmath51 will stand for cubes @xmath52 and @xmath53 forming the projections of ` two - particle ' parallelepiped @xmath54 onto single - particle configurational spaces ( and similarly with @xmath55 and @xmath56 ) .
next , @xmath57 is the laplacian in variable @xmath58 confined to cube @xmath59 , @xmath60 , with a specified boundary condition which we choose to be dirichlet s : @xmath61 in fact , a possible choice of boundary conditions is rather broad and includes periodic and ` elastic ' , in particular , neumann s .
( what we need for our method to work is that the self - adjoint extension specified by a boundary condition is half - bounded and has a compact resolvent @xmath62 for @xmath63 away from the real line . )
see @xcite , @xcite for details .
thus , operator @xmath64 in eqn ( 1.6 ) is understood in terms of the sesquilinear form @xmath65 in @xmath66 , where @xmath67 stand for the standard inner product .
a similar meaning is attributed to the multiplication operator by @xmath12 in @xmath66 ; consequently , we treat the function @xmath12 , @xmath68 , as an element of @xmath66 defined for @xmath22-a.a .
additional conditions imposed below mean that this function is bounded on @xmath38 , and the supremum @xmath69 of its absolute value has certain moments ; see below .
moreover , under the assumptions introduced in this paper , the ( self - adjoint ) operator @xmath70 is bounded from below and has , with probability one , a discrete spectrum of a finite multiplicity .
it is convenient to write its eigenvalues @xmath71 in an increasing order : @xmath72 in the ` one - volume ' wegner estimate one assesses the probability that at least one eigenvalue @xmath73 of operator @xmath70 falls in a ( narrow ) interval around a given point @xmath74 on the spectral axis : @xmath75 next , the ` two - volume ' wegner estimate addresses the probability that the eigenvalues @xmath73 and @xmath76 of operators @xmath77 and @xmath78 come near to each other in a given interval @xmath79 , for two ( distant ) parallelepipeds @xmath80 and @xmath81 .
that is , @xmath82 here and below , @xmath22 stands for the corresponding probability measure on the underlying probability space ( see section [ sectextpot ] ) . from the probabilistic point of view , the estimates for probabilities ( 1.11 ) and ( 1.12 ) are examples of concentration inequalities , albeit for rather implicit rvs @xmath73 carrying a considerable amount of dependence .
for single - particle anderson models , assuming natural properties of the random term @xmath20 , the wegner estimates are rather straightforward . for the two - particle models , the study of these estimates , with a view of localisation , started in @xcite , @xcite , @xcite and
is continued in this paper . in the main body of this work
, the interaction potential @xmath83 satisfies the following property : \(i ) _ @xmath83 is a ( measurable ) bounded real function @xmath84 obeying _ @xmath85 here
@xmath86 is the interaction radius , @xmath87 stands for the permutation of the ( vector ) variables ( @xmath88 for @xmath89 ) and @xmath90 denotes the max - norm in @xmath0 : for @xmath91 , @xmath60 : @xmath92 \eqno ( 1.14)\ ] ] we can also allow the case where @xmath83 has a hard core , i.e. , @xmath93 with @xmath94 being the diameter of the hard core .
however , we have to assume that @xmath95 remains uniformly bounded for @xmath96 with @xmath97 . in section [ sectextpot ] ,
we give formal conditions upon the structure of the potential energy terms in hamiltonian ( 1.5 ) .
a useful notion is the ` shadow ' @xmath98 of a parallelepiped @xmath99 : @xmath100 it can be a cube or a union of two cubes in @xmath0 , possibly disjoint . in what follows
we call sets of this kind ` cellular ' . throughout the paper
, @xmath101 stands for the ( euclidean ) volume of parallelepiped @xmath102 and @xmath103 for that of a projection cube @xmath104 .
in addition , we use a similar notation @xmath105 for a cellular set @xmath106 . finally , note that an extension of the results of the present paper to the general case with @xmath107 particles is also possible . to this end
, one needs to apply the technique proposed recently in @xcite .
we plan to publish such an extension in a forthcoming paper .
a common model of a is a gaussian random field on @xmath0 , with its characteristic ` linearity structure ' ; it also serves as a ` base ' for producing wider families of .
our presentation will take this fact into account : comments on properties of the external potential field will be made from a ` gaussian perspective ' .
however , the class of the external fields under consideration is much larger and includes various ` perturbations ' of and ` operations ' with gaussian .
we will not venture in this direction in the current paper but plan to address this issue elsewhere .
a convenient way to describe our assumptions on @xmath21 is as follows .
fix a ( measurable ) function @xmath108 which is ( strictly ) positive - definite : @xmath109 ( measurable ) function @xmath110 , the lebesgue integral @xmath111 and @xmath112 unless @xmath113 a.e .. in these inequalities we allow the integral in the lhs of ( 2.1 ) to equal @xmath114 .
however , given a cellular set @xmath115 , we denote by @xmath116 the set of functions @xmath117 with support in @xmath118 and finite @xmath119 : @xmath120 equipped with the inner product @xmath121 and the norm @xmath122 .
we will often suppose that an orthonormal basis in @xmath116 has been given , @xmath123 , where @xmath124 here and below , @xmath125 stands for the indicator function of set @xmath126 and the normalising constant is given by @xmath127 in the case where @xmath128 admits a bound @xmath129 where @xmath130 , we have that @xmath131 .
our first condition on @xmath21 is : \(s ) _ summability _ : @xmath109 cellular set @xmath106 and @xmath132 : @xmath133 consequently , we set @xmath134(\om ) = \int_{{\ra } } \zeta ( y)v(y;\om ) { \rm d}y , \eqno ( 2.7)\ ] ] which yields a correctly defined ( and a.s .
finite ) rv @xmath135 $ ] . in particular , given a basis @xmath136 in @xmath116 , we obtain a sequence of rvs @xmath137(\om ) $ ] . in view of condition ( s )
, we can represent a random realisation @xmath138 , @xmath139 , of external potential @xmath21 in @xmath126 as the sum @xmath140(\om ) \eta^{{\ra}}_i(x),\eqno ( 2.8.1)\ ] ] where the series is considered in space @xmath116 .
for that reason , we will employ an alternative notation @xmath141(\om ) $ ] and the decomposition @xmath142 and call rvs @xmath143 coefficient random variables in @xmath116 .
the whole sequence @xmath144 of the coefficient rvs in @xmath116 is denoted by @xmath145 ( it is , of course , basis - dependent ) .
furthermore , if @xmath146 is a member of the basis @xmath136 in @xmath116 , then we denote by @xmath147}$ ] the sequence of the remaining rvs @xmath148 . to illustrate the role of space @xmath116 , take the example of a gaussian @xmath21 in @xmath0 with the covariance kernel @xmath128 .
in this case , rvs @xmath149 $ ] and @xmath150 $ ] defined in ( 2.7 ) for functions @xmath151 with @xmath152 are independent normal @xmath153 .
when @xmath154 , with @xmath54 , the series ( 2.8.1 ) and ( 2.8.2 ) also determine the map @xmath155 as a ( random ) multiplication operator in @xmath66 .
this in turn allows us to define the ( random ) hamiltonian @xmath70 in @xmath66 ; cf .
eqn ( 1.4 ) . given a basis in @xmath156 , the randomness in @xmath77
is represented by a sequence of coefficient rvs @xmath157 .
we denote by @xmath158 the sigma - algebra ( in the space of locally square - integrable functions @xmath159 ) generated by the rvs @xmath135 $ ] , @xmath160 .
then the sigma - algebra @xmath23 is the smallest one containing @xmath158 , @xmath109 cellular sets @xmath126 .
the underlying probability distribution @xmath22 is defined on @xmath23 ; the restriction of @xmath22 to @xmath161 is denoted by @xmath162 and the expectation relative to @xmath163 by @xmath164 .
our second condition on @xmath165 is : \(t ) _ temperedness _ :
@xmath109 cellular set @xmath106 , the rv @xmath166 is @xmath163-a.s .
finite , and has a finite moment @xmath167 in this condition , we employ the representation @xmath168 following eqn ( 2.8.2 ) .
although the choice of coefficient rvs @xmath169 depends on the basis in @xmath116 , condition t is basis - independent : when it holds for a particular choice of the orthonotrmal basis in @xmath116 , it also holds for all bases . * remark 2.1.*[rem21 ] condition ( 2.10 ) holds for a gaussian @xmath170under a mild assumption that its samples are ` regular ' .
( in fact , a much stronger property takes place , guaranteeing ( 2.10 ) . )
see , e.g. , @xcite and references therein . given a pair of disjoint cellular sets @xmath171 , we will be working with the conditional distribution functions @xmath172}\right)$ ] defined as follows .
let us fix a basis @xmath173 in @xmath174 with @xmath175 .
then set : @xmath176}\right ) : = \p\left(\left[{\wh{\mathbf 1}}_{\ra}\right]<y\big| \left\{\left[\eta^{{\ra}\cup{\ra}'}_j\right]:\;j\geq 1\right\ } \right),\end{array}\eqno ( 2.12)\ ] ] for the coefficient rv @xmath177 = \left[\eta^{\ra}_0\right]$ ] , given @xmath178}_{{\ra}\cup{\ra}'}= \left\{\left[\eta^{\ra}_i\right]:\;i\geq 1\right\}$ ] , a collection of other basis - related coefficient rvs in @xmath174 .
in fact , in eqn ( 2.12 ) we are interested in sets of the form @xmath179 and @xmath180 where parallelepipeds @xmath181 and @xmath182 are distant apart , namely with @xmath183 .
\eqno ( 2.13)\ ] ] more precisely , we exploit a simple geometric fact stated in lemma [ geom ] : [ geom ] _ consider two parallelepipeds @xmath181 and @xmath184 and suppose that condition _ ( 2.13 ) _ holds true .
then there are two possibilities ( which in general do not exclude each other ) : _
\(i ) @xmath38 _ and @xmath48 are ` completely separated ' , when _
@xmath185 > 0 .
\eqno ( 2.14)\ ] ] \(ii ) @xmath38 _ and @xmath48 are ` partially separated ' . in this case one
( or more ) of the four possibilities can occur : _
@xmath186 ] & > 0,\\ { \rm{(b ) } } & \dist [ \pi_2\blam , [ \pi_1\blam \cup \pi\blam ' ] ] & > 0,\\ { \rm{(c ) } } & \dist [ \pi_1\blam ' , [ \pi\blam \cup \pi_2\blam ' ] ] & > 0,\\ { \rm{(d ) } } & \dist [ \pi_2\blam ' , [ \pi\blam \cup \pi_1\blam ' ] ] & > 0 .
\end{array } \eqno ( 2.15)\ ] ] the proof of lemma [ geom ] is straightforward ; it has been given in our earlier paper @xcite and applied to the lattice case , but geometrical arguments in @xcite actually refer to parallelepipeds in euclidean space @xmath0 .
pictorially , case ( ii ) is where one of the cubes @xmath187 , @xmath188 , @xmath60 , is disjoint from the union of the rest of the projections of @xmath189 and @xmath48 .
we note that the use of the max - norm @xmath190 is convenient here as it leads to the constant @xmath191 ( equal to @xmath192 times @xmath193 , the number of projections @xmath59 and @xmath188 , @xmath60 ) which does not depend on dimension @xmath40 .
we make use of lemma [ geom ] as follows .
first , for a given pair of bounded cellular sets @xmath171 , with @xmath194 , we set : @xmath195}}}\;\ ; \big[\;f_{{\ra}}\left(\,{\ry}+{\rb};\;\bgam _ { { \ra}\cup{\ra}'}^{\bot\left[{\wh{\mathbf 1}_{{\ra}}}\right]}\right ) -f_{{\ra}}\left(\,{\ry};\;\bgam_{{\ra}\cup{\ra}'}^{\bot\left [ { \wh{\mathbf 1}_{{\ra}}}\right]}\right)\big]\ , .
\end{array } \eqno ( 2.16)\ ] ] here , it is assumed that we have been given a basis @xmath173 in @xmath174 , with @xmath196 . in particular , with @xmath197 , we have : @xmath198}\right ) -f_{{\ra}}\left(\,{\ry};\;\bgam_{{\ra}}^{\bot { \wh{\mathbf 1}_{{\ra}}}}\right)\big]\,.\end{array } \eqno ( 2.17)\ ] ]
we then set : @xmath199 next , given @xmath200 , we denote : @xmath201\;\hbox { and } \ ; \pi\blam\cap\pi\blam'=\emptyset\big\}\ , , \end{array } \eqno ( 2.19)\ ] ] @xmath202\qquad{}\\ \hbox { and } \pi_1\blam\cap\big(\pi_2\blam\cup \pi\blam'\big)=\emptyset\big\ } \end{array } \eqno ( 2.20)\ ] ] and @xmath203\qquad{}\\ \text { and } \pi_2\blam\cap\big(\pi_1\blam\cup \pi\blam'\big)=\emptyset\big\ } \,.\end{array } \eqno ( 2.21)\ ] ] ( we suppose here that we have bases @xmath204 in @xmath205 , with @xmath206 in eqn ( 2.19 ) , @xmath207 in eqn ( 2.20 ) and @xmath208 in eqn ( 2.21 ) . ) finally , @xmath209 .
\eqno ( 2.22)\ ] ] * remark 2.2.*[rem22 ] quantity @xmath210 in eqn ( 2.16 ) describes a ` conditioned continuity modulus ' of ( the distribution function of ) rv @xmath211 , given a sample @xmath212 .
it is important to have a grasp of the magnitude of this rv and therefore of quantity @xmath210 . as an example
, we can think of variable @xmath213 $ ] as normal @xmath153 : this is the case where @xmath214 is gaussian with covariance kernel @xmath128 . in this case ,
@xmath215 and consequently , quantities @xmath216 and @xmath217 , @xmath218 , from eqns ( 2.18 ) and ( 2.19)(2.21 ) satisfy a similar bound . concluding this section
, we would like to comment on our approach to analysis of random fields with a continuous argument , particularly , on the meaning and use of quantities introduced in eqns ( 2.16)(2.22 ) .
our main intention is to provide a benefit to a readers with a physical background .
we start by referring back to eqn ( 1.3 ) specifying the structure of a two - particle potential energy under consideration . in a single - particle model ,
the hamiltonian @xmath219 is simply @xmath220 . in this case ,
wegner - type bounds were proved in @xcite , for a certain class of external potential ( including gaussian fields ) .
the proof given in @xcite is based on a special kind of decomposition of an in a cube @xmath221 ( see ( 1.8 ) ) , of the form @xmath222 where the ( scalar ) random variable @xmath223 is independent of the ` fluctuation field ' @xmath224 .
this suggests conditioning upon values of @xmath224 , so that the finite - volume hamiltonian @xmath225 ( again with dirichlet boundary conditions ) becomes a function of the scalar random parameter @xmath223 .
naturally , its eigenvalues are also viewed as functions of @xmath223 ; their analysis , based on the so - called dirichlet neumann bracketing and other techniques , leads to wegner - type bounds .
these bounds are helpful for proving the existence of the limiting density of states for single - particle model in question .
conditions on the random field @xmath138 in @xcite cover some non - gaussian , but they are somewhat restrictive even in the gaussian case . in the present paper ,
we make use of a similar decomposition ( 2.8.1)(2.8.2 ) . for @xmath226
where @xmath227 , @xmath60 , it looks like @xmath228{\wh\one}_\lam(x ) + \xi_\lam ( x;\om ) , \ ; x\in\lam,\ ; \hbox { where } \;\xi_\lam ( x;\om)=\sum_{\eta_j^\lam\in \bgam_{{\lam}}^{\bot { \wh{\one}_{{\lam } } } } } \eqno ( 2.24)\ ] ] with a ( normalised ) constant ` ground level ' @xmath229 and a coefficient @xmath230 $ ] ` moderately dependent ' on the residual ` fluctuation field ' @xmath231 .
such an idea is well - known in probability theory : in elementary courses of statistics , it is usually proved that `` the empiric mean of a gaussian sample is independent of the empiric variance '' .
see , e.g. , @xcite . in the context of this
paper , decomposition ( 2.24 ) implies a similar decomposition for two - particle finite - volume eigenvalues @xmath232 ( see ( 1.10 ) ) .
in particular , conditional on a value of @xmath233 , the eigenvalues @xmath232 are behaving , as functions of @xmath234 , in a controllable fashion , which leads to our wegner - type bounds . in the special case when random field @xmath235 is gaussian , the random variables @xmath232 are ( conditionally ) normal , with the means and variances determined by the conditions but behaving in a controllable fashion .
a one - volume wegner - type bound for finite - volume hamiltonians is given in theorem [ thm1 ] below .
let @xmath236 denote the ( random ) spectrum of operator @xmath77 from eqn ( 1.4 ) ( i.e. , the collection of its eigenvalues @xmath237 , without their multiplicities ) .
next , set @xmath238\\ \;&\leq \ou_{\blam}\ ; + \;\ov_{\pi_1\blam}(\om ) \;+\;\ov_{\pi_2\blam}(\om ) \\ \;&\leq \ou_{\blam } + 2\ov_{\pi\blam}(\om ) .
\end{array}\eqno ( 3.1)\ ] ] where @xmath239 _ assume the above conditions _
( i ) , ( e0 ) _ and _ ( e1 ) _ on function @xmath83 and external potential @xmath21 . then , @xmath109 @xmath240 , @xmath241 , @xmath242 and @xmath243 , with @xmath54 : @xmath244\cap \sigma\left(h_{\blam}\right)\neq\emptyset\big)\\ \qquad\qquad\leq c_1\,\cdot\,\left|\blam\right|\,\cdot\ , \e\left(e+2+\ow_{\blam}\right)^{d } \;\cdot\mu_{\blam } \left(4z_{\pi\blam}\cdot\eps\right ) .
\end{array}\eqno ( 3.2)\ ] ] here @xmath245 is a constant independent of @xmath74 , @xmath246 , @xmath247 and @xmath248 . _
the expression in the rhs of ( 3.2 ) includes a ` volume ' factor @xmath249 and a factor @xmath250 reflecting the growth of the ( value of the ) external potential @xmath21 .
next , we have a factor @xmath251 controlling ` singularity ' of the conditional distribution functions @xmath252}\right)\;\hbox{and}\ ; f_{\pi_j\blam}\left(\,{\ry};\;\bgam_{\pi_{\oj}\blam\cup\pi\blam'}^{\bot\left [ { \wh{\mathbf 1}_{\pi_j\blam}}\right]}\right),\;j=1,2,\;{\ry}\in\r , \ ] ] with a controlled distance between projections @xmath98 and @xmath253 . here
@xmath254 . in a ` smooth ' situation where this distribution function has a density that is bounded uniformly in @xmath255 and in @xmath256 , @xmath247 , @xmath257 satisfying ( 2.13 ) , ( see remark [ rem21 ] above )
, we have that @xmath258 where @xmath259 is a constant .
again , such a form of the bound is important for the proof of anderson s localisation , the main area of application of wegner - type inequalities .
our conditions on the random external potential field are straightforward , as well as the core argument of the proof given below .
however , they require from the reader sufficient familiarity with standard constructions used in the analysis of random fields with a continuous argument
. it may be of some help to the reader to know that these conditions are written down by postulating , in general terms , well - known qualitative properties of gaussian random fields .
on the other hand , the obtained eigenvalue concentration bounds are not optimal , and they are not supposed to be , since our main motivation here is to lay ground for the multi - particle multi - scale analysis in a euclidean space @xmath0 , using ideas and techniques similar to those used in @xcite , @xcite .
given a two - particle parallelepiped @xmath181 , the randomness in operator @xmath260 is represented by the sample of the @xmath261 .
more conveniently , we can refer to a sequence @xmath262 of random variables @xmath263 representing sample @xmath264 in @xmath265 : @xmath266 it is convenient to think of @xmath260 as a family of operators @xmath267 in @xmath66 parametrised by vectors @xmath268 ( i.e. , sample vectors of @xmath269 ) : @xmath270(\ux ) = \left[h^{(0)}_{\blam}\bphi\right](\ux)+w_\blam(\ux;\bgam ) \bphi ( \ux ) \\
= -\,\diy{\frac{1}{2}}\sum\limits_{j=1,2 } \delta^{(\lam_{l_j}(u_j))}_j\bphi(\ux ) + \left(u(\ux)+\sum\limits_{j=1,2}\sum\limits_{i=0,1,\ldots } { \bgam}_i\eta^{\pi\blam}_i(x_j)\right ) \bphi(\ux),\\ \qquad\qquad\ux=(x_1,x_2)\in\blam , \;\;\bphi\in\l_2(\blam ) .
\end{array}\eqno(3.3)\ ] ] vectors @xmath271 are then made random , subject to the distribution @xmath272 of @xmath273 .
consequently , the eigenvalues @xmath73 of @xmath267 ( written in the non - decreasing order ) are parametrised by @xmath271 , i.e. , are random variables on @xmath274 .
we can denote them by @xmath275 but in fact , to stress the dependence on function @xmath276 , we will use an alternative notation @xmath277 , @xmath278 . in this setting
, it is convenient to choose a basis @xmath279 in @xmath280 with @xmath281 and work with @xmath282 , the probability distribution for rv @xmath283\;$ ] conditional on a given vector of values @xmath284 of rvs @xmath285}_{\pi\blam}= \left\{\gam^{\pi\blam}_i , i\geq 1\right\}$ ] .
our first remark is that the additive change @xmath286 in the sample of the external potential field @xmath287 generates the change @xmath288 in the value of the potential energy function @xmath289 , @xmath68 .
correspondingly , the eigenvalues @xmath290 and @xmath291 of the operator @xmath77 before and after the change are related with @xmath292 recalling that @xmath293 , we obtain that @xmath109 @xmath294 : @xmath295 \big|\bgam^{\geq 1)}\right)\leq \mu_{\pi\blam } \left(4z_{\pi\blam_l}\eps\right).\eqno ( 3.4)\ ] ] the second remark is that , when we pass from ( 3.4 ) to ( 3.2 ) , we only have to examine a finite number of eigenvalues @xmath296 of hamiltonian @xmath77 .
more precisely , denote by @xmath297 the eigenvalues of the kinetic energy operator @xmath298 .
then the bound @xmath299 implies that , @xmath109 @xmath300 , the conditional probability @xmath244\cap \sigma\left(h_{\blam}\right)\neq\emptyset\big|\ov_{\pi\blam } \in [ l-1,l)\big)\\ \leq \#\left\{\hbox{eigenvalues of $ h^0_{\blam}$ in $ [ e-\eps -l , e+\eps + l]$}\right\}\;\cdot\mu_{\pi\blam } \left(4z_{\pi\blam}\eps\right).\end{array}\eqno ( 3.6)\ ] ] the final remark is that @xmath301$}\right\}\\ \leq \#\left(\z^d\times\z^d\cap\left\{\ux\in\r^d \times\r^d:\;\|\ux\|_{\rm{euclid}}\leq e+\eps + l\right\ } \right)\\ \leq c_1\left|\blam\right|(e+\eps + l)^{d}\end{array}\eqno ( 3.7)\ ] ] where @xmath302 is the the constant from weyl s formula .
hence , by taking expectation , we obtain from ( 3.6 ) that the lhs in ( 3.2 ) is @xmath303 i.e. , the bound ( 3.2 ) holds true .
this completes the proof of theorem [ thm1 ] .
we now turn to a two - volume wegner - type bound , dealing with a pair of parallelepipeds @xmath181 and @xmath184 ( more precisely , with the corresponding hamiltonians @xmath77 and @xmath78 ) , under an assumption that the distance between @xmath38 and @xmath48 is of the same order of magnitude as the size of these parallelepipeds . given an interval @xmath304 , set : @xmath305\\ \qquad\qquad= \inf\;\left[\ ; \left|e^{(\blam ) } _ k - e^{(\blam')}_{k'}\right|:\;\ ; e^{(\blam ) } _ k , e^{(\blam')}_{k'}\in j\ ; \right],\end{array}\eqno ( 3.8)\ ] ] where @xmath73 are the eigenvalues of @xmath77 and @xmath306 those of @xmath78 , + @xmath307 .
because the potential @xmath308\eqno ( 3.9)\ ] ] is a symmetric function of the pair @xmath309 , with @xmath310 consequently , the spectra @xmath311 and @xmath312 are identical , and the same is true for @xmath313 and @xmath314 .
@xmath109 @xmath315 , @xmath316 with @xmath317 \eqno(3.10)\ ] ] and @xmath109 @xmath318 and interval @xmath319\subset\r$ ] , where @xmath320 and @xmath321 : _
@xmath322 \leq \eps \big)\qquad{}\qquad{}\\ \qquad\leq\;c_2\,\cdot\ , \left|\blam\right|\;\left|\blam'\right|\ ; \e\big[({\rb}+\delta + 1+\ow_{\blam})^d({\rb}+\delta + 1+\ow_{\blam'})^d\big]\\ \qquad\qquad\qquad\qquad\times\ ; { \operatornamewithlimits{\max}\limits_{j=1,2}}\ ; \left[\omu_{\pi\blam}\left(4\eps\cdot z_{\pi\blam}\right),\ ; \omu_{\pi\blam'}\left(4\eps\cdot z_{\pi\blam'}\right)\right]\ , .\end{array}\eqno
( 3.11)\ ] ] * remark 3.1.*[rem31 ] as in the case of the one - volume wegner - type estimate ( 3.2 ) , the rhs in ( 3.11 ) is composed by ` volume ' factors @xmath323 and @xmath324 and factors @xmath325\eqno ( 3.12)\ ] ] and @xmath326\,.\eqno ( 3.13)\ ] ] these factors assess various aspects of randomness introduced in hamiltonians @xmath77 and @xmath78 . for us ,
an immediate use of theorems [ thm1 ] and [ thm2 ] is in proving anderson s localisation in a two - particle model ( with interval @xmath319 $ ] at the edge of the spectrum , specified by further assumptions on rf @xmath165 ) .
owing to lemma [ geom ] , parallelepipeds @xmath189 and @xmath48 obeying ( 3.10 ) satisfy either ( i ) or ( ii ) , i.e. they are either completely or partially separated .
consider first case ( i ) ( complete separation ) .
write @xmath322\leq \eps \big)\\ \qquad= \esm { \p\big(\dist\left[\sigma\left(h_{\blam}\right ) \cap j , \sigma\left(h_{\blam'}\right)\cap j\right ] \leq \eps \
, \big| \,\bfv_{\pi\blam ' } \big)}. \end{array}\eqno ( 3.14)\ ] ] note first that , under conditioning in eqn ( 3.14 ) , the eigen - values @xmath306 , @xmath327 , forming the set @xmath328 are non - random .
therefore , it makes sense to use the following inequality : @xmath329 \leq \eps \
, \big| \,\bfv_{\gamma ' } \big)\\ \qquad{}\leq n_{\blam'}\left(j;\ow_{\blam'}\right ) { \operatornamewithlimits \sup\limits_{e\in j}}\ ; \p\big(\dist\left[\sigma\left(h_{\blam}\right)\cap j , e\right ] \leq \eps\big|\bfv_{\pi\blam'}\big ) , \end{array}\eqno ( 3.15)\ ] ] where @xmath330 is the number of the eigenvalues @xmath306 that can ` eventuallly ' fall in @xmath331 . as in ( 3.11 ) , for @xmath332 $ ] and with @xmath333 standing for the eigenvalues of @xmath334 , we have : @xmath335$}\right\}\\ \quad\leq b_{2d}(l'_1l'_2)^{d}(b+\ow_{\blam'})^{d}. \end{array}\eqno ( 3.16)\ ] ] next , as in theorem [ thm1 ] , @xmath336 \leq \eps\big|\bgam_{\pi\blam'}\big)\\ \qquad\leq c_1\,\cdot\,\left|\blam\right| \e\big[\left(e+2+\ow_{\blam}\right)^{d}\big| \bgam_{\pi\blam'}\big]\quad{}\\ \;\\ \qquad\quad\times \omu_{\pi\blam}\left(4\eps z_{\pi\blam}\right ) .
\end{array}\eqno ( 3.17)\ ] ] this yields @xmath329 \leq \eps \
, \big| \,\bgam_{\pi\blam'}\big)\qquad\qquad\qquad{}\\ \qquad\leq c^{(1)}_2\;|\blam|\;|\blam'|\ ; ( b+2+\ow_{\blam'})^d\;\e\big[(b+2+\ow_{\blam})^d \big|\bgam_{\pi\blam'}\big]\qquad{}\\ \qquad\quad\times \omu_{\pi\blam}\left(4\eps z_{\pi\blam}\right ) , \end{array}\eqno ( 3.18)\ ] ] implying that @xmath337\leq \eps \big)\\ \qquad\leq c^{(1)}_2\;|\blam|\;|\blam'|\;\e\big[(b+2 + \ow_{\blam})^d(b+2+\ow_{\blam'})^d\big]\qquad{}\\ \qquad\quad\times \omu_{\pi\blam}\left(4\eps z_{\pi\blam}\right ) , \end{array}\eqno ( 3.19)\ ] ] where @xmath338 is a constant .
\(ii ) now consider the case of partial separation .
for example , assume case ( a ) from lemma [ geom ] ( see eqn ( 3.11 ) ) , when projection @xmath50 is disjoint from the union of the rest of the projections of @xmath38 and @xmath48 : @xmath339 = \emptyset.\eqno ( 3.20)\ ] ] we then estimate the probability in the lhs of ( 3.16 ) with the help of the conditional expectation @xmath337\leq \eps \big)\\ \qquad= \esm { \p\big(\dist\left[\sigma\left(h_{\blam}\right ) , \sigma\left(h_{\blam'}\right)\right ] \leq \eps \
, \big| \,\bgam_{\pi_2\blam\cup\pi\blam ' } \big)\big|\bgam_{\pi\blam'}}. \end{array } \eqno(3.21)\ ] ] note that , owing to ( 3.20 ) , the sigma - algebra generated by @xmath340 does not include any of the rvs @xmath341 forming @xmath342 .
thus , the argument used in the proof of theorem [ thm1 ] is still applicable if , instead of @xmath22 , we work with the probability distribution @xmath343 , conditional on @xmath344 and restricted to the the sigma - algebra generated by @xmath345 .
this allows us to write @xmath346 \leq \eps \
, \big| \,\bgam^{\geq 1}_{\pi_1\blam},\bgam_{\pi_2\blam\cup\pi\blam ' } \big)\\ \qquad\leq c^{(2)}_2\;|\blam|\;|\blam'|\ ; ( b+2+\ow_{\blam'})^d\\ \qquad\quad\times\e\big[(b+2+\ow_{\blam})^d \big|\bgam^{\geq 1}_{\pi_1\blam } , \bgam_{\pi_2\blam\cup\pi\blam'}\big]\ ; \omu_{\pi\blam}\left(4\eps z_{\pi\blam}\right ) \end{array}\eqno ( 3.22)\ ] ] and @xmath346 \leq \eps \,\big)\\ \qquad\leq c^{(2)}_2\;|\blam|\;|\blam'|\ ; \e\big[(b+2+\ow_{\blam})^d(b+2+\ow_{\blam'})^d\big ] \qquad{}\\ \qquad\quad\times \omu_{\pi\blam}\left(4\eps z_{\pi\blam}\right ) .
\end{array}\eqno ( 3.23)\ ] ] we are then in position to deduce the required bound for the the conditional probability in the lhs of ( 3.21 ) .
if , instead of ( 3.20 ) , we have one of the other disjointedness relations ( b)-(d ) in eqn ( 3.11 ) , then the argument is conducted in a similar fashion .
specifically , in case ( b ) we swap projections @xmath50 and @xmath347 in the above argument .
furthermore , in cases ( c ) and ( d ) , we should exchange @xmath247 and @xmath257 as compared to arguments in cases ( a ) and ( b )
. we now briefly comment on the case where the interaction potential has a hard core ( cf .
( 1.15 ) ) . in this case
hamiltonians @xmath25 and @xmath70 in eqns ( 1.1 ) and ( 1.5 ) act in subspaces @xmath348 and @xmath349 where @xmath350 more precisely , @xmath25 and @xmath70 carry additional dirihclet s boundary conditions on @xmath351 .
however , the scheme of the proof of theorems [ thm1 ] and [ thm2 ] remains unchanged , and all arguments carry through .
[ bcss1 ] a. boutet de monvel , v. chulaevsky , p. stollmann and y. suhov , _ wegner - type bounds for a two - particle continuous anderson model with an alloy - type external potential . _ - arxiv : math - ph/0821:2621 , 2008 . [ cs2 ]
v. chulaevsky , y. suhov , _ eigenfunctions in a two - particle anderson tight binding model_. commun .
phys , doi @xmath353 ( 2009 ) .
[ cs3 ] v. chulaevsky , y. suhov , _ multi - particle anderson localisation : induction on the number of particles . _
math - ph / arxiv:0811.2530 .
[ fhlm ] w. fischer , t. hupfer , h. leschke , p. mller , _ existence of the density of states for multi - dimensional continuum schrdinger operators with gaussian random potentials .
phys . , * 190 * ( 1997 ) , no .
1 , 133141 . we thank the isaac newton institute , university of cambridge , for hospitality during the programme `` mathematics and physics of the anderson localization : 50 years after '' ( july december , 2008 ) .
ys thanks ihes , bures - sur - yvette , for hospitality during visit in 2008 . | we analyse a two - particle quantum system in @xmath0 with interaction and in presence of a random external potential field with a continuous argument ( an anderson model in a continuous space ) .
our aim is to establish the so - called wegner - type estimates for such a model , assessing the probability that random spectra of hamiltonians in finite volumes intersect with a given set . for the lattice version of the two - particle model ,
a similar result was obtained in @xcite h l p 00 2 ^ 2 2 ^ 2 _ # 1#1 # 1#1 # 1\{#1 } # 1 # 1#2_#1 # 1| # 1 | # 1\{#1 } # 1 \ { # 1 } # 1#2 _#1\ { # 2 } # 1| # 1 | # 1 # 1 # 1@xmath1-s # 1@xmath1-s # 1 # 1#2#3 # 1#2 # 1 |
the spectral excess theorem , due to fiol and garriga @xcite , states that a regular ( connected ) graph @xmath0 is distance - regular if and only if its spectral - excess ( a number which can be computed from the spectrum of @xmath0 ) equals its average excess ( the mean of the numbers of vertices at maximum distance from every vertex ) , see van dam @xcite , and fiol , gago and garriga @xcite for short proofs .
recently , some local as well as global approaches to that result have been used to obtain new versions of the theorem for nonregular graphs , and also to study the problem of characterizing those graphs which have the corresponding distance - regularity property ( see , for instance , dalf , van dam , fiol , garriga , and gorissen @xcite ) .
one of these concepts is that of ` pseudo - distance - regularity ' around a vertex , introduced by fiol , garriga , and yebra @xcite , which generalizes the known concept of distance - regularity around a vertex and , in some cases , coincides with that of distance - biregularity , intended for bipartite graphs .
a bipartite graph is distance - biregular when it is distance - regular around each vertex and the intersection array only depends on the stable set such a vertex belongs to ( see delorme @xcite ) . in general
, distance - regular and distance - biregular graphs have found many ` applications ' .
some examples are completely regular codes , symmetric designs , some partial geometries , and some non - symmetric association schemes . in this note
we present a new version of the spectral excess theorem for bipartite distance - biregular graphs .
let us first introduce some basic notation and results . for more background on graph spectra and different concepts of distance - regularity in graphs see , for instance , @xcite .
let @xmath1 be a ( connected ) graph with @xmath2 vertices and diameter @xmath3 .
the set of vertices at distance @xmath4 from a given vertex @xmath5 is denoted by @xmath6 for @xmath7 , and @xmath8 .
let @xmath9 and @xmath10 . if @xmath0 has adjacency matrix @xmath11 , its spectrum is denoted by @xmath12 where the different eigenvalues of @xmath0 are in decreasing order , @xmath13 , and the superscripts stand for their multiplicities @xmath14 . from the positive ( @xmath15-)eigenvector @xmath16 normalized in such a way that @xmath17 , consider the _ weight function _
@xmath18 such that @xmath19 for @xmath20 , where @xmath21 stands for the coordinate @xmath22-th vector , and @xmath23 .
in particular , @xmath24 , so that @xmath25 assigns the number @xmath26 to vertex @xmath22 , and @xmath27 . in our study
we use some properties of orthogonal polynomials of a discrete variable ( see cmara , fbrega , fiol and garriga @xcite ) . from a mesh @xmath28 , @xmath29 , of real numbers , and a weight function @xmath30 , normalized in such a way that @xmath31 , we consider the following scalar product in @xmath32 $ ] : @xmath33 where , for short , @xmath34 , @xmath35 . then , the sequence @xmath36 of orthogonal polynomials with respect to such a scalar product , with @xmath37 , normalized in such a way that @xmath38 , and their sum polynomials @xmath39 , @xmath40 , satisfy the following properties : [ ortho - pol ] * @xmath41 and the constants of the three term recurrence @xmath42 , where @xmath43 , satisfy @xmath44 for @xmath35 . * @xmath45 , were @xmath46 , @xmath35 .
* @xmath47 , and @xmath48 for every @xmath49 . as the notation suggests , in our context the mesh @xmath50 is constituted by the eigenvalues of a graph @xmath0 , and the weight function is closely related to their ( standard or local ) multiplicities .
more precisely , when @xmath51 we have the scalar product @xmath52 and the sequence @xmath53 are called the _
predistance polynomials_. in this case , we denote them by @xmath54 , and their sums by @xmath55 .
moreover , it is known that the _ hoffman polynomial _
@xmath56 , characterized by @xmath57 and @xmath58 for @xmath49 , satisfies @xmath59 if and only if @xmath0 is regular ( see hoffman @xcite ) .
a ` local version ' of the above approach is the following .
let @xmath60 , @xmath61 , be the ( principal ) idempotent of @xmath0 .
then , for a given vertex @xmath22 , the _ @xmath22-local multiplicity of @xmath62 _ is defined as @xmath63 . as was shown by fiol , garriga and yebra @xcite , the local multiplicities
behave similarly as the ( standard ) multiplicities when we ` see ' the graph from the base vertex @xmath22 .
indeed , * @xmath64 ( vs. @xmath65 ) . * @xmath66 . * @xmath67 ( vs. @xmath68 ) . then , given a vertex @xmath22 with @xmath69 different local eigenvalues ( that is , those with nonzero local multiplicity ) , the choice @xmath70 , @xmath35 , leads us to the scalar product @xmath71 ( note that the sum has at most @xmath72 nonzero terms . )
then , the corresponding orthogonal sequence @xmath73 is referred to as the _ @xmath22-local predistance polynomials _ , and they are denoted by @xmath74 . notice that , by the above property of the local multiplicities , @xmath75 let @xmath0 be a bipartite graph on @xmath76 vertices , with stables sets @xmath77 and @xmath78 , on @xmath79 and @xmath80 vertices , respectively .
let @xmath0 have diameter @xmath3 , and let @xmath81 denote the maximum eccentricity of the vertices in @xmath82 , @xmath83 , so that @xmath84 . as it is well known , @xmath85 and , without loss of generality
, we here suppose that @xmath86 .
given some integers @xmath87 and @xmath88 , we will use the averages @xmath89 and @xmath90 .
recall also that the spectrum of a bipartite graph is symmetric about zero : @xmath91 and @xmath92 , @xmath61 .
moreover its local eigenvalues and multiplicities also share the same property . in our context
, we also have the following result . [ loc - mul - bip ]
let @xmath93 be a bipartite graph with vertex bipartition @xmath94 . then
, the local multiplicities satisfy : * @xmath95 , @xmath96 .
* @xmath97 , * @xmath98 . by counting in two ways the number of closed @xmath99-walks rooted at a vertex we have that @xmath100 .
then , @xmath101 and , hence , @xmath102 and @xmath103 follows since @xmath104 .
to prove @xmath105 , we use that @xmath106 and @xmath103 .
then , @xmath107 similarly for @xmath108 .
this result suggests to define the scalar products @xmath109 and @xmath110 , by taking the weights @xmath111 and @xmath112 , @xmath40 , respectively . in this case ,
notice that @xmath113 then , by lema [ loc - mul - bip ] , @xmath114 and similarly for @xmath110 .
the corresponding sequences of orthogonal polynomials and their sums are denoted by @xmath115 and @xmath116 , @xmath83 , respectively .
there are four cases to be considered : * if @xmath117 , then @xmath118 , and @xmath119 , with @xmath120 being the predistance polynomial for @xmath35 . *
if @xmath121 is odd , then @xmath122 is not an eigenvalue of @xmath0 , @xmath123 , and @xmath124 .
then , from the normalization conditions , @xmath125 and @xmath126 , @xmath35 . * if @xmath121 is even and @xmath127 , then @xmath128 , and the expressions for @xmath109 , in , and @xmath110 have @xmath129 terms . * if @xmath121 is even and @xmath130 , say @xmath131 , then @xmath132 but @xmath133 .
thus , the scalar product @xmath110 is defined on the mesh of the @xmath121 nonzero eigenvalues and , in this case , we denote the corresponding sequence of orthogonal polynomials by @xmath134 .
a bipartite graph @xmath0 with stable sets @xmath135 is _ distance - biregular _ when it is distance - regular around each vertex @xmath22 and the intersection array only depends on the stable set @xmath22 belongs to . for some properties of these graphs , see delorme @xcite .
an example is the subdivided petersen graph @xmath136 , obtained from such a graph by inserting a vertex into each of its edges , see fig .
[ fig : bibipetersen ] . in particular ,
every distance - biregular graph @xmath0 is _ semiregular _ , that is , all vertices in the same independent set have the same degree .
if vertices in @xmath82 have degree @xmath137 , @xmath83 , we say that @xmath0 is _ @xmath138-semiregular_. in this case , the normalized positive eigenvector of @xmath0 is @xmath139 , where @xmath140 a more general framework for studying distance - biregular graphs , which we use here , is that of pseudo - distance - regularity , a concept introduced by fiol , garriga , and yebra in @xcite .
a graph @xmath0 is said to be _ pseudo - distance - regular around a vertex _
@xmath5 with eccentricity @xmath141 if the numbers , defined for any vertex @xmath142 , @xmath143 depend only on the value of @xmath144 . in this case , we denote them by @xmath145 , @xmath146 , and @xmath147 , respectively , and they are referred to as the _ @xmath22-local pseudo - intersection numbers _ of @xmath0 . in the main result of this paper we use the following theorems of fiol and garriga @xcite .
the first one can be considered as the spectral excess theorem for pseudo - distance - regular graphs : [ teo caract c - c - r ] let @xmath22 be a vertex of a graph @xmath0 , with @xmath69 distinct local eigenvalues , and sum of predistance polynomials @xmath148 .
then , for any polynomial @xmath149 $ ] with @xmath150 , @xmath151 and equality holds if and only if @xmath152 for any @xmath153 .
moreover , equality holds with @xmath154 if and only @xmath22 has maximum eccentricity @xmath155 and @xmath0 is pseudo - distance - regular around @xmath22 .
a graph @xmath0 is said to be _ pseudo - distance - regularized _ when it is pseudo - distance - regular graph around each of its vertices .
generalizing a result of godsil and shawe - taylor @xcite , the author proved the following : [ teo caract dist - bireg ] let @xmath0 be a bipartite @xmath138-regular on @xmath156 vertices , and with @xmath129 distinct eigenvalues .
let @xmath0 have average numbers @xmath157 , @xmath158 , @xmath159 , @xmath160 , and highest degree polynomials @xmath161 , @xmath162 , @xmath163 , @xmath164 defined as above .
then , @xmath0 is distance - biregular if and only if some of the following conditions holds : @xmath103 if @xmath117 , then @xmath0 is regular and the result is just the spectral excess theorem for distance - regular graphs . if @xmath121 is odd we are in case @xmath170 discussed above , and @xmath162 and @xmath163 are multiples of the predistance polynomial @xmath171 . then , reasoning as in the proof of case @xmath172 ,
see below , we have that @xmath0 is distance - biregular if and only if @xmath173 and @xmath174 .
but , as @xmath121 is odd , by counting in two ways the number of pairs of vertices at distance @xmath121 , we have that @xmath175 .
then , putting all together , we get that @xmath176 , and @xmath0 must be again regular . to prove @xmath172
, we first claim that , with @xmath177 , the inequality can be written as @xmath178 indeed , if @xmath179 the result is obvious , whereas if @xmath180 , then @xmath181 $ ] . here
equality must be understood in the quotient ring @xmath182/i$ ] , where @xmath183 is the ideal generated by the polynomial with zeros the @xmath69 eigenvalues having non - null @xmath22-local multiplicity .
thus , @xmath184 , where @xmath185 , and the inequality in still holds .
now , by taking the average over all vertices of @xmath77 , we have @xmath186 where we used ( [ average - scal - prod ] ) .
consequently , @xmath187 where we used that , as @xmath121 is even , @xmath188 , and @xmath189 stands for the harmonic mean of the numbers @xmath190 .
furthermore , since the harmonic mean is at most the arithmetic mean , @xmath191 this , together with @xmath192 ( recall that in the scalar product @xmath193 we have @xmath194 ) , gives @xmath195 then , in case of equality all inequalities in must be also equalities and , from theorem [ teo caract c - c - r ] , @xmath0 is pseudo - distance - regular around each vertex @xmath196 . in the same way , we prove that , if @xmath197 , then @xmath0 is pseudo - distance - regular around each vertex @xmath198 .
consequently , @xmath0 is a pseudo - distance - regularized graph and , by theorem [ pdrg = drg ] , it is also distance - biregular .
* notice that , in cases @xmath103 and @xmath172 the diameters of @xmath0 turn to be @xmath201 , whereas in case @xmath199 we have @xmath202 and @xmath203 . as an example of the latter we have again the subdivided petersen graph of fig .
[ fig : bibipetersen ] .
* by using lemma [ ortho - pol]@xmath172 we can give explicit formulae for the values of the predistance polynomials at @xmath15 appeared in theorem [ teo caract dist - bireg ] .
for instance , in case @xmath199 , @xmath204 where the moment - like parameters @xmath205 are defined as before .
* since the conclusions in @xmath172 and @xmath199 are derived from two inequalities , we can add up the corresponding equalities to give only one condition , as in @xmath103 , and the result still holds .
for instance , the condition in case ( b ) can be @xmath206 * acknowledgments . *
research supported by the ministerio de educacin y ciencia ( spain ) and the european regional development fund under project mtm2011 - 28800-c02 - 01 , and by the catalan research council under project 2009sgr1387 .
m. cmara , j. fbrega , m.a .
fiol , and e. garriga , some families of orthogonal polynomials of a discrete variable and their applications to graphs and codes , _ electron .
j. combin .
_ * 16(1 ) * ( 2009 ) , # r83 . c. dalf , e.r .
van dam , m.a .
fiol , e. garriga , and b.l .
gorissen , on almost distance - regular graphs , _
j. combin .
theory ser . a _ * 118 * ( 2011 ) , no .
3 , 10941113 . e.r .
van dam , the spectral excess theorem for distance - regular graphs : a global ( over)view , _ electron .
j. combin .
_ * 15(1 ) * ( 2008 ) , # r129 . m.a .
fiol , s. gago , and e. garriga , a simple proof of the spectral excess theorem for distance - regular graphs , _ linear algebra appl . _ * 432 * ( 2010 ) , no . 9 , 24182422 .
fiol and e. garriga , from local adjacency polynomials to locally pseudo - distance - regular graphs , _
j. combin .
theory ser .
b _ * 71 * ( 1997 ) , 162183 . | the spectral excess theorem for distance - regular graphs states that a regular ( connected ) graph is distance - regular if and only if its spectral - excess equals its average excess .
a bipartite graph @xmath0 is distance - biregular when it is distance - regular around each vertex and the intersection array only depends on the stable set such a vertex belongs to . in this note
we derive a new version of the spectral excess theorem for bipartite distance - biregular graphs . |
the ubiquitous camassa holm ( ch ) equation @xcite @xmath8 where @xmath9 is a constant , has been extensively studied due to its many intriguing properties .
the aim of this paper is to construct a metric that renders the flow generated by the camassa
holm equation lipschitz continuous on a function space in the conservative case . to keep the presentation reasonably short , we restrict the discussion to properties relevant for the current study .
more precisely , we consider the initial value problem for with periodic initial data @xmath10 .
since the function @xmath11 satisfies equation with @xmath12 , we can without loss of generality assume that @xmath13 vanishes . for convenience
we assume that the period is @xmath14 , that is , @xmath15 for @xmath16 .
the natural norm for this problem is the usual norm in the sobolev space @xmath6 as we have that @xmath17 ( by using the equation and several integration by parts as well as periodicity ) for smooth solutions @xmath3 . even for smooth initial data
, the solutions may develop singularities in finite time and this breakdown of solutions is referred to as wave breaking . at wave
breaking the @xmath18 and @xmath19 norms of the solution remain finite while the spatial derivative @xmath20 becomes unbounded pointwise .
this phenomenon can best be described for a particular class of solutions , namely the multipeakons . for simplicity
we describe them on the full line , but similar results can be described in the periodic case .
multipeakons are solutions of the form ( see also @xcite ) @xmath21 let us consider the case with @xmath22 and one peakon @xmath23 ( moving to the right ) and one antipeakon @xmath24 ( moving to the left ) . in the symmetric case ( @xmath25 and @xmath26 ) the solution @xmath3 will vanish pointwise at the collision time @xmath27 when @xmath28 , that is , @xmath29 for all @xmath30 .
clearly the well - posedness , in particular , lipschitz continuity , of the solution is a delicate matter .
consider , e.g. , the multipeakon @xmath31 defined as @xmath32 , see figure [ fig : peakcol ] .
for simplicity , we assume that @xmath33 . then , we have @xmath34 and the flow is clearly not lipschitz continuous with respect to the @xmath18 norm . , which vanishes at @xmath27 , for @xmath35 ( on the left ) and @xmath36 ( on the right ) .
the solid curve depicts the multipeakon solution given by @xmath32.,title="fig:",width=257 ] , which vanishes at @xmath27 , for @xmath35 ( on the left ) and @xmath36 ( on the right ) .
the solid curve depicts the multipeakon solution given by @xmath32.,title="fig:",width=257 ] [ fig : peakcol ] our task is here to identify a metric , which we will denote by @xmath2 for which conservative solutions satisfy a lipschitz property , that is , if @xmath3 and @xmath4 are two solutions of the camassa
holm equation , then @xmath37\ ] ] for any given , positive @xmath38 . for nonlinear partial differential equations
this is in general a quite nontrivial issue .
let us illustrate it in the case of hyperbolic conservation laws @xmath39 in the scalar case with @xmath40 , @xmath16 , it is well - known @xcite that the solution is @xmath41-contractive in the sense that @xmath42 in the case of systems , i.e. , for @xmath43 with @xmath44 it is known @xcite that @xmath45 for some constant @xmath46 . more relevant for the current study , but less well - known , is the recent analysis @xcite of the hunter
saxton ( hs ) equation @xmath47 or alternatively @xmath48 which was first introduced in @xcite as a model for liquid crystals .
again the equation enjoys wave breaking in finite time and the solutions are not lipschitz in term of convex norms . the hunter
saxton equation can in some sense be considered as a simplified version of the camassa
holm equation , and the construction of the semigroup of solutions via a change of coordinates given in @xcite is very similar to the one used here and in @xcite for the camassa holm equation .
in @xcite the authors constructed a riemannian metric which renders the conservative flow generated by the hunter
saxton equation lipschitz continuous on an appropriate function space .
for the camassa
holm equation , the problem of continuation beyond wave breaking has been considered by bressan and constantin @xcite and holden and raynaud @xcite ( see also xin and zhang @xcite and coclite , karlsen , and holden @xcite ) . both approaches
are based on a reformulation ( distinct in the two approaches ) of the camassa
holm equation as a semilinear system of ordinary differential equations taking values in a banach space .
this formulation allows one to continue the solution beyond collision time , giving either a global conservative solution where the energy is conserved for almost all times or a dissipative solution where energy may vanish from the system .
local existence of the semilinear system is obtained by a contraction argument .
going back to the original function @xmath3 , one obtains a global solution of the camassa
holm equation . in @xcite ,
bressan and fonte introduce a new distance function @xmath49 which is defined as a solution of an optimal transport problem .
they consider two multipeakon solutions @xmath50 and @xmath51 of the camassa
holm equation and prove , on the intervals of times where no collisions occur , that the growth of @xmath52 is linear ( that is , @xmath53 for some fixed constant @xmath46 ) and that @xmath52 is continuous across collisions .
it follows that @xmath54 for all times @xmath55 that are not collision times and , in particular , for almost all times . by density
, they construct solutions for any initial data ( not just the multipeakons ) and the lipschitz continuity follows from . as in @xcite ,
the goal of this article is to construct a metric which makes the flow lipschitz continuous .
however , we base the construction of the metric directly on the reformulation of the equation which is used to construct the solutions themselves , and we use some fundamental geometrical properties of this reformulation ( relabeling invariance , see below ) .
the metric is defined on the set @xmath56 which includes configurations where part of the energy is concentrated on sets of measure zero ; a natural choice for conservative solutions . in particular , we obtain that the lipschitz continuity holds for all times and not just for almost all times as in @xcite .
let us describe in some detail the approach in this paper , which follows @xcite quite closely in setting up the reformulated equation .
let @xmath57 denote the solution , and @xmath58 the corresponding characteristics , thus @xmath59 .
our new variables are @xmath58 , @xmath60 where @xmath61 corresponds to the lagrangian velocity while @xmath62 could be interpreted as the lagrangian cumulative energy distribution . in the periodic case
one defines @xmath63 then one can show that @xmath64_0^\xi , \end{aligned } \right.\ ] ] is equivalent to the camassa
holm equation .
global existence of solutions of is obtained starting from a contraction argument , see theorem [ th : global ] .
the issue of continuation of the solution past wave breaking is resolved by considering the set @xmath56 ( see definition [ def : d ] ) which consists of pairs @xmath65 such that @xmath66 if @xmath67 and @xmath68 is a positive radon measure with period one , and whose absolutely continuous part satisfies @xmath69 . with three lagrangian variables @xmath70 versus two eulerian variables @xmath65 , it is clear that there can be no bijection between the two coordinate systems .
if two lagrangian variables correspond to one and the same solution in eulerian variables , we say that the lagrangian variables are relabelings of each other . to resolve the relabeling issue we define a group of transformations which acts on the lagrangian variables and lets the system of equations invariant .
we are able to establish a bijection between the space of eulerian variables and the space of lagrangian variables when we identify variables that are invariant under the action of the group .
this bijection allows us to transform the results obtained in the lagrangian framework ( in which the equation is well - posed ) into the eulerian framework ( in which the situation is much more subtle ) . to obtain a lipschitz metric in eulerian
coordinates we start by constructing one in the lagrangian setting . to this end
we start by identifying a set @xmath71 ( see definition [ def : f ] ) that leaves the flow invariant , that is , if @xmath72 then the solution @xmath73 of with @xmath74 will remain in @xmath71 , i.e. , @xmath75 .
next , we identify a subgroup @xmath76 , see definition [ def : g ] , of the group of homeomorphisms on the unit interval , and we interpret @xmath76 as the set of relabeling functions . from this
we define a natural group action of @xmath76 on @xmath71 , that is , @xmath77 for @xmath78 and @xmath79 , see definition [ def : group ] and proposition [ prop : action ] .
we can then consider the quotient space @xmath80 .
however , we still have to identify a unique element in @xmath71 for each equivalence class in @xmath80 . to this end
we introduce the set @xmath81 , see , of elements in @xmath71 for which @xmath82 and @xmath83 .
this establishes a bijection between @xmath80 and @xmath81 , see lemma [ lemma : bijection ] , and therefore between @xmath81 and @xmath56 .
finally , we define a semigroup @xmath84 on @xmath81 ( definition [ def : bars ] ) , and the next task is to identify a metric that makes the flow @xmath85 lipschitz continuous on @xmath81 .
we use the bijection between @xmath81 and @xmath56 to transport the metric from @xmath81 to @xmath56 and get a lipschitz continuous flow on @xmath56 . in @xcite ,
the authors define the metric on @xmath81 by simply taking the norm of the underlying banach space ( the set @xmath81 is a nonlinear subset of a banach space ) .
they obtain in this way a metric which makes the flow continuous but not lipschitz continuous . as we will see ( see remark [ rem : compmetric ] ) , this
metric is stronger than the one we construct here and for which the flow is lipschitz continuous . in @xcite , for the hunter
saxton equation , the authors use ideas from riemannian geometry and construct a semimetric which identifies points that belong to the same equivalence class .
the riemannian framework seems however too rigid for the camassa
holm equation , and we have not been able to carry out this approach . however , we retain the essential idea which consists of finding a semimetric which identifies equivalence classes . instead of a riemannian metric , we use a discrete counterpart . note that this technique will also work for the hunter saxton and will give the same metric as in @xcite .
a natural candidate for a semimetric which identifies equivalence classes is ( cf . )
@xmath86 which is invariant with respect to relabeling .
however , it does not satisfy the triangle inequality .
nevertheless it can be modified to satisfy all the requirements for a metric if we instead define , see definition [ def : alberto ] , the following quantity @xmath87 where the infimum is taken over all finite sequences @xmath88 which satisfy @xmath89 and @xmath90 .
one can then prove that @xmath91 is a metric on @xmath81 , see lemma [ lemma : distance ] .
finally , we prove that the flow is lipschitz continuous in this metric , see theorem [ th : stab ] . to transfer this result to the eulerian variables we reconstruct these variables from the lagrangian coordinates as in @xcite : given @xmath92
, we define @xmath66 by ( see definition [ def : ftoe ] ) @xmath93 for any @xmath94 such that @xmath95 , and @xmath96 .
we denote the mapping from @xmath71 to @xmath56 by @xmath97 , and the inverse restricted to @xmath81 by @xmath98 .
the natural metric on @xmath56 , denoted @xmath2 , is then defined by @xmath99 for two elements @xmath100 in @xmath56 , see definition [ def : dd ] .
the main theorem , theorem [ th : main ] , then states that the metric @xmath2 is lipschitz continuous on all states with finite energy . in the last section , section [ sec : topology ] ,
the metric is compared with the standard norms .
two results are proved : the mapping @xmath101 is continuous from @xmath6 into @xmath56 ( proposition [ prop : cont1 ] ) .
furthermore , if @xmath102 is a sequence in @xmath56 that converges to @xmath65 in @xmath56 .
then @xmath103 in @xmath7 and @xmath104 ( proposition [ prop : cont2 ] ) .
the problem of lipschitz continuity can nicely be illustrated in the simpler context of ordinary differential equations .
consider three differential equations : @xmath105 straightforward computations give as solutions @xmath106 we find that @xmath107 thus we see that in the regular case we get a lipschitz estimate with constant @xmath108 uniformly bounded as @xmath55 ranges on a bounded interval . in the second case
we get a lipschitz estimate uniformly valid for all @xmath109 . in the final example , by restricting attention to strictly increasing solutions of the ordinary differential equations , we achieve uniqueness and continuous dependence on the initial data , but without any lipschitz estimate at all near the point @xmath110 .
we observe that , by introducing the riemannian metric @xmath111 an easy computation reveals that @xmath112 let us explain why this metric can be considered as a riemannian metric .
the euclidean metric between the two points is then given @xmath113 where the infimum is taken over all paths @xmath114\to{\mathbb{r}}$ ] that join the two points @xmath115 and @xmath116 , that is , @xmath117 and @xmath118 . however , as we have seen , the solutions are not lipschitz for the euclidean metric .
thus we want to measure the infinitesimal variation @xmath119 in an alternative way , which makes solutions of equation lipschitz continuous .
we look at the evolution equation that governs @xmath119 and , by differentiating with respect to @xmath120 , we get @xmath121 and we can check that @xmath122 let us consider the real line as a riemannian manifold where , at any point @xmath16 , the riemannian norm is given by @xmath123 for any tangent vector @xmath124 in the tangent space of @xmath125 . from , one can see that at the infinitesimal level , this riemannian norm is exactly preserved by the evolution equation .
the distance on the real line which is naturally inherited by this riemannian is given by @xmath126 where the infimum is taken over all paths @xmath114\to{\mathbb{r}}$ ] joining @xmath115 and @xmath127 .
it is quite reasonable to restrict ourselves to paths that satisfy @xmath128 and then , by a change of variables , we recover the definition .
the riemannian approach to measure a distance between any two distinct points in a given set ( as defined in ) requires the existence of a smooth path between points in the set . in the case of the hunter
saxton ( see @xcite ) , we could embed the set we were primarily interested in into a convex set ( which is therefore connected ) and which also could be regularized ( so that the riemannian metric we wanted to use in that case could be defined ) . in the case of the camassa
holm equation , we have been unable to construct such a set
. however , there exists the alternative approach which , instead of using a smooth path to join points , uses finite sequences of points , see .
we illustrate this approach with equation .
we want to define a metric in @xmath129 which makes the semigroup of solutions lipschitz stable .
given two points @xmath130 , we define the function @xmath131 as @xmath132 \frac{\bar{x}-x}{x^{1/2}}&\text { if $ x<\bar{x}$}. \end{cases}\ ] ] the function @xmath133 is symmetric and @xmath134 if and only if @xmath135 , but @xmath133 does not satisfy the triangle inequality .
therefore we define ( cf . ) @xmath136 where the infimum is taken over all finite sequences @xmath137 such that @xmath138 and @xmath139 .
then , @xmath140 satisfies the triangle inequality and one can prove that it is also a metric .
given @xmath141 such that @xmath142 , we denote @xmath143 and @xmath144 the solution of with initial data @xmath145 and @xmath146 , respectively . after a short computation , we get @xmath147 hence , @xmath148 so that @xmath149 and the semigroup of solutions to is a contraction for the metric @xmath140 .
it follows from the definition of @xmath133 that , for @xmath150 with @xmath151 , we have @xmath152 it implies that @xmath153 satisfies @xmath154 where @xmath155 , which is also the definition of the riemann integral , so that @xmath156 and the metric we have just defined coincides with the riemannian metric we have introduced . note that if we choose @xmath157 then does not hold ; we have instead @xmath158 , which is the triangle inequality .
thus , for @xmath159 as defined by with @xmath133 replaced by @xmath160 , we get @xmath161 it is also possible to check that , for @xmath160 , we can not get that @xmath162 for any constant @xmath46 for any @xmath145 and @xmath146 and @xmath163 $ ] ( for a given @xmath38 ) , so that the definition of @xmath160 is inappropriate to obtain results of stability for .
the camassa holm equation for @xmath12 reads @xmath164 and can be rewritten as the following system nonzero , equation is simply replaced by @xmath165 . ]
@xmath166 we consider periodic solutions of period one .
next , we rewrite the equation in lagrangian coordinates .
therefore we introduce the characteristics @xmath167 we introduce the space @xmath168 defined as @xmath169 functions in @xmath168 map the unit interval into itself in the sense that if @xmath3 is periodic with period 1 , then @xmath170 is also periodic with period 1 .
the lagrangian velocity @xmath61 reads @xmath171 we will consider @xmath172 and @xmath61 periodic .
we define the lagrangian energy cumulative distribution as @xmath173 for all @xmath55 , the function @xmath62 belongs to the vector space @xmath174 defined as follows : @xmath175 equip @xmath174 with the norm @xmath176)}+{\left\vertf_\xi\right\vert}_{l^1([0,1])}.\ ] ] as an immediate consequence of the definition of the characteristics we obtain @xmath177 this last term can be expressed uniquely in term of @xmath61 , @xmath178 , and @xmath62 .
we have the following explicit expression for @xmath179 , @xmath180 thus , @xmath181 and , after the change of variables @xmath182 , @xmath183\,d\eta.\end{gathered}\ ] ] we have @xmath184 note that @xmath185 is periodic with period one .
then , can be rewritten as @xmath186 where the @xmath55 variable has been dropped to simplify the notation .
later we will prove that @xmath178 is an increasing function for any fixed time @xmath55 .
if , for the moment , we take this for granted , then @xmath187 is equivalent to @xmath188 where @xmath189 and , slightly abusing the notation , we write @xmath190 the derivatives of @xmath188 and @xmath179 are given by @xmath191 respectively . for @xmath192
$ ] , using the fact that @xmath193 and the periodicity of @xmath185 and @xmath61 , the expressions for @xmath188 and @xmath179 can be rewritten as @xmath194 and @xmath195 thus @xmath196 and @xmath197 can be replaced by equivalent expressions given by and which only depend on our new variables @xmath61 , @xmath62 , and @xmath178 .
we obtain a new system of equations , which is at least formally equivalent to the camassa
holm equation : @xmath198_0^\xi .
\end{aligned } \right.\ ] ] after differentiating we find @xmath199 from and , we obtain the system @xmath200 we can write more compactly as @xmath201 let @xmath202 we equip @xmath203 with the norm of @xmath174 , that is , @xmath204)}+{\left\vertf_\xi\right\vert}_{l^1([0,1])},\ ] ] which is equivalent to the standard norm of @xmath203 because @xmath205)}\leq{\left\vertf\right\vert}_{l^\infty([0,1])}\leq{\left\vertf\right\vert}_{l^1([0,1])}+{\left\vertf_\xi\right\vert}_{l^1([0,1])}$ ] .
let @xmath206 be the banach space defined as @xmath207 we derive the following lipschitz estimates for @xmath179 and @xmath188 .
[ lem : pq ] for any @xmath208 in @xmath206 , we define the maps @xmath209 and @xmath210 as @xmath211 and @xmath212 where @xmath188 and @xmath179 are given by and , respectively .
then , @xmath210 and @xmath209 are lipschitz maps on bounded sets from @xmath206 to @xmath203 .
more precisely , we have the following bounds .
let @xmath213 then for any @xmath214 , we have @xmath215 and @xmath216 where the constant @xmath217 only depends on the value of @xmath97 .
let us first prove that @xmath210 and @xmath209 are lipschitz maps from @xmath218 to @xmath219 .
note that by using a change of variables in and , we obtain that @xmath210 and @xmath209 are periodic with period @xmath14 .
let now @xmath208 and @xmath220 be two elements of @xmath218 . since the map @xmath221 is locally lipschitz , it is lipschitz on @xmath222 $ ] .
we denote by @xmath217 a generic constant that only depends on @xmath97 . since , for all @xmath223 in @xmath224 $ ] we have @xmath225 , we also have @xmath226 it follows that , for all @xmath192 $ ] , @xmath227 and the map @xmath228 which corresponds to the first term in is lipschitz from @xmath218 to @xmath219 and the lipschitz constant only depends on @xmath97 .
we handle the other terms in in the same way and we prove that @xmath210 is lipschitz from @xmath218 to @xmath219 . similarly , one proves that @xmath229 is lipschitz for a lipschitz constant which only depends on @xmath97 .
direct differentiation gives the expressions for the derivatives @xmath230 and @xmath231 of @xmath179 and @xmath188 .
then , as @xmath210 and @xmath209 are lipschitz from @xmath218 to @xmath219 , we have @xmath232 hence , we have proved that @xmath233 is lipschitz for a lipschitz constant that only depends on @xmath97 .
we prove the corresponding result for @xmath210 in the same way .
the short - time existence follows from lemma [ lem : pq ] and a contraction argument .
global existence is obtained only for initial data which belong to the set @xmath71 as defined below .
[ def : f ] the set @xmath71 is composed of all @xmath234 such that [ eq : lagcoord ] @xmath235 the set @xmath71 is preserved by the equation , that is , if @xmath73 solves for @xmath163 $ ] with initial data @xmath236 , then @xmath237 for all @xmath163 $ ] . the proof is basically the same as in @xcite .
[ th : global ] for any @xmath238 , the system admits a unique global solution @xmath239 in @xmath240 with initial data @xmath241 .
we have @xmath237 for all times .
let the mapping @xmath242 be defined as @xmath243 given @xmath244 and @xmath245 , we define @xmath218 as before , that is , @xmath246 then there exists a constant @xmath217 which depends only on @xmath97 and @xmath38 such that , for any two elements @xmath247 and @xmath248 in @xmath218 , we have @xmath249 for any @xmath163 $ ] . by using lemma [ lem : pq ]
, we proceed using a contraction argument and obtain the existence of short time solutions to .
let @xmath38 by the maximal time of existence and assume @xmath250 .
let @xmath251 be a solution of in @xmath252 with initial data @xmath253 .
we want to prove that @xmath254 from , we get @xmath255 hence , @xmath256
. this identity corresponds to the conservation of the total energy .
we now consider a fixed time @xmath257 which we omit in the notation when there is no ambiguity . for @xmath94 and @xmath258 in @xmath224
$ ] , we have @xmath259 because @xmath178 is increasing and @xmath260 . from , we infer @xmath261 and , from , we obtain @xmath262 hence , @xmath263 for some constant @xmath46 . similarly , one prove that @xmath264 and therefore @xmath265 and @xmath266 are finite . since @xmath267 , it follows that @xmath268 and @xmath269 . since @xmath270
, we have that @xmath271 is also finite .
thus , we have proved that @xmath272 is finite and depends only on @xmath38 and @xmath273 .
let @xmath274 . using the semi - linearity of with respect to @xmath275
, we obtain @xmath276 where @xmath46 is a constant depending only on @xmath277 .
it follows from gronwall s lemma that @xmath278 is finite , and this concludes the proof of the global existence .
moreover we have proved that @xmath279 for a constant @xmath217 which depends only on @xmath38 and @xmath280 .
let us prove .
given @xmath38 and @xmath281 , from lemma [ lem : pq ] and , we get that @xmath282 where @xmath217 is a generic constant which depends only on @xmath97 and @xmath38 . using again and lemma [ lem : pq ]
, we get that for a given time @xmath163 $ ] , @xmath283 hence , @xmath284 where @xmath285 is defined as in .
then , follows from gronwall s lemma applied to .
we denote by @xmath76 the subgroup of the group of homeomorphisms on the unit interval defined as follows : [ def : group ] [ def : g ] let @xmath76 be the set of all functions @xmath286 such that @xmath286 is invertible , @xmath287 the set @xmath76 can be interpreted as the set of relabeling functions .
note that @xmath288 implies that @xmath289 for some constant @xmath290 .
this condition is also almost sufficient as lemma 3.2 in @xcite shows . given a triplet @xmath291 , we denote by @xmath292 the total energy @xmath293 .
we define the subsets @xmath294 of @xmath71 as follows @xmath295 the set @xmath296 is then given by @xmath297 we have @xmath298 .
we define the action of the group @xmath76 on @xmath71 .
we define the map @xmath299 as follows @xmath300 where @xmath301 .
we denote @xmath302 .
[ prop : action ] the map @xmath303 defines a group action of @xmath76 on @xmath71 . by the definition it is clear that @xmath303 satisfies the fundamental property of a group action , that is @xmath304 for all @xmath79 and @xmath305 , @xmath306 .
it remains to prove that @xmath307 indeed belongs to @xmath71 .
we denote @xmath308 , then it is not hard to check that @xmath309 , @xmath310 , and @xmath311 for all @xmath312 .
by definition we have @xmath313 , and we will now prove that @xmath314 almost everywhere .
let @xmath315 be the set where @xmath178 is differentiable and @xmath316 the set where @xmath317 and @xmath286 are differentiable . using rademacher s theorem
, we get that @xmath318 . for @xmath319
, we consider a sequence @xmath320 converging to @xmath94 with @xmath321 for all @xmath322 .
we have @xmath323 since @xmath286 is continuous , @xmath324 converges to @xmath325 and , as @xmath178 is differentiable at @xmath325 , the left - hand side of tends to @xmath326 , the right - hand side of tends to @xmath327 , and we get @xmath328 for all @xmath329 . since @xmath330 is lipschitz continuous , one - to - one , and @xmath331 , we have @xmath332 and therefore holds almost everywhere .
one proves the other identity similarly . as @xmath333 almost everywhere , we obtain immediately that and are fulfilled .
that is also satisfied follows from the following considerations : @xmath334 , as @xmath335 is periodic with period @xmath14 .
the same argument applies when considering @xmath336 and @xmath337 . as @xmath61 is periodic with period @xmath14
, we can also conclude that @xmath338 .
as @xmath288 , one obtains that @xmath339 is bounded , but not equal to @xmath340 .
note that the set @xmath218 is invariant with respect to relabeling while the @xmath206-norm is not , as the following example shows : consider the function @xmath341 , and @xmath288 , then this yields @xmath342)}={\left\vertf(\xi)\right\vert}_{l^\infty([0,1])}.\ ] ] hence , the @xmath19-norm of @xmath343 will always depend on @xmath286 .
since @xmath76 is acting on @xmath71 , we can consider the quotient space @xmath80 of @xmath71 with respect to the group action .
let us introduce the subset @xmath81 of @xmath296 defined as follows @xmath344 it turns out that @xmath81 contains a unique representative in @xmath71 for each element of @xmath80 , that is , there exists a bijection between @xmath81 and @xmath80 . in order to prove this we introduce two maps @xmath345 and @xmath346 defined as follows @xmath347 with @xmath348 and @xmath208 , and @xmath349 with @xmath350 .
first , we have to prove that @xmath286 indeed belongs to @xmath76 .
we have @xmath351 and this proves .
since @xmath291 , there exists a constant @xmath352 such that @xmath353 for almost every @xmath94 and therefore follows from an application of lemma 3.2 in @xcite . after noting that the group action lets the quantity @xmath354 invariant , it is not hard to check that @xmath355 indeed belongs to @xmath296 , that is , @xmath356 where we denote @xmath357 .
let us prove that @xmath358 belongs to @xmath81 for any @xmath359 . on the one hand , we have @xmath356 because @xmath360 and @xmath361 as @xmath359 . on the other hand , @xmath362 and , since @xmath193 , we obtain @xmath363 thus @xmath364 .
note that the definition of @xmath365 can be rewritten as @xmath366 where @xmath367 denotes the translation of length @xmath368 so that @xmath369 is a relabeling of @xmath370 .
we denote by @xmath371 the projection of @xmath71 into @xmath81 given by @xmath372 .
one checks directly that @xmath373 .
the element @xmath374 is the unique relabeled version of @xmath370 which belongs to @xmath81 and therefore we have the following result .
[ lemma : bijection ] the sets @xmath80 and @xmath81 are in bijection .
given any element @xmath375\in{{{\ensuremath{\mathcal{f}}}/{g}}}$ ] , we associate @xmath376 .
this mapping is well - defined and is a bijection .
[ lem : equivpi ] the mapping @xmath377 is equivariant , that is , @xmath378 for any @xmath379 and @xmath380 , we denote @xmath381 , @xmath382 , and @xmath383 .
we claim that @xmath384 satisfies and therefore , since @xmath384 and @xmath385 satisfy the same system of differential equations with the same initial data , they are equal .
we denote @xmath386 .
then we obtain @xmath387(\eta)d\eta.\ ] ] as @xmath388 and @xmath389 for almost every @xmath312 , we obtain after the change of variables @xmath390 , @xmath391(\eta)d\eta.\ ] ] treating similarly the other terms in , it follows that @xmath392 is a solution of .
thus , since @xmath392 and @xmath393 satisfy the same system of ordinary differential equations with the same initial conditions , they are equal and is proved . from this lemma
we get that @xmath394 [ def : bars ] we define the semigroup @xmath85 on @xmath81 as @xmath395 the semigroup property of @xmath85 follows from . using the same approach as in @xcite , we can prove that @xmath85 is continuous with respect to the norm of @xmath206 .
it follows basically of the continuity of the mapping @xmath371 but @xmath371 is not lipschitz continuous and the goal of the next section is to improve this result and find a metric that makes @xmath85 lipschitz continuous .
let @xmath396 , we define @xmath397 as @xmath398 note that , for any @xmath396 and @xmath399 , we have @xmath400 it means that @xmath133 is invariant with respect to relabeling
. the mapping @xmath133 does not satisfy the triangle inequality , which is the reason why we introduce the mapping @xmath140 .
[ def : alberto ] let @xmath396 , we define @xmath401 as @xmath402 where the infimum is taken over all finite sequences @xmath88 which satisfy @xmath403 and @xmath404 . for any @xmath396 and @xmath399 , we have @xmath405 and @xmath140 is also invariant with respect to relabeling .
the definition of the metric @xmath401 is the discrete version of the one introduced in @xcite . in @xcite ,
the authors introduce the metric that we denote here as @xmath406 where @xmath407 where the infimum is taken over all smooth path @xmath408 such that @xmath409 and @xmath410 and the triple norm of an element @xmath174 is defined at a point @xmath370 as @xmath411 where @xmath412 is a scalar function , see @xcite for more details .
the metric @xmath406 also enjoys the invariance relabeling property .
the idea behind the construction of @xmath140 and @xmath406 is the same : we measure the distance between two points in a way where two relabeled versions of the same point are identified .
the difference is that in the case of @xmath140 we use a set of points whereas in the case of @xmath406 we use a curve to join two elements @xmath247 and @xmath248 .
formally , we have @xmath413 we need to introduce the subsets of bounded energy in @xmath296 .
we denote by @xmath414 the set @xmath415 and let @xmath416 .
the important propery of the set @xmath414 is that it is preserved both by the flow , see , and relabeling .
let us prove that @xmath417 for @xmath418 so that the sets @xmath419 and @xmath420 are in this sense equivalent
. from , we get @xmath421 which implies @xmath422 . by
, we get that @xmath423 and therefore @xmath424 . since @xmath425 and @xmath426 , the set @xmath427\mid
y_\xi(\xi)\geq\frac12\}$ ] has strictly positive measure . for a point @xmath428 in this set
, we get , by , that @xmath429 .
hence , @xmath430 and , finally , @xmath431 which concludes the proof of .
let @xmath432 be the metric on @xmath420 which is defined , for any @xmath433 , as @xmath434 where the infimum is taken over all finite sequences @xmath435 which satisfy @xmath403 and @xmath404 .
[ lem : linfbdj ] for any @xmath433 , we have @xmath436 for some fixed constant @xmath217 which depends only on @xmath97 . first , we prove that for any @xmath433 , we have @xmath437 for some constant @xmath217 which depends only on @xmath97 . by a change of variables in the integrals , we obtain @xmath438 we have @xmath439 from the definition of @xmath420 we get that , for any element @xmath440 , we have @xmath441 .
since @xmath442 , from , it follows that @xmath443 .
thus , yields @xmath444 we denote by @xmath217 a generic constant which depends only on @xmath97 .
the identity will be proved when we prove @xmath445 by using the definition of @xmath81 , we get that @xmath446 let @xmath447 .
similar to and , we can conclude that @xmath448 thus we have @xmath449 and analogously @xmath450 .
hence , @xmath451 by , we get that @xmath452 then , since @xmath453 we obtain that @xmath454 then , yields @xmath455 from and , and therefore follows . for any @xmath456
, we consider a sequence @xmath457 in @xmath420 such that @xmath403 and @xmath404 and @xmath458 .
we have @xmath459 since @xmath460 is arbitrary , we get . from the definition of @xmath140
, we obtain that @xmath461 so that the metric @xmath140 is weaker than the @xmath206-norm . [
lemma : distance ] the mapping @xmath462 is a metric on @xmath420 . the symmetry is embedded in the definition of @xmath133 while the construction of @xmath432 from @xmath133 takes care of the triangle inequality . from lemma [ lem : linfbdj ] , we get that @xmath463 implies that @xmath464 , @xmath465 and @xmath466 .
then , the definition of @xmath296 implies that @xmath467 .
[ rem : compmetric ] in @xcite , a metric on @xmath81 is obtained simply by taking the norm of @xmath206 .
the authors prove that the semigroup is continuous with respect to this norm , that is , given a sequence @xmath468 and @xmath370 in @xmath81 such that @xmath469 , we have @xmath470 .
however , @xmath85 is not lipschitz in this norm . from , we see that the distance introduced in @xcite is stronger than the one introduced here .
( the definition of @xmath206 in @xcite differs slightly from the one employed here , but the statements in this remark remain valid ) .
we can now prove the lipschitz stability theorem for @xmath85 .
[ th : stab ] given @xmath245 and @xmath244 , there exists a constant @xmath217 which depends only on @xmath97 and @xmath38 such that , for any @xmath433 and @xmath163 $ ] , we have @xmath471 by the definition of @xmath432 , for any @xmath456 , there exists a sequences @xmath457 in @xmath420 and functions @xmath472 , @xmath473 in @xmath76 such that @xmath403 , @xmath404 and @xmath474 since @xmath475 for @xmath476 , see , and @xmath477 is preserved by relabeling , we have that @xmath478 and @xmath479 belong to @xmath480 . from the lipschitz stability result given in , we obtain that @xmath481 where the constant @xmath217 depends only on @xmath97 and @xmath38 .
introduce @xmath482 and @xmath483 then rewrites as @xmath484 while rewrites as @xmath485 we have @xmath486 and similarly @xmath487 .
we consider the sequence in @xmath420 which consists of @xmath488 and @xmath489 .
the set @xmath414 is preserved by the flow and by relabeling .
therefore , @xmath490 and @xmath489 belong to @xmath420 .
the endpoints are @xmath491 and @xmath489 . from the definition of the metric @xmath432 , we get @xmath492 by using the equivariance of @xmath377
, we obtain that @xmath493 hence , by using , that is , the invariance of @xmath133 with respect to relabeling , we get from that @xmath494 after letting @xmath460 tend to zero , we obtain .
we now introduce a second set of coordinates , the so called eulerian coordinates .
therefore let us first consider @xmath495 .
we can define the eulerian coordinates as in @xcite and also obtain the same mappings between eulerian and lagrangian coordinates . for completeness we will state the results here .
[ def : d ] the set @xmath56 consists of all pairs @xmath65 such that 1 .
@xmath496 , and 2 .
@xmath68 is a positive radon measure whose absolute continuous part , @xmath497 , satisfies @xmath498 we can define a mapping , denoted by @xmath98 , from @xmath56 to @xmath499 : [ defl ] for any @xmath65 in @xmath56 , let @xmath500 where @xmath501 then @xmath502 .
we define @xmath503 .
thus from any initial data @xmath504 , we can construct a solution of in @xmath71 with initial data @xmath505 .
it remains to go back to the original variables , which is the purpose of the mapping @xmath97 , defined as follows .
[ def : ftoe ] for any @xmath79 , then @xmath65 given by @xmath506 belongs to @xmath56 .
we denote by @xmath97 the mapping from @xmath71 to @xmath56 which for any @xmath79 associates the element @xmath66 given by .
the mapping @xmath97 satisfies @xmath507 the inverse of @xmath98 is the restriction of @xmath97 to @xmath81 , that is , @xmath508 where the lagrangian variables are defined is represented by the interior of the closed domain on the left .
the equivalence classes @xmath375 $ ] and @xmath509 $ ] ( with respect to the action of the relabeling group @xmath510 ) of @xmath370 and @xmath511 , respectively , are represented by horizontal curves . to each equivalence class
there corresponds a unique element in @xmath81 and @xmath56 ( the set of eulerian variables ) .
the sets @xmath81 and @xmath56 are represented by the vertical curves.,width=377 ] next we show that we indeed have obtained a solution of the ch equation . by a weak solution of the camassa
holm equation we mean the following .
let @xmath512 .
assume that @xmath3 satisfies + ( i ) @xmath513 , + ( ii ) the equations @xmath514 and @xmath515 hold for all @xmath516 .
then we say that @xmath3 is a weak global solution of the camassa
holm equation . given any initial condition @xmath517 ,
we denote @xmath518 .
then @xmath519 is a weak global solution of the camassa
holm equation .
after making the change of variables @xmath520 we get on the one hand @xmath521y_\xi(t,\xi)d\xi
dt\\{\nonumber } & = -\iint_{{\mathbb{r}}_+\times { \mathbb{r}}}[u(t,\xi)y_\xi(t,\xi)(\phi(t , y(t,\xi)))_t-\phi_\xi(t , y(t,\xi))u(t,\xi)^2]d\xi dt\\ & = \int_{\mathbb{r}}u(0,\xi)\phi(0,y(0,\xi))y_\xi(0,\xi)d\xi\\{\nonumber } & \quad + \iint_{{\mathbb{r}}_+\times { \mathbb{r } } } [ u_t(t,\xi)y_\xi(t,\xi)+u(t,\xi ) y_{\xi t}(t,\xi)]\phi(t , y(t,\xi))d\xi dt \\{\nonumber } & \quad + \iint_{{\mathbb{r}}_+\times { \mathbb{r } } } u^2(t,\xi)\phi_\xi(t , y(t,\xi))d\xi dt\\{\nonumber } & = \int_{\mathbb{r}}u(0,x)\phi(0,x)dx\\{\nonumber } & \quad -\iint_{{\mathbb{r}}_+\times { \mathbb{r } } } ( q(t,\xi)y_\xi(t,\xi)+u_\xi(t,\xi)u(t,\xi))\phi(t , y(t,\xi))d\xi dt , \end{aligned}\ ] ] while on the other hand @xmath522 which shows that is fulfilled .
equation can be shown analogously @xmath523\phi(t , y(t,\xi))d\xi dt\\{\nonumber}&= \iint_{{\mathbb{r}}_+\times { \mathbb{r}}}[\frac{1}{2 } u_x^2(t , x)+u^2(t , x)-p(t , x)]\phi(t , x ) dx dt .
\end{aligned}\ ] ] in the last step we used the following @xmath524 \mid y_\xi ( t,\xi)>0\ } } u^2y_\xi + \frac{u_\xi^2}{y_\xi}d\xi=\int_0 ^ 1 \nu dx , \end{aligned}\ ] ] the last equality holds only for almost all @xmath55 because for almost every @xmath525 the set @xmath427 \mid y_\xi ( t,\xi)>0\}$ ] is of full measure and therefore @xmath526 which is bounded by a constant for all times .
thus we proved that @xmath3 is a weak solution of the camassa
holm equation .
next we return to the construction of the lipschitz metric on @xmath56 .
let @xmath527 note that , by the definition of @xmath85 and , we also have that @xmath528 next we show that @xmath529 is a lipschitz continuous semigroup by introducing a metric on @xmath56 . using the bijection @xmath98 transport the topology from @xmath81 to @xmath56 .
[ def : dd ] we define the metric @xmath530 by @xmath531 the lipschitz stability of the semigroup @xmath529 follows then naturally from theorem [ th : stab ] .
the stability holds on sets of bounded energy that we now introduce in the following definition .
given @xmath244 , we define the subsets @xmath532 of @xmath56 , which corresponds to sets of bounded energy , as @xmath533 on the set @xmath532 , we define the metric @xmath534 as @xmath535 where the metric @xmath432 is defined in .
the definition is well - posed as we can check from the definition of @xmath98 that if @xmath536 then @xmath537 .
we can now state our main theorem .
[ th : main ] the semigroup @xmath538 is a continuous semigroup on @xmath56 with respect to the metric @xmath2 .
the semigroup is lipschitz continuous on sets of bounded energy , that is : given @xmath244 and a time interval @xmath539 $ ] , there exists a constant @xmath46 which only depends on @xmath97 and @xmath38 such that , for any @xmath65 and @xmath540 in @xmath532 , we have @xmath541 for all @xmath163 $ ] . first , we prove that @xmath529 is a semigroup . since @xmath85 is a mapping from @xmath81 to @xmath81
, we have @xmath542 where we also use and the semigroup property of @xmath85 .
we now prove the lipschitz continuity of @xmath529 . by using theorem [ th :
stab ] , we obtain that @xmath543
[ prop : cont1 ] the mapping @xmath544 is continuous from @xmath6 into @xmath56 . in other words , given a sequence @xmath545 converging to @xmath496 , then @xmath546 converges to @xmath547 in @xmath56 .
let @xmath548 be the image of @xmath549 given as in and @xmath208 the image of @xmath550 given as in .
we will at first prove that @xmath551 converges to @xmath3 in @xmath6 implies that @xmath468 converges against @xmath370 in @xmath206 .
denote @xmath552 and @xmath553 , then @xmath554 and @xmath412 are periodic functions . moreover , as @xmath468 , @xmath555 , we have @xmath556 and @xmath557 , where @xmath558 and @xmath354 . by definition [ defl ]
, we have that @xmath559 and @xmath560 , and hence @xmath561 by assumption @xmath562 in @xmath563 , which implies that @xmath564 in @xmath19 , @xmath565 in @xmath41 , and @xmath566 .
therefore we also obtain that @xmath567 in @xmath19 .
we have @xmath568 then , since @xmath564 in @xmath19 , also @xmath569 in @xmath19 and as @xmath3 is in @xmath563 , we also obtain that @xmath570 in @xmath19 .
hence , it follows that @xmath571 in @xmath19 . by definition , the measures @xmath572 and @xmath573 have no singular part , and we therefore have almost everywhere @xmath574 hence @xmath575 in order to show that @xmath576 in @xmath577 , it suffices to investigate @xmath578 and @xmath579 as we already know that @xmath566 and therefore @xmath580 and @xmath335 are bounded .
since @xmath581 , we have @xmath582 for the second term , let @xmath583 .
then for any @xmath584 there exists a continuous function @xmath4 with compact support such that @xmath585 and we can make the following decomposition @xmath586 this implies @xmath587 and analogously we obtain @xmath588 . as @xmath567 in @xmath19 and @xmath4
is continuous , we obtain , by applying the lebesgue dominated convergence theorem , that @xmath589 in @xmath41 , and we can choose @xmath590 so big that @xmath591 hence , we showed , that @xmath592 and therefore , using , @xmath593 combing now , , and , yields @xmath594 in @xmath41 , and therefore also @xmath595 in @xmath41 .
because @xmath596 and @xmath597 are bounded in @xmath19 , we also have that @xmath576 in @xmath598 and @xmath595 in @xmath598 .
since @xmath580 , @xmath597 and @xmath599 tend to @xmath335 , @xmath185 and @xmath61 in @xmath598 and @xmath600 and @xmath601 , are uniformly bounded , it follows from that @xmath602 once we have proved that @xmath603 converges weakly to @xmath604 , this will imply that @xmath605 in @xmath598 . for any smooth function @xmath606 with compact support in
@xmath224 $ ] we have @xmath607 by assumption we have @xmath608 in @xmath598 .
moreover , since @xmath567 in @xmath19 , the support of @xmath609 is contained in some compact set , which can be chosen independently of @xmath590 .
thus , using lebesgue s dominated convergence theorem , we obtain that @xmath610 in @xmath598 and therefore @xmath611 form we know that @xmath603 is bounded and therefore by a density argument holds for any function @xmath606 in @xmath598 and therefore @xmath605 weakly and hence also in @xmath598 .
using now that @xmath612 shows that we also have convergence in @xmath41 .
thus we obtained that @xmath613 in @xmath206 .
as a second and last step , we will show that @xmath365 is continuous , which then finishes the proof .
we already know that @xmath567 in @xmath19 and therefore @xmath614 converges to @xmath615 .
thus we obtain as an immediate consequence @xmath616 and hence the same argumentation as before shows that @xmath617 in @xmath19 .
moreover , @xmath618 and again using the same ideas as in the first part of the proof , we have that @xmath619 in @xmath41 , which finally proves the claim , because of let @xmath621 and @xmath622 . by the definition of the metric @xmath2 , we have @xmath623 .
we immediately obtain that @xmath624 by lemma [ lem : linfbdj ] .
the rest can be proved as in ( * ? ? ?
* proposition 5.2 ) .
* acknowledgments . *
k. g. gratefully acknowledges the hospitality of the department of mathematical sciences at the ntnu , norway , creating a great working environment for research during the fall of 2009 . | we study stability of conservative solutions of the cauchy problem for the periodic camassa holm equation @xmath0 with initial data @xmath1 .
in particular , we derive a new lipschitz metric @xmath2 with the property that for two solutions @xmath3 and @xmath4 of the equation we have @xmath5 . the relationship between this metric and usual norms in @xmath6 and @xmath7
is clarified . |
hipparcos parallaxes for clumping he burning red giants in the solar neighborhood have recently raised a large interest in the astronomical community , providing a new valuable ( though controversial ) tool to approach the problem of magellanic clouds distances ( see , e.g. , udalski et al .
1998 , stanek , zaristski et al .
1998 , cole 1998 , girardi et al . 1998 ) . however , on theoretical grounds one has to notice that these parallaxes provide us for the first time with direct observational evidences for the luminosity of he burning stars whose red giant progenitors experienced electron degeneracy in the stellar core . therefore providing a relevant test for the rather sophisticated input physics supporting current theoretical predictions concerning galactic globulars and , more in particular , concerning some relevant issues as the luminosity of rr lyrae stars and the ages of halo stellar clusters .
such a test appears now of particular interest because of the growing rumor about a possible overluminosity of theoretical models for he burning stars with degenerated progenitors . as a matter of example , pols et al .
( 1998 ) found that the he clump luminosity in the open cluster m67 appears 0.2 - 0.3 mag fainter than predicted on the basis of their evolutionary computations .
such a discrepancy appears larger than the expected uncertainties on the bolometric correction and it appears further supported by independent evolutionary computations recently presented by castellani et al .
( 1999 ) , hereinafter c99 , which have discussed the severe difficulties in fitting the cm diagram of that clusters . however , in the meantime , girardi et al .
( 1998 ) presented a careful discussion on the luminosities of hipparcos he burning giants , which were found in splendid agreement with the adopted theoretical predictions . to throw light on such a scenario , in the next section we will discuss current predictions about the luminosity of he burning stars in a suitable mass range , disclosing the occurrence of sensitive differences among various authors .
section 3 will be devoted to a preliminary investigation of the uncertainties in theoretical results due to uncertainties on both the amount of extramixing from convective cores and the amount of mass loss in the pre - he burning phases . on this basis , in section 4 we will compare theoretical results with the observed mean luminosity of the red giant clump in the hipparcos sample , concluding for the possible need of a revision of the `` most updated '' input physics used in recent evolutionary models .
the origin of differences in the predicted luminosities are finally discussed in section 5 , by comparing the results of selected evolutionary codes to the light of the adopted physical ingredients .
a short section of concluding remarks will close the paper .
according to girardi et al .
( 1998 ) , the hipparcos sample of neighboring he burning giants is largely populated by stars with masses below or in the range of the so - called red giant branch phase transition ( rgb - pt ) . since the pioneering paper by sweigart et al .
( 1990 ) this range of stellar masses has been the subject of several careful evolutionary investigations . as well known ,
as we go from stars with m@xmath01 m@xmath1 to higher masses , we progressively find h - shell burning stars with a lower degree of electron degeneracy in their cores .
eventually , he is quietly ignited in the center of the structure . as a consequence ,
the mass of the he core at the he ignition progressively decreases , reaching a minimum at a star mass which depends on the original chemical composition . after this minimum ,
the he core grows again with mass , following the increasing size of the central convective core in the main sequence structures .
1 gives selected quantities concerning the behavior of he burning models across the rgb - pt as computed for z=0.02 and the two alternative assumptions y=0.27 or 0.23 .
all models have been computed according to the theoretical scenario already presented in c99 , which incorporates all the most recent evaluations of the input physics . as everywhere in the following ,
quantities given in fig.1 refer to the first model which , after igniting central he , has already reached the hr diagram location where it will spend the major phase of central he burning .
let us here notice that the luminosity of these he burning models is not related only to the mass of the he core .
initially this luminosity increases in spite of the decreasing he core , since the increased efficiency of the h burning shell overcomes the decreased output of he - burning reactions . however , eventually the decrease of the he core dominates and the luminosity reaches its minimum , whereas the lifetime in the central helium burning phase increases following the decrease of the efficiency of the he burning reactions .
the subject of the rgb - pt has been already widely debated in the literature ( see , e.g. , renzini & buzzoni 1986 , corsi et al .
1994 , girardi & bertelli 1998 and references therein ) and here it does not deserve further comments . however , fig . 2 compares the luminosities given in fig.1 with similar results but by girardi et al .
( 1998 : g98 hereinafter ) . to our surprise
, one finds that g98 luminosities appear systematically fainter than in c99 by about @xmath2log l / l@xmath1 @xmath0 0.1 @xmath3 0.15 , which is far from being a negligible amount and , in turn , it appears of the right amount to solve the already quoted m67 discrepancy . such an evidence prompted us to investigate similar data in the literature , aiming to find the origin of such a difference .
to this purpose , the same fig .
[ girardi ] shows the predictions already given in the literature on the basis of of the frascati or padua evolutionary codes before the last updating of the input physics .
the figure gives the comforting evidence that luminosities from castellani et al .
( 1992 ) appear in rather good agreement with similar data by bressan et al .
( 1993 ) . as we will further discuss in the next section , the slight underluminosity and the little difference in the rgb - pt mass of bressan et al .
models is only the expected consequence of their adoption of a moderate core overshooting scenario .
0.27 by various authors , as labeled ; `` over '' marks evolutions with core overshooting . ]
the same figure shows that the updated input physics adopted in c99 has the effect of increasing the luminosity of the models with a degenerate progenitors , according to the discussion already given in cassisi et al .
( 1998 ) and in reasonable agreement with stellar models recently presented by pols et al .
however , one also finds that the new input physics in the padua models ( girardi & bertelli 1998 , girardi et al .
1999 ) has the opposite effect , sensitively decreasing the predicted luminosities . as a whole
, one finds that uncertainties on predicted luminosities can be even larger than @[email protected] , leaving an unpalatable uncertainty in the current evolutionary scenario . on general grounds ,
one expects that the quoted differences are the results of differences either in the input physics or in the assumptions about the efficiency of macroscopic mechanisms , like core overshooting , which can affect the evolution of stellar structures . to discuss this point , in the next sections we will investigate the range of variability in current theoretical predictions , as produced by the various assumptions governing the evolutionary behaviour
making reference to the set of models presented in c99 , in this section we will explore the influence on central he burning models of several assumptions concerning these structures , namely , i ) the efficiency of core overshooting mechanisms , and , ii ) the effect of mass loss . in this way we aim to reach a clear insight on the `` solidity '' of the result
one is dealing with in the literature .
[ over ] ( upper panel ) shows the effect on the model luminosity of selected choices about the efficiency of core overshooting when the original stellar mass is varied between 1 and 3 m@xmath1 , while the lower panel in the same figure adopts the g98 representation to show the run of the same models in the hr diagram .
labels in these figure give the adopted amount of extramixing ( in unity of the local pressure scale height ) around the convective cores . in passing , note that comparison between this figure and fig.1 gives the already known evidence that the rgb phase transition shifts to lower masses as the metallicity decreases .
as already known , one finds that overshooting decreases the mass of the rgb - pt ( although it then occurs at a larger age ) and , correspondingly , that the maximum luminosity reached by the models before the transition decreases .
however , one finds that for moderate amounts of overshooting such a decrease is rather small and , in any case , models with masses of the order of 1.2 m@xmath1 or lower are little affected by such a mechanism . in addition the minimum luminosity attained at the rgb - pt varies by only @xmath2logl / l@xmath4 between a standard model and a model with @xmath5=0.25 .
ov ) , see text .
lower panel : the run of the same models but in the hr diagram . ] thus the differences in the assumptions about the efficiency of overshooting can hardly be at the origin of the differences in fig .
[ girardi ] and , in turn , they can not be used to reconcile pols et al .
( 1998 ) or c99 computations with m67 or hipparcos constraints . is shown for the four labeled values of the original mass .
dashed line gives the expected distribution for models having lost 10% of their original mass . ]
the effect of mass loss deserves a bit more discussion . here
we will assume that mass loss occurs in the advanced phase of h shell burning , so that the internal structure of the he burning star is not affected by such an occurrence , which only decreases the amount of envelope surrounding the central he core .
under this assumption , the effect of mass loss on he burning models can be easily computed by simply decreasing the envelope of the constant - mass model .
[ massloss ] maps the effect in the hr diagram of different amount of mass loss from the selected models .
the behavior depicted by data in this figure can be easily understood as follows : \i ) as long as models develop strong electron degeneracy ( i.e. , for masses lower or of the order of 1.5 m@xmath1 ) the mass of the he core at the he ignition is the result of rgb evolution . as a consequence it is largely independent of the evolving mass and , in addition ,
it is little affected by mass loss ( see , e.g. , the discussion in castellani & castellani 1993 ) , thus he burning models with mass losing progenitors behave like models without mass loss but starting their evolution with the actual mass of the he burning model . in this case the evolutionary behavior with mass loss can be easily predicted in terms of canonical models without mass loss .
\ii ) for larger stellar masses the he core at the he ignition is largely connected to the extension of the convective cores in the previous ms structures . as a consequence , the mass of the he core in he burning structures is now a sensitive function of the original stellar mass , whereas it keeps being little affected by mass loss , which mainly occurs in the post ms phases . in such case
theoretical expectations for the luminosity abandon the location of canonical models , as shown in fig .
4 . as an example , fig .
[ massloss ] gives theoretical predictions assuming for all stars a common mass loss of 10% .
this is intended to be a useful illustration of the effect we get assuming reasonable mass loss along the rgb .
however , one should keep in mind that the real situation may well be more complicate . assuming the reimers ( 1975 ) mass - loss rates , for instance ,
one finds that stars of higher mass lose less mass along the rgb .
this is so because they have lower rgb - tip luminosities , evolve at higher effective temperatures , and have a shorter rgb lifetime . in this case , stars more massive than about 1.5 @xmath6 would lose negligible amounts of mass on the rgb , contrarily to the lowest - mass ones ( see fig .
6 in girardi 1999 ) . in the context of the present discussion
, this means that mass - loss is expected to be less effective exactly in the mass range in which it would more affect the luminosity of the he - burning models , i.e. in the left part of fig .
3 we have explored the influence of physical effects which are badly understood and the treatment of which is usually done in parametrized form in stellar evolution codes . having discussed the scenario of these theoretical uncertainties , let us here attempt to use hipparcos constraints to test the already quoted different predictions concerning the luminosities of he burning stars . following the careful discussion given by g98 about the clump population ,
one finds that for any reasonable assumption about the age spread of field stars one predicts the bulk of the clump to be populated by stars below the transition mass , whose luminosity is practically insensitive to assumptions about the efficiency of overshooting .
coming back to fig.2 , one finds that at a solar metallicity c99 predictions give for these stars a luminosity larger than g98 by about @xmath2logl=0.1 .
more in general , comparison of data in g98 and c99 shows that the two evolutionary scenarios have a rather similar dependence of luminosities on the chemical composition , thus with the quoted systematic difference at any given metallicity .
`` sic stantibus rebus '' , the already quoted evidence that g98 evolutionary scenario appears able to nicely fit the hipparcos mean magnitude of clumping he burning stars , implies that c99 must predict too luminous giants , running against the hipparcos evidence .
this has been confirmed by independent simulation of the clump population based on c99 evolutionary tracks , as transferred into the cm diagram by adopting model atmospheres by castelli et al.(1997a , b ) .
data in fig.1 gives the additional evidence that reasonable variations in the assumed original he content can not decrease the c99 predicted luminosity by the required amount .
therefore one concludes that g98 evolutionary scenario works better .
however , for the sake of the discussion one has to notice that there is -at least in principle - a way to reconcile c99 prediction with hipparcos observations .
4 in this paper shows that a substantial amount of mass loss could lower the c99 predictions by the required amount of about @[email protected] . as an example
, one would require a mass loss by about 0.9 m@xmath1 for a 2.0 m@xmath1 , and by about 0.6 m@xmath1 for a 1.5 m@xmath1 model .
this , however , appear a too large requirement vis - a - vis current estimates for mass loss . taking also into account the uncertainties on evolutionary parameters of the field population in the solar neighborhood , we regard the previous discussion not as a proof , but at least as a suggestion that the most updated models , as c99 are , when dealing with the progeny of degenerated rg tend to give too luminous he burning models .
it appears of obvious relevance to address the problem of the discrepancies between the models by g98 and c99 , trying to understand them in terms of different descriptions of the input physics .
we leave out differences in the actual treatment of the input physics ( e.g. interpolation in and between opacity tables ) , which we think are of smaller influence and might contribute to a minor part of the variations in the clump luminosities as shown in fig . 2 .
since the most important quantity determining the core helium burning luminosity is the core mass at the helium flash , we will concentrate on this parameter .
[ f : mccomp ] shows that the 1 m@xmath1 g98 models have core masses lower by @xmath7 compared to c99 at the time of helium ignition at the rgb tip .
[ f : mccomp ] is made for solar composition but we checked that even for other metallicities and helium abundances the differences in the he core mass are very similar .
[ f : mccomp ] we have identified the following differences in the input physics used in the two evolutionary programs ( `` padua '' for g98 and `` franec '' for c99 models ) under consideration ( other aspects such as reaction rates , electron screening , mixing - length formalism , etc . , are to great extent identical ) : 1 . _ plasma neutrino emission _ :
g98 use munakata et al . ( 1985 ) for @xmath8 , while c99 use haft et al . (
1994 ) ; 2 .
_ electron conduction
_ : g98 employ hubbard & lampe ( 1969 ) , c99 itoh et al .
( 1983 ) ; 3 . _ radiative opacities _ :
the 92 tables by opal ( rogers & iglesias 1992 ) are used in g98 , while the 96 ones ( iglesias & rogers 1996 ) are those in c99 ; 4 .
_ equation of state _ : c99 use the new opal - eos tables ( rogers et al.1996 ) extended by that of straniero ( 1988 ) in those regions where the opal - eos does not exist ( see cassisi et al .
1998 for more details ) .
g98 use their own analytic eos , which takes into account partial ionization of hydrogen and helium , degeneracy and coulomb - effects , and which has been described in girardi et al .
( 1996 ) . the influence of some of these differences could be investigated quite easily because the two codes to some extent allow the selection of several sources of input physics .
the tests were done for different cases of initial chemical composition and mass . [ [ neutrino - emission . ] ] neutrino emission .
+ + + + + + + + + + + + + + + + + + for the composition @xmath9 , @xmath10 and an initial mass of @xmath11 we find that @xmath12 decreases from 0.482 to 0.476 ( -0.006 ) @xmath6 , if we switch from the haft et al .
( 1994 ) back to the older munakata et al . ( 1985 ) neutrino emission rates ( franec code ) .
this is the same result as cassisi et al .
( 1998 ; models 8 and 7 in their table 1 ) obtained for an @xmath13 model with @xmath14 , @xmath15 and also identical to what we find in the case of the padua - code for the same model . in the latter case , we also verified that using the munakata et al .
( 1985 ) emission down to @xmath16 increases the he core mass by @xmath17 .
therefore , the total budget of core mass reduction due to the g98 treatment of neutrino emission physics amounts to @xmath18 . recall that for the @xmath11 model , the g98 value for @xmath12 is @xmath19 smaller than that of c99 .
[ [ opacities . ] ] opacities .
+ + + + + + + + + + we checked the effect of electron conduction opacity by switching in the franec code from hubbard & lampe ( 1969 ) to itoh et al .
( 1983 ) electron conduction opacities .
the radiative opacities in these test cases are of a generation older than opal , but the differential effect of changing condution opacities can safely be assumed to be largely independent of the radiative opacities .
the test model was a @xmath20 star of @xmath9 , @xmath10 . the use of the older conductive opacities ( i.e. those used by g98 ) leads to a reduction of @xmath21 in @xmath12
. the influence of switching from the 92 to the 96 opal opacities was not tested , but according to our experience it should be minor compared to that of the electron conduction . [ [ equation - of - state . ] ] equation of state . + + + + + + + + + + + + + + + + + + the eos being a complicated part of both programs , we could not easily exchange one for the other .
however , we could perform the following test : we took the pressure and temperature stratification of a c99-model on the rgb and applied the g99-eos to it in order to obtain the density .
we compared with the original c99 density stratification .
the result is shown in fig .
[ f : eoscomp ] for a rgb model with y=0.238 , z=0.004 : while the core would have densities higher by about 1.5% in the padua code , the envelope would be less dense by 12% .
both effects reflect the evolutionary changes along the rgb , such that a g98-model would appear to be more evolved than a c99 one .
therefore , also the difference in the eos is expected to lead to a lower core mass at helium ignition for the g98 calculations .
[ f : eoscomp ] from these three investigations we conclude that with the differences in neutrino emission rates and conductive opacities we can explain almost 60% of the discrepancy in core mass .
this moves the g98 core masses already within the general spread of results .
at least part of the remaining difference can be ascribed to the eos . finally , we have compared results with three further codes , which use almost identical input physics as c99 , in particular with regard to radiative opacities , neutrino emission and electron conduction . with regard to the garching stellar evolution code
( see , e.g. , schlattl & weiss 1999 ) we find that @xmath12 is @xmath22 larger in the c99 models for a @xmath13 star of @xmath14 and two metallicities , @xmath23 .
this small difference can be understood as being a consequence of the fact that more ( @xmath24 ) helium is dredged up in the c99 models .
finally , evolutionary calculations by pols et al .
( 1998 ) and dominguez et al . ( 1999 ) with their respective codes ( pols et al . 1998 ; straniero et al . 1997 ) reveal for various chemical compositions a very high degree of agreement with those of c99 as shown in fig [ f : spscomp ] , including @xmath12 , which differs by less than 0.01 m@xmath25 between dominguez and c99 . model with solar compositions from pols et al .
( 1998 ) and from c99 ; lower panel : comparison between the 1.5 m@xmath25 model with solar compositions from dominguez et al . 1999 and from c99 . ]
[ f : spscomp ] we therefore can conclude that more than half of the difference in the core mass at helium ignition between g98 and c99 can be removed by adopting the same neutrino emission rates and electron conduction opacities ; using identical eos would lead to further convergence , although we can not quantify this point . at the same time , codes with identical physics do indeed result in very similar core masses which differ at a level of a few @xmath26 only .
therefore , the differences between the g98 and c99 results concerning this quantity are well understood as a consequence of different physical inputs .
note that taking from the literature ( see e.g. sweigart & gross 1978 ) @xmath2logl / l@xmath1 @xmath27 for stars with degenerate progenitors , * the difference in @xmath12 explains the difference in luminosity among the various models in fig.2 , with the exception of predictions by bressan et al .
( 1993 ) that reveal the contribution of some other difference than the core mass only . * as a conclusion , the different predicted luminosities we are dealing with appear as the natural results of evolutionary codes with different but in both cases reasonable input physics and thus as an example of the intrinsic unavoidable uncertainties in any current evolutionary scenario .
as it is well known , and as recently explicitly reiterated , e.g. , in castellani ( 1999 ) , the widespread use of the henyey algorithm in producing stellar models assures that different codes with identical input physics must produce quite similar results . in this paper we presented several evidences , as taken from the literature , supporting such an occurrence .
thus evolutionary models are as `` good '' as the adopted input physics is . in this paper
we have however shown that models based on different , but all `` reasonable '' , physics inputs may have sizable differences in the predicted luminosity of he burning low mass stars .
testing these theoretical predictions on the absolute magnitudes given by the hipparcos satellite for he burning stars clumping in the field , one finds several indications for preferring the luminosity predicted by the padua code , as used in girardi et al .
even if the argument is still open to further investigation , we feel that the discussion presented in this paper raises - at least - serious doubts on the capability of the most recent `` updated input physics '' of correctly predicting the luminosity of he burning stars as originated from progenitors which underwent strong electron degeneracy . therefore giving a serious warning about the uncritical acceptance of theoretical predictions concerning the luminosity of hb stars in galactic globulars and , in turn , all the related quantities . in this respect ,
we note that observational evidences concerning the luminosity of globular hb stars are far from being clear .
caputo et al . ( 1999 ) have recently discussed rr lyrae variables in the globular m5 showing that their pulsational properties possibly suggest a luminosity lower than predicted by c99 models . however , cassisi et al .
( 1999 ) and salaris & weiss ( 1998 ) used hb stars brighter than g98 predictions to derive cluster distance moduli which appear in excellent agreement with some independent evaluations based on hipparcos parallaxes for field subdwarfs . here
we can only conclude that the matter deserves further investigation , to decide whether or not the mass loss can reconcile `` most updated '' stellar models with hipparcos constraints for the clump of he burning stars in the solar neighborhood . before closing the paper ,
let us notice that all the computations discussed in the paper agree in predicting he burning stars systematically brighter when the metallicity is decreased .
the theoretical predictions are illustrated , e.g. , in the figure 1 of g98 , and figure 12 of c99 .
this has important implications for the method of distance determination based on clump stars , since it indicates that the clump population observed in nearby galaxies may have an intrinsic luminosity different from the local stars sampled by hipparcos . in the case of the magellanic clouds , this difference may amount to 0.2 - 0.3 mag , as inferred by cole ( 1998 ) and g98 .
it is a pleasure to thank o.r .
pols and o. straniero for kindly providing us with the evolutionary tracks from pols et al.(1998 ) or dominguez et al .
a. chieffi and m. limongi are acknowledged for helpful discussions . the finantial support of the `` ministero della universit e della ricerca scientifica e tecnologica '' to the project _ stellar evolution _ is kindly acknowledged .
bressan a. , fagotto f. , bertelli g. , chiosi c. , 1993 a&as 100,647 caputo f. , castellani v. , marconi m. , ripepi v. 1999 , mnras 306,815 cassisi s. , castellani v. , deglinnocenti s. , weiss a. , 1998 , a&as 129 , 267 cassisi s. , castellani v. , deglinnocenti s. , salaris m. , weiss a. , 1999 , a&as 134 , 103 castellani m. , castellani v. 1993 , apj 407 , 649 castellani v. , 1999 , in proceedings of `` evolving evolution '' , carloforte ( cagliari ) , t.zanzu , v. testa & m. bellazzini eds .
castellani v. , chieffi a. , straniero o. 1992 , apjs 78 , 517 castellani v. , deglinnocenti s. , marconi m. , 1999 , mnras 303,265 ( c99 ) castelli f. , gratton r.g .
, kurucz r.l . , 1997a , a&a 318 , 841 castelli f. , gratton r.g .
, kurucz r.l . , 1997b , a&a 324 , 432 cole a.a . , 1998 ,
apj 500 , l137 corsi c.e .
, buonanno r. , fusi pecci f. et al . , 1994 ,
mnras 271 , 385 dominguez i. , chieffi a. , limongi m. , straniero o. , 1999 , apj , 524,226 girardi l. , 1999 , mnras 308 , 818 girardi l. , bertelli g. 1998 , mnras , 300 , 533 girardi l. , bressan a. , chiosi c. , bertelli g. , nasi e. , 1996 , a&as 117 , 113 girardi l. , groenewegen m.a.t . , weiss a. , salaris m. 1998 , mnras 301 , 149 ( g98 ) girardi l. , bressan a. , bertelli g. , chiosi c. , 1999 , a&as , in press haft m. , raffelt g. , weiss a. , 1994 , apj 425 , 222 hubbard w.b .
, lampe m. , 1969 , apjs 163 , 297 iglesias c.a . , rogers f.j . , 1996 , apj 464 , 943 itoh n. , mitake s. , iyetomi h. , ichimaru s. , 1983 , apj 273 , 774 munakata h. , kohyama y. , itoh n. , 1985 , apj 296 , 196 pols o.r . ,
schroeder k - p , hurley j.r .
, tout c.a . , eggleton p.p .
1998 , mnras 298 , 525 reimers d. , 1975 , mem .
lige , ser .
6 , vol . 8 , p. 369
renzini a. , buzzoni a. , 1986 , in `` spectral evolution of galaxies '' , eds .
chiosi c. , renzini a. , dordrecht : reidel , p. 195
rogers f.j . ,
iglesias c.a . , 1992 ,
apjs 79 , 507 rogers f.j . , swenson , f.j .
, iglesias c.a . , 1996 , apj 465 , 902 salaris m , weiss a , 1998 , a&a 335 , 943 schlattl , h. , weiss , a. , 1999 , a&a , 347 , 272 stanek k.z . , zaritsky d. , harris j. , 1998 , apj 500 , l141 straniero o. , 1988 , a&as 76 , 157 straniero o. , chieffi a. , limongi m. , 1997 , apj 490 , 425 sweigart a.v . , gross p.g .
1978 , apjs 36 , 405 sweigart a.v .
, greggio l. , renzini a. 1990 , apj 364 , 527 udalski a. , szymaski m. , kubiak m. et al . , 1998 , acta astr . | this paper deals with theoretical predictions for he burning models in a range of masses covering the so - called red giant branch phase transition . taking as a guideline the observational constraints given by hipparcos parallaxes to the predicted luminosity of models originated from red giant progenitors with he core undergoing electron degeneracy , we compare models by various authors as recently appeared in the literature , disclosing sensitive differences in the predicted luminosity .
the `` solidity '' of these theoretical predictions is investigated by exploring the effects of varying the assumptions about the efficiency of core overshooting or the amount of mass loss , giving quantitative estimates of the related uncertainties .
however , one finds that theoretical predictions concerning the luminosity of the red giant clump in the hipparcos sample is scarcely affected by these mechanisms . *
a comparison among theoretical predictions as recently given by different authors convincingly demonstrates that the different luminosity predictions are the natural results of evolutionary codes with different - but all reasonable- input physics . * in this context observations suggests that stellar models based on the `` most updated '' input physics are possibly overestimating the luminosity of these structures , raising doubts on several current predictions concerning the luminosity of hb stars in galactic globulars . |
novel metamaterials @xcite , in particular those for which the spatial distribution of optical parameters can be specifically tailor made , have proven to be a viable route to realize various linear , nonlinear or tunable optical properties and processes . amongst them , the hyperbolic metamaterials ( hm ) - indefinite metamaterials , for which the permittivity and permeability tensors are negative along only certain of the principal axes of the metamaterial @xcite - attract interest because of possibilities of optical steering and manipulation . closely related are optical effects like negative refraction @xcite , hyperlensing @xcite , cancellation of reflection and transmission @xcite .
optical devices based on indefinite media are designed , like , e.g. , polarization beam splitters @xcite , angular filters @xcite , optical transmission modulator @xcite .
theoretical approaches were worked out to handle the negative refraction in anisotropic indefinite media have also been presented @xcite .
tunability and simple production become important requirements .
one of the few metamaterials which offer tunability @xcite , is nematic liquid crystal ( nlc ) doped with coated core - shell spheres ( ndlc ) @xcite .
recently we have demonstrated a negative refraction in near infra - red ( ir ) in this system with a planar nlc configuration for a wide interval of incident angles of electromagnetic ( em ) wave , using finite elements ( fe ) calculations ( comsol ) @xcite .
similar results in visible light range were presented in ref .
@xcite where it was shown that a hm consisting of nlc doped with a sufficiently high concentration of silver nanoparticles can become a hm ; all - angle negative refraction for homeotropic nlc orientation was demonstrated based on geometric arguments .
the point we would like to emphasize here is that a medium is indefinite is neither necessary @xcite nor a sufficient condition for negative refraction ; we shall show below that for a hm a positive refraction as well as reflection can take place .
thus , additional quantitative calculations are necessary to trace the path of em in an indefinite medium .
the aim of this paper is to design , using effective medium approximation and fe calculations , an electric field driven device for switching in ndlc hm between three scenarios : negative refraction , positive refraction and reflection for wide intervals of angles of incidence of em and of nlc director orientations . to this end ,
the approach based on negative refraction in anisotropic metamaterials @xcite is used .
consider a rectangular nlc cell with thickness @xmath0 along @xmath1 direction , filled with ndlc metamaterial described below , with planar alignment of nlc molecules in @xmath2 plane ( fig .
[ model](a ) ) .
the incident light with transverse magnetic ( tm ) polarization ( @xmath3 in @xmath2 plane and @xmath4 along @xmath5 direction ) propagates in @xmath2 plane and impinges as an extraordinary wave onto the nlc host .
the permittivity tensor for nlc reads @xcite : @xmath6 where @xmath7 denotes the angle between the @xmath8 axis and the director @xmath9 , @xmath10 , @xmath11 , @xmath12 .
the ndlc metamaterial @xcite comprises the host nematic liquid crystal containing uniformly distributed non magnetic spheres with core radius @xmath13 , made of polaritonic material and semiconductor shell ( drude material ) with thickness @xmath14 .
the effective permittivity and permeability of ndlc were calculated in ref .
@xcite using mie theory and maxwell garnet mixing rule : @xmath15 where @xmath16 , @xmath17 denotes the free space wavelength , @xmath18 permittivity of the nlc host along main axes , @xmath19 number density of the spheres , and @xmath20 scattering coefficients .
formula ( [ eq ] ) is valid in long - wavelength limit for low values of filling factor @xmath21 .
parameters used in this paper are : @xmath22 , @xmath23 , @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 , @xmath30 , @xmath31 , see ref .
@xcite for the definitions .
plots of the real parts @xmath32 , @xmath33 in function of @xmath34 are shown in fig .
[ model](b ) .
( color online ) ( a ) geometry of the system .
( b ) dependence of effective permittivity and permeability on permittivity of the lc host @xmath35 , eq .
( [ eq ] ) . ]
the propagation of a tm incident wave in this medium can be described by specifying @xmath36 in the propagation plane along principal axes and @xmath37 @xcite .
those components were calculated from eq .
( [ eq ] ) using the tensor components of @xmath35 along main axes : @xmath38 and @xmath39 @xcite . in this paper
we put @xmath40 .
then , @xmath41 for @xmath42 and @xmath43 for @xmath44 .
as the real parts of effective permittivities along principal axes are opposite in signs , ndlc is an indefinite medium .
the effective permeability is independent on @xmath35 in the interval of values relevant for this design ( fig .
[ model](b ) ) and we put @xmath45 . the tensor components @xmath46 in coordinates from fig .
[ model](a ) , calculated by rotating the diagonal tensor in principal axes around @xmath5 axis by angle @xmath7 , are shown in fig .
[ rotation ] . as angle
@xmath7 characterizes the direction of the optic axis which can be varied through the external electric field , we conclude that the ndlc becomes ( within the limits of validity of the effective medium theory ) a tunable hyperbolic medium . ,
@xmath47 and @xmath48 on rotation angle @xmath7 . ]
rotation of the director of nlc accompanied by the variation of the effective permittivity tensor changes the refraction at the boundary isotropic / indefinite medium . in geometric approach
this rotation promotes the rotation of the hyperbola of equifrequency contour in ndlc material @xcite . in the principal axes of effective permittivity tensor
this hyperbola determines the dispersion relation for tm waves in ndlc @xcite : @xmath49 where @xmath50 denotes the velocity of light in vacuum .
the case of @xmath51 is shown as inset in fig . [ all ] , where the equifrequency contour of an isotropic material is represented by a circle .
based on this diagram , the geometrical analysis of refraction can be done in a standard way @xcite by observing the continuity of the tangential components of incident ( @xmath52 ) and refracted ( @xmath53 ) wave vectors .
the direction of energy flow described by the poynting vector @xmath54 within the indefinite medium is indicated by the arrow drawn normal to the hyperbolic isofrequency surface .
the refraction angles for the wave beam @xmath55 and for the wave vector @xmath56 can be calculated analytically .
the same geometric approach is used in the case of rotated nematic / equifrequency contour ( see , e.g. , appendix in ref .
@xcite ) . in the case of uniform configuration of nlc ( @xmath51 )
the geometric approach predicts that negative refraction of the tm wave occurs for all angles of incidence @xmath57 ( solid line in fig .
[ all ] ) .
this prediction is confirmed by fe calculations ( symbols in fig . [ all ] ) .
[ gamm40 ] characterizes the energy flow and electromagnetic wave in the system for angle of incidence @xmath58 .
part ( a ) shows the magnitude of the electric field and poynting vectors , part ( b )
the @xmath5-component of magnetic field .
direction of vector @xmath53 for refracted wave and angle of negative refraction for poynting vector are in agreement with analytical results from fig .
[ all ] .
( color online ) plots of refraction angles @xmath59 ( solid line ) and @xmath60 ( dashed line ) .
symbols represent fe results @xcite . large symbol - see fig .
[ gamm40 ] .
inset : isofrequency contours for the tm wave incident from an isotropic material to an indefinite one with @xmath61 and @xmath62 for the angle of incidence @xmath63 . ]
( color online ) finite element simulations of the electric and magnetic field mapping for gaussian beam with tm polarization , @xmath58 .
magnitude of the electric field and poynting vectors ( a ) , @xmath5-component of magnetic field ( b ) . ]
using the formalism presented in the last section we have studied the behaviour of a plane tm wave incident from vacuum on the indefinite ndlc for a few chosen angles of incidence @xmath57 and for an arbitrary orientation of nlc director ( @xmath64 ) .
the dependence of refraction angle @xmath55 on @xmath7 is shown in fig .
[ diagram ] .
we have singled out three distinct regions .
we first discuss region ii that was calculated using analytical geometric approach and extends from @xmath65 to @xmath66 .
the location of its boundaries is independent on @xmath57 .
the negative refraction for the poynting vector is present for all angles of incidence @xmath57 , while positive refraction requires that @xmath67 .
for example , for @xmath63 negative refraction is present for @xmath68 , positive refraction for @xmath69 .
we have used the approach from ref .
@xcite to show that in this region the wave vector of the wave propagating in ndlc hm metamaterial is purely real which implies that the plane wave impinging from vacuum excites an homogeneous wave . in regions i ( @xmath70 ) and
iii ( @xmath71 ) the geometric construction breaks down , the wave vector becomes imaginary , which implies that the impinging plane wave excites an evanescent wave . to verify analytical geometric predictions
we have used fe calculations . in regions
i and iii total reflection occurs ( there was no transmitted wave ) while in region ii the simulations for @xmath63 ( symbols ) confirm theoretical analysis .
small discrepancies close to boundaries of region ii have numerical origin .
these results make possible a design of a novel electrooptic effect : external electric field driven real time switch between negative refraction , positive refraction and total reflection . to illustrate its functionality we have performed fe simulations of a tm gaussian beam with @xmath72 m impinging a ndlc hm cell with thickness @xmath73 m at angle of incidence @xmath63 . in fig .
[ fe ] the magnetic field mapping is shown for three operating states of the switch , which depend on the orientation of the nlc host controlled by external electric field : negative refraction for @xmath51 ( top ) ; positive refraction for @xmath74 ( middle ) and total reflection for @xmath75 ( bottom ) .
refraction angle @xmath55 in function of the orientation @xmath7 of nlc director for @xmath76 ( full circles ) , @xmath77 ( solid line ) , @xmath78 ( squares ) and @xmath79 ( triangles ) .
large circles represent fe calculations for @xmath63 . ]
( color online ) operating principle of the switch . left : orientation of nlc host , right : magnetic field mapping for a gaussian tm beam incident at @xmath63 .
negative refraction ( @xmath51 , top ) ; positive refraction ( @xmath74 , middle ) ; total reflection ( @xmath75 , bottom ) . ]
we have shown using effective medium approach and fe calculations ( comsol ) that nanosphere dispersed liquid crystal metamaterial becomes a tunable indefinite medium in near ir .
the tunability is provided by varying the external electric field which changes the orientation of the nlc director , the overall effective permittivity tensor and the orientation of the hyperbola of the dispersion relation of extraordinary waves through the medium .
the latter allows to control the character of electromagnetic field in the medium : propagating or evanescent .
we have exploited this tunability to design a new electrooptic effect for the manipulation of light : electric field driven switch between negative and positive refraction and reflection , and have illustrated its functionality .
compared to negative refraction in pure nlc medium @xcite the ndlc device operates in much wider intervals of incidence / refraction angles .
this work is supported by the air force office of scientific research ( afosr ) .
thanks polish national science centre for financial support under grant nn507 322440 .
k.t . acknowledges support of the foundation for polish science start program . | we show that nanosphere dispersed liquid crystal ( ndlc ) metamaterial can be characterized in near ir spectral region as an indefinite medium whose real parts of effective ordinary and extraordinary permittivities are opposite in signs . based on this fact we design a novel electrooptic effect : external electric field driven switch between normal refraction , negative refraction and reflection of tm incident electromagnetic wave from the boundary vacuum / ndlc . a detailed analysis of its functionality is given based on effective medium theory combined with a study of negative refraction in anisotropic metamaterials , and finite elements simulations . |
in @xcite kauffman defines an extension of classical knot diagrams to virtual knot diagrams , motivated by gauss codes on the one hand and knots in thickened surfaces on the other hand .
several classical knot invariants can be generalized to the virtual theory without much effort , e.g. , the knot group and derived invariants such as the alexander polynomial , the bracket and jones polynomials , and vassiliev invariants ( which can be introduced in different ways , see @xcite and @xcite ) . the present paper deals with a polynomial invariant that is derived from an invariant of links in thickened surfaces introduced by jaeger , kauffman , and saleur in @xcite .
the determinant formulation of the polynomial immediately generalizes to virtual link diagrams .
it is a laurent polynomial in two variables with integral coefficients that vanishes on the class of classical link diagrams but gives non - trivial information for diagrams that represent non - classical virtual links .
especially , examples can be given that the invariant is sensitive with respect to changes of orientation of a virtual knot .
furthermore , the polynomial fulfills a conway - type skein relation in one variable and thus it is denoted by the term conway polynomial . in the same way as in the classical case ,
the one - variable alexander polynomial of a virtual link can be derived from the virtual link group , but the skein - relation for ( a normalized version of ) the classical alexander polynomial can not be extended to the class of virtual links . therefore , this alexander polynomial is different from the conway polynomial mentioned above in a non - trivial way in contrast to the classical case ( and certain generalizations to links 3-manifolds , see for example @xcite , theorem 5.2.11 ) .
this paper is organized as follows .
after , in section [ virtknots ] , a short introduction into the field has been given , the determinant formulation of the conway polynomial for virtual links is described in section [ invariant ] .
some properties of the polynomial are deduced , especially , that the invariant fulfills a conway - type skein relation , and several example calculations are given .
then , in section [ alex ] , the alexander invariants derived from the link group , namely , alexander matrix , alexander ideal , and alexander polynomial , are defined and it is shown that the alexander polynomial does not fulfill any linear skein relation
. finally , in section [ conrem ] , general problems in extending certain invariants of classical links to the virtual category are described and the direction of further investigations is indicated .
in classical knot theory , knots and links in 3-dimensional space are examined . as a main tool ,
projections of such links to an appropriate plane are considered , namely , the so - called _ link diagrams _ ( see , for example , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
the idea of virtual knot theory is to consider link diagrams where an additional crossing type is allowed .
thus , for an oriented diagram , there are three types of crossings : the classical positive or negative crossings and the virtual crossings ( see fig . [ crossings ] ) . [ cols="^,^,^ " , ] since the changes of sign corresponding to reidemeister moves of type i are compensated by the factor @xmath0 in @xmath1 and since the definition of @xmath2 and @xmath3 does not depend on the virtual crossings of the diagram , the statement on @xmath4 follows immediately .
@xmath5 a normalization of the polynomial @xmath1 using rotation numbers , as done in @xcite , is not possible in the case of virtual link diagrams because of reidemeister move i. but due to the change of the variable `` @xmath6 '' to `` @xmath7 '' in comparison with @xcite , at least the sign of the polynomial is determined .
define the _ normalized polynomial _ @xmath8 as follows .
if @xmath1 is a non - vanishing polynomial and @xmath9 is the lowest exponent in the variable @xmath6 then define @xmath10 . otherwise let @xmath11 .
@xmath12 $ ] is an invariant of virtual links .
@xmath5 let @xmath13 be the virtual knot diagram depicted in fig .
[ trefoil ] .
( 4.5,4 ) ( 0,-0.5)y ( 2,3.7)1 ( 2,2.2)2 the diagram arises from a diagram of the right - handed trefoil where a positive crossing has been replaced by a virtual one .
the corresponding polynomial @xmath1 can be calculated by using the definition : @xmath14 -xy^{-1 } & 0 & -1 & 0 \\[1ex ] -1 & 0 & 1-x & -y \\[1ex ] 0 & -1 & -xy^{-1 } & 0 \end{array } \right ) = \ , x^2+x^2y^{-1}+xy - xy^{-1}-y-1\ ] ] the normalized polynomial @xmath8 is identical to the polynomial @xmath1 in this case .
[ vanish ] let @xmath13 , @xmath15 , @xmath16 be virtual link diagrams and let @xmath17 denote the disconnected sum of the diagrams @xmath15 and @xmath16 . then the following hold .
@xmath18 if @xmath13 has no virtual crossings @xmath19 , @xmath20 part a ) follows as in @xcite .
part b ) is an immediate consequence of the definition of the matrix @xmath21 .
@xmath5 for a connected sum @xmath22 of virtual link diagrams @xmath15 and @xmath16 , a formula of the form @xmath23 with a constant @xmath24 does not hold in general , in contrast to what could be expected at first glance .
for example , let @xmath15 be a diagram with non - vanishing @xmath4polynomial and let @xmath16 be a diagram of the trivial knot
. then the equation above obviously gives a contradiction .
the opposite of theorem [ vanish ] a ) does not hold in general .
a counter - example is given in fig .
[ balance ] . b the diagram represents a non - classical link since reversing the orientation of one of the two components yields a virtual link diagram with non - trivial @xmath4polynomial , whereas the @xmath4polynomial of the diagram with the original choice of orientations vanishes .
[ skeinrel ] let ( @xmath25 , @xmath26 , @xmath27 ) be a skein triple of virtual link diagrams . then the following skein relation holds : @xmath28 the formula can easily be checked by verifying the corresponding relation for every state @xmath29 in the state sum model used in @xcite to define the `` partition function '' @xmath1 .
@xmath5 since a skein relation as in theorem [ skeinrel ] is fulfilled , in the following @xmath1 and @xmath8 will be called _ conway polynomial _ and _ normalized conway polynomial _ , respectively .
a more familiar version of the conway skein relation can be achieved by setting @xmath30 and defining @xmath31 .
then , considering @xmath32 , theorem [ skeinrel ] immediately gives the following .
let ( @xmath25 , @xmath26 , @xmath27 ) be a skein triple of virtual link diagrams .
then the following skein relation holds : @xmath33 @xmath5 applying the skein relation of theorem [ skeinrel ] is , of course , very helpful for calculating the conway polynomial . for example , the virtual link depicted in fig .
[ balance ] can immediately be seen to have vanishing conway polynomial since changing an arbitrary crossing in the diagram yields the diagram of a hopf link and cutting open the same crossing gives a diagram of the trivial knot . for the diagram that arises from changing the orientation of one link component
, the corresponding skein tree has a branch that ends in a diagram with two virtual and two classical crossings which can not be simplified by any changes of the classical crossings .
therefore , latter diagram has to be calculated by using the definition of the conway polynomial . [ basis ]
let @xmath13 be a virtual link diagram .
then : @xmath34 @xmath35 @xmath36 does not depend on the over - under information of the diagram s classical crossings .
part a ) and b ) follow from the fact that summing up the columns ( rows ) of the determinant belonging to @xmath37 ( @xmath38 ) gives the trivial column ( row ) vector .
part c ) is an immediate consequence of setting @xmath39 in theorem [ skeinrel ] ( or in the definition ) .
@xmath5 by theorem [ basis ] , @xmath40 $ ] is an invariant of virtual links that is invariant with respect to changes of classical crossings , too .
thus @xmath36 can give information to distinguish those `` basic links '' whose conway polynomials must be calculated by the determinant formulation instead of applying the skein relation , i.e. , generating elements of the virtual skein module related to the conway skein relation . for example , the value of @xmath36 for the non - trivial diagram at the leaf of the skein tree that has been mentioned in the previous remark is non - zero .
therefore , the corresponding generating element is different from the trivial knot . for a virtual link diagram @xmath13 ,
let @xmath41 denote the _ mirror image _ of @xmath13 , i.e. , the diagram that arises from @xmath13 by changing the over - under information of every classical crossing .
if @xmath13 and @xmath41 are equivalent then @xmath13 is called _ amphicheiral _ and otherwise _
chiral_. furthermore , let @xmath42 be the _ inverse _ of @xmath13 , i.e. , the diagram with inverse orientation for every component of @xmath13 .
if @xmath13 and @xmath42 are equivalent then @xmath13 is called _ invertible _ and otherwise _ non - invertible_. in the case of classical links , it is known that kauffman and homfly polynomials , and therefore the conway polynomial too , are insensitive with respect to invertibility of a knot or link ( indeed , this is true for any invariant derived from quantum groups , see @xcite , @xcite ) .
surprisingly , the conway polynomial of a virtual knot can be different from the polynomial of its inverse .
the virtual knot with diagram @xmath13 that is depicted in fig .
[ notinv ] x is chiral as well as non - invertible . calculating normalized conway polynomials of @xmath13 , @xmath41 , @xmath42 , @xmath43 gives the following result : @xmath44 @xmath45 thus the four diagrams under consideration represent pairwise different virtual knots .
in the same way as for classical links , the _ ( virtual ) link group _ can be defined via the wirtinger presentation of a virtual link diagram , i.e. , a group generator is assigned to each arc of the diagram and a group relation can be read off at each classical crossing corresponding to the rules shown in fig .
[ wirtinger ] ( 3,2.5 ) ( 0,1)(1,0)2 ( 1,0)(0,1)0.8 ( 1,1.2)(0,1)0.8 ( 2.2,0.9)@xmath46 ( 1.2,0.3)@xmath47 ( 1.2,1.7)@xmath48 ( 3,2.5 ) ( 2,1)(-1,0)2 ( 1,0)(0,1)0.8 ( 1,1.2)(0,1)0.8 ( -0.4,0.9)@xmath46 ( 1.2,0.3)@xmath47 ( 1.2,1.7)@xmath49 ( for details see @xcite ) .
the link group is an invariant of virtual links .
the above definition of the link group is a purely combinatorial one and it has , in general , nothing to do with the complement of a link in 3-space as in classical knot theory . indeed , it can be shown that there exists a virtual knot group which is not the fundamental group of any 3-manifold . related work on virtual knot groups can be found in @xcite and @xcite .
as explained in @xcite , the _ alexander matrix _ corresponding to a group presentation can be calculated via a differential calculus . for a presentation with @xmath50 generators and @xmath51 relations ,
it is a @xmath52 matrix with entries from the ring @xmath53 $ ] .
the _ ( first ) alexander ideal _ @xmath54 of a diagram @xmath13 is generated by the @xmath55 minors of the alexander matrix for the corresponding link group and the _ ( first ) alexander polynomial _ @xmath56
$ ] is the greatest common divisor of @xmath54 .
the alexander ideal is an invariant of virtual links and the alexander polynomial is an invariant up to sign and up to multiplication by powers of @xmath57 . in @xcite alexander polynomials derived from an _ extended alexander group _ of a virtual link diagram
are considered .
indeed , the conway polynomial @xmath58 is related to the _ 0th virtual alexander polynomial _ of @xcite , see remark 4.2 therein .
in contrast to the classical alexander polynomial , the alexander polynomial for virtual links does not fulfill any linear skein relation as stated in the next theorem .
therefore it is crucially different from the conway polynomial defined in section [ invariant ] .
[ noskein ] for any normalization @xmath59 of the polynomial @xmath60 , i.e. , @xmath61 with some @xmath62 and @xmath63 , the equation @xmath64\ ] ] has only the trivial solution @xmath65 .
assume the above skein relation has a non - trivial solution .
consider the skein triples @xmath66 where @xmath13 is a classical knot diagram with one crossing and @xmath67 where @xmath68 is a standard diagram , with arbitrary orientations of the components , of the hopf link ( the latter triple corresponding to an arbitrary of the diagram s two crossings ) .
then the related alexander polynomials have values , up to normalization , as follows .
@xmath69 inserting these values , each multiplied by a factor @xmath70 , into the skein relation immediately shows that , up to normalization , @xmath71 and @xmath72 .
thus the skein relation is equivalent to the classical alexander skein relation : @xmath73 finally , let @xmath74 be the virtual link diagram arising from @xmath68 by changing an arbitrary classical crossing to a virtual crossing
. then @xmath75 and @xmath76 .
inserting these values , each multiplied by a factor @xmath77 , into the alexander skein relation yields a contradiction .
@xmath5 the proof of theorem [ noskein ] shows that the classical alexander skein relation is , up to a factor , the unique one that holds for the alexander polynomial of classical links , and this skein relation can not be extended to the class of virtual links . the virtual knot with diagram @xmath13 which is depicted in fig .
[ notinv ] can not be distinguished from its inverse by alexander polynomials since the alexander ideals are identical : @xmath78 c the virtual knot with trivial knot group and trivial jones polynomial investigated in @xcite , see fig .
[ kaufex ] , has non - vanishing normalized conway polynomial @xmath79 and therefore it does not represent a classical knot .
this is a result that could not be achieved with the means of @xcite .
as mentioned in @xcite , the alexander - conway polynomial for links in thickened surfaces can be defined in the multivariable case as well and the construction analogously yields a multivariable conway polynomial for virtual links . also the multivariable alexander polynomial can be derived from the virtual link group and it is an invariant that is different from the conway polynomial as has been seen above for the one - variable case .
it is quite natural to consider generalizations of homfly and kauffman polynomials to virtual links next .
besides the definition of these polynomials via skein relations , there are several state models for them , see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and also @xcite and citations therein .
but , when trying to generalize any of these approaches , one meets with at least one of the following three obstacles .
first , a virtual link diagram in general can not be unknotted by classical crossing changes .
therefore it is necessary to find a basis for the virtual skein module corresponding to the homfly and kauffman skein relations , respectively , which is not yet known .
secondly , some models make use of the signed graph corresponding to a black - and - white colouring of the plane , but it is not clear how to handle virtual crossings in a generalized model . and finally , state models mostly rely on rotation numbers which are not invariant with respect to reidemeister moves of type i. it should be mentioned that missing invariance under reidemeister moves of type i and i is a more serious problem when defining virtual link invariants than missing invariance under reidemeister moves of type i when defining classical link invariants . for example , having in mind theorem [ noskein ] , it is not possible to derive a linear skein relation for the alexander polynomial from the virtual link group .
the proof that is given by hartley @xcite for the classical case can not be extended to virtual link diagrams because it intrinsically uses a normalization of the alexander polynomial via rotation numbers during the proof .
observe that in classical knot theory a result by b. trace @xcite assures that a regular isotopy invariant can always be made to an ambient isotopy invariant using writhe and rotation number of an oriented link diagram . a more general class of invariants that may be generalized to virtual links
are _ quantum link invariants_. in @xcite it is described how to define such invariants via @xmath80-matrices and the abstract tensor approach , see also @xcite .
again , the invariance under reidemeister moves of types i and i causes difficulties and thus the virtual quantum link invariants under consideration are only invariant with respect to _ virtual regular isotopy _ ,
i.e. , the two troublemaking moves are avoided .
the conway polynomial for virtual links that has been introduced in the present paper also arises from a quantum link invariant ( see @xcite ) but it is not difficult to get control over its behaviour with respect to reidemeister moves of types i and i. it is left to further investigations which ( quantum link ) invariants have similar properties and can be extended to full invariants of virtual links .
the author would like to thank saziye bayram for finding an error in a calculation , and dan silver for informing him about his results . | a polynomial invariant of virtual links , arising from an invariant of links in thickened surfaces introduced by jaeger , kauffman , and saleur , is defined and its properties are investigated .
examples are given that the invariant can detect chirality and even non - invertibility of virtual knots and links .
furthermore , it is shown that the polynomial satisfies a conway - type skein relation in contrast to the alexander polynomial derived from the virtual link group .
+ _ keywords : _ virtual knot theory , conway skein relation , alexander invariants + _ ams classification : _ 57m25 * on alexander - conway polynomials + for virtual knots and links * + jrg sawollek + december 21 , 1999 ( revised : january 3 , 2001 ) |
just like our sun , the line profile of hd209458 shows a double - horn emission with the distinguishing feature of intervening interstellar gas strongly absorbing the line core . from the signal left at the earth s orbit ,
@xcite ( hereafter ) initially claimed that hst / stis medium resolution observations of hd209458 are showing a strong absorption of 15% during transit and that a large asymmetry appears on the blue side of the absorption profile .
those authors used the 15% drop - off of the stellar signal during transit as an indication of the presence of a hydrogen cloud that extends beyond the roche lobe of the planet .
the next step for was to use the asymmetry claimed in the blue side of the line during transit as an indication that the evaporated gas is accelerated by the radiation pressure from the star , an acceleration that results into the build - up of a cometary shape of the nebulosity that escapes from the planet away from the star .
+ later on , @xcite rejected the acceleration process by radiation pressure , arguing that neutrals will not have the time to be accelerated by the radiation pressure before they get ionized .
furthermore , they proposed that observations thus far reported by on hd209458b could be explained by the presence of a population of energetic neutral atoms produced by charge exchange between the neutral atmosphere and the stellar wind ions beyond an obstacle region corresponding to the planetary magnetopause assumed around @xmath1 r@xmath2 from the center of the planet . in the meantime
, we revisited the hst / stis medium resolution archive data to report that exoplanet hd209458b absorbs @xmath3 of the parent star flux during its transit , much less than previously reported by ( @xcite , hereafter ) .
we also reported that no sign of asymmetry could be found in the stellar line profile during transit , a finding that directly contradicts the claims initially reported in . soon thereafter , the absorption drop - off reported by for a specific wavelength range was used out of the spectral information provided in fig.3 of to claim that evaporation was strengthened with a factor @xmath4 asymmetry in the absorption between the blue ( @xmath5 ) and red ( @xmath6 ) sides of the line during transit ( @xcite hereafter ) . in any case , it appears that our initial selection of the wavelength window to estimate the transit absorption created some confusion as to the reading of the data analysis presented in , and this confusion adversely affects efforts to discriminate between the processes thus far proposed to explain the hydrogen distribution around hd209458b ( , @xcite ) . in the following , we revisit available hst / stis medium resolution archive data to conduct a sensitivity study that should dissipate the existing confusion , an important step in formulating clear observational constraints on the steady - state properties of the hydrogen nebula around hd209458b .
+ data analysis is carefully described in section 2 so that light curves may be extracted ( in selected wavelength windows within the stellar line ) and studied in section 3 . in section 4 , time variability of the hd209458 emission is examined and further evidence is provided on the variable behaviour of the hd209458 emission line , a property that may not be well - described by proxies derived for the solar line .
the absorption spectral profile is then derived as the ratio between the stellar flux during transit to the unperturbed spectrum of the star with no transit effect .
finally , the adequacy of models thus far proposed to explain the hydrogen distribution around hot giant exoplanets is discussed in the light of the constraints derived here , with new solutions proposed for the h nebula distribution .
four hst visits during the planetary transit of hd209458 are considered , each composed of two to three long exposures ( @xmath7 ) distributed in time over the transit period ( e.g. , table 1 ) .
all exposures were obtained with the hst / stis g140 m medium resolution grating using the 52``x0.1 '' slit in the time - tag mode .
it is important to emphasize that the time - tag mode is a powerful tool that keeps track every @xmath8 of photon events during each exposure .
this mode is generally used to improve the spectral and imaging resolutions by correcting for the different motions that may affect the recorded signal ( @xcite ) .
doppler shifts due to hst motion , thermal drifts , and spacecraft jitter / drift are the main sources of motions . in the standard mode ( accum ) used in several studies ( , ) , correction is made only for the doppler motion .
for the other sources of motion , we used the time - tag mode , which offers the possibility of cancelling these motions by taking short sub - exposures that may be cross - correlated .
in addition , with its high time resolution , the time - tag mode also offers the opportunity to monitor the time variation of the recorded signal , allowing the study of fast events and/or accurate correction for variable features by also considering short time exoposures .
+ size , making each pixel covering 0.024 " over the sky .
the dispersion direction is horizontal so each line of the image is a spectrum .
the spatial direction is vertical .
the position of the stellar spectrum along the slit is indicated by an arrow .
the bright vertical band is the sky background distribution over the sky .
two gaps , corresponding to the stis fiducial bars ( @xcite ) , appear at the bottom and top of the slit .
[ fig1 ] ] following , our set of eleven exposures are sampled here using a time bin of @xmath9 from the available time - tag information .
next , each sub - exposure is calibrated through the stis pipeline .
because the hst / stis / g140 m grating is a spectro - imaging instrument , every exposure provides an image of the sky along the slit and the corresponding spectra on the cross dispersion direction ( fig .
1 ) . compared to high resolution observations ( @xcite ) , the overlap between the stellar and sky signals is worse in the medium resolution , thereby making the corresponding correction more difficult ( fig .
2a ) . on average ,
the altitude of the hst orbit is high ( @xmath10 km ) but still within the earth s hydrogen geocorona .
therefore , any observation is contaminated by the emission and absorption from that upper layers of the earth s atmosphere and by the absorption and emission from the interstellar medium . insofar as the star is a point - like source that has few pixels extented over the slit spatial direction ( fig .
1 ) , the other areas of the slit have only the sky background emission which may a priori be used to correct for that contamination . as shown in figure 2b , the difficulty of this correction comes from the uncertainty about whether the geocorona s signal strength when estimated from a detector sector along the stis slit is different from the one where the stellar signal was recorded . comparing the sky background signal from different sectors of the detector along the slit and for different conditions of observation , we confirm the conclusion of that the stis mamma detector has an inherent non - uniformity corresponding to an incompressible uncertainty of @xmath11% on extended sources that is comparable to photon statistical errors .
+ at this level , the correction for the sky contamination is not yet complete .
indeed , we still need to correct for the absorption by the geocoronal gas . while it may not be necessary when the target s signal is nearly zero in the wavelength range around the spectral position of the line in the reference frame tied to earth even so
, we must check how the line spread function ( lsf ) of the instrument expands the absorption signature over the spectral domain of interest .
in addition , the spectral position of the geocoronal absorption changes from one exposure to another in the heliocentric reference frame depending on the day of observation ( see table .
1 ) . assuming standard models for the earth s upper atmosphere , the geocoronal opacity was evaluated for every exposure before the corresponding absorption profile is derived , convolved with the stis / g140 m lsf , and then applied to the signal after the sky background emission was subtracted ( @xcite ) .
fortunately , most of the geocoronal absorption takes place within the spectral domain that we excluded from our data analysis so that its impact is less than 1% in the useful domain . finally , because the interstellar absorption does not really change from one exposure to another , and should not affect our analysis of the transit effect
, no correction is applied for this feature .. + after all corrections were applied , the whole data set resulted in 53 bins time series of the hd208458 planet - star system versus the orbital phase angle that is quite similar to our initial time series within the range of statistical errors ( e.g. , fig . [ fig1 ] of ) . to ensure that all the effects cumulated ( lsf , long slit , thermal motion , etc . )
did not corrupt the stellar signal per wavelength pixel at the @xmath12% level , the wavelength domain of contamination by the sky background was defined as @xmath13 $ ] nm , the same as in .
to obtain the trend of the planetary transit in , integrated the time series spectra in the range @xmath14 $ ] nm on the blue side of the stellar line together with the range @xmath15 $ ] nm on the red side . accumulating the signal inside the planetary transit period , previously derived a @xmath16% drop - off of the stellar intensity .
the question then is : does the absorption rate depend on the selected wavelength range ?
for example , using low resolution hst / stis observations of hd209458 , @xcite found an absorption rate of @xmath17 % for neutral hydrogen , a rate much smaller than they derived when using hst / stis medium resolution observation .
recently , claimed that the difference between our estimate ( @xmath18 ) and their initial absorption rate ( @xmath19 ) comes from the distinct integrated wavelength ranges considered .
these authors also claimed that they can reconcile medium- and low - resolution observations using the same argument of difference in the wavelength range namely , full line in the low - resolution case compared to restricted windows around the stellar line peaks in the medium - resolution observations analysis .
therefore , it is important to verify these claims and to estimate the sensitivity of the transit absorption to the selected wavelength range . + starting from the time series of the line emission ( e.g. , figure 1 in ) , we have generated light curves ( lcs ) for several wavelength windows , including those considered by and more recently by . as an initial step , we tried to compare lcs corresponding separately to the blue and red line sides ( fig .
accumulating the signal from three bins inside transit with the corresponding time weights , our analysis shows an absorption @xmath20 for the blue side and @xmath21 for the red side ( see table 2 for details ) .
as shown in figure 3 , the two lcs behave similarly over time , except for the last bin around time @xmath22 where the red side seems even more absorbed . despite the noise level
, the observed trend for the two lcs supports the idea that both sides of the stellar line are absorbed almost equally during transit .
by contrast , for the same wavelength domains , recently reported absorptions of @xmath5 and @xmath6 , respectively , for the blue and red sides of the line , suggesting a rather large asymmetry between the two sides .
evidently , our results differ significantly from those of when considering the two sides of the stellar line separately .
however , when merging signals from the two sides of the line , the corresponding absorptions seem comparable for the two analyses , yet this similarity ( @xmath23 here and in compared to @xmath24 in ) is only incidental . + in a second step , we varied the wavelength domain of integration over the stellar line .
as shown in figure 4 , both our selected wavelength window and the one of give quite similar lcs with a transit absorption rate around @xmath25 for our spectral domain and @xmath26% ( compared to @xmath27 in ) for the domain selected by .
in addition , when considering the whole stellar line , our absorption rate remains around @xmath28 , consistent with our initial estimate and also with recent analysis using hst / acs observations ( @xcite ) .
+ what , then , is the physical origin of this weak sensitivity of the absorption during transit to the selected wavelength range ? before we proceed to answer this question , in the following we study the time variability of the stellar in order to explore its potential impact on the transit absorption line profile .
time variability of the stellar signal is considered here as a potential cause for the discrepancy between the different data analysis related to the hst / stis / g140 m medium resolution observations . and used the solar variability as a proxy to catch the variability of the emission line of hd209458 ( solar type star ) .
this approach may not be adequate because no two stars of the same type behave similarly over short time scales .
moreover , hd209458 has a nearby orbiting exoplanet that may induce tidal and electromagnetic modifications of the stellar signal , effects that do not exist for our sun . as a matter of fact
, hd209458 was previously suspected to have a relatively moderate chromospheric activity from caii h and k lines that were recorded over full orbits of the system ( @xcite ) .
more recently , time variability of the stellar flux was noted and showed both for a specific wavelength range and at different orbital phases ( see fig .
2 & 3b in ) . however , the noise level was relatively high making difficult the diagnostic on the origin of the signal variability . another possible way to characterize
time variability is to carefully inspect the individual data set so far obtained by comparing the stellar signal at different phase positions to test any correlation with the orbital position of the planet .
such correlation may result from unkown electromagnetic interactions suspected to operate between the planet and the star during transit ( @xcite ) , much like what we know from the jupiter - io or jupiter - ganymede systems ( @xcite ) .
+ a careful inspection of the four visits thus far available in the archive reveals that , indeed , the data subset obtained during the third transit of the observing program ( data files o6e203010 , o6e203020 , and o6e203030 ; see table 1 ) show extra features that are not observed in the other visits of the hd209458 system ( fig . 5 ) .
some of the observed features are well above the noise level and are suspected to correlate with the planet orbital position , particularly for the spectral feature that appears around @xmath29 nm in the far blue side of the stellar line .
if we restrict the analysis to the solely line profile in the range @xmath30 km / s from the line center , it is difficult to descriminate a time variability from statistical variations at low .
yet we can estimate such time variability by comparing integrated fluxes from the different exposures as shown for example in fig . 3 of and fig . 2 of .
in these figures , the data points corresponding to the third transit ( the so called c points ) behave strangely with a departure of up to 3 @xmath31 far from the theoretical transit light curve .
all these indications tend to support that data set o6e2030x0 bears an unusual variability that is not found in the other exposures .
the extra features are not related to any earth s energetic particles activity along the hst orbit , as all observational parameters are comparable to the other data sets .
we suspect that the data set o6e2030x0 may have corrupted the diagnostic on the transit depth and absorption profile in analysis . in our case , the impact of the same data set was largely diminished by the splitting of the exposures into our time series and by having extra observations ( data files o4zea40x0 , see table 1 ) strengthening the weight ( @xmath32% more ) of the pool of `` quiet '' exposures .
finally , the uniqueness of the disturbed data set o6e2030x0 and the noise level make it difficult to definitely assess a correlation with the orbital position of the planet , a question that remains open until future observation of the hd209458 system can be obtained after the fuv capabilities of the hst are restored .
in practice , to obtain a self - consistent absorption profile , both steady - state unperturbed ( out - of - transit ) and in - transit line profiles are required and will be processed by iteration . in the present case ,
the iteration process converged in only a few steps , providing the absorption profile as shown in figure 6 .
the obtained absorption spectral profile is typical of damped absorption systems , though the spectral domain around the line center is hidden by , respectively , the interstellar medium absorption and the contamination from the sky background .
the first thing to note is that the atmospheric absorption is statistically flat far into the stellar line wings down to @xmath33 nm on the blue and up to @xmath34 nm on the red , including the spectral position of the stellar line peaks .
as the peaks intensity provides the highest contribution to the lc shown in figures 3 - 4 , it becomes apparent that a flat absorption of @xmath3 proposed in our initial study ( ) is a rather good approximation . beyond that frequency domain ,
it is difficult to accurately assess the spectral variation due to the noise level . + binning the absorption line profile in wavelengths improves the signal quality , yet it does not reveal any asymmetry between the blue and red sides at a reduced spectral resolution of @xmath35 . as previously shown in figure 3
, fluxes integrated , respectively , on the blue and red sides of the emission line behave quite similarly over time , supporting the contention that both wings are absorbed almost symmetrically during transit in the spectral domain @xmath36 from line center . any difference , if at all , must be weak and/or occur outside the available spectral domain such that current data noises and/or the sky background contamination hide it .
+ in the following , we take a theoretical approach to verify whether such observational properties of the absorption profile are predictable , particularly its lorentzian shape ( fig .
first , we recall some theoretical background on the opacity properties within a strong conservative line like . typically for a resonance line ,
the absorption is dependent on wavelengths with strong opacities in the line core and much less in the ( absorption ) line wings . for a typical temperature of @xmath37k for the hd209458b inflated atmosphere ( @xcite ) ,
the absorption doppler width is almost @xmath38 nm ( or @xmath39 ) , which is comparable to the spectral resolution ( @xmath39 ) of the hst observations used here . referring back to the hd209458 line profile ( , ) , the line s peaks appear at @xmath40 , or equivalently , @xmath41 doppler units far from the line center , placing them well in the wings of the absorption line profile of the planetary atmosphere . because of the steep fall off of the flux in the stellar wings , the contribution of the signal around the peaks positions is dominant in any estimate of the flux attenuation during transit
also , one should note that for a target that is not spatially resolved , such as transiting exoplanets , the observed signal extinction is a weighted average of absorption from different regions of the exoplanetary extended atmosphere , with the outer layers having the maximal weight .
this could be examined by writing the basic definition of the observed ( transmitted ) stellar line profile during full transit : + @xmath42 where @xmath43 is the absorption profile by h atoms versus the velocity @xmath44 ( or wavelength ) from the line center , @xmath45 is the unperturbed stellar line profile , and @xmath46 is the opacity along the line of sight that goes through the cloud s volume at the impact parameter @xmath47 from the star s disc center and a polar angle @xmath48 . the integration is evaluated over the whole extent of the stellar disc , allowing various distributions of the gas around the planet that may lead to the same observed extinction .
this degeneracy would not appear if the atmospheric layer were thin compared to the size of the planet .
+ to illustrate the multiplicity of solutions , a standard atmospheric model of hd209458b ( @xcite ) was used . as shown in figure 6 , a rather good fit to the absorption line profile during transit
is obtained for the _
div1 _ model of @xcite .
the div1 model includes heavy elements contribution to the h production and appears up - to - date regarding chemistry reactions .
if using the standard model of @xcite that does not include heavy elements chemistry , the same good fit is obtained when the h abundance is scaled by a constant factor . beyond the difference between @xcite and @xcite models , the scaling constant may also reflect other processes of h atomic production that are not accounted for in classical photochemistry models . in the past ,
the study of the thermosphere of giant planets such as jupiter and saturn and their satellites showed the need for such enhancement of the h production in order to fit the observations thus far obtained by voyager 1 & 2 , iue , galileo , or hst ( see , for instance , @xcite ) .
for example , in the case of jupiter , a planetwide enhancement ( @xmath49 times ) of the h production is expected from magnetospheric protons and electrons precipitation in the polar regions ( mainly h2 or heavy molecules dissociation by energetic particles ) . for reference
, the same process produces the classical auroral emissions .
the constituents produced by auroral processes and the input energy and momentum in the polar regions are then transported and distributed over all latitudes / longitudes ( @xcite ) . as of today ,
the h abundance and distribution on jupiter are not yet fully uncovered ( @xcite ) .
the idea here is that the magnetospheric energy input in the auroral regions may play a key role in the upper atmosphere of hd209458b s composition and structure .
such effects are not taken into account in atmospheric models thus far published in the literature . on that basis , for a planet such as hd209458b orbiting at 0.05au from its parent star , thereby receiving much more energy through its polar regions than expected on jupiter , the conditions may favor a rather strong enhancement of the h production that may lead to larger atmospheric opacities at than shown in fig .
15 of @xcite .
+ in the model calculations shown in figure 6 , a voigt function was assumed for the h atomic absorption profile in order to account for the natural broadening of the line . as expected from the line shape ,
the atmosphere is opaque and consistent with the line - of - sight opacities used ( see figure 15 in @xcite ) .
the line shape reflects the effect of the broadening in line wings by the large h column along the line of sight that leads to a large and relatively flat absorption far from the line center .
however , it is important to point out that the interpretation of the absorption level versus wavelengths is not straightforward because the absorption is a weighted contribution from different atmospheric layers located at different distance from the planet s center ( e.g. , equation 1 ) .
for that reason , we stress that the fit to the absorption profile and the corresponding atmospheric opacities correspond to one possible sketch of the h distribution that may not contain all of the key physical processes at play in the real spatially extended system .
other scenarios of gas distribution may lead to the same observed absorption profile ( e.g. fig .
for example , we are able to find a more extended atmosphere for which the extinction by the exoplanet atmosphere fits rather well with the observed absorption profile ( see fig . 6 ) . as it is impossible to obtain any information on the line core because of the interstellar medium absorption and the sky background contamination , only very high fuv observation at a level better than @xmath50 may disentangle the different processes in play .
at this point , we are able to compare our result with previous analysis that used most of the same data set ( , ) .
the immediate question is , what could be the origin of the difference in the transit absorption rate as derived for the same wavelength domain @xmath51 [ v1,v2]=[121.515 , 121.61 ] nm namely , @xmath52 in this study compared to @xmath53 in the case of and
? claimed that a difference in the selected wavelength domains was the cause of the difference in the absorption rates .
to support their claim , those authors derived a transit absorption of @xmath19 from their analysis for the wavelength domain @xmath54 , while a different transit absorption of @xmath55 was derived for the wavelength domain @xmath56 [ v3,v4]=[121.484 , 121.643 ] nm .
the problem is that the coherency claimed by is merely illusory because the provided rates are simply inconsistent .
indeed , one may note that the considered wavelength window @xmath54 is embedded in @xmath57 , with @xmath58 [ v1,v3 ] & [ v4,v2]=[121.484 , 121.515 ] nm & [ 121.61 , 121.643 ] nm as the difference domain .
considering the out - of - transit stellar line profile as provided by and , one may wonder : how could an absorption rate possibly be cut by a factor of two when extending the wavelength domain from dw1 to dw2 and only adding the falling - off signal from the very limited dw3 stellar wings ?
in other words , what could be the transit absorption rate in the complementary domain @xmath59 if 15% is obtained in the first domain @xmath54 and 7.2% in the second one @xmath57 as estimated by ? + starting from the definition of the transmitted flux in equation 1 , we write that the absorption @xmath60 integrated over the wavelength domain @xmath54 is a weighted sum of the absorption @xmath61 integrated in the domain @xmath57 and @xmath62 in the domain @xmath59 . for a sake of clarity
we introduce the following notation for the signal attenuation @xmath63 corresponding to a prescribed wavelength domain @xmath64 : @xmath65 where @xmath66 is the fraction of transmitted signal with @xmath67 and @xmath68 in above equations , the subscripts @xmath69 stands for numerator and @xmath70 for denominator , and @xmath71 is the unperturbed stellar flux .
note that the contribution to the integration from the sky background wavelength domain is set to zero .
+ starting with the most extended @xmath57 domain , we may write : @xmath72 splitting the integration in the numerator over the different domains , we obtain @xmath73 taking into account the definition of the rates in the @xmath57 and @xmath59 domains , we can write : @xmath74 we now may derive the absorption in the @xmath59 wavelength domain : @xmath75 all integrals @xmath76 are known and directly derived from the unperturbed ( out - of - transit ) line profile .
now , using the numbers provided by and for the attenuation in the different wavelength domains of interest here , namely that @xmath77 ( or @xmath78 ) and @xmath79 ( or @xmath80 ) , we can derive the fraction of the signal transmitted in the @xmath59 wavelength domain should be @xmath81 .
this means that in order to make consistent the 15% and 7.2% rates derived by respectively for the @xmath54 and @xmath57 spectral domains , an enhancement of 8% of the stellar signal is required during transit in the wavelength window @xmath59 .
surprisingly , when the absorption rate proposed by is divided by a factor 1.08 ( 8% enhancement ) , then we recover an absorption rate of 8.7% for the @xmath54 spectral domain , consistent with our present estimation ( see table 2 ) .
we believe that the normalization used by to compensate stellar variability by using the stellar wings signal as a proxy may be one of the causes of the discrepancy between their results and ours . indeed , and repeatedly claimed that because most of the absorption should , in their estimation , occur in the line core , then the stellar line wings ( out - domain in definitions ) should not show any spectral absorption .
as shown here ( e.g. , fig .
6 ) , the transit absorption profile has very extended wings with a substantial absorption up to the out - domain used by , which casts doubt upon their approach to estimating the transit attenuation and to handling stellar time variability using the line wings signal as a reference .
despite serious limitations of current technology to resolve exoplanets from their parent stars , we learned a lot on their atmospheric structure thanks to the transit technique .
for the extrasolar planet hd209458b , we are close to sketching its atmospheric structure over several scales with detection , respectively , of neutral sodium in the lower atmosphere ( @xcite ) , very extended corona of hydrogen in the upper atmosphere , and an intermediate layer of temperature jump ( @xcite ) . for exoplanets orbiting very close to their host stars ,
these results confirm the scenario of atmospheric inflation by the huge stellar flux , a fact that raises the question of atmospheric evaporation and the survival of these planets over several stages of stellar evolution .
+ therefore , it may be worthwhile to discuss the adequacy of thus - far proposed models to explain the hydrogen distribution around hot giant exoplanets in the light of the constraints derived here and to propose new hints that may help investigate this fundamental problem .
since disseminated their results on hd209458 , many studies followed regarding the mass loss from exoplanets , particularly that class of planets that closely orbit to their parent stars .
initially , proposed evaporation followed by radiation pressure acceleration of hydrogen atoms beyond the roche lobe leading to a comet - like shape of the hydrogen nebula that escape away from the star . as we know from and the present study , this scenario
is not confirmed by the hst / stis observations because no signature of asymmetry could be found as initially claimed by .
the second difficulty of the model was pointed out by @xcite , namely , that the radiation pressure has not enough time to operate before the atoms are ionized .
+ these conclusions do not mean that evaporation is not operating in the hd209458 system .
in fact , the only process being dismissed is the acceleration by radiation pressure because , consistently with the conclusions of the present analysis , it does not seem to be efficient enough to produce the necessary population that must leave a clear spectral asymmetry in the absorption profile . @xcite tried 3d hydrodynamical simulation of the hydrogen nebula around hd209458 and predicted several light curves for the planetary transit , depending on the escape efficiency of the atmosphere .
unfortunately , these authors underestimated the absorption by the hydrogen cloud by assuming a gaussian absorption profile for the h atoms at , an assumption that is not adequate for this atomic transition . in all cases ,
the natural width of the upper level of the h atom induces a natural broadening , described by a lorentz profile that must be included in the absorption calculations ( @xcite ) . when included
, this natural broadening should significantly enhance the absorption in the line wings and consequently the absorption for the total line .
if the velocity distribution of atoms is maxwelian , then the attenuation is enhanced by the ratio @xmath82 of a voigt to a gaussian function that may be several order of magnitude beyond @xmath83 doppler units from the line center ( @xcite ) .
+ in a later study , @xcite proposed that the inflated atmosphere of hd209458b should be extended beyond the magnetopause region of the planet so that charge exchange with the stellar wind protons would necessarily produce energetic neutral atoms ( enas ) . in this process ,
the size of magnetospheric cavity ( obstacle region in @xcite ) is one of the key parameters that defines the enas content .
the more the obstacle region is extended , the less enas are produced . posing the magnetopause position at @xmath84 ,
these authors were able to account for the hd209458b transit s absorption features as claimed by assuming a stellar wind with a density of @xmath85 and a bulk velocity @xmath86 .
as far as the conclusions of and have been invalided , the fit proposed in @xcite is no more adequate and one may question if other parameters of the stellar wind could account for the constraints derived in and in the present study .
+ before we conclude on this question , it is important to discuss some difficulties of the enas production model as proposed by @xcite .
because of the nature of the charge exchange reaction , we must have two populations : one is the enas and the second is a family of ions flowing outward in the same direction as the planetary atoms that produced them . in their publication , @xcite just removed that population from their simulation box . also , the stellar wind should slow down when approaching the assumed obstacle .
a self - consistent formulation of the problem should allow for a cascade of enas and ions productions through the charge exchange process ( see @xcite ) .
such ions will induce their own ram pressure against the stellar wind , pushing away the magnetospheric cavity ( the obstacle ) outward at larger distances ( @xcite ) .
how great that distance is depends on self - consistent simulations that need to be done . according to @xcite ( fig . 7 in their supplementary draft ) , moving the obstacle ( magnetopause ) further out will reduce the enas production and , consequently , the attenuation in the red and blue part of the spectrum .
the conclusion here is that the current version of the the enas model as proposed by @xcite lack the coherent treatment of magnetospheric structure and response that is much needed to evaluate whether this process is important in shaping the hydrogen distribution around hd209458b .
+ another difficulty the @xcite model faces is the formulation of the attenuation process of the emission due the atomic hydrogen . as in @xcite
, these authors evaluated their attenuation profile during transit while overlooking the fact that the second level of a hydrogen atom has a finite life time that induces a natural broadening of the line .
the natural broadening is represented by a lorentz function that describes the absorption profile of h atoms in their rest frame .
therefore , for a gas with a given velocity distribution ( like the one shown in fig . 2 of @xcite )
, the absorption profile that should be used for attenuation calculations must be obtained as the convolution between the lorentz profile and the velocity distribution ( @xcite ) .
for the sake of clarity , we reproduced the velocity profile from @xcite and convolved it with a single lorentz profile for a gas temperature of @xmath87k .
in general , when dealing with a non - homogeneous medium , for every position across the planetary disc ( see equation 1 ) , one needs to evaluate the atomic absorption profile locally for the corresponding temperature .
the same caution is required when dealing with several populations of particles . as shown in fig .
7 , the velocity distribution field derived by @xcite , when compared with the same distribution including the effect of natural broadening , leads to distinct results with stronger and extended line wings and a smoothing out of the red side , making the new distribution much symmetric in the star s reference frame than initially set .
finally , we remark that the effect of natural broadening is even more visible when the enas contribution is smaller .
+ with that in mind , one may well imagine a hybrid solution where a thin layer of superthermal atoms ( enas production , turbulence energized particles , etc . )
is on top of a classical thermal atmosphere that may absorb the stellar flux in the way it is observed ( e.g. , fig .
the [ h ] thickness of the thin layer should not exceed few percent of the column of the underlying atmosphere .
this configuration takes its origin in similar scenarios that also occur in the so - called bulge region in the upper atmosphere of jupiter ( @xcite ) .
as previously noted , however , enas and ions production as proposed by @xcite must be evaluated in conjunction with the evolving size of the magnetopause cavity so that the effective enas population that contributes to the absorption is consistent with the magnetopause position and the gas structure around .
also , the corresponding absorption profile should account for the effect of natural broadening of the line as described here ( eg fig .
7 ) . in practice
, many solutions may be found that may require spanning the stellar and atmospheric models parameters in a self - consistent way .
this effort is out of the scope of this study .
hst archive observations of the emission of hd209458 are revisited to test the absorption rate sensitivity to the spectral range considered inside the emission line and to derive the absorption line profile during transit .
our first conclusion is that the absorption depth during transit is weakly dependent on the assumed spectral range assumed , with rates in the range of @xmath88 for all possible wavelength windows as reported in our initial study ( ) .
self - consistent derivation of the absorption line profile during transit shows a symmetrical lorentzian shape , typical of optically thick conditions of the hydrogen nebula around hd209458b . whether using direct light curves from the time series or inspecting the absorption line profile during transit , neither method shows any indication of preferred absorption on the blue side of the stellar line in the spectral domain @xmath89 from line center . finally , we stress that because the planet is not resolved spatially , several scenarios of the gas distribution may lead to the observed absorption line profile . in that context , two scenarios that fit rather well to the absorption profile
are suggested .
a first model assumes an optically thick atmosphere of hd209458b obtained from the standard model of @xcite .
a second class of models consists of a standard model that may have a thin layer of superthermal neutrals ( enas , turbulence energized particles , etc . ) on top of it , much like what we observe for the upper atmosphere of jupiter ( @xcite ) . in all cases , the natural broadening that was missed in most past studies
is required in the modelling of the absorption profile .
moreover , for enas production , a self - consistent treatment of the charge exchange process in conjunction with the size of the magnetospheric cavity is much needed to test the validity of such process . in any case , high resolution observations with a _
s / n _ better than @xmath50 are required to efficiently constrain the hydrogen distribution and its time and phase variations along the planetary orbit .
the author acknowledges support from universit pierre et marie curie ( upmc ) and the centre national de la recherche scientifique ( cnrs ) in france .
he also warmly thanks prof .
w. harris for his hospitality during his 6 months stay at the department of applied science at uc davis .
this work is based on observations with the nasa / esa hubble space telescope , obtained at the space telescope science institute , which is operated by aura , inc .
ballester , g. , sing , d. , herbert , f. 2006 , nature , 445 , 511 .
ben - jaffel , l. , c. magnan , and vidal - madjar , a. 1988 , , 204 , 319 .
ben - jaffel , l. , et al .
1993 , geophys .
lett . , 20 , 747 .
ben - jaffel , l. , et al .
1995 , icarus , 113,91 .
ben - jaffel , l. 2007 , , 671 , l61 .
ben - jaffel , l. , kim , y. , and clarke , j. 2007 , icarus , 190 , 504 .
bougher , s. , waite , h. , majeed , t. , and gladstone , r. 2005 , j. geophys .
, 110 , citeid e04008 .
brown , t. , et al .
2002 , in hst stis data handbook , b. mobasher , ed .
( baltimore , stsci , v 4 ) , pp 131 - 192 .
chamberlain , d. and hunten d. 1987 , theory of planetary atmospheres : an introduction to their physics and chemistry , academic press .
charbonneau , d. , et al .
2002 , , 568 , 377 .
clarke , j. , gladstone , r. , and ben - jaffel , l. 1991 , geophys .
lett . , 16 , 587 .
clarke , j. , et al .
2002 , nature , 415 , 997 .
cranmer , s. , and saar , h. 2007 , arxiv : astro - ph/0702530v1 .
cuntz , m. , et al .
2000 , , 533 , l51 .
ehrenreich , d. et al .
2008 , , 483 , 933 .
emerich , c. , et al .
1996 , science , 273 , 1085 .
holmstrom , m. et al .
2008 , nature , 451 , 970 .
knutson , h. , et al .
2007 , , 655 , 564 .
munoz , g. a. 2007 , planet .
space sci .
, 55 , 1426 . hihalas , d. 1978 , stellar atmosphere , second edition , freeman & company , san francisco .
neufeld , d. 1990 , , 350 , 216 .
osterbart , r , and fahr , h. 1992 , , 264 , 260 .
ratkiewicz , r. , and ben - jaffel , l. 2002 , , 107 , ssh 2 - 1 .
sandel , b. , broadfoot , l. , and strobel , d. 1980 , geophys .
schneiter , m. et al .
2007 , , 671 , l57 .
shkolnik , e. , et al .
2005 , , 622 , 1075 .
sommeria , j. , ben - jaffel , l. , and prange , r. 1995 , icarus , 118,2 .
vidal - madjar , a. et al .
2003 , nature , 442 , 143 .
vidal - madjar , a. et al .
2004 , , 604 , l69 .
vidal - madjar , a. et al .
2008 , , 676 , l57 .
yelle , r. 2004 , icarus , 170 , 167 .
wood , b.e .
, et al . 2005 , , 159 , 118 . | all far ultraviolet observations of hd209458 tend to support a scenario in which the inflated hydrogen atmosphere of its planetary companion strongly absorbs the stellar flux during transit . however , it was not clear how the transit absorption depends on the selected wavelength range in the stellar line profile , nor how the atomic hydrogen cloud was distributed spatially around hd209458b . here
we report a sensitivity study of observed time and spectral variations of the stellar flux .
in particular , the sensitivity of the absorption depth during transit to the assumed spectral range in the stellar line profile is shown to be very weak , leading to a transit depth in the range @xmath0 for all possible wavelength ranges , and thereby confirming our initially - reported absorption rate .
taking the ratio of the line profile during transit to the unperturbed line profile , we also show that the spectral signature of the absorption by the exoplanetary hydrogen nebula is symmetric and typical of a lorentzian , optically thick medium .
our results question the adequacy of models that require a huge absorption and/or a strong asymmetry between the blue and red side of the absorption line during transit as no such features could be detected in the hst fuv absorption profile . finally , we show that standard atmospheric models of hd209458b provide a good fit to the observed absorption profile during transit .
other hybrid models assuming a standard model with a thin layer of superthermal hydrogen on top remain possible . |
the interpretation of the second solar spectrum requires taking into account a number of physical processes that are not relevant for the intensity spectrum .
some of the lines observed in the second solar spectrum , like the nai d@xmath4 and d@xmath0 , cr i triplet at 5206 , baii d@xmath4 and d@xmath0 , mg ii h and k , and caii h and k , are affected by quantum interference between states of different total angular momentum ( @xmath5 or @xmath1 states ) .
the importance of quantum interference was first demonstrated both observationally and theoretically by .
an approximate theoretical approach for treating @xmath5-state interference with partial frequency redistribution ( prd ) was proposed by @xcite .
this theory was then applied to model the linear polarization profiles of the cr i triplet at 5204 - 5208 by .
an atom with a nuclear spin @xmath6 exhibits hyperfine structure splitting ( hfs ) .
the quantum interference between these hyperfine structure ( @xmath1 ) states produces depolarization in the line core .
nai d@xmath0 , baii d@xmath0 , and scii line at 4247 are some examples of spectral lines whose linear polarization profiles show signatures of @xmath1-state interference .
the ba ii d@xmath0 line is due to the transition from the upper @xmath7 and the lower @xmath8 fine structure levels ( see figure [ smitha - level - diag](a ) ) . in the odd isotopes of ba , both the upper and lower levels undergo hfs due to the nuclear spin @xmath9 .
this gives rise to four upper and two lower @xmath1-states ( see figure [ smitha - level - diag](b ) ) .
in modeling the baii d@xmath0 line it is necessary to account for quantum interference between the upper @xmath1-states .
the odd isotopes constitute about 18% of the total ba abundance in the sun .
the rest 82% is from the even isotopes , which are not subject to hfs ( because @xmath10 ) . and
@xmath11 , with their hyperfine structure components ( b ) .
the energy levels are not drawn to the scale.,width=377 ] high precision spectro - polarimetric observations by using zimpol clearly revealed the existence of three distinct peaks in the linear polarization @xmath12 profiles of the ba ii d@xmath0 line . using the last scattering approximation , explained the physical origin of these three peaks .
he showed that the central @xmath2 peak arises from the even isotopes of ba , and the two side peaks from the odd isotopes . with a similar last scattering approach ,
the magnetic sensitivity of the ba line was explored by @xcite .
the above mentioned authors however did not account for radiative transfer and prd effects .
these effects were taken into account by , but the treatment was limited to the even isotopes of ba , for which hfs is absent .
in @xcite we considered radiative transfer with prd in both even and odd isotopes , and the full effects of hfs with @xmath1-state interferences .
here we present a brief summary of our attempts to model the baii d@xmath0 line profile observed in a quiet region close to the solar limb .
further we present center - to - limb variation ( clv ) observations of this line and our initial attempts to model the same .
the paper is organized as follows : in section [ smitha - transfer - equations ] we present the polarized radiative transfer equation suitable to handle several isotopes of ba simultaneously . in section [ smitha - obs - details ]
we present the details of the observations . in section [ smitha - modeling - pro ]
we discuss the model atom and the model atmosphere used .
the results are presented in section [ smitha - results ] with concluding remarks in section [ smitha - conclusions ] .
the polarized transfer equation in a 1-d non - magnetic axisymmetric atmosphere is given by @xmath13 , \label{smitha - transfer}\end{aligned}\ ] ] in the reduced stokes vector basis .
positive @xmath14 represents the linear polarization oriented parallel to the solar limb .
the total opacity @xmath15 , with @xmath16 and @xmath17 being the continuum scattering and continuum absorption coefficients , respectively and @xmath18 being the wavelength averaged absorption coefficient for the @xmath19 transition . here
@xmath20 and @xmath21 are the lower and upper level electronic angular momentum quantum numbers respectively . for the case of ba ii
d@xmath0 , the appropriate expression for the profile function @xmath22 is @xmath23 here the voigt profile function for the even isotopes of ba ii is denoted as @xmath24 .
this corresponds to the @xmath25 transition in the absence of hfs .
@xmath26 is the profile function for the odd isotopes .
it is the weighted sum of individual voigt profiles @xmath27 representing each of the @xmath28 absorption transition .
the initial and the intermediate hyperfine split levels have total angular momentum quantum numbers @xmath29 and @xmath30 respectively .
@xmath26 is given by @xmath31 .
\label{smitha - prof - combined}\end{aligned}\ ] ] in equation ( [ smitha - phi - g ] ) the contributions from both the @xmath32ba ( 6.6% ) and @xmath33ba ( 11.2% ) odd isotopes add upto the @xmath34 of @xmath26 . in equation ( [ smitha - transfer ] ) , @xmath35 is the reduced total source vector for a two - level atom with an unpolarized lower level and is defined as @xmath36 showed that , for the case of ba ii d@xmath0 any ground level polarization would be destroyed by elastic collisions with hydrogen atoms . in equation ( [ smitha - s - total ] ) , @xmath37 , where @xmath38 is the planck function .
the line source vector @xmath39 is @xmath40 with @xmath41 here @xmath42 is a @xmath43 diagonal matrix , including the effects of hfs in odd isotopes . @xmath44
diag @xmath45 , with @xmath46 being the redistribution function components for the multipolar index @xmath47 , containing both type - ii and type - iii redistribution of @xcite .
the quantum number replacement @xmath48 in equation ( 7 ) of gives the expression for @xmath49 .
we use crd functions in place of the type - iii redistribution functions as in @xcite . the contributions from the individual redistribution matrices for the odd isotopes @xmath32ba and @xmath33ba are also taken into account .
@xmath50 is also a @xmath43 diagonal matrix for the even isotopes without hfs .
its elements @xmath51 are the redistribution functions corresponding to the scattering transition @xmath52 .
they are obtained by setting the nuclear spin @xmath10 in @xmath53 .
an expression for @xmath50 in the stokes vector basis can be found in @xcite and in .
we here use the angle - averaged versions of these quantities .
@xcite showed that in the non - magnetic case the use of the angle - averaged redistribution matrix is sufficiently accurate for all practical purposes .
we use the two branching ratios ( see * ? ? ?
* ; * ? ? ?
* ) @xmath54 where @xmath55 and @xmath56 are the radiative and inelastic collisional rates , respectively .
@xmath57 is the elastic collision rate computed from @xcite .
the depolarizing elastic collision rates are given by @xmath58 with @xmath59 .
the @xmath60 is computed using @xmath61 where @xmath62 is the neutral hydrogen number density in ( @xmath63 ) , @xmath64 the temperature in kelvin , and @xmath65 the energy difference between the @xmath66 and @xmath67 fine structure levels in ( @xmath68 ) and @xmath69 is the boltzmann constant in ( @xmath68/k ) .
we neglect the collisional coupling between the @xmath67 level and the metastable @xmath70 level ( see * ? ? ?
. pointed out the importance of @xmath67 - @xmath70 collisions for the line center polarization of ba ii d@xmath0 .
he showed that the neglect of such collisions would lead to an overestimate of the line core polarization by @xmath71 .
it was shown by that this in turn would lead to an overestimate of the microturbulent magnetic field @xmath72 by @xmath73 .
our aim is to explore the roles of prd , hfs , quantum interferences , and the atmospheric temperature structure in the modeling of the triple peak structure of the ba ii d@xmath0 linear polarization profile .
we assume frequency coherent scattering in the continuum with its source vector given by @xmath74 the matrix @xmath75 is the rayleigh scattering phase matrix in the reduced basis .
@xmath76 is the line thermalization parameter defined by @xmath77 .
the stokes vector @xmath78 can be computed from the irreducible stokes vector @xmath79 by simple transformations .
the observed polarization profiles of the ba ii d@xmath0 line shown in this paper for @xmath80 are same as those presented in @xcite .
the observations for all @xmath3 values were acquired by the eth team of stenflo on june 3 , 2008 , using their zimpol-2 imaging polarimetry system at the themis telescope on tenerife .
the polarization modulation was done using ferroelectric liquid crystal modulators .
the observations were recorded at various limb distances with the slit being placed parallel to the heliographic north pole .
figure [ smitha - ccd - images ] shows the ccd images of @xmath81 and @xmath2 for different @xmath3 values .
these essentially represent the clv of the intensity and degree of linear polarization . in the section [ smitha - results ]
we attempt to model these clv observations .
and @xmath2 observed at different limb distances on the solar disk.,title="fig:",width=211 ] and @xmath2 observed at different limb distances on the solar disk.,title="fig:",width=219 ]
the observed @xmath82 profiles of the ba ii d@xmath0 line can be modeled using a two - stage process .
this process is similar to the one described in . in this two - stage process ,
the intensity is first computed which is then used as an input to compute the polarization .
the code that is used for computing the intensity was developed by ( * ? ? ?
* referred to as the rh - code ) . to compute the intensity and polarization
we need a model atom that is representative of the atom under consideration and a one - dimensional model atmosphere that mimics the solar atmosphere . for the case of ba ii d@xmath0 ,
since the effects of different isotopes are to be taken into account , we construct three different atom models . out of these ,
two are for the odd and one for the even isotope . the atom model for the even isotope ( @xmath83ba )
is given by the five levels of figure [ smitha - level - diag](a ) , while for the odd isotopes ( @xmath32ba and @xmath33ba ) the model is extended to include the hyperfine splitting as described by figure [ smitha - level - diag](b ) .
the hfs of the @xmath66 level and the metastable levels are neglected .
the effects of hfs are taken into account only for the d@xmath0 transition in the intensity computations .
for the odd isotopes , the radiative transitions between @xmath67 and the metastable levels are treated in the same way as for the even isotope . however , in the polarization computations , under the two - level atom approximation , the coupling to the metastable levels are neglected for both even and odd isotopes .
we neglect the contribution from other less abundant even isotopes .
the wavelengths of the six hyperfine transitions for the odd isotopes are taken from kurucz database and are also listed in table 1 of @xcite .
the line strengths of these transitions are given in equation ( [ smitha - prof - combined ] ) .
model atmospheres.,width=377 ] some of the commonly used standard 1-d solar model atmospheres are the fala , falf , falc ( see * ? ? ?
* ) and the falx ( see * ? ? ?
* ) . among these four models
falf is the hottest and falx the coolest .
however , as will be discussed below , we find that a model atmosphere that is cooler than falx is needed to fit the observed profiles .
the new model , denoted @xmath84 , is obtained by reducing the temperature of the falx model by about 300 k in the height range 500 1200 km above the photosphere .
the temperature structures of the standard model atmospheres and the modified model are shown in figure [ smitha - temp - struc ] .
the three atom models of ba ii atom described in the above section give us three different sets of physical quantities ( two for the odd isotopes and one for the even isotope ) when used in the rh code to compute intensity .
these quantities are the line opacity , line emissivity , continuum absorption coefficient , continuum emissivity , continuum scattering coefficient , and the mean intensity .
see @xcite for the mathematical expressions that are used in the rh - code to compute these quantities for the even isotopes . for the odd isotopes ,
the profile functions in these expressions are replaced by @xmath26 defined by equation ( [ smitha - prof - combined ] ) .
profiles computed using different model atmospheres.,width=264 ] the three sets of quantities obtained from three atom models are combined in the ratio of their respective isotope abundances .
the combined quantities are then used as inputs to the polarization code .
the @xmath2 profiles so computed for the various model atmospheres are shown in figure [ smitha - no - fit ] .
the figure shows that the @xmath2 profiles are quite sensitive to the temperature structure of the model atmosphere and that the central peak can not be reproduced from any of the standard model atmospheres .
all the models produce a dip in @xmath2 at the line center .
earlier studies have shown that such a central dip in @xmath2 is generally due to the effects of prd ( see for example , figure 3 of , and figure 6 of ) . the main contribution to
this dip comes from the type - ii frequency redistribution . in the case of ba ii d@xmath0 ,
the central dip is formed mainly through contributions from the even isotopes ( see figure 5 of * ? ? ?
. however the profiles computed using only crd do not show any central dip .
it is generally known that @xmath2 profiles computed with crd and prd in realistic model atmospheres differ not only in the line wings but also in the line core .
this is contrary to the behaviour expected of @xmath2 profiles computed in isothermal atmospheres , where they nearly match .
a study of the mechanisms that cause the differences in the line core , in the case of realistic atmospheres , would be taken up in a forthcoming publication .
show that the magnitude of the @xmath2 central dip is strongly dependent on the choice of atmospheric parameters .
the central dip becomes shallower for cooler models .
this behavior is also evident from our figure [ smitha - no - fit ] .
such dips in @xmath2 are common in line profile modeling with prd , and are often smoothed out by applying instrumental and macro - turbulent broadening .
the profiles presented in figure [ smitha - no - fit ] have already been smeared with a gaussian function having a full width at half maximum ( fwhm ) of 70 m .
though it is possible to smoothen out the dip by increasing the smearwidth , it will suppress the side peaks which are due to the odd isotopes and also disturbs the fit to the intensity profile . a smearwidth value of 70 m has been chosen by us to optimize the fit and it is also consistent with what we expect based on the observing parameters and turbulence in the chromosphere .
since all the standard model atmospheres failed to provide a fit , we had to modify the temperature structure of the falx model to get a new much cooler model .
this modified model is called the @xmath84 and the details of its construction are given in @xcite .
this new model atmosphere succeeds in fitting both the intensity and the linear polarization profiles .
the fit to the @xmath2 profile is shown as the dashed line in figure [ smitha - fit ] .
the theoretical profile in this figure also includes the contribution from the spectrograph stray light which is 4% of the continuum intensity .
this is done to mimic the observed profiles which contain the contribution from stray light .
finally , the effect of microturbulent magnetic field with a strength of 2 g is included to obtain a good @xmath2 fit at the line center .
profiles using the @xmath84 model atmosphere and also the comparison between profiles computed with prd and crd.,width=377 ] the ba ii d@xmath0 4554 line is a resonance line and is strongly influenced by the effects of prd .
hence it is very important to take account of the prd effects while modeling .
its importance in case of the even isotopes of ba ii is demonstrated in . in the right panel of figure
[ smitha - fit ] , we present a comparison between the observed and the theoretical @xmath2 profiles based on crd alone ( dotted line ) and on full prd ( dashed line ) for the @xmath85 model .
the intensity profile can be fitted well using either prd or crd ( see * ? ? ?
* ) , but the polarization profile can not be fitted with crd alone .
this shows that the effects of prd are essential to model the @xmath2 profiles of the ba ii d@xmath0 line .
the left panel of figure [ smitha - clv ] shows the clv of the observed @xmath2 profiles .
the right panel of this figure shows the clv of the theoretical @xmath2 profiles computed using the @xmath84 model atmosphere .
as seen from the right panel , the newly constructed model atmosphere , though successfully fits the observed profiles for @xmath80 , fails to fit the observations for @xmath86 .
in fact for @xmath86 , we again obtain a central dip instead of a peak in the @xmath2 profiles . the central dip becomes deeper as @xmath87 .
this behavior was noticed also for the case of na i d@xmath0 by .
thus the modification in the temperature structure of the falx model atmosphere , presented in @xcite and in this paper , helps only for @xmath80 .
for other @xmath3 positions , the temperature structure has to be further modified .
thus we are unable to find a single 1d model atmosphere that fits the clv observations of the ba ii d@xmath0 line
. positions .
for example , when modeling the cr i triplet in , we had to modify the temperature structure of the falf model atmosphere in the deeper layers to obtain a fit to the @xmath82 observations at @xmath88 . ]
profiles and the theoretically computed clv .
the theoretical profiles are computed using the @xmath84 model atmosphere.,width=377 ]
in this paper we briefly describe the results of our previous paper on modeling the polarization profiles of the ba ii d@xmath0 line by taking full account of prd , radiative transfer , and hfs effects .
also , we extend the same to model the observed clv of the @xmath2 profiles . to model the clv profiles , we use the same approach that was used in @xcite .
we find that some of the standard model atmospheres like the falf , falc , fala , and falx fail to reproduce the central peak , and instead give a central dip which is mainly due to prd effects . also , this central dip is quite sensitive to the temperature structure of the model atmosphere . by slightly modifying the temperature of the standard falx model atmosphere ,
we obtain a fit to both @xmath81 and @xmath2 profiles .
this new model is called the @xmath84 .
however this new model successfully fits the observations only for @xmath80 and fails at other limb distances .
hence within the sample of the model atmospheres tested by us , no single one dimensional , single component model atmosphere succeeds in reproducing the observed @xmath2 profiles at all limb distances .
this possibly indicates that multi - dimensional transfer effects are required to model the clv observations of this line .
finally , we show that since the ba ii d@xmath0 is a strong resonance line , it has damping wings that are governed by prd .
it is therefore necessary to take full account of prd in order to correctly model the @xmath2 profile .
in contrast , the intensity profiles computed with prd or crd do not differ substantially from each other through out the entire line profile .
we would like to thank dr .
michele bianda for useful discussions .
we acknowledge the use of hydra cluster facility at the indian institute of astrophysics for computing the results presented in the paper .
we would like to thank the referee for a careful reading of the manuscript and for giving useful comments and suggestions . | the ba ii d@xmath0 line at 4554 is a good example , where the @xmath1-state interference effects due to the odd isotopes produce polarization profiles , which are very different from those of the even isotopes that do not exhibit @xmath1-state interference .
it is therefore necessary to account for the contributions from the different isotopes to understand the observed linear polarization profiles of this line . in this paper
we present radiative transfer modeling with partial frequency redistribution ( prd ) , which is shown to be essential to model this line .
this is because complete frequency redistribution ( crd ) can not reproduce the observed wing polarization .
we present the observed and computed @xmath2 profiles at different limb distances .
the theoretical profiles strongly depend on limb distance ( @xmath3 ) and the model atmosphere which fits the limb observations fails at other @xmath3 positions .
[ authorguide ] |
both indirect evidence @xcite and later direct imaging @xcite , have shown that disks are quite common in young stellar systems .
these disks are commonly thought @xcite to be sites for planet or brown dwarf formation . a large fraction of stars are formed in binary systems @xcite and in the same star formation regions as single stars .
theory suggests that the most likely mechanisms responsible for forming jovian mass planets or low mass brown dwarfs are either gravitational collapse of large scale spiral structure or coagulation of small solid grains followed by later accretion of additional gas ( ` core accretion ' ) in the disks of forming stellar systems .
evaluating the effectiveness of these mechanisms is important for understanding the origin of our own solar system as well as planetary systems in other mature single or multiple systems .
the l1551 irs 5 system serves as a useful observational testbed for comparison to theoretical modeling because of its relative youth ( @xmath1 yr , * ? ? ?
* ) and many previous detailed observations ( see e.g. * ? ? ?
this system consists of an extended nebulosity some 2400@xmath21100 au in size with an inner core of 220@xmath276 au @xcite .
two bipolar jets flow outward in each direction from the core in the plane perpendicular to its long axis .
the core has been resolved into two sources with projected separation of about 50 au and inferred disk masses of @xmath3@xmath4 , each @xmath520 - 25 au in diameter @xcite .
the total mass in the core has been estimated to contain 0.51.0@xmath4 of material @xcite , which produces @xmath530@xmath6 in luminous output @xcite .
we present a numerical simulation of a binary star / disk@xmath7star / disk system using a two dimensional ( @xmath8 ) smoothed particle hydrodynamic ( sph ) code .
the dimensions of the disks and semi - major axis of the binary are chosen to be similar to the inner core region of l1551 irs 5 . in the absence of strong constraints on the constituents of the binary (
e.g. the masses of the two stars ) , we choose to set up a binary system consisting of identical components , obtained by setting up a single system in isolation , then duplicating it exactly .
we assume each star and disk have mass @xmath9@xmath4 and @xmath10@xmath4 , respectively .
the disk radius is set to @xmath11 au which , for a semi - major axis of @xmath12 au , is comparable to the largest stable streamline @xcite .
the mass and temperature of the disk are distributed according to @xmath13 and @xmath14 power laws respectively .
the absolute scale of each power law is determined from the disk mass , the radial dimensions of the disk and the condition that the toomre stability parameter , @xmath15 , is no smaller than @xmath16=1.5 over the entire disk .
this value ensures that the simulation begins in a state marginally stable against the growth of spiral structure , so that we do not accidently ` discover ' a collapsed object early in the evolution which in reality is an artifact of our initial condition .
both density and temperature are free to vary in time and space , so the initial condition will not prevent spiral structure growth or fragmentation , if the evolution leads to such .
the gas is set up on circular orbits around the star so that pressure and gravitational forces exactly balance centrifugal forces .
radial motion is zero .
the magnitudes of the pressure and self - gravitational forces are small compared to the stellar gravity , so the disk is nearly keplerian in character .
approximately 60000 equal mass particles are set on a series of concentric rings around the star in a single , star / disk system , then duplicated , bringing the total number of particles to @xmath5120000 .
the two stars and disks are offset equal distances in the @xmath17 and @xmath18 directions .
we define the binary semi - major axis to be @xmath12 au , similar to l1551 irs 5 .
only weak constraints on eccentricity exist in l1551 irs 5 , primarily consisting of the sizes of the observed disks : eccentricities larger than @xmath19 would lead to rapid roche lobe overflow .
we set @xmath19 to be the initial value in this simulation .
the system is at apoapse at time @xmath20 with the orbital velocities defined by approximating each star@xmath7disk system as a point mass , so that the orbit determination reduces to the solution of the two body problem .
the disks are self gravitating and each star is modeled as a point mass free to move in response to gravitational forces from the rest of the system .
the stellar gravitational forces are calculated using a plummer potential with a softening radius of 0.2 au , which also serves as an accretion radius , @xmath21 .
sph particles with trajectories that pass closer than @xmath21 to a star are absorbed , and the star s mass and momentum increase accordingly . the thermodynamic evolution is identical to that described in @xcite .
thermal energy is added to the gas due to active hydrodynamic processes using an artificial viscosity scheme , which approximately models shocks and turbulence .
this heating is roughly equivalent in magnitude to an alpha model with @xmath22 .
thermal energy is removed from the disk gas by radiative cooling due to passive blackbody emission from the disk s photosphere surfaces .
the blackbody temperature is calculated at each time step and for each sph particle .
this treatment remedies a major shortcoming of previous models @xcite which used a ` locally isothermal ' or ` locally adiabatic ' approximation to show that relatively low mass disks can undergo fragmentation and/or collapse , despite earlier claims @xcite that a very massive disk is required .
the system is evolved for eight binary orbits , or 2700 yr ( @xmath23=350 yr ) .
figure [ fig : disk - pair ] shows the system shortly before and after the fourth periapse passage of the two components ( measured from the beginning of the simulation ) . before each periapse ,
the two disks are smooth and exhibit no visible spiral structure , although they are no longer perfectly ` round ' . during and after periapse
, each disk develops strong , two armed spiral structures due to the mutual tidal interactions of the binary .
the structures decay to a smooth condition like that in the top panel over the next @xmath24@xmath23 .
the cycle repeats with little variation as the system again approaches periapse , and we expect that further evolution will be similar . the spiral structures decay because internal heating in the disk increases the stability of the disks against spiral arm growth , as measured by the toomre stability ( fig .
[ fig : temp - prof ] top ) of each disk .
the minimum value of @xmath15 increases from its initial @xmath16=1.5 to @xmath16@xmath25 before periapse and @xmath16@xmath26 afterwards .
these values are the same before and after each successive periapse passage and both are well above the @xmath27 values for which spiral structures are expected to grow .
therefore , if we suppose that jovian planets form via gravitational collapse or fragmentation of spiral structure in disks , their formation will be unlikely in this system .
the high stability is due to an increase in the disk temperature .
in fact , the temperatures are high enough to cause some grain species to be vaporized .
this is important because the core accretion model for planet formation requires that solid grains can coagulate and are not instead repeatedly returned to vapor state .
water ice may be particularly important because it composes 40% by mass of the solid material in the disk and is among the most volatile grain species @xcite , vaporizing at @xmath28 k. the bottom panel of figure [ fig : temp - prof ] shows the disk midplane temperatures obtained in this simulation . only in a region with @xmath29 au does the temperature reach low enough values that water ice can form , even temporarily .
however in this region , the matter is most subject to shocks generated by the spiral structure produced by the binary interaction , which raise its temperature by as much as a factor three over the azimuth average .
the spiral patterns co - rotate with the orbit of the binary and the orbital period at 15 au is @xmath582 yr , so material everywhere in the disk will have time to travel through the spiral arms several times before they decay .
a simulation run with zero eccentricity also produces spiral structure . in this
run the azimuth averaged temperatures were similar to fig .
[ fig : temp - prof ] , but the temperature in the shocked regions are not as extreme .
water ice would still be vaporized everywhere but more refractory species may not be . in the outer disk ,
grains less than @xmath30 mm in size which pass through such a shock and which contain water ice will be vaporized on a timescale of @xmath31 s depending on the temperature @xcite and the remaining more inert species may dissagregate
. in this region , passage through the warmest part of a spiral arm requires @xmath32 yr , so sufficient time exists to return grains of this size to the gas phase .
grain growth may still occur between the spiral arms and when the spiral arms have decayed , but must begin with gaseous material each time , so growth of solid material into larger entities will be suppressed .
temperatures at high altitudes are lower , but grains which form there will tend to sink to the midplane as they grow larger and also be destroyed
. therefore jovian planet formation by the core accretion mechanism will occur much more slowly , if at all in this system .
the weak link remaining in the argument against the core accretion mechanism is the lack of knowledge of the microphysics important for dust coagulation .
for example , one could imagine that growth of silicate and iron grains is catalyzed by temporarily enhanced cross sections as mantles of more volatile material form on their surfaces and the gravitational torques produced by the binary interaction enhance mixing throughout the disks .
even if this type of interaction takes place , the eventual formation of planet sized objects remains in doubt .
as the rocky aggregates grow , conditions appropriate for dust coagulation ( e.g. ` perfect sticking ' ) break down and collisions between particles become increasingly disruptive due to the finite strength of the aggregates .
the disruption of solid bodies depends strongly on the relative velocity of the impactor and target particles , with disruption occurring for velocities
@xmath331 - 3 km / s for planetesimal sized targets ( @xmath34 km ) @xcite . on average in an accretion disk , the relative velocity of planetesimals will be proportional to their eccentricities , @xmath35 . for our model , a relative velocity of 1
km / s corresponds to an eccentricity of @xmath36 at 1 au , or @xmath37 at 10 au .
we have seen that gravitational torques are strong enough to generate large amplitude spiral structure as they drive the gas onto eccentric orbits .
gas eccentricities are quickly damped due to shock dissipation , but planetesimals are only weakly coupled to the gas and will have time to encounter other objects and collide or to increase their eccentricities still further as the evolution proceeds .
if particle eccentricities can grow to @xmath38 , we expect that the growth of kilometer sized bodies will be suppressed in binary systems such as the one modeled here .
however , a more detailed analysis must be done in order to constrain this possibility .
the conclusions regarding planet formation will remain valid as long as the temperatures determined from the model are lower limits on the temperatures present in real systems .
if the model produces temperatures which are too low , then real systems will be even more stable against spiral structure growth and fragmentation and less likely to produce large , coagulated grains .
if they are too high , the model could inaccurately portray the disk as too stable .
are the temperatures produced in the model too low ?
we can constrain the temperatures in the model by comparing the radiated energy from observed systems ( here specifically to l1551 irs 5 ) , to that produced by the simulation in various wavelength bands .
this comparison requires that we relate the luminous output to the temperature . in regions where the optical depth is high , as it is in the accretion disks ,
the radiated emission can be approximated as a blackbody with a temperature of the disk photosphere .
the disk s midplane and photosphere temperatures are then related to each other by a given rosseland opacity and the local vertical density / temperature profile .
we determine such a profile as by - product of the cooling model in this work , under the assumption that the vertical structure is instantaneously adiabatic
. other work @xcite has shown that the structure may instead be super adiabatic .
if this is the case then the midplane temperatures will be higher , and our conclusion is stronger .
the opposite case may be true instead : large relative heating can occur at high altitudes , even though the high altitude heating is small in an absolute sense @xcite .
this means that vertical temperature structure may become distorted and an incorrect midplane temperature could be inferred .
high altitude dynamical heating will play a role similar to high altitude passive heating from stellar photons .
@xcite show that this process produces a temperature inversion high above the photosphere , but this region contributes negligibly to the radiated flux .
therefore , we can rely on the modeled radiated output of the simulation to represent accurately the temperatures at the disk midplane , given the physical processes included in the calculation .
our conclusions about the formation of planets will be confirmed if we find that the energy output from the disks is equal to or less than that observed . to obtain a valid comparison between the observed and modeled fluxes
, we must be certain that the flux from other parts of the system ( e.g. the circumbinary disk and envelope ) is not a significant contributor to the observed flux used in the comparison .
we must also be certain that extinction between the source and the observer has not altered the emitted flux .
we therefore require very high spatial resolution photometry at long wavelengths , not affected by extinction . in figure [ fig : freq - cmp ] , the highest available resolution , long wavelength observations of l1551 irs 5 are plotted and compared to the flux densities produced from the simulation . for all wavelengths between 1.3 cm and 870 @xmath39 m
, the observed fluxes exceed those obtained from the simulation by a factor of @xmath40 .
the differences at 1.4 mm and 870 @xmath39 m are larger , a factor @xmath41 , however they do not resolve the binary and may contain some contaminating flux from the circumbinary environment .
given these comparisons , we can conclude that the temperatures in the disk are conservatively estimated by the simulation ( too low ) and that the disks in l1551 irs 5are more stable against spiral arm growth and fragmentation and less likely to allow dust coagulation than in the simulation .
our conclusion that planet formation is unlikely in the l1551 irs 5 system is secure .
the inventory of physical processes considered in this calculation is not complete .
the total luminosity will include direct contributions at the short wavelength end of the spectrum from not only the two circumstellar disks , but also the two stars .
accretion of material from the inner disk edge ( 0.2 au for this simulation ) onto the stellar surface will also add to the total . at longer wavelengths , the circumbinary disk ,
the infalling envelope and radio emission from bipolar polar outflows will add to the total .
of these processes , only outflows remove thermal energy from the disks , but such outflows are thought to originate very close to the star .
they will not remove energy from the part of the disk important for planet formation .
the rest will either add directly to the short wavelength spectrum or heat the cooler parts of the disk by absorption of short wavelength radiation , which subsequently reradiates at longer wavelengths ( see e.g. * ? ? ? * ) . including them in the calculation
can only raise disk temperatures and strengthen our conclusions .
estimates of the total luminosity ( not shown ) from all these sources are within a factor of two of the observed 30@xmath6 luminosity of l1551 irs 5 .
we are encouraged by this agreement , and expect that the overall agreement between model and observations would become closer if radiative reprocessing were included .
a number of questions remain .
what is the distribution of planetesimal random velocities ?
will lower mass disks be more susceptible to planet formation because of increased efficiency of radiative energy losses ? what happens for binary systems with different separations or with unequal stellar mass ratios ?
how distant is distant enough , so that one component of the binary does not strongly influence the other ?
these questions will be addressed more completely in a followup paper .
i gratefully acknowledge david koerner for the generous release of the 1.3 cm fluxes before their publication .
i thank brian pickett and willy kley for helpful conversations during the development of this paper and for brian s later comments as referee . | we show that planet formation via both gravitational collapse and core accretion is unlikely to occur in equal mass binary systems with moderate ( @xmath0 au ) semi - major axes .
internal thermal energy generation in the disks is sufficient to heat the gas everywhere so that spiral structures quickly decay rather than grow or fragment . this same heating will inhibit dust coagulation because the temperatures rise above the vaporization temperatures of many volatile materials .
we consider other processes not included in the model and conclude that our temperatures are conservatively estimated ( low ) , i.e. planet formation is less likely in real systems than in the model . |
since the seminal work by vilfredo pareto@xcite , it is known that the wealth distribution in capitalist economies shows a very peculiar and somehow universal functional form . in the range of low income
, the cumulative distribution of wealth ( the probability that the wealth of an individual is greater than a given value ) may be fitted by an exponential or log - normal decreasing function , while in the region containing the richest part of the population , generally less than the @xmath0 of the individuals , this distribution is well characterized by a power - law ( see for example @xcite for a review ) .
this empirical behavior has been confirmed by a number of recent studies on the economy of several corners of the world .
the available data is coming from so far apart as australia @xcite , japan @xcite , the us @xcite , continental europe @xcite or the uk @xcite .
the data is also spanning so long in time as ancient egypt @xcite , renaissance europe @xcite or the @xmath1 century japan@xcite .
most of these data are based on the declaration of income of the population , which is assumed to be proportional to the wealth .
there are however some other databases obtained from different sources like for instance the area of the houses in ancient egypt @xcite , the inheritance taxation or the capital transfer taxes @xcite .
the results mostly back pareto s conjecture on the shape of the wealth distribution .
the interesting problem that remains to be answered is the origin of the peculiar functional trend .
the answer to this question is a long standing problem , which even motivated some of the initial mandelbrot s and simon s work fifty years ago .
let @xmath2 be the probability of having a wealth higher than @xmath3 .
pareto s law then establishes that the tail of @xmath2 decays as @xmath4 where @xmath5 is the so called pareto index and @xmath6 the normalized wealth distribution function . typically , the presence of power - law distributions is a hint for the complexity underlying a system .
it is however important to notice that in spite of what happens with most exponents in statistical physics @xmath5 may change in time depending on the economical circumstances @xcite , making thus impossible the definition of some sort of universal scaling in this problem .
this aspect is a key characteristic that any model on wealth distribution should be able to reproduce .
economical models are essentially composed by a group of agents placed on a lattice that interchange money following pre - established rules .
the system will eventually reach a stationary state where some quantities , as for instance the distribution @xmath2 , may be measured . following these ideas , bouchaud and mzard @xcite and solomon and
richmond @xcite separately proposed a very general model for wealth distribution .
this model is based on a mean field type scenario with interactions among all the agents and on the existence of multiplicative fluctuations acting on each agent s wealth .
their results on the wealth distributions adjust well to the phenomenological @xmath2 .
roughly the same conclusions were obtained by scaffeta @xcite , who considered a nonlinear version of the model and from other regular lattice based models as those in refs . @xcite and @xcite .
this kind of models defined on pre - established regular lattices is however unable , by construction , to account for the complexity of the interaction network observed in real economical systems . in parallel to the previous efforts to characterize economical systems , the study of complex networks has experienced a burst of activity in the last few years ( see ref .
@xcite for a recent review ) .
social networks , in particular , are of paramount interest for economy since everyday economical transactions actually produce a network of this type .
the topology of the economical network can indeed condition the output of any economical model running on it .
such effect has been documented for example in refs .
@xcite , in which the models described above were simulated on small - world or scale free networks .
one of the main characteristics of social networks is the positive correlation existing between the node degrees @xcite , i.e. , the high connected individuals commonly tend to connect with other well connected people .
the way of constructing this type of networks is precisely the main topic of a recent work by bogua and coworkers @xcite . in what follows
, we are going to use a somehow similar approach to grow our working network .
bogua s method is based on the existence of hidden variables characterizing each agent state . in this work , and in the spirit of ref .
@xcite , we present a simple economic model where those hidden variables are identified with the wealth of each agent .
this introduces a coupling between the dynamics of the network structure and the evolution of the wealth distribution .
each value of the external parameters thus determines not only the final wealth distribution but also the structure of the underlying interchange network .
this paper is organized as follows . in section 2
we introduce our model , in which the agents are identified as families linked by first - degree family relationship . in section 3 we present computational results on this model . for a wide range of the parameters of the model we study both the wealth distribution and the structure of the underlying network . in section 4
we discuss our results from several viewpoints . in this section
the results are compared with real data on wealth distribution , the correlation between the wealth and connectivity of the agents is studied , and the dynamics leading to wealth diversification is investigated .
section 5 is then dedicated to conclusions .
in modeling the wealth distribution in societies we identify as main entities ( agents ) the families . in the framework of our model ,
the families are nodes in a complex network , and the links of this network are first - degree family relations . beside its links ,
each node is characterized by its `` age '' , @xmath7 , and wealth , @xmath8 .
the age of a node is proportional to the simulation time - steps elapsed from its birth ( @xmath9 , where @xmath10 denotes the time - step when node @xmath11 was born ) , and the wealth is a positive real number that will change in time .
we consider both the total wealth of the system , @xmath12 , and the number of families ( nodes ) , @xmath13 , conserved .
the structure of the network is not a - priori fixed , and will also change during the evolution of the system .
initially we start with nodes arranged on a regular hierarchical network ( as sketched in figure 1 ) where the age of node @xmath11 is simply @xmath14 . in this manner
node @xmath15 will be the oldest and node @xmath13 the youngest one .
it is worth mentioning here that the final statistics of wealth distribution and the final network topology are rather independent on how the initial network topology was chosen .
we verified this by choosing several other qualitatively different initial network structures .
initially we also assign random wealth to each node according to a uniform distribution on the @xmath16 interval . in this manner we constructed the start - up society with a simple network structure ( family relationships ) and
randomly distributed wealth values .
the time - evolution of the system is then chosen to be as simple as possible , but capturing the realistic wealth exchange processes between families .
for each simulation time - step the dynamics is as follows : 1 . the oldest node
( let this be @xmath17 ) is taken away from the system .
the wealth of this node is uniformly redistributed between its first neighbors ( nodes that are linked to it ) , and all its links are deleted .
node @xmath17 is re - introduced in the network with age @xmath18 .
it is linked to two randomly selected nodes ( let these be @xmath19 and @xmath20 ) that have wealth greater than a minimal value @xmath21 .
the wealth @xmath21 is taken away from the wealth of the selected @xmath19 and @xmath20 nodes , and it is redistributed in a random and preferential manner in the society .
the preferential redistribution is realized by splitting the @xmath22 wealth in @xmath23 parts and choosing the the nodes which will benefit from these parts with a probability proportional to their actual wealth .
this preferential redistribution will favor a rich - get - richer effect .
after the redistribution of the @xmath22 amount , a @xmath24 part ( @xmath25 ) of the remaining wealth of node @xmath19 and @xmath20 is given as start - up wealth for node @xmath17 .
after these wealth redistribution processes the wealth of node @xmath19 and @xmath20 will be thus @xmath26(1-p)$ ] and @xmath27= [ w(l)-q](1-p)$ ] , respectively .
node @xmath17 will start with @xmath28 $ ] wealth .
the age of all nodes is increased by unity .
let us now explain the socio - economic phenomena that are modeled by the above dynamics .
step 1 models the inheriting process following the death of one family .
the wealth of this family is redistributed among its first - degree relatives ( children ) .
step 2 models the formation of a new family . in order to create a new family two other families have to raise one child . for raising a child a minimum amount of wealth
is needed ( @xmath21 ) .
this cost is paid to the society ( for food , clothes , services ... ) , and the members of the society will benefit unevenly from it .
families with bigger wealth control more business , so they will naturally benefit more . the preferential redistribution of the @xmath22 wealth models this uneven profit , and it is the main ingredient necessary to reproduce the pareto distribution . finally ,
when a new family is born it is linked by first - degree relations to two existing families and gets a given part ( @xmath24 ) of the parents wealth as start - up money .
the time - scale of the simulation is governed by the time needed to change all nodes , which we call one generation or one monte carlo step ( mcs ) . by fixing @xmath13 and @xmath12 , and studying the thermodynamic limit @xmath29 , the model becomes essentially a two - parameter model ( @xmath21 and @xmath24 ) , which is suitable for extended computer simulations .
although very simple in nature , the chosen dynamics incorporates , we believe , the main socio - economic factors that influence the redistribution of wealth between families .
as time passes the families will be able to gather more and more wealth due to the @xmath22 wealth redistribution process in the society .
when their wealth becomes big enough they can create new families , and donate a part of this wealth to the new family .
this process is costly and will therefore lower their wealth .
very poor nodes will not likely reach the @xmath21 threshold and will not be able to create new families , becoming isolated nodes .
there is no clear determinism however , since the redistribution in step 2 is realized in a random manner , and the selection of the two nodes to which the new family links is also random .
so in principle there is the chance that nodes that start with low wealth will become very rich , or rich nodes do not increase their wealth as expected .
the actual way how the preferential redistribution in step 2 is implemented is by dividing the @xmath22 value in many ( usual several hundred ) equal parts , and each part is assigned to a randomly chosen node , biased proportionally with the wealth of the node . to do this biased redistribution
the use of a bkl @xcite type monte carlo algorithm is very helpful .
another possibility ( leading to the same results ) for doing this preferential redistribution would be to select @xmath23 nodes with the same probability , independently of their wealth , and then to split the @xmath22 amount between the selected nodes proportionally with their actual wealth .
it is also important to note that in realizing step 1 , one can get to a situation where the selected node has no links ( a family dies out without children ) . for simplicity reasons , in this case
we have also chosen to redistribute the wealth of the node in the whole society by using the same preferential rule .
of course , this model is a rough description of the reality and it should be viewed only as a first `` mean - field '' approximation . in real societies the number of families and also the total amount of wealth should not be considered fixed .
many other social aspects could be of interest , the actual value of @xmath21 and @xmath24 should vary from family to family within quite broad distributions , the nodes must not die out according to their age and many cultural and religious factors can influence the dynamics of the underlying social network . in spite of all the neglected effects we will see that this simple model is able to reproduce the observed wealth - distributions and generates reasonable first - degree family relation networks .
the main advantage of this model is that the network structure on which the wealth - exchange is realized is not a - priori put in the system .
the network forms and converges to a stable topology in time , together with the wealth diversification in the system and the appearance of the pareto distribution .
extensive computer simulations were done to study the wealth distribution and the generated family - network for various values of the model parameters . in order to minimize the statistical fluctuations we averaged over @xmath30 realizations for each parameters values .
the model as defined above has several parameters : @xmath13 , the number of nodes in the network , @xmath12 , the total wealth of the system , @xmath31 the number of simulation steps done to reach a given state , the number @xmath23 giving the parts on which the @xmath22 wealth is divided , and the value of the @xmath21 and @xmath24 wealth - exchange parameters . by simple simulations
it is easy to show that the results are independent of the chosen value of @xmath23 , provided that @xmath23 is big enough ( @xmath32 gives already stable results ) . in the results that will be presented we always used @xmath33 .
we will argue in the following that the main free parameters are @xmath21 and @xmath24 , since the model converges rapidly both as a function of time and as a function of the number of nodes to a stable limiting distribution and network - structure .
it is easy to realize that the chosen value of @xmath12 will not change the nature of the results , but it simply rescales the values of the wealth . a simple computer exercise will also convince us that the above defined family - network model converges in time very quickly to a statistically invariant state both for the wealth distribution and network structure .
results for a relatively big lattice ( @xmath34 ) and for realistic @xmath35 and @xmath36 values are presented in figure 2 .
we see that roughly after 5 mcs , both the cumulative wealth distribution and the first two moments of the degree distribution converge to their stable limit .
on the other hand one can also check that the model has a well defined thermodynamic limit . as @xmath13 increases , we obtain again that both the cumulative wealth distribution and the statistical properties of the network reach a stable limit .
characteristic results for this variation are presented in figure 3 . as we can see from the figure , for reasonably big lattices @xmath37
, a stable limit is reached .
we will study now the influence of the @xmath24 and @xmath21 wealth - exchange parameters .
since we have verified that the model converges relatively quickly to a stable limit , we will consider in all simulations 10 mcs .
the number of nodes in the network will be chosen @xmath34 , which ensures that the thermodynamic limit is approached .
the @xmath24 parameter can theoretically vary in the @xmath16 interval ; we consider however , realistic a variation in the @xmath38 interval .
since start from wealth values distributed randomly and uniformly on the @xmath16 interval , the minimal @xmath21 value needed to raise a child should thus also be in the @xmath16 interval , otherwise no new family could be linked to the network .
first we present our results on the @xmath2 cumulative wealth distribution curves . for two fixed values of @xmath24 ( @xmath39 and @xmath35 ) the curves
are given in figure 4 .
the curves in figure 4 suggest that the good scale - free pareto tail is obtained for @xmath21 values in the @xmath40 interval , and we will thus focus in the following on this parameter region .
it is also evident that results for @xmath35 have a better trend .
the pareto index ( power - law exponent ) in this region varies in the @xmath41 interval , depending on the chosen @xmath24 and @xmath21 values and fitting intervals .
the @xmath2 curves have the right shape , they show the power - law trend for the rich nodes and the exponential behavior in the low and medium wealth limit ( figure 5 ) .
moreover , one can also observe that in good agreement with the reality , roughly @xmath42 of the nodes have wealth in the pareto regime .
the network generated by the model is a simple exponential one . considering the realistic @xmath43 and @xmath44 parameter region , in figure 6 we present results obtained for the @xmath45 degree distribution ( probability - distribution that one node has a given number of links ) . from the degree distribution
we conclude that the network is an exponential one .
the most probable connectivity of a node is around @xmath46 , and we obtained that in this parameter region @xmath47 varies between @xmath48 , which are reasonable values for real first - degree family relation networks .
no relevant clusterization was observed in these networks .
it is also instructive to study different kinds of correlations in the generated networks .
first one can study the correlation between the @xmath19 connectivity of the nodes , and mean - connectivity of the neighbors @xmath49 for nodes with @xmath19 links .
if there exist a positive degree - degree correlation , i.e. if well connected nodes tend to connect with well connected ones , then @xmath50 must increase with @xmath19 . in the relevant parameter region results in this sense ,
are plotted in figure 7 . for low values of @xmath24
there are no obvious correlations , but as @xmath24 increases one can observe a positive correlation effect , @xmath50 increases roughly linearly with @xmath19 .
this means that , if the new family gets a bigger portion of the parents wealth , the number of links parents and children have are positively correlated .
the effect is simply understandable , taking into account that for higher values of @xmath24 the wealth of the parents and children should be also correlated , creating similar conditions for accepting links .
the correlation between the wealth @xmath3 of one node and the average wealth of the neighbors @xmath51 , should follow a similar trend .
indeed , as expected , this correlation also has an increasing trend as @xmath24 is increased ( figure 8a ) .
this positive correlation effect is more clear again for not too high wealth values , since in the high @xmath3 limit there are few nodes and the statistics is poor .
a similar correlation trend can be observed if one studies the correlation between the wealth of the nodes and the total wealth of the neighbors . in figure 8
we plotted the results only for @xmath52 , since for higher values of @xmath3 the curves are very noisy due to the poor statistics .
finally , one can study the correlation between the wealth and connectivity of a node , either by plotting @xmath53 ( the average number of links for nodes with wealth around @xmath3 in a given @xmath54 interval ) as a function of @xmath3 , or by simply calculating @xmath55 . in figures 9a and
9b we plot the values of @xmath56 as a function of time , and in figure 9c we show the @xmath57 curves .
( for constructing the curves in figure 9c we used boxes of size @xmath58 . ) from figures 9a and 9b one
notices again , that both the network structure and wealth distribution approach quickly ( less than 5 mcs ) a statistically stable limit .
it is interesting to observe that the @xmath56 correlation is stronger for low @xmath24 values , which makes sense since as @xmath24 increases the availability of a wealthy node to accept more links decreases .
as @xmath24 increases the @xmath56 trend suggests that we deal with a clear anti - correlation between the wealth and number of links of a node , which means that nodes which do not get too many links will in general become wealthy .
the trend of the @xmath59 curves ( figure 9c ) suggests similar conclusions , but here we can also see this correlation effect differentiated as a function of the @xmath3 value . in the low and medium wealth limit
there is a clear anti - correlation between wealth and number of links , while for the wealthy nodes ( much fewer in number ) there is a positive correlation trend . in figure 9c .
we plotted again the data only for @xmath52 , since for higher wealth values the curves are rather noisy due to poor statistics .
let us now analyze real wealth distribution data in societies in order to check the quantitative agreement with our results .
we use estimates for the distribution of personal wealth in united kingdom ( available on the internet ) @xcite , based on inheritance tax , capital transfer tax and other data ( the methods used for getting these estimates are also described in @xcite ) . plotting the cumulative wealth distribution for a chosen year ( 2001 in our case ) , one gets the graph in figure 10 ( distributions for other years are quite similar , even quantitatively ) . on the data presented in figure 10
, one can nicely identify the exponential regime for low and medium wealth values , and the pareto power - law distribution in the high wealth limit . as emphasized in the introduction
, the pareto tail describes the upper @xmath0 of the society .
the uk-2001 data suggests a pareto index @xmath60 ( figure 10 ) . an immediate comparison with the distribution obtained for our family - model ( figure 5 ) ,
shows that for the reasonable @xmath36 and @xmath35 parameters the model offers a fair description .
the pareto index for these parameters is around @xmath61 , in the low and medium wealth limit the @xmath2 curve is exponential , and the pareto law is valid for the upper @xmath42 of the society . concerning the shape of the @xmath2 curve , the model thus seems to work well .
the network structure generated by the model also seems to be realistic .
the exponential nature of the network , the most probable value of the connectivity @xmath62 , and the average connectivity @xmath63 are all reasonable for real first - degree family relation networks .
the correlations @xmath64 , @xmath65 , @xmath66 and @xmath67 , presented in figures 7.-9 . , and described in the previous section , are also reasonable .
this kind of correlations could bee expected , since our model is somehow similar to the ideas of hidden variables proposed by bogua et al @xcite , and their model also generated correlated networks . within the proposed model we can also identify the wealth - diversification mechanism that finally leads to pareto s law .
the time evolution of the @xmath56 correlations ( figure 9a and 9b ) , and the time - evolution for the @xmath2 cumulative distribution functions ( figure 2 ) viewed in parallel give us important clues in this sense . in the beginning of the dynamics
there is usually a strong anti - correlation effect ( @xmath68/dt<0 $ ] ) between wealth and number of links .
this means that in this regime those nodes will become wealthier which have fewer number of links .
the pareto tail here does not exist , and this is where the strong wealth - diversification starts .
after this initial transient regime , the @xmath69 correlation will converge to a stable limit , and simultaneously the stable @xmath2 cumulative distribution function with the pareto tail is formed .
the main mechanism leading to the strong wealth - diversification in our model is thus the initial strong anti - correlation between the wealth and the number of links of one node .
one can also simply verify that the main necessary ingredient that will produce the power - law tail is the preferential wealth redistribution in the system . without the preferential wealth - redistribution of the @xmath22 amounts ,
the model will not generate power - law tails for @xmath2 .
this rich - gets - richer effect seems to be thus the main mechanism leading to power - law wealth distribution in the richer part of the societies .
we have presented a family - network model designed to explain the cumulative wealth distribution in societies . in our model the wealth - exchange is realized on a first - degree family relation network , and it is governed by two parameters .
the dynamics is defined through realistic rules and generates both the underlying family network and wealth distribution .
the model has a stable thermodynamic limit , and the dynamics quickly leads to a network structure and wealth - distribution which are stable in time .
extended computer simulations show , that for reasonable parameter values both the obtained cumulative wealth distribution function and network structure are realistic : ( i ) in good agreement with real measurement data we were able to generate cumulative wealth - distribution functions with pareto - like power - law tails , ( ii ) the obtained pareto index is close to the measured values , ( iii ) the cumulative wealth distribution function for the low and medium wealth values is exponential as found in social data ( iv ) the pareto regime is valid for the upper @xmath0 of the society , ( v ) the generated first - degree family relation network is realistic .
we observed that in our model the initial wealth - diversification is realized through a strong anticorrelation between the wealth of the nodes and their number of links .
as the main mechanism leading to the formation of the pareto power - law tail we identified the preferential redistribution of wealth in the society . in the generated networks many interesting correlations have also been revealed . in spite of its strengths the proposed model is still a rough approximation to reality
. one may argue that many important cultural , social or economic phenomena have been neglected .
we consider this model as a first , mean - field type approximation .
the novel aspect of our approach is however that the network structure was not a - priori introduced in the model , but it got formed during the postulated wealth - exchange dynamics .
subscribing to the ideas presented in @xcite , we also feel that such type of approach should be considered for explaining many other social or economic phenomena and complex network structures .
z. neda acknowledges a nato fellowship and the excellent working atmosphere at the centro de fsica do porto .
r. coelho was partially supported by a junior research grant ( bi ) from centro de fsica do porto .
j. ramasco acknowledges a post - doctoral grant from fct ( portugal ) . | a model based on first - degree family relations network is used to describe the wealth distribution in societies .
the network structure is not a - priori introduced in the model , it is generated in parallel with the wealth values through simple and realistic dynamical rules .
the model has two main parameters , governing the wealth exchange in the network .
choosing their values realistically , leads to wealth distributions in good agreement with measured data .
the cumulative wealth distribution function has an exponential behavior in the low and medium wealth limit , and shows the pareto - like power - law tail for the upper @xmath0 of the society .
the obtained pareto indexes are in good agreement with the measured ones .
the generated family networks also converges to a statistically stable topology with a simple poissonian degree distribution . on this family - network
many interesting correlations are studied , and the main factors leading to wealth - diversification and the formation of the pareto law are identified . , , wealth distribution , random networks , econophysics , pareto s law 89.75-k , 89.65.gh , 89.75.hc , 87.23.ge |
cool stars with a spectral type later than f5 ( cooler than about 6500k ) have convective outer layers , which when coupled with rotation , create a dynamo that amplifies magnetic fields and brings them to the stellar surface .
radial magnetic field lines trap the stellar wind , forcing it to co - rotate with the star out to large radii and removing angular momentum from the star .
@xcite first applied such a model to the solar angular momentum loss , and variants of their model can successfully explain the observed rotational spin - down of cool stars in open clusters of increasing age @xcite .
low - mass secondary stars in cataclysmic variables ( cvs ) should exhibit strong ( 0.1 - 0.3 t ) magnetic fields since they have deep convection zones and are rapidly rotating as a result of being tidally locked with their companion .
( tidal interaction ensures that the secondary star is brought into synchronous rotation with the orbit . )
the currently accepted hypothesis is that magnetic braking by the secondary star is responsible for removing angular momentum from the system , perpetuating the mass transfer @xcite .
magnetic braking is also crucial in the standard explanation for the dearth of cvs with orbital periods between 2 and 3 hours ( the `` period gap '' ) . as cvs evolve towards shorter orbital periods
the secondary star becomes less massive , until finally , at masses of 0.25m@xmath0 and periods around 3 hours , it should be fully convective @xcite .
at this point it is said that either magnetic activity may cease @xcite , or perhaps the magnetic field changes topology @xcite , and thus magnetic braking becomes ineffectual . in turn
, mass transfer will cease , and the secondary star , which is slightly bloated due to the effect of mass loss , shrinks to within its roche - lobe .
gravitational radiation is now the only angular momentum loss mechanism , and only when the binary reaches a period of two hours does mass transfer begin again .
there are two worrying problems with this scenario .
first that it relies on applying single star models to semi - detached binaries at somewhat shorter rotation periods .
the danger of this assumption is underlined by the case of the w uma binaries , whose magnetic fields are a factor of order four or five below those for analogous single stars @xcite .
the second problem is the lack of good observational evidence for magnetic fields in cv secondaries .
strong ( 0.1 - 0.3 t ) magnetic fields covering a large fraction of the stellar surface are well established in rapidly rotating single and binary g to m - type stars ( e.g. * ? ? ?
these fields can be detected directly by zeeman splitting @xcite and inferred indirectly from the presence of cool starspots on the stellar surface , or from the non - radiative magnetic heating in outer atmospheres diagnosed by coronal x - ray emission @xcite . applying these techniques to cvs
is difficult , due in part to the variability of the accretion disc masking that of the secondary star , and in part due to their intrinsic faintness .
there are , however , other indications that spots and/or magnetic activity may be present in cvs . magnetic activity cycles
are a plausible explanation for changes seen on timescales of decades in the overall brightness , inter - outburst periods and orbital periods of cvs .
a critical account of these is given in @xcite , and given that other explanations are possible , they do not amount to a strong case for magnetic activity .
more convincing is to use starspots as an explanation for subtle differences in secondary star line strengths , either in comparison with single stars of similar spectral type , or as a function of orbital phase ( e.g. * ? ? ?
* ; * ? ? ?
however , there are always other possible explanations for these effects , including light from the accretion disc , or the peculiar evolutionary history of the secondary star . in summary
, the evidence to date for magnetic activity in cv secondary stars amounts to using them as possible post - facto explanations of extant observational data .
what is required is an observational test of a prediction based on the idea that the secondary stars have strong magnetic fields .
starspots are the key to the technique we present here . their presence
can be deduced from rotational modulation of stellar brightness or more directly from doppler imaging of the surfaces of rapidly rotating stars ( e.g. * ? ? ?
@xcite have used zeeman doppler imaging to show that starspots do indeed coincide with regions of strong magnetic field , and starspots are now known to exist in both single cool stars and non - interacting binaries which have rotation periods from about 12 hours to a few weeks .
@xcite and @xcite have developed a new way to measure the spot area and temperature , which relies on a typical starspot being approximately 1000k cooler than the unspotted photosphere .
the total spectrum from the star is then the flux - weighted mean from the spotted and unspotted areas . in particular ,
when one observes magnetically active k stars , molecular absorption features due to tio are seen , which can only come from much cooler m - type atmospheres . by modeling the total spectrum as the sum of template k - type and m - type stars , an estimate of the spot area and temperature
is recovered , assuming that the star is uniformly covered by spots . in this letter
we present the first results using a similar method for a cv secondary star
. although the application of this technique to cvs is straightforward in principle , there are observational complications which make it difficult in practice .
first , unlike @xcite , we can not normalise the spectra such that the continuum flux is unity , as the presence of non - stellar components ( e.g. the accretion disc ) whose flux varies with wavelength , would make the relative fluxes in the lines incorrect .
thus we must flux our spectra , ensuring that the relative levels across the spectra are accurate to approximately one percent .
it is well known that normal techniques can introduce problems at this level , and worse still the tio bands are contaminated by absorption from the earth s atmosphere . however , preserving this flux information means that our fitting technique can utilize the overall continuum shape of the template spectra , in addition to the strengths and widths of the absorption lines . for the latter information to be useful , very high signal - to - noise spectra
are required .
we chose to investigate the 6.6 hour orbital period cv , ss cygni . since the secondary star in ss cyg is an early k star ( @xcite and references therein ) , the mere detection of m - type tio features would be a positive indication of spot activity on its surface .
four ss cyg spectra totaling 24 mins exposure time were taken within 0.3 in phase of superior conjunction of the white dwarf ( i.e. when the secondary star is at its closest to the earth ) , using the intermediate dispersion spectrograph on the 2.5 m isaac newton telescope . with the 1.6 arcsec
slit used , the spectrograph yielded a resolution of 2.5 .
the spectra covered the wavelength range 6820 - 8190 at a dispersion of 1.22 per pixel .
however , we only used the region 7000 - 7590 , since this was well corrected for telluric absorption , and contains the tio band of interest .
the reduced spectra were shifted so that the spectral lines lay at their rest wavelength and added , using a weighted mean , to give a spectrum with a signal - to - noise of about 200 .
we also observed 35 single k and m stars to use as template spectra , which were also shifted so that the spectral lines lay at their rest wavelength positions .
we ensured good relative flux calibration by observing an o or b - star close to each target , after the target had been observed .
this hot star was itself fluxed ( using data from flux standards ) , and the resulting spectrum ( excluding telluric absorption bands ) was fitted with a quadratic ( a good approximation to the flux distribution of hot stars in this region of the spectrum ) , which removed any residual fluxing errors on scales of less than 300 .
we next divided the unfluxed spectrum of the target by the unfluxed spectrum of the o - star , thus removing the telluric absorption lines .
finally we multiplied this spectrum by the quadratic fit to the hot star spectrum to obtain a fluxed spectrum of the target .
our mean ss cyg spectrum is shown in figure [ fig1 ] .
there are two strong emission lines at 7065 and 7280 ( both hei ) , whose broadness implies they originate in the accretion disc .
( they are broadened both by the rotation of the disc , and our shifting of the spectra into the rest frame of the secondary star . )
the rest of the spectrum is a host of weak , rotationally broadened absorption lines from the secondary star .
it is immediately evident , even without modeling , that there is a broad absorption trough between 7100 and 7200 , which is normally associated with tio ( although the bandhead itself , at 7055 is obscured by an emission line ) .
however , the secondary star in this cv is of spectral type early k ( 4800 - 5300k ) , and is thus too hot for tio to exist in significant quantity .
indeed tio only appears in stars of spectral type k7 and later , where the temperature is below 4300k .
it would appear , therefore , that we are observing a k - type star , with cooler , m - type regions , which are presumably starspots .
we modeled the mean ss cyg spectrum using varying proportions of our template k and m , giant and dwarf stars , with the addition of a shallow quadratic to represent the non - stellar components .
the spectral lines of the template stars were broadened to match the lines of ss cyg which are rotationally broadened to about 90 km s@xmath1 @xcite .
we restricted the fit to the region 7000 - 7590 , since it contained one tio band and was well corrected for telluric absorption . to begin with we tried a fit which contained just one template star , and the quadratic ( figure [ fig1 ] ) .
the best fitting spectral type was k4v .
when we added a second star ( figure [ fig2 ] ) @xmath2 fell by a factor of 2.6 .
( there were 126 degrees of freedom for the two star fit , which returned a @xmath2 of 3.9 . ) the two star model not only fits the area around tio well , it also fits the region around 7400 better than the single star model .
this second fit gave spectral type of k4v ( 4800k ) for the unspotted parts of the photosphere , and a spot spectral type m2v ( 3900k ) .
the hemisphere we observed had approximately 22% of its area covered in spots . to assess the errors in these figures we fitted a two star model to the individual spectra which had been combined to create our mean ss cyg spectrum .
the resulting scatter implied our error is about a sub - type for both the `` normal '' and spotted spectral types .
about 50% of the light was calculated to come from the secondary star .
surprisingly there was little variation in the implied spot coverage with changes in either the normal or spotted spectral type .
this is because the fall in flux ( as opposed to equivalent width ) across the tio bandhead changes very little with spectral type , when calculated per unit area of atmosphere .
the fall in continuum flux being matched almost exactly by the increasing depth of the absorption . using the data of @xcite and @xcite we find a 10 percent change in the flux drop across the tio bandhead between m2 and m4 . to prove the method
does not have any propensity to introduce tio features where there should not be any , we performed the same experiment on the known spotted star hd 82443 . in this star
the temperature of the unspotted areas is known to be 4925k , with a spot temperature of 4300k @xcite ; thus this object should not show detectable tio .
we fitted the spectrum of this star in an identical fashion to ss cyg , allowing both spectral types to run free .
we found no evidence for tio , and the fit returned a figure of less than 5% for the non - stellar contribution .
the simplest explanation of our observations is that we have observed starspots on the surface of the secondary star in ss cyg .
however , there are three other obvious possibilities which should be considered : gravity darkening of the secondary star ; a peculiar abundance pattern , or an effect caused by the non - stellar component . to examine the possibility that gravity darkening may be responsible for the cooler region(s ) , we modeled a star with a pole temperature of 4800k , observed at an inclination of 60 degrees ( probable for ss cyg ) , using the code of @xcite .
gravity darkening will have its greatest effect at phase 0.5 , when the inner lagrangian point is facing the observer .
this was our reason for preferring to observe phases close to zero , but to ensure our modeling is conservative , we modeled phase 0.3 , the closest that we observed to 0.5 .
we found that only four percent of the flux weighted area lay below the temperature of 4500k where tio first becomes visible . only if the tio band depth at this and lower temperatures were 100 percent , could our observed depth of 4 percent be explained .
in fact the depth at these temperatures is only a few percent .
peculiar abundances are also unlikely to cause the observed effect .
the problem is not merely changing the relative strengths of certain lines , but producing simultaneously the features of an early k star and tio , which only occurs in cooler atmospheres . modeling a range of abundance possibilities to prove this lies outside the scope of this letter , but to be credible , such modeling would have to be backed with an explanation as to why the star has the implied abundances .
finally , could the subtraction of the non - stellar component cause the tio features ?
although the exact nature of the non - stellar component is controversial , it does seem to be smooth on scales of 1000 and more @xcite . given that our residuals are also smooth
, it seems that only if the non - stellar component itself has tio could it explain our results .
our conclusion is therefore that the most convincing explanation of our data is that starspots do exist on the surfaces of cv secondary stars .
these are likely to be associated with equipartition magnetic fields of strength 0.1 - 0.3 t .
this puts the idea that magnetically controlled wind braking is responsible for cv evolution on a much firmer footing and opens the way for further studies of these magnetic fields in ways similar to those being pursued in rapidly rotating single stars .
it also shows that magnetic fields and spot activity do not disappear at extremely high rotation rates , that are an order of magnitude faster than the claimed dynamo or angular momentum saturation threshold seen in the k stars of young open clusters @xcite .
however , much crucial work still needs to be done , including establishing whether spots disappear at rotation periods of 2 - 3 hours , thus providing a natural explanation for the cv period - gap .
cv secondaries at these periods are themselves m - type stars and so the mere detection of tio bands will not be sufficient to prove the presence of starspots .
the isaac newton telescope is operated on the island of la palma by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofisica de canarias .
this work made use of peter van hoof s atomic line list ( http://www.pa.uky.edu/ peter / atomic ) .
the majority of this work was carried out when naw was supported by a pparc studentship at keele university , and tn was in receipt of a pparc advanced fellowship at the same institute .
we are grateful to steve saar for comments on an earlier version of this paper .
barnes j.r .
, collier - cameron a. , james d.j .
, donati j .- f
. , 2001 , mnras , 324 , 231 bouvier j. , forestini m. , allain s. , 1997 , , 326 , 1023 dhillon v.s .
, littlefair s.p . ,
marsh t.r . , sarna m. , boakes e. , 2002 , , accepted donati j - f . , collier cameron a. , 1997 , mnras , 291 , 1 haisch b.m . , schmitt j.h.m.m . , 1996 ,
pasp , 108 , 113 howell s.b . ,
ciardi d.r . ,
dhillon v.s . ,
skidmore , w. , 2000 , 530 , 904 johns - krull c.m . ,
valenti , j.a .
, 1996 , apj , l95 jones h.r.a . ,
longmore a.j . ,
jameson r.f . ,
mountain c.m . , 1994 , , 267 , 413 krishnamurthi a. , pinsonneault m.h . , barnes s. , sofia s. , 1997 , apj , 480 , 303 harrison t.e . , mcnamara b.j . ,
szkody p. , gilliland r.l .
, 2000 , , 120 , 2649 martez - pias i.g . , giovannelli f. , rossi , c. , gaudenzi s. , 1994 , , 291 , 455 messina s. , guinan e.f . ,
lanza a.f .
, ambruster c. , 1999 , , 347 , 249 oneal d. , saar s.h .
, neff j.e . , 1996 , apj , 463 , 766 oneal d. , saar s.h .
, neff j.e . , 1998 , , 501 , l73 rappaport s. , joss p.c . , verbunt , f. , 1983 , apj , 275 , 713 reid i.n . ,
hawley s.l .
, gizis j.e . , 1995 , , 110 , 1838 richman h.r . , applegate j.h . , patterson j. , 1994 , apj , 106 , 1075 robinson e.l . ,
barker e.s . ,
cochran a.l .
, cochran w.d . , nather r.e . , 1981 , apj , 251 , 611 saar s.h . , linsky j.l . , 1985 , apj , 299 , l47 shahbaz t. , naylor t. , charles p.a . , 1993 , mnras , 265 , 655 sills a. , pinsonneault m.h .
, terndrup d.m . , 2000 , apj , 534 335 spruit h.c .
, ritter h. , 1983 , , 124 , 267 stauffer j.r .
, caillault j - p .
, gange m. , prosser c.f . ,
hartmann l.w .
, 1994 , apjss , 91 , 625 stpie k. , schmidtt j.h.m.m . , voges w. , 2001 , a&a , 370 , 157 strassmeier k.g .
, rice j.b . , 1998 , , 330 , 685 taam r.e . , spruit h.c . , 1989 , apj , 345 ,
972 tinker j. , pinsonneault m. , terndrup d. , 2002 , apj , 564 , 877 verbunt f. , zwaan c. , 1981 , 100 , l7 weber e. , davis l.d . , 1967 , apj , 148 , 217 | cataclysmic variables ( cvs ) are interacting binary systems where a cool , rapidly rotating secondary star passes material to a white dwarf . if this mass loss is to continue then there must be continuous angular momentum loss from the system .
by analogy with the sun and other cool stars , it has been assumed that magnetic braking is responsible , angular momentum being carried away by an ionised wind from the secondary star , threading a dynamo - generated magnetic field .
we have discovered tio absorption in the spectrum of ss cyg , whose secondary star should be too hot to show such features .
the most likely explanation of its presence is cool star spots caused by the strong ( 0.1 t ) fields required by the magnetic braking theories . |
according to textbooks , gene expression is a very complicated dynamical process which is regulated at a number of its stages during the synthesis of proteins @xcite . similar to many big cities , with heavy traffic , biological cells host complicated traffic of biochemical signals at all levels . at the nanometer scale ,
clusters of molecules in the form of proteins drive the dynamics of the cellular network that schematically can be divided into four regulated parts : the dna or genes , the transcribed rnas , the set of interacting proteins , and the metabolites .
genes can only affect other genes through specific proteins , as well as through some metabolic pathways that are regulated by proteins themselves .
they act to catalyze the information stored in dna , all the way from the fundamental processes of transcription and translation to the final quantities of produced proteins .
+ for the purpose of modeling , it is essential to generate simple models that help to understand elementary dynamical components of these complex regulatory networks as molecular tools that participate in an important way in the machinery of cellular decisions , that is to say , in the behavior and genetic program of cells .
the central importance of control theory in biology can be assessed through the recent problem of identifying control motifs ( or modules ) , which are patterns that occur in a gene network far more often than in randomized networks of biological regulators @xcite .
this hot issue has been first pinpointed in a breakthrough paper of doyle and collaborators @xcite in which the regulation of bacterial chemotaxis was interpreted in terms of the simple integral control ` adaptive module ' introduced by barkai and leibler @xcite .
since gene regulation appears to occur only at some definite states of the whole process , which in general are not well known , we are from the point of view of control engineering in the case of the reconstruction of those specific states under the condition of limited information .
+ it is quite clear that the availability of all state variables to direct measurement is an extremely rare occasion for gene expression phenomena or when it is possible it could be too expensive . for this particular task , but in completely different technological areas , the engineers have developed software sensors ( state observers ) that accurately reconstruct the state variables of various technological processes @xcite .
the basic concept of state of a system or process could have many different empirical meanings in biology .
for the particular case of gene expression , the meaning of a state is essentially that of a concentration .
the typical problem in control engineering that appears to be tremendously useful in biology is the reconstruction of some specific regulated states under conditions of limited information .
+ in general , an observer is expected to provide a good estimate @xmath0 of the natural state @xmath1 of the original system . for this , one usually can think that some distance @xmath2 ( in the sense of a norm @xmath3 in a vectorial space ) goes to zero as @xmath4 .
such softwares can be constructed using the mathematical model of the process to obtain an estimate @xmath5 of the true state @xmath6 .
this estimate can then be used as a substitute for the unknown state @xmath6 .
the usage of state observers has proven useful in process monitoring and for many other tasks .
the concept of observer is used herein in the sense of control theory , defining an algorithm capable of giving a reasonable estimation of the unmeasured variables of a process . in the case of
gene expression processes the description is made very concrete in the following by looking at quite simple mathematical models that refer to single gene cases and which in principle can be extended to some _ operons _ that are single gene clusters .
+ in this paper we will examine in detail a particularly simple observer due to gauthier and collaborators @xcite possessing arbitrary exponential decay and linear error dynamics for the case of a three - state genetic regulation process .
we were led to consider this observer because of its simplicity and its _ high gain _ property .
the gain is defined as the amount of increase in error in the dynamics of the observer .
this amount is directly related to the velocity with which the observer recovers the unknown signal .
for the observer of gauthier _ et al _ the amount of increase in error is constant and usually of high values leading to a fast recover of the unmeasurable states .
a kinetic model of a simple genetic regulation process was first developed by goodwin as long ago as 1963 @xcite .
it has been further generalized by tyson and othmer @xcite and clearly explained by de jong in his recent review @xcite .
we consider here the most simple version of this kinetic model . for three concentrations @xmath7 , @xmath8 , @xmath9 , corresponding to the messenger rna ( mrna ) that codes for the unstable enzyme , the enzyme , and the metabolite , respectively , we write tyson s model in the form @xmath10 the parameters @xmath11 are all strictly positive and represent production constants , whereas @xmath12 are also strictly positive degradation constants .
these rate equations express a balance between the number of molecules appearing and disappearing per unit time .
notice that the model assumes that the concentration @xmath8 increases linearly with @xmath7 and the concentration @xmath9 linearly with @xmath8 , which are natural assumptions . in the case of @xmath7 ,
the first term is the production term involving a nonlinear nondissipative _ regulation function _ @xmath13 that we take of the @xmath14-steepen hill form ( @xmath15 is the steepness parameter ) in common use @xmath16 for the activation and inhibition cases , respectively .
the parameter @xmath17 gives the threshold for the regulatory influence of the concentration of the metabolite on the target gene , whereas the steepness parameter @xmath14 is a measure of the collective effect of groups of metabolite molecules and also defines the shape of the hill curve .
this nonlinear parametrization describes the ` biological regulation process ' that includes the production of the mrna by transcription of its structural gene , its possible intranuclear processing by cleavage , its enzymatic degradation within the nucleus , and its migration to the cytoplasm by some form of diffusion or biological transport .
once in the cytoplasm , the mrna is both translated into the unstable enzyme and enzymatically degraded .
system @xmath18 and its trivial chain generalization in the linear part is considered to be a good model for the simplest type of allosteric regulation in biochemistry , i.e. , the inhibition or activation of an enzyme or protein by a small regulatory molecule that interacts with the enzyme at a site ( allosteric site ) other than the active site at which catalytic activity occurs .
the interaction changes the shape of the enzyme , thus affecting the active site of the standard catalysis .
this change of shape of the enzyme is sufficient to change its ability to catalyze a reaction in either negative or positive way and enables a cell to regulate needed metabolites .
the allosteric regulation has the typical features of a feedback loop in control theory if the regulatory protein acts on the enzyme in the pathway of its own synthesis .
many attempts have been made to develop nonlinear observer design methods .
one could mention the industrially popular extended kalman filter , whose design is based on a local linearization of the system around a reference trajectory , restricting the validity of the approach within a small region in the state space @xcite,@xcite .
the first systematic approach for the development of a theory of nonlinear observers was proposed some time ago by krener and isidori @xcite . in further works ,
nonlinear transformations of the coordinates have also been employed to put the considered nonlinear sytem in a suitable `` observer canonical form '' , in which the observer design problem may be easily solved @xcite,@xcite,@xcite . the main idea in this case is to find a state transformation to represent the system as a linear differential equation plus a nonlinear term , which is a function of the measured state . in this section ,
we present the design of a nonlinear software sensor in which the metabolite concentration is the naturally measured state ( the most easy to measure ) and corresponds to the mathematical state @xmath9 in the model introduced in the previous section .
therefore , it seems logical to take @xmath9 as the output of the system @xmath19 we now apply the technique of high - gain observers that works for many nonlinear systems and guarantees that the output feedback controller recovers the performance of the state feedback controller when the observer s gain is sufficiently high .
the model given by the aforementioned system @xmath18 has the form @xmath20 in which @xmath21 , and moreover there is a `` physical subset '' @xmath22 where the system lies . to make this mathematically precise
we must introduce some further mathematical terminology .
let us construct the @xmath23th time derivative of the output .
this can be expressed using lie differentiation of the function @xmath24 by the vector field @xmath25 , @xmath26 .
@xmath27 is the @xmath23th lie derivative of @xmath24 by @xmath25 and a function of @xmath6 defined inductively as follows @xmath28 when @xmath29 is observable , the map @xmath30 is a diffeomorphism where @xmath31~.\end{aligned}\ ] ] for @xmath32 to be a local diffeomorphism in a region @xmath33 , it is necessary and sufficient that the jacobian @xmath34 should be nonsingular on @xmath33 and moreover that @xmath32 is one - to - one from @xmath33 to @xmath35 , see @xcite .
notice , that no matter if we choose @xmath36 or @xmath37 , the coordinate transformation is the same .
this means that the structure of the observer will be the same for both cases : gene activation or inhibition .
+ when the system is observable on @xmath33 , it can be rewritten in the global coordinate system defined by @xmath32 in the following matrix form : @xmath38 = \left [ \begin{array}{c } { { \xi}}_2\\ { { \xi}}_3\\ \varphi \left ( { \xi } \right ) \\ \end{array } \right ] \\ y = c \ \xi = [ 1\ 0\ 0]\ \left [ \begin{array}{c } { { \xi}}_1\\ { { \xi}}_2\\ { { \xi}}_3 \\ \end{array } \right ] \end{array } \right . ~,%\nonumber\end{aligned}\ ] ] where , moreover , @xmath39 can be extended from @xmath33 to the entire @xmath40 by a @xmath41 function globally lipschitz on @xmath40 .
the latter form allows us to make use of the following result proven by gauthier and collaborators @xcite : + _ consider the system @xmath42 where @xmath43 is the solution of the matrix equation @xmath44 for @xmath17 large enough , with @xmath45 a matrix of brunovsky form ( @xmath46 ; @xmath47 is the kronecker symbol ) , which plays the role of a shift operator on @xmath48 . then , eq .
( [ observ coord trans ] ) defines an observer for @xmath49 , with @xmath50 _ in our case , an observer is a dynamical system as given by eq .
( [ observ coord trans ] ) that hill track the trajectory of the original system ( here @xmath49 ) .
notice that both systems are identical unless an additional term that compensate the error in the observer , where the error is given by the difference @xmath51 , which is seen to be exponentially decreasing in time .
the gauthier observer is particularly simple since it appears to be only a copy of @xmath52 , together with a correction term that depends only on the dimension of the state space and not on the system @xmath53 itself . in others
words , the structure of the observer does not depend on the hill steepness parameter @xmath14 ( eq . ( [ eqn1 ] ) ) . + for the sake of concreteness we will construct the observer only for the activation case .
however , one should notice that only the function @xmath54 will change for the inhibition case .
the gauthier observer in eq .
( [ observ coord trans ] ) in the original coordinates is given by @xmath55 where @xmath56 for the particular three - dimensional state space of @xmath57 we get @xmath58~.\right.\end{aligned}\ ] ] the matrix @xmath43 in the three dimensional case can be easily computed by means of eq .
( [ ricc ] ) given in the gauthier theorem and its inverse @xmath59 appears to be @xmath60~.\end{aligned}\ ] ] plugging the matrices ( [ upspart ] ) and ( [ invs ] ) in eq . ( [ gobso ] ) , we get the following equation for the observer introduced by gauthier and collaborators as applied to our biological case @xmath61\left ( x_3 - \hat{x}_3 \right)~.\label{observ originals coord } \end{aligned}\ ] ] we use this form of the gauthier observer to estimate the states @xmath7 and @xmath8 of the dynamical system @xmath18 .
we work with @xmath62 and the values of the parameters given in table 1 that are not necessarily the experimental values but are consistent with the requirements of the model .
figure ( [ simulaciones ] ) shows the results of a numerical simulation , where the solid lines represent the true states and the dotted lines stand for the estimates , respectively .
in addition , for the real system we have taken @xmath63 whereas for the observer @xmath64 in order to show the robustness of this type of nonlinear observer with respect to the steepness parameter .
lll symbol & meaning & value ( arb .
units ) + @xmath65 & production constant of mrna & @xmath66 + @xmath67 & production constant of protein & @xmath68 + @xmath69 & production constant of metabolite & @xmath68 + @xmath70 & degradation constant of mrna & @xmath71 + @xmath72 & degradation constant of protein & @xmath68 + @xmath73 & degradation constant of metabolite & @xmath68 + @xmath74 & hill s threshold parameter & @xmath68 + ) for an activation case .
plot ( a ) represents the evolution of mrna concentration in time and plot ( b ) the variation of protein ( enzyme ) concentration in time.,title="fig:",height=170 ] ( -0,-10)time ( -140,110)@xmath75 ( 35,110)@xmath76 ( -30,110)(a ) ( -175,10 ) ) for an activation case .
plot ( a ) represents the evolution of mrna concentration in time and plot ( b ) the variation of protein ( enzyme ) concentration in time.,title="fig:",height=170 ] ( -170,10 ) ( -30,110)(b )
we presented here the mathematical exercise of designing a high - gain observer for a simple one - gene regulation dynamic process involving end - product activation ( inhibition leads to similar results ) , which is able to rebuild in an effective way the non - measured concentrations of mrna and the involved protein .
thus , the limitation of those experiments in which one has available only the metabolite can be overcome by employing this simple observer .
in addition , this type of nonlinear observer could be used on line and is robust with respect to @xmath14 , i.e. , it does not need the exact value of the hill steepness parameter .
however , for more complex inputs of more complicated observable dynamical systems , this constant gain observer could have less performance and be overcome by some _ adaptive _ observers that can change in order to work better or provide more fit for a particular purpose . in the case of more limited information ,
e.g. , for unknown functional form of the regulation function and high noise levels that can spoil the performance of the observer , the completely different mathematical procedure of creating dynamical extensions of the observer system are required @xcite .
this work was partially sponsored by grants from the mexican agency _ consejo nacional de ciencia y tecnologa _ through project 46980 .
j.j . tyson and h.g .
othmer , _ the dynamics of feedback control circuits in biochemical pathways _ , prog .
biol . 5 ( 1978 ) 1 - 62 .
h. de jong , _ modeling and simulation of genetic regulatory systems : a literature review _ , j. comput .
( 2002 ) 67 - 103 .
j.p . gauthier and g. bornard , _ obsevability for any u(t ) of a class of nonlinear systems _ , ieee trans . on autom .
control , ac-26 ( 1981 ) 922 - 926 .
j.p . gauthier and i.a.k .
kupka , _ observability and observers for nonlinear systems _ , siam j. control optimization 32 ( 1994 ) 975 - 994 . h. shim , y.i . son , and j.h .
seo , _ semi - global observer for multi - output nonlinear systems _ , syst .
control lett .
41 ( 2001 ) 233 - 244 . v.
ibarra - junquera , l.a .
torres , h.c .
rosu , g. argello , and j. collado - vides , _ nonlinear software sensor for monitoring genetic regulation processess with noise and modeling errors _ , phys . rev .
e 72 ( 2005 ) 011919 ( 9 pp ) . | high - gain nonlinear observers occur in the nonlinear automatic control theory and are in standard usage in chemical engineering processes .
we apply such a type of analysis in the context of a very simple one - gene regulation circuit . in general
, an observer combines an analytical differential - equation - based model with partial measurement of the system in order to estimate the non - measured state variables .
we use one of the simplest observers , that of gauthier et al . , which is a copy of the original system plus a correction term which is easy to calculate .
for the illustration of this procedure , we employ a biological model , recently adapted from goodwin s old book by de jong , in which one plays with the dynamics of the concentrations of the messenger rna coding for a given protein , the protein itself , and a single metabolite . using the observer instead of the metabolite , it is possible to rebuild the non - measured concentrations of the mrna and the protein . _ keywords : _ high - gain observer ; diffeomorphism ; gene regulation ; gene expression published : physica a 380 ( 1 july 2007 ) 235 - 240 + doi : 10.1016/j.physa.2007.02.105 + arxiv : q - bio/0409036v2 [ q-bio.qm ] |
one of the main interests in hypernuclear physics is to investigate how an addition of @xmath0 particle influences the properties of atomic nuclei . due to the absence of pauli s principle between nucleon and @xmath0 particle
, it is believed that a @xmath0 hyperon can be treated as an _ impurity _ to probe deep interior of the nuclear medium . with the presence of hyperon as an impurity , some bulk properties of nuclei such as the shape and collective motions
may be changed @xcite .
indeed , the shrinkage of @xmath5li , with respect to @xmath6li , has been observed experimentally by measuring the b(e2 ) value from the @xmath7 state to the @xmath8 state of @xmath5li @xcite . as the shape of nuclei plays a decisive role in determining their properties such as quadrupole moment and radius ,
mean - field model calculations have been performed in recent years to investigate the change of nuclear shape due to the addition of a @xmath0 hyperon .
deformed skyrme - hartree - fock(shf ) studies in ref . @xcite
have shown that the deformation parameter of the hypernuclei which they studied is slightly smaller ( within the same sign ) than that of the corresponding core nuclei , and thus no significant effect of @xmath0 hyperon on nuclear deformation was found .
on the other hand , we have performed a relativistic mean field ( rmf ) study and found that the nuclear deformation of the core nuclei completely disappears for @xmath9c and @xmath10si hypernuclei due to the addition of @xmath0 particle @xcite , although we have obtained similar results to the shf calculations in ref .
@xcite for many other hypernuclei . in ref .
@xcite , we have compared between the shf and rmf approaches and have shown that the different results with respect to nuclear deformation between the two approaches is due to the fact that the rmf yields a somewhat stronger polarization effect of @xmath0 hyperon as compared to the shf approach .
we have also shown that the disappearance of the deformation realizes also in the shf approach if the energy difference between the optimum deformation and the spherical configuration is less than about 1 mev @xcite .
all of these mean - field calculations have assumed axial symmetric deformation .
although many nuclei are considered to have axially symmetric shape , the triaxial degree of freedom plays an important role in transitional nuclei , nuclei with shape coexistence , and nuclei with gamma soft deformation @xcite .
in particular , we mention that recent studies with the constrained hartree - fock - bogoliubov plus local quasi - particle random phase approximation ( chfb+lqrpa ) method @xcite as well as the rmf plus generator coordinate method ( rmf+gcm ) @xcite have revealed an important role of triaxiality in large amplitude collective motion in sd - shell nuclei . the aim of this paper is to extend the previous mean - field studies on deformation of hypernuclei by taking into account the triaxial degree of freedom , that is , by including both the @xmath11 and @xmath4 deformations in order to investigate the effect of @xmath0 hyperon in the full ( @xmath11,@xmath4 ) deformation plane .
we particularly study the potential energy surface ( pes ) in ( @xmath11,@xmath4 ) deformation plane with the shf method for carbon hypernuclei as well as some sd - shell hypernuclei .
notice that the shape evolution of the c isotopes in the full ( @xmath1 ) plane has been studied by zhang _
_ using the shf method @xcite .
we extend the work of zhang _
et al . _ by introducing a hyperon degree of freedom .
the paper is organised as follows . in sec .
ii , we briefly summarise the skyrme - hartree - fock method for hypernuclei . in sec .
iii , we present the results for the potential energy surface for the c isotopes as well as for sd - shell nuclei , @xmath2si and @xmath3 mg
. we also discuss the softness of the energy surface along the @xmath4 deformation . in sec .
iv , we summarize the paper .
the self - consistent mean field approach provides a useful means to study the ground state properties of hypernuclei . the core polarization effect ,
that is , the change of properties of a core nucleus due to an addition of @xmath0 particle such as a change of total energy and radius can be automatically taken into account with this method @xcite .
the self - consistent non - relativistic mean field calculations with a skyrme - type @xmath0-nucleon(@xmath0n ) interaction have been performed by rayet in refs .
the relativistic mean field approach has also been applied to hypernuclei in _
e.g. , _ refs .
it has been pointed out that the neutron drip line is extended by the addition of @xmath0 hyperon @xcite . in the present paper , we employ the skyrme - type @xmath0n interaction and perform mean - field calculations by extending the computer code ev8 @xcite to @xmath0 hypernuclei .
the code solves the hartree - fock equations by discretizing individual single - particle wave functions on a three - dimensional cartesian mesh .
the pairing correlation is taken into account in the bcs approximation . with this method
, both axial and triaxial quadrupole deformations can be automatically described .
the code ev8 has been applied to the study of shape transition and deformation of several nuclei @xcite in the ( @xmath1 ) deformation plane .
the skyrme - type @xmath0n interaction is given in complete analogy with the nuclear skyrme interaction @xcite .
the skyrme part of the total hypernuclear energy thus reads @xmath12,\ ] ] where @xmath13(*r * ) is the standard nuclear hamiltonian density based on the skyrme interaction .
e.g. , _ refs . @xcite for its explicit form . @xmath14(*r
* ) is the hyperon hamiltonian density given in terms of the lambda and nucleon densities as ( with a correction for the coefficient of the @xmath15 term ) @xcite @xmath16 here , @xmath17 , @xmath18 , and @xmath19 are the particle density , the kinetic energy density , and the spin density of the @xmath0 hyperon
. these are expressed using the single - particle wave function @xmath20 for the @xmath0 particle .
@xmath21 , @xmath22 , and @xmath23 are the total densities for the nucleons .
@xmath24 , and @xmath25 are the skyrme parameters for the @xmath0n interaction .
the hartree - fock equations for the single - particle wave functions are obtained by taking variation of the energy @xmath26 .
the equation for the nucleons reads , @xmath27\phi _ q = e_q\phi _ q , \label{hf - nu}\ ] ] where @xmath28 refers to protons or neutrons , while that for the @xmath0 particle reads @xmath29\phi _ { \lambda } = e_{\lambda}\phi _ { \lambda}.\ ] ] here , @xmath30 and @xmath31 are the single - particle energies and the effective mass for the hyperon is given by @xmath32 @xmath33 is the single - particle potential originating from the nucleon - nucleon skyrme interaction @xcite . @xmath34 and @xmath35 are the single - particle potentials originating from the @xmath0n interaction .
these are expressed as @xcite @xmath36 and @xmath37 the pairing correlations among the nucleons are treated in the bcs approximation . for the pairing interaction ,
we employ a zero - range density - dependent pairing force @xcite , @xmath38 where @xmath39 is the spin - exchange operator , @xmath40 @xmath41 , and @xmath42 . in this paper
, we mainly use the sgii interaction @xcite for the nn interaction , while the set no.@xmath43 in ref .
@xcite for the @xmath0n interaction .
the latter interaction was constructed by fitting to the binding energy of @xmath44o , yielding the well depth for a @xmath0 particle , @xmath45= 29.38 mev , in infinite nuclear matter .
notice that the spin - orbit strength @xmath25 in the @xmath46 interaction is zero for the set no . 1 @xcite . for the pairing interaction ,
we follow ref .
@xcite to use @xmath47 mev@xmath48@xmath49 for both protons and neutrons for carbon hypernuclei , while we follow ref .
@xcite to use @xmath50 mev@xmath48@xmath49 for calculations of sd - shell hypernuclei
. a smooth pairing energy cutoff of 5 mev around the fermi level is used @xcite .
we assume that the @xmath0 particle occupies the lowest single - particle state . since the primary purpose of this paper is to draw the potential energy surfaces ( pes ) of @xmath0 hypernuclei as a function of @xmath11 and @xmath4 deformation parameters , the isoscalar quadrupole constraint is imposed on the total energy .
we relate the deformation parameter @xmath11 for hypernuclei with the total quadrupole moment @xmath51 using the equation @xmath52 where @xmath53 is the mass number of the core nucleus and @xmath54 fm is the radius of the hypernucleus .
c and ( b ) @xmath55c in the ( @xmath1 ) deformation plane obtained with the sgii parameter set .
each contour line is separated by 0.07mev .
the triangles indicate the absolute minima in the pes .
, title="fig : " ] + c and ( b ) @xmath55c in the ( @xmath1 ) deformation plane obtained with the sgii parameter set .
each contour line is separated by 0.07mev .
the triangles indicate the absolute minima in the pes .
, title="fig : " ] + c and @xmath55c along the axially symmetric deformation corresponding to fig .
the energy surface for @xmath55c is shifted by a constant amount as indicated in the figure . ]
we now numerically solve the hartree - fock equations and discuss the deformation properties of hypernuclei in the ( @xmath1 ) deformation plane .
we first investigate the shape of carbon hypernuclei from @xmath55c to @xmath56c . to this end , we follow refs . @xcite and reduce the strength of the spin - orbit interaction by a factor of 0.6 in the skyrme functional .
this prescription was introduced in order to reproduce an oblate ground state of @xmath57c @xcite .
c and @xmath9c .
each contour line is separated by 0.2mev.,title="fig : " ] c and @xmath9c .
each contour line is separated by 0.2mev.,title="fig : " ] c and @xmath9c . ]
figure [ 10c ] shows the potential energy surfaces for @xmath58c and @xmath55c so obtained .
the triangles in the figure indicate the ground state minimum in the energy surface .
the energy curve along the axially symmetric deformation is also shown in fig . [ 10caxial ] as a function of the quadrupole deformation parameter @xmath11 . along the axially symmetric configuration
, there are deep energy minima in both sides of prolate and oblate configurations of @xmath58c ( that is , the shape coexistence ) , having a very small energy difference of less than 40 kev between them .
the ground state corresponds to the prolate configuration with @xmath11=0.35 .
however , the energy surface is almost flat along the triaxial deformation @xmath4 as one can see in fig .
[ 10c ] ( a ) in the ( @xmath1 ) deformation plane . with the addition of a @xmath0 hyperon , the ground state configuration moves from prolate to oblate ,
although the energy surface is so flat along the @xmath4 degree of freedom ( see fig . [ 10c](b ) ) that the ground state configuration may not be well defined in the mean field approximation .
we have confirmed that this feature remains the same even if we use the skyrme @xmath0n interaction set no.2 and no .
5 @xcite , instead of no .
c and @xmath59c .
each contour line is separated by 0.15mev , title="fig : " ] c and @xmath59c .
each contour line is separated by 0.15mev , title="fig : " ] for the optimum values of the @xmath11 deformation parameters shown in fig . [
the energy surface for @xmath59c is shifted by a constant amount as indicated in the figure . ]
[ 12c ] ( a ) and [ 12caxial ] show the energy surface for @xmath57c in the @xmath60 deformation plane and that along the @xmath61 line , respectively . for this nucleus ,
the skyrme - hartree - fock method with the reduced spin - orbit interaction yields a deep oblate minimum . in this case , the addition of a @xmath0 particle shows little effect on the energy surface , although the deformation is slightly smaller than its core nucleus ( see figs .
[ 12c](b ) and [ 12caxial ] ) .
this is similar to the result for @xmath62ne discussed in ref.@xcite with the rmf method . for the case of @xmath63c
, it has a triaxial minimum at @xmath64 and @xmath65 at an energy of @xmath660.154mev with respect to the prolate configuration ( see fig .
[ 18c ] ( a ) ) .
we find that @xmath59c has a ground state configuration with similar values of @xmath11 and @xmath4 deformation parameters to those of the core nucleus , @xmath63c , as indicated in fig .
[ 18c ] ( b ) .
the triaxial minimum is again shallow with an energy difference of 0.077 mev between the optimum deformation and the prolate configuration . in fig .
[ 18c - g ] , we plot the energy curve as a function of the triaxial deformation parameter @xmath4 for the optimum values of the @xmath11 deformation parameter . with the addition of a @xmath0 particle to @xmath63c
, one sees that the energy curve becomes slightly softer ( compare also figs .
[ 18c](a ) and [ 18c](b ) ) .
c and @xmath63c nuclei in the @xmath60 deformation plane . ]
the energy difference between @xmath59c and @xmath63c at each deformation , that is , @xmath67 , in the @xmath60 deformation plane is plotted in fig .
it clearly shows that the addition of a @xmath0 particle prefers the spherical configuration even if the core nucleus has a deformed minimum .
we have confirmed that this is the case for the other carbon isotopes as well .
notice that the @xmath0 particle slightly prefers the prolate configuration for a fixed value of @xmath11 , causing the slightly softer energy curve towards the prolate configuration shown in fig .
this originates from the fact that the overlap between the deformed nuclear density and a spherical @xmath0 density is maximum at the prolate configuration , as we discuss in the appendix .
the results of our calculations are summarised in table i , together with the results for the other carbon isotopes .
si and @xmath10si .
each contour line is separated by 0.6 mev.,title="fig : " ] si and @xmath10si . each contour line is separated by 0.6 mev.,title="fig : " ] mg and @xmath68 mg .
each contour line is separated by 0.5 mev.,title="fig : " ] mg and @xmath68 mg .
each contour line is separated by 0.5 mev.,title="fig : " ] let us next discuss the deformation energy surfaces of @xmath2si , and @xmath3 mg nuclei in the ( @xmath1 ) deformation plane .
for these nuclei , we use the original strength for the spin - orbit interaction .
we first show the potential energy surface of @xmath69si and @xmath70 mg in figs . [ 28si](a ) and fig . [ 24mg](a ) , respectively .
the energy surface for @xmath69si shows a deep oblate minimum , while that for @xmath70 mg shows a deep prolate minimum .
notice that n = z=14 is an oblate magic number .
the energy curves for @xmath70 mg as a function of @xmath11 with @xmath71 , and of @xmath4 with @xmath72 are also plotted in in figs . [ 24axial](a ) and [ 24axial](b ) , respectively .
mg and @xmath68 mg along the axially symmetric deformation corresponding to fig .
( b ) the energy curve as a function of the triaxial deformation parameter @xmath4 for the optimum values of the @xmath11 deformation parameters shown in fig .
the energy surfaces for @xmath68 mg are shifted by a constant amount as indicated in the figures.,title="fig : " ] mg and @xmath68 mg along the axially symmetric deformation corresponding to fig .
( b ) the energy curve as a function of the triaxial deformation parameter @xmath4 for the optimum values of the @xmath11 deformation parameters shown in fig .
the energy surfaces for @xmath68 mg are shifted by a constant amount as indicated in the figures.,title="fig : " ] for these nuclei , as the potential minimum is deep , the addition of a @xmath0 particle does not change significantly the shape of the potential energy surface , as shown in figs . [
28si](b ) , [ 24mg](b),and [ 24axial ] .
as shown in the previous subsection , the energy gain due to the additional @xmath0 particle appears mainly around the spherical configuration .
notice , however , that the @xmath0 particle makes the energy curve slightly softer along the triaxial degree of freedom , @xmath4 .
si and @xmath73si .
each contour line is separated by 0.4 mev.,title="fig : " ] si and @xmath73si . each contour line is separated by 0.4 mev.,title="fig : " ] mg and @xmath73 mg . each contour line is separated by 0.4 mev.,title="fig : " ] mg and @xmath73 mg . each contour line is separated by 0.4 mev.,title="fig : " ] mg and @xmath73 mg along the axially symmetric deformation corresponding to fig .
( b ) the energy curve as a function of the triaxial deformation parameter @xmath4 for the optimum values of the @xmath11 deformation parameters shown in fig .
the energy surfaces for @xmath73 mg are shifted by a constant amount as indicated in the figures.,title="fig : " ] mg and @xmath73 mg along the axially symmetric deformation corresponding to fig .
( b ) the energy curve as a function of the triaxial deformation parameter @xmath4 for the optimum values of the @xmath11 deformation parameters shown in fig .
the energy surfaces for @xmath73 mg are shifted by a constant amount as indicated in the figures.,title="fig : " ] we next discuss the @xmath73 mg and @xmath74si nuclei .
@xmath75 mg has z=12 and n=14 , that is , the protons favour a prolate configuration while the neutrons favour an oblate configuration . as a competition of these two opposite effects
, the structure of @xmath75 mg may not be trivial @xcite .
@xmath75si is the mirror nucleus of @xmath75 mg , and the deformation properties are expected to be similar to @xmath75 mg .
figures [ 26si](a ) and fig . [ 26mg](a ) show the potential energy surfaces for @xmath75si and @xmath75 mg , respectively .
indeed , the two energy surfaces are similar to each other , and show an oblate minimum with a considerably flat surface along the @xmath4 direction .
the energy difference between the oblate and the prolate configurations is 0.12 mev for @xmath75si and 0.39 mev for @xmath75 mg .
the energy curve for @xmath75 mg as a function of @xmath11 with @xmath71 , and of @xmath4 with @xmath76 are plotted in in figs .
[ 26axial](a ) and [ 26axial](b ) , respectively .
the energy curves for @xmath75si are qualitatively similar , and are not shown . as in @xmath70 mg and @xmath69si , the addition of a @xmath0 hyperon does not significantly alter the potential energy surface of these nuclei , although it somewhat softens the energy surface along the @xmath4 direction ( see figs . [ 26si](b ) , [ 26mg](b ) , and [ 26axial ] ) .
we again find that the additional @xmath0 particle favours the spherical configuration .
the calculations presented in this subsection are performed with the sgii set of the skyrme interaction .
we have repeated the same calculations with another skyrme parameter , siii , and have found that the results are qualitatively the same as the results obtained with sgii . for a long time , the @xmath70 mg nucleus has been considered to be a candidate of nuclei with a triaxial shape , because of the low - lying second 2@xmath77state in the rotational spectrum @xcite .
that is , the experimental spectrum of @xmath70 mg has been interpreted as consisting of a @xmath78=0 ground state rotational band and a @xmath78=2 rotational band built upon a @xmath4 vibrational state at 4.23 mev @xcite .
the previous mean - field calculations for the ground state of @xmath70 mg using rmf @xcite and shf @xcite have shown an axially symmetric prolate ground state .
recently , it has been pointed out that the angular momentum projection is essential in order to reproduce the triaxial ground state of @xmath70mg@xcite .
the 3-dimensional angular momentum projection plus generator coordinate method ( 3damp+gcm ) calculations have shown a good agreement with the experimental data on low - spin states of @xmath70 mg @xcite . in order to see the effect of @xmath0 hyperon on @xmath4 vibration , we compute the second derivative of the energy curve with respect to @xmath4 around the minimum , @xmath79 .
that is , when we approximate the energy curve around the minimum as , @xmath80 the second derivative @xmath81 provides information on the frequency @xmath82 for the @xmath4 vibration if the vibrational moment of inertia @xmath83 is known . by numerically taking the second derivative , we obtain @xmath84 = 0.961 and @xmath85 = 0.988 .
that is , the addition of a @xmath0 particle makes the @xmath4 vibration softer if the change in the moment of inertia @xmath83 is negligible .
although the @xmath0 particle softens the energy curve both for @xmath68 mg and @xmath73 mg , the mechanism is somewhat different between the two nuclei . for the @xmath73 mg nucleus ,
the prolate configuration decreases more energy as compared to the oblate configuration ( see fig .
13 ( b ) ) , because the prolate configuration has a larger overlap between the @xmath0 particle and the nucleon densities , as is discussed in the appendix . as a consequence ,
the curvature of the energy curve at the oblate minimum becomes smaller with the addition of a @xmath0 particle .
in contrast , the effect of the smaller value of @xmath11 is more significant in @xmath68 mg .
that is , the energy curve along the axially symmetric configuration is considerably steep for @xmath86 ( see fig .
10 ( a ) ) , and even a small change in @xmath11 induces a significant energy change at the oblate configuration .
therefore , for @xmath68 mg , the energy decreases more at the oblate side as compared to the prolate side for a fixed value of @xmath11 ( see fig . 10 ( b ) ) , leading to the softer gamma - vibration . notice that the absolute value of @xmath11 is relatviely small for @xmath75 mg , lying in a `` flat '' region , and this effect is much less important in the @xmath73 mg nucleus .
[ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ]
we have investigated the shape of @xmath0 hypernuclei in the ( @xmath1 ) deformation plane with the skyrme - hartree - fock + bcs approach .
in contrast to the previous mean - field studies , we have taken into account the triaxial deformation using a 3-d cartesian mesh method . we have studied the potential energy surface for the carbon hypernuclei from @xmath55c to @xmath56c as well as sd - shell hypernuclei @xmath73si , @xmath10si , @xmath68 mg and @xmath73 mg .
the potential energy surface for @xmath58c , @xmath75 mg , and @xmath75si is characterized by a flat surface along the @xmath4 degree of freedom connecting the prolate and the oblate configurations .
we have found that the addition of a @xmath0 particle makes the energy surface slightly softer along the triaxial degree of freedom , although the gross feature of the energy surface remains similar to the energy surface for the corresponding core nuclei . in refs .
@xcite , we have argued that the influence of the addition of a @xmath0 particle is stronger in the relativistic mean - field approach as compared to the non - relativistic skyrme - hartree - fock approach employed in this paper .
this implies that the softening of the energy curve for the @xmath4-vibration may be larger than that estimated in this paper , if we employ the rmf approach instead of the shf approach
. it would be an interesting future subject to carry out three - dimensional calculations for hypernuclei with rmf in order to confirm whether it is the case
. there would be many ways to improve our calculations presented in this paper .
firstly , as the angular momentum projection is shown to be essential to yield the triaxial shape of @xmath70 mg @xcite , it may be important to carry out the angular momentum projection on top of the mean - field energy surface in order to discuss the effect of @xmath0 particle on the shape of hypernuclei . secondly , for nuclei with a flat energy surface along the @xmath4 direction
, the generator coordinate method may be required .
in particular , it will provide a more quantitative estimate for the energy change of the @xmath4 vibrational state due to the addition of @xmath0 particle . in any case , the mean - field calculations presented in this paper provide a good starting point for these calculations .
it will be an interesting subject to experimentally measure the deformation properties and collective motions of hypernuclei .
a discussion has been started for a future experiment of @xmath4-ray spectroscopy of sd - shell hypernuclei at the new generation experimental facilities , _
e.g. , _ the j - parc facility @xcite . the change of deformation will be well investigated if excitation energies in a rotational band and the b(e2 ) values can be measured experimentally .
we thank h. tamura , y. zhang , n. hinohara and a. ono for useful discussions .
this work was supported by the japanese ministry of education , culture , sports , science and technology by grant - in - aid for scientific research under the program number 22540262 .
particle densites as a function of the triaxiality @xmath4 for a fixed value of @xmath87 .
the @xmath0-particle density is assumed to be a spherical gaussian function , while a deformed woods - saxon shape is considered for the nucleon density . ] in this appendix , we discuss the overlap between @xmath0 particle and nucleon density distributions using simple parametrizations for the density distributions .
since we consider that a @xmath0 particle occupies the lowest single - particle state , we assume that the @xmath0 particle density is almost spherical and is given by , @xmath88 on the other hand , for the nucleon density , we assume that it is given by a deformed woods - saxon form , that is , @xmath89},\ ] ] where @xmath90.\end{aligned}\ ] ] here , the radius @xmath91 is determined for each @xmath11 and @xmath4 in order to satifsy the volume conservation condition , that is , @xmath92},\ ] ] is independent of @xmath11 and @xmath4 .
figure 14 shows the overlap between @xmath93 and @xmath94 , that is , @xmath95 as a function of the triaxiality @xmath4 for a fixed value of @xmath87 . to this end
, we use @xmath96 fm , and @xmath97=0.55 fm .
the value of @xmath98 is fixed so that the volume integral of @xmath94 is 24 .
we use @xmath99=1.565 fm , which corresponds to the harmonic oscillator with a frequency of @xmath100 mev . as one
can see , the overlap is the largest for the prolate configuration , although the variation is small with respect to @xmath4 . | we study the shape of @xmath0 hypernuclei in the full ( @xmath1 ) deformation plane , including both axially symmetric and triaxial quadrupole deformations . to this end
, we use the constrained skyrme hartree - fock+bcs method on the three - dimensional cartesian mesh .
the potential energy surface is analyzed for carbon hypernuclei as well as for sd - shell hypernuclei such as @xmath2si and @xmath3 mg .
we show that the potential energy surface in the ( @xmath1 ) plane is similar to each other between the hypernuclei and the corresponding core nuclei , although the addition of @xmath0 hyperon makes the energy surface somewhat softer along the @xmath4 direction . |
the grand unified theories ( guts ) , which embed the standard model ( sm ) gauge group into a large single gauge group , are quite attractive , and so a variety of models of the theories are proposed since the earliest work based on the su(5 ) symmetry group was presented by georgi and glashow in 1974 @xcite . among them , the supersymmetric grand unified theories ( susy guts ) are considered to be promising candidates since they realize the gauge coupling unification with great accuracy @xcite as well as solving the hierarchy problem in the guts @xcite . the supersymmetric version of the georgi - glashow su(5 ) gut , which is the simplest among the susy guts , is called the minimal susy su(5 ) gut @xcite .
one of the most distinctive features which guts predict is the existence of baryon - number violating interactions , such as proton decay . in the susy guts with @xmath6 parity , for example ,
proton decay is induced by two different processes : the colored - higgs and the @xmath7-boson exchanging processes .
the colored higgs triplets are introduced for the higgs doublets in the minimal supersymmetric standard model ( mssm ) to be incorporated into the su(5 ) multiplets
. the colored - higgs exchange yields baryon - number violating dimension - five operators , which give rise to the dominant contribution to proton decay in the minimal susy su(5 ) guts @xcite .
on the other hand , @xmath7 bosons are the gauge bosons in the unified su(5 ) gauge group , and they induce proton decay through the dimension - six operators . since there has been no experimental signal of proton decay so far , strong limits are imposed on these interactions , especially on the former one .
super - kamiokande gives the bounds on the proton lifetime : @xmath8 yrs with 219.7 kt - yr of data @xcite and @xmath9 yrs with 172.8 kt - yr of data at 90% confidence level @xcite .
these bounds are so stringent that the minimal su(5 ) susy gut is excluded since the colored - higgs - boson exchange process yields too short lifetime in the @xmath10 channel @xcite .
this prediction is , however , quite model - dependent .
in fact , many models are proposed in order to allow them to evade the experimental constraints .
such attempts @xcite are accomplished by pushing the mass of the colored higgs much heavier than the gut scale and/or suppressing the dimension - five operators by a certain symmetry , _
e.g. _ , the peccei - quinn ( pq ) symmetry @xcite in the four - dimensional guts . in the extra - dimensional models , the @xmath6 symmetry is introduced to suppress the dimension - five proton decay @xcite . with the extensions ,
the @xmath7-boson exchanging interactions become dominant and the proton lifetime turns out to be long enough to avoid the current limit . in this work , we assume such mechanism works and concentrate on the proton decay via the @xmath7-boson exchange . nowadays , the weak - scale supersymmetry itself is severely constrained by the experiments at the large hadron collider ( lhc ) .
since there has been no signal of susy particles , the atlas and the cms collaborations give severe limits on their masses , especially those of colored particles @xcite .
moreover , the collaborations have recently announced that they detected signatures of higgs boson for the first time @xcite . according to their results , the mass of higgs boson is about 125 gev . the result is welcome to the mssm , which predicts the light higgs boson , while the substantial radiative corrections are required in order to raise the mass of higgs boson up to 125 gev .
since the corrections are increased as the susy particles become heavy , this situation implies that the susy scale might be higher than the electroweak scale @xcite , and challenges natural solutions to the hierarchy problem .
these results give constraints on models of the susy - breaking mediation .
the low - energy gauge mediation model may be required to introduce several messenger multiplets so that the susy particles are heavy enough @xcite . on the other hand
, there have been several alternatives to explain the 125 gev higgs boson in an extension of the mssm .
introducing vector - like supermultiplets to the mssm is one of the simplest way to accomplish the purpose @xcite . in this case , the quantum effects by the extra multiplets help to increase the higgs - boson mass . with the multiplets being a representation of the grand unified group ,
the perturbative gauge coupling unification is still preserved with great accuracy .
the mssm with vector - like multiplets and high - scale susy models modifies the proton decay rates . in this work ,
we examine it in detail .
we will find that vector - like matters enhance the proton decay rate , while heavy sfermions extend its lifetime .
this investigation indicates that the proton - decay experiments might provide a hint for the low - energy structure of the susy guts .
we assume that the susy guts are realized in the four - dimensional spacetime , and not consider the extra - dimensional susy guts in which the gauge symmetry is broken by the boundary conditions . in the latter models , the main decay modes of proton depend on configuration of quarks and leptons in the extra - dimensional space @xcite
. however , the decrease / enhancement of the proton lifetime we show are almost universal even in those models .
this paper is organized as follows .
we begin by reviewing the susy su(5 ) guts in sec .
[ msgut ] , and discuss the proton decay via the @xmath7-boson exchange in the subsequent section .
then , we evaluate the lifetime of proton decay in the case of the mssm with vector - like matters and heavy sfermions in sec .
[ vectorlike ] and sec .
[ heavy_susy ] , respectively .
section [ conclusion ] is devoted to conclusion . in appendix , we present formulae for evaluating the long - distance part of the renormalization factors .
first , we review the susy su(5 ) guts in order to illustrate the notation and conventions which we use in this article . just like the georgi - glashow su(5 )
model @xcite , we assume that the sm fermions , as well as their superpartners , are embedded in a @xmath11 representation .
the multiplets @xmath12 and @xmath13 , which are the matter fields of @xmath14 and @xmath15 representations , respectively , are identified as @xmath16 where all of the component fields are expressed in terms of the left - handed chiral superfields , and primes show that the above fields are the gauge eigenstates .
the super / sub scripts of @xmath17 , @xmath18 , @xmath19 , and @xmath20 denote a color index with @xmath21 .
the chiral superfields @xmath22 and @xmath23 in @xmath12 as well as @xmath17 and @xmath18 in @xmath13 form the su(2)@xmath24 doublets , respectively , @xmath25 while @xmath19 , @xmath20 , and @xmath26 are the su(2)@xmath24 singlets .
the su(5 ) gauge theory contains the 24 gauge bosons and each of them corresponds to a component of a vector superfield , @xmath27 , with @xmath28 indicating the gauge index . by exploiting the fundamental representation of the su(5 ) generators , @xmath29
, we define a @xmath30 matrix of the vector superfields : @xmath31 .
the components of the matrix are written as @xmath32 where each component is expressed by the same symbol as that used for the corresponding gauge field .
we collectively refer to @xmath33 , @xmath34 , and their hermitian conjugates as @xmath7 and @xmath35 bosons , and use the following notation for them : @xmath36 here @xmath37 denote the isospin indices .
the higgs superfields in the mssm , on the other hand , are incorporated into a pair of fundamental and anti - fundamental fields as @xmath38 where the last two components are corresponding to the mssm higgs superfields , @xmath39 and the new higgs superfields @xmath40 and @xmath41 are called the higgs color triplet superfields .
the superpotential for the yukawa couplings of quarks and leptons is given as @xmath42 where @xmath43 indicate the generations and @xmath44 represent the su(5 ) indices .
the yukawa couplings @xmath45 and @xmath46 in eq .
( [ superpotential ] ) have redundant components and much of the degree of freedom is eliminated through the field re - definition of @xmath13 and @xmath12 @xcite .
we parametrize the couplings according to ref . @xcite as @xmath47 with @xmath48 the kobayashi - maskawa matrix .
the phase factors @xmath49 are subject to a condition : @xmath50 with the parameters , we express the matter fields in terms of the mass eigenstates as follows : @xmath51 @xmath52 in the following discussion , we express interactions in the basis of mass eigenstates unless otherwise noted . here , we do not take specific assumptions for symmetry breaking of su(5 ) and the mass generation of the colored higgs .
they are related to suppression of the proton decay induced by dimension - five operators .
next we discuss the proton decay rate induced by the @xmath7 and @xmath35 boson exchange .
the couplings between the gauge bosons and the sm fermions are given as @xmath53~,\end{aligned}\ ] ] where @xmath54 is the su(5 ) gauge coupling constant and @xmath55 indicates the charge conjugation .
@xmath56 and @xmath57 denote the second and third rank totally antisymmetric tensors , respectively .
this interaction lagrangian causes proton decay . in the present case ,
the dominant decay mode is @xmath58 , and this decay process is induced by the following effective lagrangian : @xmath59~. \label{effective_lagrangian}\end{aligned}\ ] ] where @xmath60 is the mass of @xmath7 and @xmath35 bosons and @xmath61 .
the renormalization effects resulting from the anomalous dimensions of the operators are represented by @xmath62 and @xmath63 , which we will evaluate below .
the hadron matrix elements of the operators are evaluated by using the ordinary chiral lagrangian method @xcite : @xmath64p_lu^\gamma = \frac{\alpha_{\rm h}}{\sqrt{2}f_\pi}(1+d+f)p_lu_p(\bm{p},s)~,\ ] ] and @xmath65p_ru^\gamma = \frac{\alpha_{\rm h}}{\sqrt{2}f_\pi}(1+d+f)p_ru_p(\bm{p},s)~,\ ] ] where @xmath66 is the pion decay constant , and @xmath67 is defined by the equation
@xmath68p_lu^\gamma = \alpha_{\rm h}p_lu_p(\bm{p},s)~,\ ] ] with @xmath69 the charge conjugation matrix .
the value of @xmath67 is computed in ref . @xcite as @xmath70 at the renormalization scale @xmath71 gev . by using the matrix elements ,
we obtain the partial decay width @xmath72 induced by the effective lagrangian in eq .
( [ effective_lagrangian ] ) : @xmath73~ , \label{decay_width}\end{aligned}\ ] ] where @xmath74 and @xmath75 are the masses of proton and the neutral pion , respectively , and @xmath76 . in the following calculation , we take @xmath77 , @xmath78 , and @xmath79 as in ref . @xcite . now , in order to evaluate the proton decay rate , all we have to do is to determine the unified gauge coupling constant @xmath80 , the @xmath7 boson mass @xmath60 , and the renormalization factors @xmath81 and @xmath82 . they are dependent on each gut model .
the unified gauge coupling constant @xmath80 is computed by solving the renormalization group equations ( rges ) for the gauge coupling constants @xmath83 ( @xmath84 ) in the sm gauge interactions . in this article
, we consider the gauge coupling running up to the two - loop level , and exploit the @xmath85 renormalization scheme in order to respect the supersymmetry @xcite .
furthermore , we adopt the definition of @xmath80 as @xmath86 with @xmath87 gev . in the mssm ,
the two - loop renormalization group equations ( rges ) for the gauge coupling constants are given as @xcite @xmath88~,\ ] ] where @xmath89 and @xmath90 with @xmath91 the yukawa couplings .
here , we use the su(5 ) normalization for the u(1 ) hypercharge . since the yukawa couplings enter into the two - loop level contributions to the gauge coupling rges , it is sufficient to consider the rges for the yukawa couplings at one - loop level .
they are given as @xmath92 , \nonumber \\
\mu\frac{\partial}{\partial \mu}y_b & = \frac{1}{16\pi^2}y_b \biggl[6y_b^2+y_t^2+y_\tau^2 -\frac{7}{15}g^2_1 - 3g_2 ^ 2-\frac{16}{3 } g_3 ^ 2\biggr ] , \nonumber \\ \mu\frac{\partial}{\partial \mu}y_\tau & = \frac{1}{16\pi^2}y_\tau \biggl[3y_b^2 + 4y_\tau^2
-\frac{9}{5}g^2_1 - 3g_2 ^ 2\biggr ] .\end{aligned}\ ] ] in the sm , on the other hand , the coefficients for the gauge coupling beta functions are @xcite @xmath93 and @xmath94 the running of the yukawa couplings in this case is given as follows : @xmath95 , \nonumber \\
\mu\frac{\partial}{\partial \mu}y_b&=\frac{1}{16\pi^2}y_b \biggl[\frac{3}{2}y_t^2+\frac{9}{2}y_b^2 + y_\tau^2 -\frac{1}{4}g^2_1-\frac{9}{4}g_2 ^ 2 - 8 g_3 ^ 2\biggr ] , \nonumber \\ \mu\frac{\partial}{\partial \mu}y_\tau&=\frac{1}{16\pi^2}y_\tau \biggl[3y_t^2 + 3y_b^2+\frac{5}{2}y_\tau^2 -\frac{9}{4}g^2_1-\frac{9}{4}g_2 ^ 2\biggr ] .\end{aligned}\ ] ] modifications in the coefficients in particular models are mentioned to in the following sections .
now we deal with the renormalization factors , @xmath81 and @xmath82 .
the factors are expressed as the product of the long- and short - distance factors , _
i.e. _ , @xmath96 where @xmath97 and @xmath98 represent the long- and short - distance factors , respectively .
the long - distance contribution @xmath97 is common to @xmath62 and @xmath63 , and independent of the high - energy physics .
its value is evaluated as @xmath99 at two - loop level .
details of the calculation are given in appendix .
the short - distance factors are , on the other hand , model - dependent quantities . for instance , if there is no threshold between the electroweak and the gut scales they are evaluated at one - loop level in refs . @xcite as follows : @xmath100^{-\frac{\gamma_3}{b_3^{(1)}}}\biggl [ \frac{\alpha_2(m_z)}{\alpha_2(m_{\rm gut } ) } \biggr]^{-\frac{\gamma_2}{b_2^{(1)}}}\biggl [ \frac{\alpha_1(m_z)}{\alpha_1(m_{\rm gut } ) } \biggr]^{-\frac{\gamma_1^{(1)}}{b_1^{(1)}}}~ , \nonumber \\
a_s^{(2)}&=\biggl [ \frac{\alpha_3(m_z)}{\alpha_3(m_{\rm gut } ) } \biggr]^{-\frac{\gamma_3}{b_3^{(1)}}}\biggl [ \frac{\alpha_2(m_z)}{\alpha_2(m_{\rm gut } ) } \biggr]^{-\frac{\gamma_2}{b_2^{(1)}}}\biggl [ \frac{\alpha_1(m_z)}{\alpha_1(m_{\rm gut } ) } \biggr]^{-\frac{\gamma_1^{(2)}}{b_1^{(1)}}}~,\end{aligned}\ ] ] where , in the mssm , @xmath101 while in the sm , @xmath102 the extension of the result to each case is referred to in the subsequent sections . for reference
, we evaluate the proton - decay lifetime assuming the susy scale to be 1 tev , _
i.e. _ , all the superparticles are assumed to have masses of @xmath103 tev .
the result is @xmath104 here we neglect the possible effects of the threshold corrections from particles whose masses are around the gut scale .
we also neglect them in the following calculations , since the effects are completely model - dependent .
in this section , we discuss the grand unified models in which extra vector - like matters are added into the susy su(5 ) guts .
we assume that there exist @xmath105 and @xmath106 pairs of chiral supermultiplets which transform as @xmath107 and @xmath108 representations , respectively , and evaluate the proton decay rate for the cases . for brevity , all of the multiplets are assumed to have the same mass , @xmath109 .
as mentioned to in the introduction , such models are motivated by the symptoms of higgs boson with its relatively heavy mass , @xmath110 gev , reported recently at the lhc @xcite ; if the additional multiplets with mass around weak scale couple to the mssm higgs bosons , the higgs mass @xmath111 can be raised on account of the quantum corrections @xcite .
the gauge mediation requires the vector - like matters as the susy - breaking messenger .
thus , we take a wide parameter range for @xmath109 .
the existence of the extra vector - like multiplets modifies the rges for the gauge coupling constants , as well as the short - distance renormalization factors .
the beta functions of the gauge coupling constants in the mssm receive additional contributions @xcite : @xmath112 here @xmath113 and @xmath114 are the corrections to the leading and the next - to - leading order contributions for the beta functions of the gauge coupling constants , @xmath115 and @xmath116 , which are defined in eq .
( [ mssm_beta ] ) . here
, we ignore the contribution from the yukawa couplings of the vector - like matters at two - loop level for simplicity . from the one - loop contribution in eq .
( [ beta_vec ] ) , it is found that addition of vector - like matters equally changes the running of the gauge couplings , and thus maintains the gauge coupling unification .
the corrections to the one - loop contribution , @xmath113 , also modify @xmath117 and @xmath118 . to be concrete ,
the renormalization effects of the energy scale above @xmath109 are given as @xmath119^{-\frac{4}{3 [ b_3^{(1)}+ \delta b_3^{(1)}]}}\biggl [ \frac{\alpha_2(m)}{\alpha_2(m_{\rm gut } ) } \biggr]^{-\frac{3}{2[b_2^{(1)}+\delta b_2^{(1)}]}}\biggl [ \frac{\alpha_1(m)}{\alpha_1(m_{\rm gut } ) } \biggr]^{-\frac{11}{30[b_1^{(1)}+\delta b_1^{(1)}]}}~ , \nonumber \\ a_s^{(2)}&=\biggl [ \frac{\alpha_3(m)}{\alpha_3(m_{\rm gut } ) } \biggr]^{-\frac{4}{3 [ b_3^{(1)}+ \delta b_3^{(1)}]}}\biggl [ \frac{\alpha_2(m)}{\alpha_2(m_{\rm gut } ) } \biggr]^{-\frac{3}{2[b_2^{(1)}+\delta b_2^{(1)}]}}\biggl [ \frac{\alpha_1(m)}{\alpha_1(m_{\rm gut } ) } \biggr]^{-\frac{23}{30[b_1^{(1)}+\delta b_1^{(1)}]}}~.\end{aligned}\ ] ] these modifications result in alternations of the proton - decay lifetime . in order to parametrize the alternations
, we define the following ratio : @xmath120 where @xmath121 represents the proton - decay lifetime with ( without ) vector - like matters . as seen form eq .
( [ decay_width ] ) , this ratio does not depend on the @xmath7-boson mass but only on @xmath80 , @xmath117 and @xmath118 : @xmath122_{\rm w / o } } { \alpha_5 ^ 2\bigl[\bigl(a^{(1)}_r\bigr)^2 + \bigl(a^{(2)}_r\bigr)^2(1+|v_{ud}|^2)^2 \bigr]_{\rm w/}}~.\ ] ] here , w/ ( w / o ) again implies that the factor is for the case with ( without ) vector - like matters . as functions of vector - like matter mass .
left and right graphs show @xmath123 and @xmath124 in solid lines from top to bottom , respectively .
light ( dark ) shaded region is excluded by the current experimental limit , @xmath125 years at 90% confidence level @xcite , in the case of @xmath126 gev ( @xmath127 gev ) .
susy scale is set to be 1 tev.,height=245 ] in fig .
[ fig : vectorlike ] , we plot the ratio @xmath6 as functions of the masses of the vector - like multiplets . in the analysis ,
the susy scale is set to be @xmath128 tev .
each solid line in the left graph corresponds to the number of @xmath107 multiplets @xmath129 from top to bottom without @xmath130 multiplets . in the right graph , on the other hand , the solid lines represent the cases with @xmath108 multiplets ( @xmath131 , from top to bottom ) and @xmath132 .
the light ( dark ) shaded region is excluded by the current experimental limit , @xmath8 years at 90% confidence level @xcite in the case of @xmath133 gev ( @xmath127 gev ) . on the whole parameter region in this figure , the unified gauge coupling constant , @xmath80 , is less than @xmath134 , thus , perturbativity is maintained . in order to clarify the effects of the renormalization factors on the enhancement of proton decay rate , we also plot the ratio of the factors , @xmath135_{\rm w/}/ [ ( a^{(1)}_r)^2 + ( a^{(2)}_r)^2(1+|v_{ud}|^2)^2 ] _ { \rm w / o}$ ] , as functions of the masses of the vector - like multiplets in fig .
[ fig : renormalization ] . here
we set @xmath123 from bottom to top and the susy scale to be 1 tev .
the renormalization factor in the absence of the vector - like multiplets is calculated as @xmath136_{\rm w / o}=40 $ ] .
these results indicate that the mssm with vector - like matters might cause proton decay fast enough to be detected by the current or future experiments . in particular ,
if there are three @xmath107 vector - like matters or is a @xmath137 multiplet at the tev scale , the susy su(5 ) guts are excluded even by the present limit for @xmath126 gev . from bottom to top .
susy scale is set to be 1 tev.,height=245 ] as functions of the masses of the multiplets .
solid lines represent @xmath138 multiplets of @xmath139 representations from top to bottom .
light ( dark ) shaded region is excluded by the current experimental limit , @xmath125 years at 90% confidence level @xcite in the case of @xmath126 gev ( @xmath127 gev ) .
susy scale is set to be 1 tev.,height=264 ] in addition , we calculate the ratio in the case where @xmath140 pairs of @xmath141 multiplets are introduced .
this set of representations is the same as those of one generation of chiral matter fields and their corresponding complex representations . in this case , the beta functions are equal to those of @xmath142 in eq .
( [ beta_vec ] ) . after carrying out a similar calculation , we plot the ratio @xmath6 against the masses of the multiplets in fig . [
fig : gen ] . here , solid lines represent @xmath138 multiplets of @xmath143 from top to bottom .
we again set the susy scale to be 1 tev .
from this figure , it is found that a pair of extra generations which have masses below @xmath144 gev yield too large proton decay rate to be excluded by the current experimental limit in the case of @xmath126 gev . although the threshold corrections around the gut scale might alter the predicted values of proton lifetime , a growing tendency in the proton decay rate is independent of particular models .
therefore , the proton - decay experiments are extremely promising for constraining such scenario .
in turn , we discuss the susy gut with sfermions having masses much larger than the tev scale .
such a mass spectrum might be realized when the supersymmetry is broken via the anomaly - mediation mechanism @xcite , and there have been a lot of works which examine the scenario @xcite . in this scenario , although a high degree of fine - tuning is inevitable , the susy flavor and cp problems are relaxed owing to heavy masses of sfermions @xcite .
the thermal relic scenario of dark matter in the universe is still achieved @xcite , and the dark matter is to be directly detected in the future experiments @xcite .
in addition , this heavy scale susy scenario is also suggested by the recent lhc results since there has been no signal of superparticles and the atlas and cms collaborations provide stringent limits on the masses of colored particles @xcite .
moreover , in this case the sm - like higgs boson mass easily reaches @xmath145 gev as the radiative corrections are enhanced due to the heavy colored particles . in the following discussion , we assume that the masses of squarks and sleptons are around the scale of @xmath146 , which is much higher then the electroweak scale , while the masses of gauginos , higgsinos , and higgs bosons except the lightest one are @xmath147 tev , which is denoted by @xmath148 hereafter . therefore ,
between @xmath148 and @xmath146 , the latter fields are to be added to the rge analysis .
furthermore , we assume that their contributions to the gauge - coupling running through the yukawa couplings , as well as those with sfermions running in the loops , are negligible .
hence , the beta functions of the gauge couplings in this region are given as @xmath149 with @xmath150 the same as the sm ones .
since each one - loop contribution @xmath115 in the above equation differ from that in eq .
( [ mssm_beta ] ) by the same number ( 2 in this case ) , the perturbative gauge coupling unification is again preserved in the present case . .
solid , dashed , and dotted lines correspond to the cases where @xmath148 is set to be 1 , 3 , and 10 tev , respectively .
@xmath7-boson mass is taken to be @xmath151 gev .
, height=264 ] in fig .
[ fig : heavysusy ] , we plot the lifetime of proton against @xmath146 .
solid , dashed , and dotted lines correspond to the cases where @xmath148 is set to be 1 , 3 , and 10 tev , respectively .
the @xmath7-boson mass is taken to be @xmath152 gev in this figure .
it is found that the proton - decay lifetime is slightly extended , although the enhancement factor is less than two .
the contribution of the renormalization factors to the alternation of the proton lifetime is less significant in this case than that in the case discussed in sec .
[ vectorlike ] ; it is at most a few % on the whole parameter region in this figure .
compared with the case discussed in the previous section , the present situation does not so much change the proton decay rate .
hence , searching for the proton decay is still stimulating even for the heavy susy scenario .
we have studied the proton decay rate under the two situations ; one is the mssm with vector - like multiplets , and the other is the mssm with heavy sfermions .
it is found that the proton lifetime is significantly reduced in the former case , while in the latter case it is slightly prolonged . in any case , the proton - decay experiments are , together with the lhc experiment and other precision measurements , expected to shed light on the supersymmetric grand unified models , as well as the supersymmetry itself .
we thank yasumichi aoki for useful discussion . this work is supported by grant - in - aid for scientific research from the ministry of education , science , sports , and culture ( mext ) , japan , no .
20244037 , no . 20540252 , no .
22244021 and no.23104011 ( jh ) , and also by world premier international research center initiative ( wpi initiative ) , mext , japan .
the work of nn is supported by research fellowships of the japan society for the promotion of science for young scientists .
in this appendix , we demonstrate the evaluation of the long - distance contribution to the renormalization factors , _ e.g. _ , @xmath97 in eq .
( [ ar ] ) .
first , we write down the two - loop renormalization group equations for the strong coupling constant below the electroweak scale : @xmath153 with @xmath154 the strong coupling constant and @xmath155 where @xmath156 denotes the number of quark flavors in an effective theory . the long - distance factor , @xmath97 , is determined by the ratio of the coefficients for the effective operators at the scale of @xmath157 and 2 gev : @xmath158 with the coefficient @xmath159 satisfying the following rge at two - loop level @xcite : @xmath160c(\mu)~.\ ] ] the solution of the equation is @xmath161^{-\frac{2}{b_1 } } \biggl [ \frac{4\pi b_1+b_2\alpha_s(\mu ) } { 4\pi b_1+b_2\alpha_s(\mu_0 ) } \biggr]^{\bigl(\frac{2}{b_1}-\frac{6 + 2n_f}{9 b_2}\bigr)}~,\ ] ] with @xmath162 and @xmath163 given in eq .
( [ b1b2 ] ) .
thus , @xmath97 is given as @xmath164^{\frac{6}{25}}\biggl [ \frac{\alpha_s(m_b)}{\alpha_s(m_z ) } \biggr]^{\frac{6}{23 } } \biggl [ \frac{\alpha_s(2~{\rm gev})+\frac{50\pi}{77 } } { \alpha_s(m_b)+\frac{50\pi}{77 } } \biggr]^{-\frac{173}{825 } } \biggl [ \frac{\alpha_s(m_b)+\frac{23\pi}{29 } } { \alpha_s(m_z)+\frac{23\pi}{29 } } \biggr]^{-\frac{430}{2001}}~,\end{aligned}\ ] ] and numerically it turns out to be @xmath165 99 h. georgi and s. l. glashow , phys .
lett . * 32 * , 438 ( 1974 ) .
s. dimopoulos , s. raby and f. wilczek , phys .
d * 24 * , 1681 ( 1981 ) ; + w. marciano and g. senjanovi ' c , phys .
d * 25 * , 3092 ( 1982 ) ; + m.b .
einhorn and d.r .
jones , nucl .
b * 196 * , 475 ( 1982 ) ; + u. amaldi , w. de boer and h. furstenau , phys .
b * 260 * , 447 ( 1991 ) ; + p. langacker and m. -x .
luo , phys .
d * 44 * , 817 ( 1991 ) .
h. georgi , h. r. quinn and s. weinberg , phys .
lett . * 33 * , 451 ( 1974 ) ; + s. dimopoulos and s. raby , nucl .
b * 192 * , 353 ( 1981 ) ; + e. witten , nucl .
b * 188 * , 513 ( 1981 ) .
s. dimopoulos and h. georgi , nucl .
b * 193 * , 150 ( 1981 ) ; + n. sakai , z. phys .
c * 11 * , 153 ( 1981 ) .
n. sakai and t. yanagida , nucl .
b * 197 * , 533 ( 1982 ) ; + s. weinberg , phys . rev .
d * 26 * , 287 ( 1982 ) .
h. nishino , k. abe , y. hayato , t. iida , m. ikeda , j. kameda , y. koshio and m. miura _ et al .
_ , arxiv:1203.4030 [ hep - ex ] . m. miura , pos ichep * 2010 * , 408 ( 2010 ) .
t. goto and t. nihei , phys .
d * 59 * , 115009 ( 1999 ) ; + h. murayama and a. pierce , phys .
d * 65 * , 055009 ( 2002 ) . j. hisano , h. murayama and t. yanagida , phys .
b * 291 * , 263 ( 1992 ) ; + k. s. babu and s. m. barr , phys .
d * 48 * , 5354 ( 1993 ) ; + j. hisano , t. moroi , k. tobe and t. yanagida , phys .
b * 342 * , 138 ( 1995 ) ; + b. bajc , p. fileviez perez and g. senjanovic , phys .
d * 66 * , 075005 ( 2002 ) ; + b. bajc , p. fileviez perez and g. senjanovic , hep - ph/0210374 .
r. d. peccei and h. r. quinn , phys .
lett . * 38 * , 1440 ( 1977 ) .
l. j. hall and y. nomura , phys .
d * 64 * , 055003 ( 2001 ) .
[ atlas collaboration ] , arxiv:1202.1408 [ hep - ex ] ; + s. chatrchyan _ et al . _
[ cms collaboration ] , arxiv:1202.1488 [ hep - ex ] .
m. ibe and t. t. yanagida , phys .
b * 709 * , 374 ( 2012 ) ; + m. ibe , s. matsumoto and t. t. yanagida , phys . rev .
d * 85 * , 095011 ( 2012 ) .
m. a. ajaib , i. gogoladze , f. nasir and q. shafi , arxiv:1204.2856 [ hep - ph ] . t. moroi and y. okada , phys .
b * 295 * , 73 ( 1992 ) ; + k. s. babu , i. gogoladze , m. u. rehman and q. shafi , phys . rev .
d * 78 * , 055017 ( 2008 ) ; + s. p. martin , phys .
d * 81 * , 035004 ( 2010 ) ; + m. asano , t. moroi , r. sato and t. t. yanagida , phys .
b * 705 * , 337 ( 2011 ) ; + m. endo , k. hamaguchi , s. iwamoto and n. yokozaki , phys .
d * 84 * , 075017 ( 2011 ) ; + j. l. evans , m. ibe and t. t. yanagida , arxiv:1108.3437 [ hep - ph ] .
a. hebecker and j. march - russell , phys .
b * 539 * , 119 ( 2002 ) ; l. j. hall and y. nomura , phys .
d * 66 * , 075004 ( 2002 ) .
j. r. ellis , m. k. gaillard and d. v. nanopoulos , phys .
b * 80 * , 360 ( 1979 ) [ erratum - ibid .
* 82b * , 464 ( 1979 ) ] .
j. hisano , h. murayama and t. yanagida , nucl .
b * 402 * , 46 ( 1993 ) .
y. aoki , c. dawson , j. noaki and a. soni , phys .
d * 75 * , 014507 ( 2007 ) .
y. aoki , talk at gut2012 , 16 march , 2012 .
y. aoki _ et al .
_ [ rbc - ukqcd collaboration ] , phys . rev .
d * 78 * , 054505 ( 2008 ) .
w. siegel , phys .
b * 84 * , 193 ( 1979 ) .
j. e. bjorkman and d. r. t. jones , nucl . phys .
b * 259 * , 533 ( 1985 ) .
m. e. machacek and m. t. vaughn , nucl .
b * 222 * , 83 ( 1983 ) .
l. f. abbott and m. b. wise , phys .
d * 22 * , 2208 ( 1980 ) . c. munoz , phys .
b * 177 * , 55 ( 1986 ) .
d. ghilencea , m. lanzagorta and g. g. ross , nucl .
b * 511 * , 3 ( 1998 ) .
l. randall and r. sundrum , nucl .
b * 557 * , 79 ( 1999 ) ; + g. f. giudice , m. a. luty , h. murayama and r. rattazzi , jhep * 9812 * , 027 ( 1998 )
. n. arkani - hamed and s. dimopoulos , jhep * 0506 * , 073 ( 2005 ) ; + g. f. giudice and a. romanino , nucl .
b * 699 * , 65 ( 2004 ) [ erratum - ibid .
b * 706 * , 65 ( 2005 ) ] ; + n. arkani - hamed , s. dimopoulos , g. f. giudice and a. romanino , nucl .
b * 709 * , 3 ( 2005 ) .
f. gabbiani , e. gabrielli , a. masiero and l. silvestrini , nucl .
b * 477 * , 321 ( 1996 ) .
j. hisano , s. matsumoto , m. nagai , o. saito and m. senami , phys .
b * 646 * , 34 ( 2007 ) .
j. hisano , k. ishiwata and n. nagata , phys .
b * 690 * , 311 ( 2010 ) ; + j. hisano , k. ishiwata , n. nagata and t. takesako , jhep * 1107 * , 005 ( 2011 ) ; + t. moroi and k. nakayama , phys .
b * 710 * , 159 ( 2012 ) .
t. nihei and j. arafune , prog .
* 93 * , 665 ( 1995 ) . | in the supersymmetric grand unified theories ( susy guts ) , gauge bosons associated with the unified gauge group induce proton decay .
we investigate the proton decay rate via the gauge bosons in the susy guts under the two situations ; one is with extra vector - like multiplets , and the other is with heavy sfermions .
it is found that the proton lifetime is significantly reduced in the former case , while in the latter case it is slightly prolonged .
determination of the particle contents and their mass spectrum below the gut scale is important to predict the proton lifetime .
the proton decay searches have started to access to the @xmath0 gev scale .
ipmu12 - 0074 1.35 cm 1.2 cm junji hisano@xmath1 , daiki kobayashi@xmath2 , and natsumi nagata@xmath3 + 0.4 cm _
@xmath2department of physics , nagoya university , nagoya 464 - 8602 , japan _ + _
@xmath4ipmu , todias , university of tokyo , kashiwa 277 - 8568 , japan _ + _ @xmath5department of physics , university of tokyo , tokyo 113 - 0033 , japan _ |
the interaction between matter and radiation is one of the fundamental mechanisms shaping the distribution of the baryonic component in the universe , from stellar to cosmological scales .
this process often couples wildly different scales and the large dynamical range makes accurate modelling difficult .
a notable example is given by the reionization of cosmic hydrogen at redshift @xmath0 ( for a review , see , e.g. , meiksin 2009 ) . during this epoch , the hii regions generated by the first ionizing sources expand to intergalactic scales and overlap , leaving most of the universe highly ionized and re - heated by several thousand kelvin degrees .
the brightest , high - redshift quasars , may have produced even larger hii regions before the end of reionization , with linear sizes extending up to a hundred comoving mpc . nonetheless , most of the evolution in the gas temperature and ionization state still happened on much smaller scales , i.e. within the ionization - fronts , whose sizes are comparable to the local photon mean - free - path just outside the hii regions , about three orders of magnitude smaller than the hii regions themselves for gas at mean cosmic density . even without considering such an extreme case in dynamical range , the numerical solution of the full radiative transfer equations still represents a big computational challenge .
this is mostly due to two factors .
the first is the high - dimensionality : seven variables are required for a full specification of the radiation field ( three spatial variables , two angular directions , photon frequency , and time ) .
the second is the intrinsic `` non - locality '' of the problem : the radiation field at a given point is determined by the gas properties along the lines of sight towards all the sources of radiation ( including the diffuse medium itself ) . for these reasons ,
current numerical models rely on approximations aimed to decrease the problem dimensionality and , at the same time , the computational costs associated with the `` non - locality '' of the full radiation transfer ( e.g. , abel , norman & madau 1999 , gnedin & abel 2001 , maselli , ferrara & ciardi 2003 , razoumov & cardall 2005 , mellema et al .
2006 , rijkhorst et al .
2006 , ritzerveld & icke 2006 , whalen & norman 2006 ; for a review , see also iliev et al .
2006 and references therein ; more recent codes include , e.g. , trac & cen 2007 , semelin et al .
2007 , aubert & teyssier 2008 , altay et al .
2008 , pawlik & schaye 2008 , finlator et al .
2009 , petkova & springel 2009 ) .
a widely used approximation is the so called `` method of characteristics '' which is particularly suited for cases where the radiation field is completely dominated by individual sources ( with no contribution from the radiative recombinations produced by the medium itself ) . in this scheme , the radiation field
is determined simply from the column densities along a line of sight between the sources and a given point in the computational domain , discretized in cells ( or particles ) . in the `` long - characteristics '' flavour , the column densities
are directly calculated with a single ray from the sources to the cell centre , summing up all the contributions cut through the cells in between . in the faster , but less accurate , `` short - characteristics '' version , the column densities are determined in an ordered way starting from the cells closer to the source and moving outwards after summing up ( using some interpolation scheme ) previous contributions . in both cases ,
sampling the radiation field with a single ray from the sources to the cell centre may be problematic in particular situations , i.e. when the cells are optically thick and in presence of strong density gradients .
practically , the consequences may be the loss of photon conservation and numerical artifacts in the i - front shapes ( see , e.g. , mellema et al .
2006 ) . in a cosmological situation , e.g. for the reionization scenario described before
, we would like to have an accurate method for the very typical case in which numerical resolution imposes optically thick cells .
this may be obtained sampling the radiation field in a statistical , homogeneous way , using the widely used monte carlo techniques . in this case , a series of rays is randomly casted around the sources in order to obtain the right solid angle distribution , averaging the contributions to the radiation field within individual ( three - dimensional ) volume elements .
the drawbacks of this method are the computational costs : many rays ( per cell ) are required to avoid statistical noise and the method becomes less and less efficient as one moves outwards from the sources ( but see , abel & wandelt 2002 ) .
moreover , if the grid is composed by cells of different sizes , the required number of rays to be casted is determined by the size of the smallest cells , irrespectively of the volume fraction they actually occupy . in this work
, we develop a novel approach aimed to obtain an accurate solution of the radiative transfer problem in cosmological situations in an efficient way : combining the monte carlo scheme with the `` cell - by - cell '' approach typical of the characteristic methods .
as we mentioned at the beginning of this section , despite the large scales associated with , e.g. , the reionization process , the medium properties are very often determined within a small fraction of its volume , i.e. within the i - fronts .
the basic idea is to concentrate the computational efforts onto this part of the computational volume , where they are actually needed .
this is achieved thanks to a new algorithm which casts , with the correct solid angle distribution , a series of rays for each cell _ individually_. in this way , we are able to use an algorithm that adaptively determines : i ) the number of rays needed for a particular cell to achieve the convergence of the radiation field ( irrespectively of the cell size , or distance from the source ) , ii ) in which part of the volume to apply the ( expensive ) monte carlo ray - tracing .
the result is that the computing time of the rt scales now with the number of cells to be evolved during a simulation - step , i.e. , typically the cells contained within the i - fronts , with a huge gain in computational speed ( and accuracy ) with respect to a classical monte carlo . with the new `` cell - by - cell '' approach we are also able to gain the full advantages of adaptive mesh refinement ( amr ) methods ( see , e.g. berger & oliger 1984 ) that increase the spatial resolution where needed .
in particular , we implemented a scheme , based on the original clustering algorithm by berger & rigoutsos ( 1991 ) , to adaptively refine the mesh in correspondence of the i - fronts during the course of the simulation .
this allows us to fully resolve the small scales associated with the i - fronts within the large , cosmological simulation boxes that follow the reionization process .
resolving the i - fronts has a profound impacts for a large range of astrophysical applications , e.g. , to accurately predict the temperature state of the gas surrounding a bright quasar during reionization .
in the validating test section ( section 4 ) , we will present an example of how resolving the i - front may be important in the determination of the gas temperature within the _ whole _ hii region .
the paper is organized as follows . in section 2 ,
we review the basic radiative transfer equation .
the computational method used by our new rt code , ` radamesh ` ( * * r**adiative - transfer on * * ada**ptive * mesh * ) , is presented in section 3 , in section 4 , we show the validating tests of the code . we conclude in section 5 .
the cosmological radiative transfer equation in comoving coordinates ( e.g. , norman , paschos & abel 1998 ) is given by @xmath1 where @xmath2 is the monochromatic specific intensity of the radiation field , @xmath3 is a unit vector along the direction of propagation of the ray , @xmath4 is the ( time - dependent ) hubble constant , @xmath5 is the ratio of cosmic scale factors between photon emission at frequency @xmath6 and the present time t , @xmath7 denotes the opacity at frequency @xmath6 and @xmath8 is the source function of the medium . if the scale of interest @xmath9 is much smaller than the hubble radius , @xmath10 , ( as always in our case ) and the medium properties are changing on a time scale shorter than the light crossing time @xmath11 , equation ( [ generalrteq ] ) reduces to the classical , static radiative transfer equation : @xmath12 this equation admits the following solution , @xmath13 where @xmath14 is the optical depth along @xmath3 .
this solution is in general adequate for cosmological simulations except on small distances from the sources ( or , equivalently , for very bright sources ) . in this case
, the approximations break down allowing the ionization - fronts to expand faster than the speed of light . in section [ gammasec ]
we discuss an approximate correction to solve this unphysical behaviour .
it is often convenient to express @xmath15 as a sum of the attenuated , direct radiation from individual sources ( @xmath16 ) and the diffuse radiation ( @xmath17 ) generated within the medium : @xmath18 using equation ( [ ieq ] ) , the radiation field intensity can be specified at any point of the simulation box given the value of the optical depth to the individual sources and the source term of the medium . from the radiation field , it is then possible to derive for each species @xmath19 the photoionization rate ( per particle ) : @xmath20 and the gas photoheating rate ( per unit volume ) : @xmath21 where @xmath22 and @xmath23 are the frequency threshold and the cross - section for the ionization of species @xmath19 , respectively , and @xmath24 is the physical number density .
analogously to eq .
( [ isplit ] ) , we can split @xmath25 and @xmath26 into the sum of direct ( @xmath27,@xmath28 ) and diffuse ( @xmath29,@xmath30 ) components .
finally , @xmath31 and @xmath32 are used to compute the chemistry evolution of the neutral fraction of hydrogen ( @xmath33 ) , neutral helium ( @xmath34 ) , singly ionized helium ( @xmath35 ) , and the temperature ( expressed in terms of the total energy density @xmath36 ) according to the following rate equations : @xmath37 here @xmath38 , @xmath39 and @xmath40 are the temperature - dependent radiative recombination ( to all levels ) , collisional ionization , and dielectronic recombination coefficients , respectively ; @xmath41 is the number density of electrons , @xmath42 is the gas clumping factor for simplicity , we assume the same value of @xmath43 for each species in the current version of the code . ,
and @xmath44 is the total cooling rate ( including recombinations , collisional excitations , compton , brehmstrahlung and hubble cooling ) .
we use the analytical fits of hui & gnedin ( 1998 ) for the ionization , recombination and cooling rates .
one of the main challenges in the calculation of the photoionization / heating rate is represented by the presence of the diffuse term in eq .
( [ isplit ] ) , since it requires the transport of the diffuse photons generated from atomic recombinations in every point of the medium .
a widely used solution to this problem is the so called `` on - the - spot '' approximation ( ots or _ case b _ , originally proposed by baker & menzel 1938 ) , which assumes that every diffuse , ionizing photon is absorbed exactly in the same point where it is generated ( i.e. , the photon mean - free - path is much smaller than the resolution scale ) . in this case , assuming for the moment a pure hydrogen medium , the diffuse part of the photoionization rate can be written as : @xmath45 where @xmath46 is the recombination coefficient to the hydrogen ground level , and @xmath47 is the analogous coefficient for the levels @xmath48 . substituting this relation in eq .
( [ fhieq ] ) , we obtain : @xmath49 i.e. , a relation that depends only on the photoionization rate from individual sources ( @xmath50 ) and , thus , greatly simplifies the radiative transfer problem .
analogous relations can be obtained for hei and heii , assuming that every ionizing photon from recombinations is absorbed by the same species from which it originates ( obviously , this approximation breaks down if we consider that non - self - ionizing photons from heii recombinations can actually ionize both hei and hi , see section [ recradsec2 ] for a detailed discussion ) . as also noticed by ritzerveld ( 2005 ) , despite being widely used , the ots approximation is based on an incorrect argument : if the local mean - free - path is very small as assumed , i.e. , the local optical depth is very high for the diffuse photons generated locally , even higher the optical depth would be for the radiation coming from discrete , non - local sources ( unless their spectral energy distribution is much harder ) .
for instance , in the classical strmgren sphere situation with a monochromatic source ( see test 1 below ) , the regime where the ots approximation may hold corresponds to regions where the directional flux from the central source is not able to penetrate , i.e. only at the extreme edge of the strmgren sphere .
the opposite case ( called _ case a _ ) is represented by the situation in which the mean - free - path of the diffuse photons goes to infinity and , effectively , @xmath51 . despite the loss of photon conservation for a bounded region like a strmgren sphere ,
this approximation is actually more correct than the ots in most situations as we will show later .
the computational volume in ` radamesh ` is discretized in a ( series of ) regular , block - structured grid(s ) where the physical properties of the medium are associated with a zone - centered grid element , i.e. with a _ cell_.
the current possible choices for the computational domain in ` radamesh ` are : i ) single , uniform cartesian grid , ii ) _ static _ multi - mesh structure , iii ) _ evolving _ multi - mesh structure . in the
case , the grid is spatially fixed to the initial , single or multi - mesh structure obtained , e.g. , from the output of an adaptive mesh refinement ( amr ) hydro - simulation . in the _ evolving _ case , this initial structure is adaptively refined ( unrefined ) each time the cells satisfy a chosen refinement ( unrefinement ) criterion .
the refining procedure is described in details in section [ amrsec ] .
the radiation field ( from individual sources and from diffuse radiation ) is discretized in a series of rays that are propagated through this single or multi - mesh structure using a ray - tracing algorithm . before discussing the ray - tracing method ,
it is necessary to describe more in detail the multi - mesh implementation in ` radamesh ` .
multi - mesh domains in ` radamesh ` are composed of a nested hierarchy of rectangular grids of different sizes and levels of refinement , following the implementation called `` patch - based amr '' , originally described in berger & oliger ( 1984 ) .
the position and aspect ratios of this patches are optimized in order to increase the spatial ( and temporal ) resolution where needed within the computational box .
other possible multi - mesh implementations have been proposed in the past , like , e.g , the `` tree - based amr '' ( see khokhlov 1998 and references therein ) . in this case , each cell ( or @xmath52 group of cells in three - dimensions ; typically @xmath53 ) is refined into children cells , on a cell - by - cell basis , and a `` grid '' is the analogous of a single cell ( or a @xmath52 group of cells ) .
these methods were originally developed for the numerical solution of hydrodynamical equation and they have advantages and disadvantages with respect to each other for this particular problem . in our case , patch - based amr presents a significant advantage with respect to other implementations : it reduces the total number of grids .
one of the major problem presented by a ray - tracing algorithm is the fact that a significant part of the computational time may be spent in `` crossing '' the simulation box .
this is in a sense unavoidable , because solving the radiative transfer equations is a highly non - local problem .
optimizing the data and memory structure is thus fundamental in order to have an efficient algorithm .
minimizing the number of `` crossing '' grids helps reducing the overhead associated with the ray - tracing in a multi - mesh structure . in order to increase the efficiency in `` crossing '' the computational volume , grids and cells in ` radamesh ` at different levels of refinement are connected with two separate hierarchical trees and linked lists .
grids are strictly nested , i.e. , each grid at level @xmath54 is fully contained by a single grid at level @xmath55 ( the `` parent '' grid ) .
note that , although this may increase the total number of grids , it allows a more efficient tree - search for both grids and cells .
grids that share the same `` parent '' are connected via a linked list .
thus , each grid @xmath56 at level @xmath54 may have a maximum of three associated pointers : i ) the `` parent '' grid at level @xmath55 , ii ) the `` next brother '' grid in the linked list at level @xmath54 , iii ) the `` son '' grid at level @xmath57 ( i.e. , the head of the linked list of grids fully contained within @xmath56 ) .
grids without an associated `` son '' are called `` leaf '' grids .
this hierarchical tree is a light structure that requires small amounts of memory ( if the number of grids is significantly smaller than the total number of cells , as it is always the case for `` patch - based amr '' ) and that can be reconstructed or saved easily given its strict nesting .
if refined , a cell at level @xmath54 contains a group of @xmath58 subcells , where @xmath59 is the refinement factor . in principle
, @xmath59 may vary with levels , although commonly is fixed to a constant value ( typically @xmath60 ) .
analogously to the grid case , a cell without refinement is called a `` leaf '' cell .
combined with the grid - tree , there is another simple hierarchical structure in ` radamesh ` associated directly with the cells : each refined cell at level @xmath54 points to the grid at @xmath57 that contains its @xmath58 subcells .
this simple ( but more memory - consuming ) cell - tree is built efficiently when needed from the grid - tree and , for this reason , is not saved during output or back - up in order to save memory usage .
the use of the cell - tree allows to recover very quickly the location of the relevant subgrid ( and thus the subcell ) at level @xmath57 when crossing the tree towards higher levels of refinement , without searching through the `` next brother '' linked list of the subgrids .
when searching for cells at coarser ( or at the same ) level , the grid - tree is used instead . the combination of the grid - tree and the simple cell - tree represents a good balance between flexibility ( that translate into computational speed ) of the box crossing algorithm and memory consumption of the tree - structure .
ray - tracing is a well - studied problem in computational geometry , with several efficient methods commonly used in computer graphics .
one of the most efficient solution is the simple and fast grid - traversal algorithm of amanatides & woo ( 1987 ) , particular suited in the case of a single , uniform cartesian grid .
basically , this simple algorithm determines the intersection points between the boundaries of the cells and a ray ( defined from its initial position and two direction angles ) as it traverses the computational grid .
we employ the grid and cell hierarchical trees discussed above to extend this fast algorithm from the single , uniform grid to the multi - mesh structure .
this is done simply adding to the original algorithm , after each cell boundary crossing , a ( recursive ) check for the need of changing the current traversing grid ( either at the same or at a different level ) .
let us consider , for example , the case in which a ray is currently traversing a grid at level @xmath61 .
there are two situations in which a change of grid is required : a ) the ray is still within the boundaries of the current grid but is entering a subgrid at level @xmath62 ; b ) the ray is exiting the boundaries of the current grid . we know that we are in the first situation simply checking the current cell tree : if the cell is associated with a subgrid , we immediately recover from the tree the memory location of the new grid at @xmath63 and , from the intersection coordinates , the corresponding cell within this grid .
we recursively continue the search until we reach a `` leaf '' cell . in the second case ,
the search consists of two phases .
first , we cross the grid tree towards the `` parent '' grids at level @xmath64 until we find a grid where the ray is fully contained ( and not at the grid boundaries ) or where the ray is _ _ entering _ _ the grid boundaries .
note that usually this search is very fast given that , for a typical nested structure , we only need to go to the level @xmath65 . when the grid at @xmath64 is found , we check if the cell where the ray is currently located is a `` leaf '' .
if this is not the case , we apply the same procedure as in case a ) . when traversing a `` leaf '' grid we can apply the original traversal algorithm without the need for the recursive check ( and change of grids ) described above . in this respect ,
the use of a `` patch - based amr '' , given its larger grids , result in a reduction of the computational costs of the traversing algorithm . `
radamesh ` is based on a ray - tracing algorithm that propagates the ( discretized ) radiation field through the computational volume .
however , in the same spirit of the amr method , the key element in ` radamesh ` is not represented by the _ ray _ or photon package but , instead , by the computational _ cell_. in other words , the propagation of the radiation field ( and the solution of the radiative transfer equation ) is performed on a `` cell - by - cell '' basis rather than `` ray - by - ray '' .
computationally , this translates into the change of the main loop in the algorithm , from rays to cells : the radiation field from the ( discrete and diffuse ) sources is propagated with a photon - conserving method separately for each cell in the computational volume . as we explain below , at the heart of the algorithm
there is a new method that allows to draw a series of rays from a source , located at any position inside ( or outside ) the box , through a cubic cell with the _ correct _ solid angle distribution .
this extends the `` monte carlo '' methods , where a series of rays is uniformly casted around a source to obtain the correct solid angle distribution , from the computational box to the cell level .
this `` cell - by - cell monte carlo '' is a more natural approach when most of the physical evolution of the medium happens rapidly only in a small portion of the simulated box at a time .
this is a typical case in cosmological simulation , either because of a difference in density ( e.g. , self - shielded dense clumps in a mostly ionized medium ) or in the ionized state ( e.g. , the expansion of an ionization - front in a mostly neutral igm during reionization ) with the cell - based approach , we can split the computational volume , either single or multi - mesh , grouping the cells with different evolving times or properties and focusing the computational efforts only where actually needed .
this is obtained associating to each cell an individual time - step ( @xmath66 ) that is used to determine whether the cell can be considered _ active _ or not during the current simulation step .
we discuss how we derive @xmath66 and how we define the _ active _ cell in section [ timestepsec ] . in detail ,
the computational algorithm consists of an iterative method divided into four main parts for each simulation step ( and for each _ active _ cell ) : i ) finding the ionization and photo - heating rates , ii ) choosing the time - step , iii ) solving the chemistry equation for the evolution of the medium properties , and , if needed , iv ) performing an adaptive refinement of the computational mesh .
the first part of the algorithm is the most time - consuming ( and important ) part of the method and is described in detail in the following section .
the photo - ionization ( @xmath31 ) and photo - heating rates ( @xmath26 ) are computed for each _ active _ , leaf cell ( during the current time - step ) with an iterative monte carlo procedure until convergence for both quantities is reached .
the convergence level is typically set at 1% , but can be increased ( decreased ) depending on the requested level of accuracy .
we compute separately , with two different methods , the contribution to the total photoionization / heating rate from discrete sources ( @xmath67 ) and from diffuse emission ( @xmath68 ) .
most of the effort is dedicated to the calculation of ( @xmath67 ) , since in many ( cosmological ) cases the directional flux from discrete sources represents the major contribution to the total photoionization rate .
the procedure to obtain the values of @xmath67 and @xmath69 is the following : for each iteration step , a packet of rays is propagated from selected points within the cell to the sources .
these points are chosen with a monte carlo procedure based on a rejection algorithm that insures an uniform solid - angle distribution within the cell .
the total flux is then re - scaled according to the fraction of the solid angle covered by the cell .
we present in the appendix the analytical approximation that gives the solid angle of a ( cubic ) cell as seen by any point inside ( or outside ) the box . for each ray ,
we compute the hi , hei , and heii column densities both for the path length within the cell ( @xmath70 ) and for the path length connecting the cell edge to the source ( @xmath71 ) .
the column densities are then used to calculate the frequency - dependent optical depths @xmath72 and @xmath73 . given the optical depths , we derive the probability that a photon with frequency @xmath6 is absorbed by the species @xmath19 within the cell : @xmath74 } { \sum\limits_{j=1}^{3}\{1-\mathrm{exp}[-\delta\tau_j(\nu)]\ } } \{1-\mathrm{exp}[-\delta\tau_i(\nu)]\}\ .\ ] ] the optical depths and the spectral energy distribution of the sources are sampled into @xmath75 ( logarithmically - spaced ) frequency bins .
the probability distributions @xmath76 are used to compute the photoionization rate , including for the moment only the first term in eq .
( [ ieq ] ) , for each ray of the packet and for each source : @xmath77\ , \ ] ] where @xmath78 denotes the number of ionizing photons per unit time emitted by the source in the frequency bin @xmath79 .
finally , the value of @xmath67 is obtained by averaging @xmath80 over the number of rays in the packet , the cell volume and the fraction of the solid angle covered by the cell ( see appendix ) .
several packets of rays are generated until the value of @xmath67 converges to the required level .
the procedure is repeated for each source and the single values of @xmath67 are added .
the photoionization rates @xmath69 are obtained in a similar way , starting from @xmath81\ .\ ] ] in order to avoid redundant calculations of the column densities @xmath71 from the cell edges to the sources for each ray packet , we first evaluate @xmath71 on the cell vertexes `` visible '' from each source .
if the difference between these values is less than a given ( small ) threshold , we do not use the full ray - casting algorithm .
instead , we use the @xmath71 values at the vertexes to interpolate ( linearly ) the needed values on the cell faces . typically , the cells that need the full ray - casting algorithm are a few percent of the total volume and correspond to the ionization - front regions ( where the column - density varies rapidly ) . for the remaining cells , interpolating the column - densities
is a good approximation and allows to substantially speed - up the computation .
note that in this case , the algorithm is similar to a classical long - characteristic ray - tracing . to achieve a better performance
it is also possible to choose a column - density threshold ( or , equivalently , an optical depth threshold at the ionization limit ) above which ` radamesh ` skips the calculation of the @xmath67 and @xmath69 values .
this is computationally convenient , e.g. , in the early stages of the expansion of an ionization - front in a neutral and dense medium , when most of the volume is still optically thick to the source radiation .
analogously , it is possible to choose a minimum value of @xmath70 and @xmath71 below which an optically - thin approximation is used instead of the full ray - tracing algorithm .
this is typically the situation for the ionized region in proximity of the sources , where the optical depth is negligible .
as mentioned in section 2 , eq .
( [ ieq ] ) permits i - fronts to propagate faster than light in near proximity of a bright source .
an exact solution for this problem that does not require to fully solve the time - dependent radiative transfer equation and avoids the loss of photon conservation is only available in the case of an homogeneous medium ( e.g. , shapiro et al .
2006 ) as we have also discussed elsewhere ( cantalupo et al .
2008 ) . in the more general case of an inhomogeneous medium
, we fix this unphysical behaviour by disregarding the ionizing flux of a given source at a distance @xmath82 , where @xmath83 is the source current lifetime .
note that , the loss of photon conservation due to this approximation is typically restricted to the very early phases of the expansion of the i - front produced by very bright sources .
once the value of the directional photoionization / heating rate have been obtained , we calculate ( with a simpler procedure ) the diffuse component rates for the same , _ active _ cells .
also in this case we follow a `` cell - by - cell '' approach : given a particular ( _ active _ ) cell for which we want to compute @xmath84 and @xmath85 , we generate with a monte carlo procedure a set of rays that propagates outwards from the cell centre .
for each cell encountered by the ray , we compute the medium opacity and the source term as a function of the @xmath86 , @xmath87 , @xmath88 , @xmath89 , and temperature . in particular ,
we include : i ) hi , hei and heii free - bound continuum ( from osterbrock 1989 ) , ii ) heii balmer continuum ( from ercolano & storey 2006 ) , iii ) heii two - photon continuum ( from nussbaumer & schmutz 1984 ) , and iv ) the heii ly@xmath90 line .
we neglect the hei balmer continuum and hei emission lines given their ( relatively ) small contribution to the total emissivity .
given the source function , we compute and add the respective @xmath84 and @xmath85 to the direct photoionization / heating rate , according to the equation presented in section 2 . in this method , each ray represents a fixed fraction of the solid angle ( @xmath91 ) of the sky as seen by the cell , determined by the chosen number of rays ( that we leave in the current version as a free parameter ) . note that this procedure is accurate if the three - dimensional diffuse field is homogeneous over a scale corresponding to @xmath92 , where @xmath93 is the distance from the cell
. obviously ,
if the medium is clumpy , this method becomes more and more inaccurate at increasing distances from the cell , unless the number of rays is increased accordingly .
note , however , that in most cosmological situations , the diffuse field represents a small component with respect to the direct radiation from sources .
as mentioned above , we associate an individual time - step @xmath66 to each cell in the computational volume .
we first estimate @xmath66 for the cells with the newly calculated @xmath25 by taking a fixed fraction @xmath94 of the minimum ionization timescale from the rate equations . the minimum value found for @xmath66
is stored ( as the new global time - step @xmath95 ) . at this point
we check which cells can be considered _ active _ during the next simulation step comparing the individual @xmath66 with @xmath95 . if @xmath96 , where @xmath97 is a user - defined parameter , the cell is considered _ active _ and we assign @xmath98 . if a previously _ active _ cell does not meet this criterion we de - activate the cell until a simulation time @xmath99 ( or a number @xmath100 of simulation steps ) has elapsed . if @xmath101 all ( leaf ) cells are always considered active . since the cells within the ionization - front have the shortest @xmath66 , they are always selected as _ active_. in practice , the factor @xmath102 controls how broad is the region of _ active _ cells around the i - front .
cells outside of this region are updated only when necessary , typically after several simulation time - steps , given their much longer ionization / recombination time - scales .
even for very large values of @xmath102 ( e.g. , @xmath103 ) the gain in computational speed is huge , since most of the volume has evolution time - scales several orders of magnitude larger than the i - fronts . in this part of the computational algorithm , we solve the rate equation and we evolve the medium properties for the _ active _ cells with the newly calculated @xmath25 and @xmath26 . for the integration of the rate equations
, we use the radau iia method ( hairer & wanner 1996 ) , an implicit runge - kutta scheme of variable , adaptive order .
this allows this part of the code to be computationally stable and reasonably fast for the required accuracy .
if activated , the adaptive mesh refinement is performed with the recursive clustering algorithm of berger & rigoutsos ( 1991 ) . before applying this algorithm ,
we flag the cells that satisfy a given refinement criterion .
the currently implemented refinement criterion is based on a combination of three variables for each atomic species : i ) the cell neutral fraction , ii ) the cell ionization rate ( @xmath25 ) , and iii ) the ( frequency dependent ) cell optical depth @xmath104 .
the ultimate goal is to flag and refine the fast evolving cells that are inside the i - front ( i.e. , the cells with the highest @xmath25 ) , until their @xmath105 is below the desired value . in practice
, the refinement criterion should produce a multi - mesh grid where the front is well resolved , i.e. where the @xmath104 at the species energy threshold is well below unity .
moreover , we would like also to have a pre - refined region just beyond the current position of the i - front , and a _ sharp _ cut corresponding to the regions where the i - front is just passed leaving the medium ( highly ) ionized .
the size and shape of the refining region can be easily tuned with a criterion that includes the three variables discussed above , see test 6 for an example .
the recursive amr algorithm starts from the base grid at the coarser level and it is composed by three main steps : i ) flagging , ii ) clustering , and iii ) refining / unrefining .
once the cells to be refined at the coarser level ( @xmath106 ) are flagged as discussed above , they are clustered into new rectangular patches .
these are splitted until the clustering efficiency ( i.e. , the ratio between the number of flagged and total number of cell in the new patch ) is greater than a chosen minimum threshold , typically 80% . in order to avoid the creation of grids that are too small ( and thus inefficient for the ray - tracing algorithm ) , it is also possible to chose a minimum size for the newly created patches . in most of the situations ,
we find that a minimum size of two ( parent ) cells gives the best balance between number of newly created cells and efficiency of the algorithm . we then _ propagate _ the physical properties from the coarser grid ( @xmath107 ) to the newly created patches at level @xmath108 ( refinement ) .
currently , only straight injection refinement has been implemented . if a cell at level @xmath107 that was previously refined has not been flagged during the current time - step , the properties of its subcells at level @xmath108 are _ propagated _ back to the parent cell that becomes now a leaf cell ( un - refinement ) .
temperature during un - refinement is weighted according to the subcell electron number densities , while the species neutral fractions ( @xmath109 ) are weighted according to the mass densities .
this algorithm is recursively repeated , patch by patch , at finer levels , until there are newly flagged cells or the maximum number of refinement levels is reached . during clustering and refinement ,
the grid - tree and the cell - tree are updated , new memory is dynamically allocated for the newly created grids and , at the same time , the memory slots associated with the un - refined grids are released . by the algorithm construction , parent and son grids
have memory slots allocated in close parts of the memory map .
this ensures an efficient memory usage and reduces computational overhead during ray - tracing . in case where the amr is performed on an existing multi - mesh grid ( e.g. , the output of an hydrodynamical amr code )
, there is the very important option in ` radamesh ` to consider this initial grid as _ fixed _ : i.e. , the amr builds the new patches on the top of the existing multi - mesh grid without unrefining it below the original , initial level .
this allows us to increase the resolution where needed for the radiative transfer ( on the i - front ) without losing the resolution achieved with the hydro code .
the capacity of ` radamesh ` of adding further levels of refinement on an existing amr grid , without loosing previous information , is of great importance for several physical applications .
for example for the study of the highly ionized ( but denser than the average ) regions in proximity of a qso that where already touched by its i - front and that would appear as ly@xmath90 forest in the quasar spectrum .
myr in the plane @xmath110 ( top panel ) and @xmath111 ( bottom panel).,title="fig : " ] myr in the plane @xmath110 ( top panel ) and @xmath111 ( bottom panel).,title="fig : " ] myr ( top panel ) and @xmath112 myr ( bottom panel).,title="fig : " ] myr ( top panel ) and @xmath112 myr ( bottom panel).,title="fig : " ] with respect to the strmgen radius @xmath113 ( bottom panel ) and the analytical solution ( top panel ) .
see text for details . ]
in this section we present the validating tests of the code based on the radiative transfer code comparison project ( iliev et al . 2006 ; i06 thereafter ) .
these tests have been designed in order to compare all the important aspects of several radiative - transfer codes present in the literature .
they include the correct tracking of both slow and fast ionization - fronts in homogeneous and inhomogeneous density fields , i - front trapping , spectrum hardening and the solution of the temperature state .
the original tests in i06 ( test 1 to 4 ) are performed for a single , uniform grid , pure hydrogen medium and without recombination radiation ( using the ots approximation ) . in order to compare our results with the other rt codes
, we use the same single grid ( pure - hydrogen ) configuration as used in i06 to reproduce the results of test 1 to 4 . in the second part of this section
, we present a set of case studies aimed at substantiating the new characteristics of our code .
in particular , we verify in test 5 and test 6 the multi - mesh , adaptive capability of ` radamesh ` . in test 7 and test 8
we show the effects of the diffuse radiation transfer for hydrogen only medium ( test 7 ) and including helium ( test 8) . and times @xmath114 myr ( top panel ) and @xmath115 myr ( bottom panel).,title="fig : " ] and times @xmath114 myr ( top panel ) and @xmath115 myr ( bottom panel).,title="fig : " ] myr ( top panel ) and @xmath115 myr ( bottom panel ) .
the results from ` radamesh ` are presented as solid black lines .
the other lines represent the results from a sample of four different codes taken from the rt code comparison project ( i06 ) , in particular : ` c^2-ray ` ( red , short - dashed line ) , ` crash ` ( cyan , dotted line ) , ` ftte ` ( blue , long - dashed line ) , and ` rsph ` ( green , dot - dashed line ) .
, title="fig : " ] myr ( top panel ) and @xmath115 myr ( bottom panel ) .
the results from ` radamesh ` are presented as solid black lines .
the other lines represent the results from a sample of four different codes taken from the rt code comparison project ( i06 ) , in particular : ` c^2-ray ` ( red , short - dashed line ) , ` crash ` ( cyan , dotted line ) , ` ftte ` ( blue , long - dashed line ) , and ` rsph ` ( green , dot - dashed line ) .
, title="fig : " ] and times @xmath114 myr ( top panel ) and @xmath115 myr ( bottom panel).,title="fig : " ] and times @xmath114 myr ( top panel ) and @xmath115 myr ( bottom panel).,title="fig : " ] myr ( top panel ) and @xmath115 myr ( bottom panel ) . the results from `
radamesh ` are presented as solid black lines .
the other lines represent the results from a sample of four different codes taken from the rt code comparison project ( i06 ) , see caption in figure [ t2f16 ] .
, title="fig : " ] myr ( top panel ) and @xmath115 myr ( bottom panel ) .
the results from ` radamesh ` are presented as solid black lines .
the other lines represent the results from a sample of four different codes taken from the rt code comparison project ( i06 ) , see caption in figure [ t2f16 ] .
, title="fig : " ] this test represents the classical problem of the expansion of a hii region in an uniform ( pure - hydrogen ) medium around a single ionizing source .
we assume that a steady , monochromatic ( @xmath116 ev ) source emitting @xmath117 ionizing photons per unit time turns on in an initially - neutral , uniform - density , static medium with hydrogen number density @xmath118 . likewise in i06
, we use for this test the ots approximation .
the temperature is fixed at @xmath119 k. under these conditions , and assuming that the front is sharp ( i.e. that it is infinitely - thin , with the gas inside fully - ionized and the gas outside fully - neutral ) , there is a well - known analytical solution for the evolution of the i - front radius , @xmath120 , and velocity , @xmath121 , given by @xmath122^{1/3}\,,\\ \rm v_i&=&\frac{r_{\rm s}}{3t_{\rm rec}}\frac{\exp{(-t / t_{\rm rec } ) } } { \left[1-\exp(-t / t_{\rm rec})\right]^{2/3}}\ , , \label{strom0}\end{aligned}\ ] ] where @xmath123^{1/3}\,,\ ] ] is the strmgren radius , i.e. the radius at which recombinations balance the ionizations and the hii region expansion stops .
here @xmath124 is the case b recombination coefficient and @xmath125^{-1}\,,\ ] ] is the recombination time .
the hii region initially expands quickly and then slows down as the evolution time approaches the recombination time , @xmath126 . at a few recombination times , the i - front stops and in absence of gas motions remains static thereafter .
the numerical parameters for this test are the followings : computational box dimension @xmath127 kpc ( the source is at one corner of the box ) , gas number density @xmath128 @xmath129 , initial ionization fraction ( given by collisional equilibrium ) @xmath130 , and ionization rate @xmath131 photonss@xmath132 .
for these parameters the recombination time is @xmath133 myr . assuming a recombination rate @xmath134 at @xmath119 k , then @xmath135 kpc .
note that the value of @xmath113 is actually independent from the use of the ots or case b approximation ( see test 7 ) .
myr ( top panel ) and @xmath136 myr ( bottom panel).,title="fig : " ] myr ( top panel ) and @xmath136 myr ( bottom panel).,title="fig : " ] myr ( top panel ) and @xmath136 myr ( bottom panel).,title="fig : " ] myr ( top panel ) and @xmath136 myr ( bottom panel).,title="fig : " ] in figure [ t1 ] , we show the images of the hi fraction in the plane y=0 and z=0 at time @xmath112 myr , when the equilibrium strmgen sphere is reached .
the hii region is nicely spherical in both the planes , demonstrating that our algorithm produces an uniform coverage of the solid angle around the source . in figure
[ t1f8 ] , we plot the spherically averaged radial profiles of the ionized ( @xmath137 ) and neutral fraction ( 1-@xmath137 ) at times @xmath138 and 500 myr .
the thickness of the transition between the hii and hi regions is in agreement with both analytical and numerical expectations from most of the other codes in i06 . in particular , thanks to our new adaptive algorithm ( that ensures the convergence of the radiation field in any cell ) , the i - front thickness does not suffer from diffusive effects shown by other `` classical '' monte carlo methods ( cfr . ,
e.g. , the results of ` crash ` in i06 ) .
finally , in figure [ t1f7 ] , we show the time evolution of the i - front position ( defined as the point of 50% ionization ) .
the code tracks the i - front correctly , with the position never varying by more than few percent from the analytical solution .
myr ( top panel ) and @xmath136 myr ( bottom panel ) .
the results from ` radamesh ` are presented as solid black lines .
the other lines represent the results from a sample of four different codes taken from the rt code comparison project ( i06 ) , see caption in figure [ t2f16 ] .
, title="fig : " ] myr ( top panel ) and @xmath136 myr ( bottom panel ) .
the results from ` radamesh ` are presented as solid black lines .
the other lines represent the results from a sample of four different codes taken from the rt code comparison project ( i06 ) , see caption in figure [ t2f16 ] .
, title="fig : " ] myr ( top panel ) and @xmath136 myr ( bottom panel ) .
the results from ` radamesh ` are presented as solid black lines .
the other lines represent the results from a sample of four different codes taken from the rt code comparison project ( i06 ) , see caption in figure [ t2f16 ] .
, title="fig : " ] myr ( top panel ) and @xmath136 myr ( bottom panel ) .
the results from ` radamesh ` are presented as solid black lines .
the other lines represent the results from a sample of four different codes taken from the rt code comparison project ( i06 ) , see caption in figure [ t2f16 ] .
, title="fig : " ] in this test we use the same parameters of test 1 , but now the ionizing source has a @xmath139 black - body spectrum and we allow the gas temperature to evolve .
initially , the gas is fully neutral with a temperature t=100 k. there are no analytical solutions for this test , therefore we compare our results to what obtained by the other codes in i06 . in figure [ t2hi ] , we show the images of the neutral hydrogen fraction ( on the @xmath111 plane ) at times @xmath114 and @xmath115 myr .
the spherically averaged hi profiles , for the same time - snapshots , are presented in figure [ t2f16 ] ( black , solid lines ) , together with a sample of the results obtained by the other codes in i06 ( see caption in the figure ) .
the overall size of the hii region and the internal structure agree very well .
the temperature images and the spherically averaged profiles at times @xmath114 and @xmath115 myr are presented in figures [ t2temp ] and [ t2f17 ] .
also in this case , the resulting temperature structure agree well with most of the other codes in i06 . in particular , we are able to obtain a large pre - heated region without any sign of spatial anisotropy , since our algorithm does not suffer from the under - sampling the radiation field ( like in a `` classical '' monte carlo ) , even a large distances from the sources .
myr obtained by ` radamesh ` ( top - left panel ) and by three of the four codes that performed the same tests in i06 , in particular : ` crash ` ( bottom - left panel ) , ` c^2-ray ` ( top - right panel ) , and ` ftte ` ( bottom - right panel ) . ]
myr obtained by ` radamesh ` ( top - left panel ) and by three of the four codes that performed the same tests in i06 ( see caption in figure [ t4hi_005 ] ) . ] myr obtained by ` radamesh ` ( top - left panel ) and by three of the four codes that performed the same tests in i06 ( see caption in figure [ t4hi_005 ] ) . ] myr obtained by ` radamesh ` ( top - left panel ) and by three of the four codes that performed the same tests in i06 ( see caption in figure [ t4hi_005 ] ) . ]
this test verifies that the propagation of an i - front within a dense clump is slowed down at the point of being stopped ( _ trapped _ ) and the production of a correct shadowing effect ( in both the ionization and the temperature state ) behind the clump .
we use the same set - up as described by i06 : a spherical ( hydrogen - only ) uniform cloud of radius @xmath140 kpc , is located at the position @xmath141 kpc within a box of length @xmath127 kpc . the hydrogen number density and the initial temperature outside of the clump are , respectively , @xmath142 @xmath129 and @xmath143k , while inside the clump we have @xmath144 @xmath129 and @xmath145 k. the radiation has a black - body spectrum with @xmath146 k and a constant ionizing photon flux @xmath147 s@xmath132 @xmath148 , incident to the @xmath110 box side .
given these parameters , we expect that the i - front should be trapped slightly beyond the clump centre , as discussed in i06 . in figure [ t3hi_ima ]
, we show slices of the gas neutral fraction at the box mid - plane @xmath149 kpc ( passing through the centre of the clump ) at times @xmath150 myr ( top panel ) and @xmath136 myr ( bottom panel ) .
the corresponding temperature slices are presented in figure [ t3temp_ima ] . at @xmath150 myr ,
the i - front is not yet trapped and it is still moving supersonically from the edge of the clump .
the images show that the shadow is sharp and produced correctly behind the clump .
as expected , the i - front is trapped slightly beyond the clump centre at @xmath136 myr .
the position of the i - front and the hi profiles inside the clump , presented in figure [ t3hi_prof ] , agree well with the other codes in i06 , altough , especially at later times , the results are rather code dependent . the same is true for the temperature profiles , as shown in figure ( [ t3temp_prof ] ) . in this test
, we follow the propagation of the ionization - fronts from multiple sources in a static cosmological density field .
the initial conditions for the density and source spatial distribution / luminosities are provided by i06 . in particular
, we use a time - slice at @xmath151 from a hydro - simulation with a box size of @xmath152 comoving mpc and @xmath153 cells .
the initial temperature is fixed to t=100 k everywhere .
the ionizing sources correspond to the 16 most massive halos in the box , with a luminosity proportional to the halo mass and a black - body spectrum with @xmath154 k ( see i06 for more details ) . in figures [ t4hi_005 ] and [ t4hi_02 ] , we present slices of neutral hydrogen fraction cut through the simulation box at coordinate @xmath155 , in box units , at time @xmath156 myr and @xmath157 myr , respectively . comparing our results with three of the four codes that performed this test in i06 ( ` c^2-ray ` , `
crash ` and ` ftte ` ) , we find a general agreement , although all the codes produce somewhat different morphologies .
this general agreement is confirmed also examining the temperature slices in figures [ t4temp_005 ] and [ t4temp_02 ] . here
the differences between the codes presented in i06 is higher and due to different assumption about spectral hardening and partially to the different algorithm used .
the spectral hardening effect , namely the increased temperature in the regions that are still mostly neutral , is traced in detail by ` radamesh ` .
myr and coordinate @xmath155 ( box units ) .
the multi - mesh structure in the plane @xmath155 is overlaid . ] with this test , not present in the original set of i06 , we show and verify the multi - grid capability of ` radamesh ` in the case of a static multi - mesh .
in particular we use the same initial condition of test 1 changing the original grid with three nested , concentric grids with @xmath158 cells , corresponding to two levels of refinement ( with a factor two and four increased resolution with respect to the base mesh ) . with this configuration ,
the centre of the box has a resolution equivalent to a @xmath159 cells grid .
we limit the number of levels and the resolution of the base grid in this test for illustrative purposes .
the source position is @xmath160 in box units . in figure ( [ t5 mg ] ) , we show the image of hi fraction and the computational meshes corresponding to the slice at coordinate @xmath155 and time @xmath161 myr . as we can see from the image ,
the front is tracked very well despite of the different grids ( and resolution ) , with no spurious effect introduced by the multi - grid structure . in the original set of tests in i06 ,
the initial conditions ( e.g. , box size , hydrogen density , source luminosity ) have been chosen in such a way to properly resolve the i - front with a single , uniform grid of @xmath153 cells .
for example , in the final test of i06 , test 4 ( multiple sources in a `` cosmological '' density field ) , the box size is fixed to @xmath152 comoving mpc , corresponding to @xmath162 physical kpc at the simulation redshift ( @xmath151 ) for @xmath163 . however , in a more realistic cosmological situation , e.g. for the study of the reionization process or the expansion of the i - front around a bright , high - redshift qso , the box size must be at least 2 orders of magnitude larger than test 4 ( see , e.g. , meiksin 2009 ) .
this test , not present in the original set of i06 , demonstrate the ability of ` radamesh ` to resolve the i - front of a bright quasar in a large cosmological box , with the help of an adaptively evolving multi - mesh .
in particular we show that properly resolving the i - front is essential to accurately predict the temperature state of the gas , both in the i - front and in the qso hii region . for this purpose
we run a series of simulation increasing the maximum level of refinement until convergence is reached for the temperature state of the gas . in the same spirit of test 2
, we use an uniform ( hydrogen - only ) medium with evolving temperature and we locate the source at one corner of the box . the box size is @xmath164 comoving mpc , the medium density is equal to the mean hydrogen density of the universe and it evolves with redshift .
the initial redshift is @xmath165 , the total evolution time is @xmath166 yr ( corresponding to a final redshift @xmath167 ) .
the source has a power - law spectrum with a ( frequency ) spectral slope of @xmath168 , sampled in 60 logarithmically spaced bins from 1 to 60 rydberg .
the total ionization rate is @xmath169 ph s@xmath132 .
these parameters are comparable to the respective quantities associated with the observed high - redshift quasars in sdss ( e.g. , fan et al .
2006 ) . for the assumed spectrum ( which has a mean ionizing photon energy of @xmath170 ryd ) the photon mean - free path is @xmath171 physical kpc at the initial redshift for a fully neutral patch of gas composed of hydrogen only .
to resolve the i - front properly we would ideally need that @xmath172 is sampled by at least two cells , i.e. a spatial resolution of @xmath173 physical kpc . with an uniform grid
this would correspond to a @xmath174 cells mesh , effectively out of the reach for current computational facilities .
instead , we use a @xmath153 base grid and 5 additional levels of adaptive grid refinement that follow the i - front expansion , effectively achieving the required spatial resolution . in figure [ t6image ]
, we show an image slice of the hydrogen neutral fraction at @xmath175 yr ( in the quasar rest - frame ) .
the i - front scale is so small compared to the box size that it visually appears perfectly thin in the large image .
zooming in by a large factor in a small region containing the i - front ( see inset in the same figure ) , and now the i - front thickness appears . in box units ,
the i - front size ( as measured from the points where the neutral fractions are 0.1 and 0.9 ) is roughly @xmath176 , i.e. @xmath177 physical kpc , corresponding to @xmath178 .
overlaid , we show the adaptively refined multi - mesh structure that closely follows the i - front , with the highest level of refinement ( @xmath179 ) that encompasses the i - front itself . at this level of refinement ,
the i - front is fully resolved by at least 10 simulation cells . in figure
[ t6plot ] , we show how resolving the i - front changes the predicted temperature profile inside the qso hii bubble and in the i - front region . from bottom to top ,
the lines represent the temperature profiles obtained varying the maximum level of refinement from @xmath180 ( i.e. , single uniform grid ) to @xmath181 .
as expected , temperature convergence is reached for @xmath182 , i.e. at spatial resolution comparable or smaller than @xmath172 , indicating that we are effectively resolving the i - front .
in fact , fully resolving the i - front improves the sampling of the spectral hardening ( within the i - front itself ) and a harder spectrum is able to produce a large _ increase _ in the gas temperature , as observed in figure [ t6plot ] note that the current temperature of a cell within the hii region at @xmath115 myr is mainly determined at an earlier epoch , i.e. , when the cell was located within the i - front .
therefore , also the temperature increase _ within _ the hii region can be ascribed to the ( i - front ) spectral hardening effect .
numerical effects on the gas temperature due to the different resolution are negligible , as we show in appendix b. .
the prediction of the correct temperature state in the i - front and in the qso bubble is fundamental , e.g. , for the study of the reionization epoch with the near - zone ly@xmath90 forest ( see e.g. , bolton et al .
2010 ) or with the i - front ly@xmath90 emission ( cantalupo et al .
2008 ) .
comoving mpc ) and i - front adaptive mesh refinement .
the qso was turned on at @xmath165 and keep it at a constant ionizing rate of @xmath169 ph s@xmath132 for @xmath175 yr ( the final redshift is thus @xmath167 ) . from bottom to top , the lines correspond to increasing maximum level of refinement ( from @xmath180 to @xmath181 ) . the base grid is composed by @xmath153 cells .
this plot clearly shows that the high - resolution achieved through amr is essential to correctly resolve the i - front and , thus , to properly recover the gas temperature profile around the qso in a cosmological context . ]
we repeat test 1 dropping the ots approximation used in i06 and including now the full radiative transfer of diffuse photons produced by hydrogen recombinations .
since all the cell in the strmgren sphere may be now source of ionizing photons , we must change the original test configuration , placing the ( discrete ) source at the centre of the box and increasing the box size by a factor of two . apart from these modifications
, we use the same parameter set as in test 1 . for a monochromatic spectrum , homogeneous medium and for a fixed temperature , we know that the radius of the strmgren sphere ( @xmath183 ) obtained with the ots ( case b ) approximation must be equal to the one obtained with the full radiative transfer of the diffuse radiation .
this is a fundamental constraint deriving from photon conservation .
moreover , once the ionization equilibrium has been reached , there is an analytical approximation for the ratio between the diffuse and directional radiation field needed to compensate , at a given radius @xmath93 , the recombinations , as found by ritzerveld ( 2005 ) : @xmath184^{1-\alpha/\alpha_{{\mathrm}{b}}-1 } \ , \ ] ] where the first equality is only valid for a monochromatic spectrum .
note that eq .
( [ ritz ] ) is exact for an `` outward - only '' system , i.e. when the contribution to @xmath185 is given by the recombinations at @xmath186 ( indeed , @xmath187 in ritzerveld s solution ) .
therefore , we expect that this approximation should slightly underestimate @xmath188 for @xmath189 ( note , however that within this region the total radiation field will be dominated by @xmath190 ) . in figure [ gammadiff ]
, we present the result obtained by ` radamesh ` including the full radiative transfer of diffuse photons ( black solid line ) . in very good agreement with the analytical expectations
( red dashed line ) , we found that the diffuse radiation field strength equals the directional field from the central source at @xmath191 , confirming that diffuse radiation is dominant in the outer parts of the strmgren sphere .
the full rt results produce , as expected , a slightly higher value of diffuse radiation in the central parts of the hii regions with respect to the `` outward - only '' solution .
because of photon conservation , this is correctly balanced by a lower diffuse field at the edge of the strmgren sphere when compared to the analytical approximation . in figure [ hifracdiff ] ,
we show how the hydrogen neutral fraction profile ( @xmath192 ) at `` equilibrium '' ( @xmath112 myr ) is modified by the full radiative transfer of diffuse photons ( red solid line ) with respect to the ots ( case b ) approximation ( blue , long - dashed line ) and the other extreme represented by the case a approximation ( black , short - dashed line ) .
as expected , in the interior of the hii regions , where the gas is highly ionized , the mean - free - path of the diffusion photons is very large and thus the profile obtained by the full rt closely follows the one obtained assuming case a. however , as the local opacity increases moving outward , diffuse photons start to be absorbed and the hi fraction tends to become closer to the ots case . in the outer parts of the hii region ( @xmath193 ) , diffusion radiation becomes the dominant component and the gas is more ionized than in the ots case b approximation .
however , in accordance with photon conservation , the final size of the strmgren sphere is very close to the one obtained with case b : the recombination radiation escaped from the inner region has been absorbed ( creating an `` excess '' ) closer to the strmgren radius , but the final photon balance is the same . as a remark , it is interesting to note that the ots approximation is actually never valid inside the strmgren sphere for this very simple , standard test ( apart , obviously , at the strmgren radius ) . actually , case a is a much better approximation if one is interested in the inner part of the hii region , or , analogously , in the early phases of the ionization - front expansion .
( see text for details ) . in this test , the medium temperature has been fixed to @xmath119k ] myr for case a ( short dashed black line ) , `` on - the - spot '' case b ( long dashed blue line ) approximations and including the full radiative transfer of diffuse emission from hii recombinations ( solid red line ) . in this test
, the medium temperature has been fixed to @xmath119k . ]
the previous test has been limited to the very simple case of a monochromatic spectrum ( for both diffuse and direct radiation ) and a hydrogen only , fixed temperature medium . in this way , we were able to compare our results to the analytical prediction from photon conservation arguments .
however , the real effect induced by the full rt of recombination radiation is much more complex since its characteristic spectral energy distribution ( sed ) is in general different from the sed of the discrete sources .
moreover , the presence of helium may have a profound impact also on the neutral hydrogen distribution , given that both the bound - free and the balmer continuum of hei and heii ( together with two - photon and line emissions ) are able to ionize hi .
in this test , we verify these effects using the same configuration of test 2 , including now the full rt of recombination radiation from hydrogen and helium . the only parameters , apart the inclusion of the diffuse component rt , that have been modified with respect to test 2 are the following : i ) the discrete source has been placed in the centre of the box and the box size has been increased by a factor 2 , ii ) the medium include hydrogen and helium , iii ) we evolve the simulation to @xmath112 myr .
note that we do not expect in this case a full equilibrium to be reached , given the complex interplay between hydrogen and helium recombination emission and their ionization state . in figure [ hifracdiff_t2b ] , we present the hi profile obtained by the full rt of diffuse radiation ( red solid line ) in comparison with the case a approximation ( black , short - dashed line ) and the ots case b ( blue , long - dashed line ) at @xmath112 myr . in figure
[ hefracdiff_t2b ] , we show the analogous profiles for the hei , heii and heiii fractions ( see labels in the figure ) . as evident from figure [ hifracdiff_t2b ] , there is now an excess of hi photoionizations with respect to case b , that leads to a larger strmgren sphere size .
this excess is due to diffuse photons produced by helium recombinations , in particular the balmer continuum ( plus ly@xmath90 and two - photon continuum ) produced in the heiii region , that alters the total photon budget available for the photoionization of hi . in effect , a fraction of the high - energy , heii - ionizing photons from the central source are `` converted '' via heiii recombinations to lower energy , hi - ionizing photons ( and also hei - ionizing photons ; indeed , a similar but less conspicuous effect is also visible for the hei fractions in figure [ hefracdiff_t2b ] ) .
therefore , assuming case b for hydrogen and helium actually results in the loss of this important component of the photon budget .
this enlightens the importance of including heii and heiii recombination emission also for studies that are only interested in the properties of the hydrogen component .
myr for case a ( short dashed black line ) , case b ( long dashed blue line ) approximations and including the full radiative transfer of ionizing diffuse emission from hii , heii and heiii recombinations ( solid red line ) . ] myr for case a ( short dashed black lines ) , `` on - the - spot '' case b ( long dashed blue lines ) approximations and including the full radiative transfer of ionizing diffuse emission from hii , heii and heiii recombinations ( solid red lines ) . ]
we have presented a new , three - dimensional radiative transfer code , called ` radamesh ` , based on an adaptive monte carlo ray - tracing scheme .
the algorithm has been specifically developed to efficiently resolve the small scales associated with the ionization - fronts within large , cosmological simulations , with the help of an adaptive mesh refinement ( amr ) scheme combined with a new `` cell - by - cell monte carlo '' approach : rays are casted separately from selected cells within the computational box to the sources with a method that ensures their correct solid angle distribution .
this method has several advantages with respect to a classical monte carlo ray - tracing scheme where a series of rays is uniformly casted around a source to obtain the correct solid angle distribution . , in particular for multi - mesh domains .
indeed , in a classical monte carlo , the number of rays to cast is determined by the smallest , most distant cells to the source , i.e. from the ( few ) cells at the highest refinement level , oversampling the rest of the box . in ` radamesh ` , the number of rays is proportional to the number of cells to be evolved during the current time - step , i.e. the fast evolving cells typically located within the i - front .
this translates into a huge gain in computational speed and efficiency , given the small fraction of the computational volume occupied by the i - front .
moreover , we are now able to choose a local criterion that adaptively determines the number of rays needed in a particular cell , e.g. set by the convergence of the radiation field .
this ensures to obtain the required accuracy in the prediction of the temperature or ionization state of the gas in a very efficient way : concentrating the computational efforts only where actually needed . in the same spirit of hydrodynamical amr codes , `
radamesh ` is also able to increase the spatial resolution where required , adaptively refining the mesh in correspondence of the ionization - fronts .
the implemented algorithm is based on the original patch - based amr method of berger & rigoutsos ( 1991 ) . `
radamesh ` is able to trace the ionization - fronts from multiple sources as well as the diffuse ionizing radiation produced by hydrogen and helium recombinations with a multi - frequency approach .
the time - evolution of six different species ( hi , hii , hei , heii , heiii , e ) and the gas temperature is followed with a time - dependent , non - equilibrium chemistry solver based on an implicit runge - kutta scheme of variable , adaptive order .
we performed all the four tests present in the radiative transfer code comparison project of iliev et al .
2006 , plus four additional , new tests aimed to substantiate and show the new characteristic of ` radamesh ` .
the first four tests include the correct tracking of the i - front in homogeneous ( test 1 ) and in - homogeneous density fields with multiple sources ( test 4 ) , i - front trapping behind a dense clump ( test 3 ) , and the effect of photo - heating on the gas temperature state ( test 2 ) .
these tests , like in the original i06 work , have been performed on a single , static mesh , without emission from atomic recombinations in order to better compare our results with the other codes . `
radamesh ` results are in very good agreement with the majority of the other codes present in i06 . in particular , thanks to the new adaptive algorithm , we have shown ( test 1 ) that the recovered i - front thickness and structure does not suffer from any diffusive effects and numerical broadening typical of other monte carlo codes ( given their poor sampling at large distances from the source ) .
the same is true when the gas temperature is examined ( test 2 ) .
the second set of tests ( test 5 to test 8) shows the ability of ` radamesh ` to deal with both static ( test 5 ) and adaptively evolving ( test 6 ) multi - mesh structures , and with diffuse radiation produced by hydrogen ( test 7 ) and helium ( test 8) recombinations . in test 5 , we reproduce the same results of the classical strmgren sphere situation ( test 1 ) with a nested hierarchy of meshes at different refinement levels , showing that no spurious effects are introduced by the multi - mesh structure .
most importantly , we have shown in test 6 that ` radamesh ` is able to fully resolve an expanding i - front from a bright source ( e.g. , a quasar ) within a large , cosmological box ( 100 comoving mpc size ) , thanks to an adaptively evolving mesh with several levels of refinement .
the ability of resolving the i - front on cosmological scales is fundamental for a large range of applications .
for instance , recovering the correct gas temperature within the i - front and the qso bubble is important , e.g. , for the study of the reionization epoch with the near - zone ly@xmath90 forest ( see e.g. , bolton et al .
2010 ) or with the ly@xmath90 emission generated within the i - front ( cantalupo et al .
2008 ) . in test 6
, we have demonstrated that a proper treatment of the spectral hardening inside a resolved i - front may result in a substantial increase of the gas temperature ( @xmath194 k ) within the _ whole _ hii region surrounding the bright , central source . the effect of diffuse radiation generated by hydrogen recombinations on the classic strmgren sphere case ( with monochromatic radiation ) has been presented in test 7 . here
, we have verified that our treatment of diffuse radiation produces the same strmgren sphere size with respect to the case b approximation , in agreement with photon conservation . moreover , in accordance with the analytical prediction by ritzerveld ( 2005 ) , we have verified that the diffuse field becomes the dominant component at a distance from the source corresponding to 87% of the strmgren radius .
the neutral fraction profile in the inner part of the hii region follows the same profile obtained with case a approximation , while in the outer part the gas is more ionized and the i - front narrower than the result obtained with the widely used , case b approximation . in test 8
, we have also considered the effect of hei and heii recombination radiation on the hydrogen and helium ionization state .
we have found that heii diffuse radiation , especially the hi - ionizing , heii balmer continuum ( together with heii two - photon continuum and ly@xmath90 emission ) may have an important effect also on the hydrogen ionization state , substantially increasing the strmgren sphere size . at present , ` radamesh ` is not yet coupled with hydro - dynamics but is able to post - process the output of three different hydrodynamical codes , both grid - based , e.g. , ` ramses ` ( teyssier 2002 ) , ` charm ` ( miniati & colella 2007 ) and particle - based , e.g. ` gadget ` ( springel 2005 ) .
these outputs are efficiently converted to ` radamesh ` patch - based structures with a fast algorithm based on the clustering method of berger & rigoustous ( 1991 ) , also used into the amr module .
the default format is very similar to the widely used ` chombo hdf5 ` structure https://seesar.lbl.gov/anag/chombo , allowing ` radamesh ` outputs to be directly visualized with the most recent , high - performance visualization packages ( e.g. , ` visit ` https://wci.llnl.gov/codes/visit ) .
currently , ` radamesh ` is efficiently parallelized with openmp .
all the results presented in this paper have been obtained with a few hours of computational time on a 8-core intel - xeon workstation .
sc thanks francesco miniati for useful discussions and martin haehnelt for comments on an earlier version of this manuscript .
cp acknowledges the role of piero madau who nearly 10 years ago sensitised him to the cosmological radiative - transfer problem and followed the development of some unpublished algorithms .
part of the figures presented in this paper have been realized with the visualization software visit , we are grateful to llnl to have made public this powerful tool .
abel , t. , norman , m. l. , & madau , p. 1999 , apj , 523 , 66 abel , t. , & wandelt , b. d. 2002 , mnras , 330 , l53 altay , g. , croft , r. a. c. , & pelupessy , i. 2008 , mnras , 386 , 1931 amanatides , j. , & woo , a. , `` a fast voxel traversal algorithm for ray tracing '' , proc .
eurographics 87 , amsterdam , the netherlands , august 1987 , pp 1 - 10 aubert , d. , & teyssier , r. 2008 , mnras , 387 , 295 baker , j. g. , & menzel , d. h. 1938 , apj , 88 , 52 berger , m. j. , and oliger , j. , `` adaptive mesh refinement for hyperbolic partial differential equations '' , j. comput .
phys . , 53 , pp .
484 - 512 , 1984 berger , m. j. , and rigoustsos , i. , `` an algorithm for point clustering and grid generation '' , new york university - cims report nyu-501 , 1991 bolton , j. s. , becker , g. d. , wyithe , j. s. b. , haehnelt , m. g. , & sargent , w. l. w. 2010 , arxiv:1001.3415 cantalupo , s. , porciani , c. , & lilly , s. j. 2008 , apj , 672 , 48 ercolano , b. , & storey , p. j. 2006 , mnras , 372 , 1875 gnedin , n. y. , & abel , t. 2001 , new astronomy , 6 , 437 fan , x. , at al . 2006 , aj , 132 , 117 finlator , k. , zel , f. , & dav , r. 2009 , mnras , 393 , 1090 hairer , e. , & wanner , g. , `` solving ordinary differential equations ii .
stiff and differential - algebraic problems '' , springer series in comput .
mathematics , vol .
14 , springer - verlag 1991 , second revised edition 1996 . , l. and gnedin , n. y. , 1997 , mnras , 292 , 27 khokhlov , a. m. , 1998 , j.comput.phys . , 143 , 519 , i. t. , ciardi , b. , alvarez , m. a. , maselli , a. , ferrara , a. , gnedin , n. y. , mellema , g. , nakamoto , t. , norman , m. l. , razoumov , a. o. , rijkhorst , e .-
j . , ritzerveld , j. , shapiro , p. r. , susa , h. , umemura , m. , whalen , d. j. , 2006 , mnras , 371 , 1057 maselli , a. , ferrara , a. , & ciardi , b. 2003 , mnras , 345 , 379 meiksin , a. a. 2009 , reviews of modern physics , 81 , 1405 mellema , g. , iliev , i. t. , alvarez , m. a. , & shapiro , p. r. 2006 , new astronomy , 11 , 374 miniati , f. , & colella , p. 2007
, journal of computational physics , 227 , 400 norman , m. l. , paschos , p. , & abel , t. 1998 , memorie della societa astronomica italiana , 69 , 455 nussbaumer , h. , & schmutz , w. 1984 , a&a , 138 , 495 osterbrock , d. e. , 1989 , astrophysics of gaseous nebulae and active galactic nuclei .
university science books , mill valley , ca , p. 422
pawlik , a. h. , & schaye , j. 2008 , mnras , 389 , 651 petkova , m. , & springel , v. 2009 , mnras , 396 , 1383 razoumov , a. o. , & cardall , c. y. 2005 , mnras , 362 , 1413 rijkhorst , e .-
plewa , t. , dubey , a. , & mellema , g. 2006 , a&a , 452 , 907 ritzerveld , j. 2005 , a&a , 439 , l23 ritzerveld , j. , & icke , v. 2006 , physical review e , 74 , 026704 semelin , b. , combes , f. , & baek , s. 2007 , a&a , 474 , 365 shapiro , p. r. , iliev , i. t. , alvarez , m. a. , & scannapieco , e. 2006 , apj , 648 , 922 springel , v. 2005 , mnras , 364 , 1105 teyssier , r. 2002 , a&a , 385 , 337 trac , h. , & cen , r. 2007 , apj , 671 , 1 whalen , d. , & norman , m. l. 2006 , apjs , 162 , 281
starting from the explicit form of the solid angle for each face of a cubic cell , we have found the following approximation for the total cell solid angle as seen by any source inside ( or outside ) the computational box : @xmath195 \ , \end{split}\ ] ] where @xmath196 is the three - dimensional levi - civita symbol ( note the implicit sum over the indices @xmath197 and @xmath198 ) , @xmath199 is the @xmath200 matrix containing the cell vertexes coordinates ( translated to the cartesian system with the source at the origin ) , @xmath201 and @xmath202 are , respectively , the cell centre and the source coordinates , @xmath54 is the linear cell size , @xmath203 is the @xmath204 unit matrix ( i.e. , the matrix of ones ) , and @xmath205 is the following @xmath200 matrix : @xmath206 .
all lengths and coordinates are in box units .
note that the presence of the factors containing @xmath207 and @xmath196 reduce the total number of terms in the overall sum to four per cell face .
the function @xmath208 removes from the sum the solid angle of a face _
invisible _ to the source , since this has a negative sign . in figure
[ satest ] , we show the comparison between the recovered cell solid angle obtained by a full monte carlo simulation with @xmath153 cubic cells and @xmath166 rays ( black circles ) , and the above analytical approximation ( red solid line ) . the solid angle is shown as a function of the distance from the source , placed at @xmath209 , and for two different directions : parallel to the @xmath137-axis ( i.e. , normal to two cell faces ; lower line ) and along the line of sight that connects the source to the box vertex @xmath210 ( upper line ) .
note that these two directions correspond to the minimum and maximum possible solid angle of a cubic cell at a given distance from the source .
the analytical approximation is in very good agreement with the monte carlo results
. grid .
the solid angle is shown as a function of the distance from the source , as seen by two different angular directions ( see labels ) . ]
in test 6 , we have shown that the spectral hardening inside a resolved i - front may result in a substantial increase of the gas temperature ( @xmath194 k ) within the _ whole _ hii region surrounding a bright quasar . to better demonstrate that this result is not affected by numerical effects due to the different grid resolutions , we have repeated test 6 assuming a source with similar ionizing rate but with a soft spectrum ( a black - body with @xmath211 k ) . in this case ,
spectral hardening is minimal and therefore we do not expect that increasing the resolution on the i - front should substantially change the gas temperature as in test 6 . as we show in figure [ t6b ] ( line colors and style have the same meaning as in figure 18 ) , this is indeed the case : different maximum levels of refinement produce now very similar temperatures inside the hii region showing that numerical effects are not important .
note that the ( larger ) shift in the i - front position in figure [ t6b ] ( with respect to figure 18 ) is due to the fact that , at the lowest resolution , cells are very optically thick to radiation with frequencies right above the ionization threshold ( the vast majority , given the soft spectrum assumed here ) .
this slightly alters photon - conservation , which holds to high accuracy when the front is resolved into less optically thick elements , as shown in test 1 .
k ) . in this case ,
spectral hardening is minimal and different maximum levels of refinement produce now very similar temperatures inside the hii region .
this demonstrates that numerical effects are not substantially affecting the results of test 6 . ] | we present a new three - dimensional radiative transfer ( rt ) code , ` radamesh ` , based on a ray - tracing , photon - conserving and adaptive ( in space and time ) scheme . `
radamesh ` uses a novel monte carlo approach to sample the radiation field within the computational domain on a `` cell - by - cell '' basis .
thanks to this algorithm , the computational efforts are now focused where actually needed , i.e. within the ionization - fronts ( i - fronts ) .
this results in an increased accuracy level and , at the same time , a huge gain in computational speed with respect to a `` classical '' monte carlo rt , especially when combined with an adaptive mesh refinement ( amr ) scheme . among several new features , ` radamesh ` is able to adaptively refine the computational mesh in correspondence of the i - fronts , allowing to fully resolve them within large , cosmological boxes . we follow the propagation of ionizing radiation from an arbitrary number of sources and from the recombination radiation produced by h and he . the chemical state of six species ( hi , hii , hei , heii , heiii , e ) and gas temperatures are computed with a time - dependent , non - equilibrium chemistry solver .
we present several validating tests of the code , including the standard tests from the rt code comparison project and a new set of tests aimed at substantiating the new characteristics of ` radamesh ` . using our amr scheme ,
we show that properly resolving the i - front of a bright quasar during reionization produces a large increase of the predicted gas temperature within the whole hii region .
also , we discuss how h and he recombination radiation is able to substantially change the ionization state of both species ( for the classical strmgren sphere test ) with respect to the widely used `` on - the - spot '' approximation .
[ firstpage ] radiative transfer - methods : numerical - hii regions - intergalactic medium - diffuse radiation - cosmology : theory |
in the non perturbative regime of strongly interacting theories effective lagrangians play a dominant role since they efficiently describe the non perturbative dynamics in terms of the relevant degrees of freedom .
symmetries , anomalous and exact , are used to constrain the effective lagrangians .
an important point is that the effective lagrangian approach is applicable to any region of the qcd or qcd - like phase diagram whenever the relevant degrees of freedom and the associated symmetries are defined . at zero temperature and quark chemical potential the simplest effective lagrangian describing a relevant part of the nonperturbative physics of the yang - mills ( ym )
theory is the glueball lagrangian whose potential is : @xmath0 \ . \end{aligned}\ ] ] the latter is constrained using trace anomaly and @xmath1 $ ] with @xmath2 the gluon field stress tensor .
it describes the vacuum of a generic yang - mills theory .
a similar effective lagrangian ( using superconformal anomalies ) can be written for the non perturbative super yang - mills ( sym ) theory .
this is the celebrated veneziano - yankielowicz @xcite lagrangian . in yang mills and super yang - mills theories
no exact continuous global symmetries are present which can break spontaneously and hence no goldstones are present .
the situation is different when flavors are included in the theory . here
the spontaneous breaking of chiral symmetry leads to a large number of goldstone s excitations .
we note that in we were able , using string techniques , to derive a number of fundamental perturbative and non perturbative properties for supersymmetric qcd such as the beta function , fermion condensate as well as chiral anomalies .
recently in @xcite we constructed effective lagrangians of the veneziano - yankielowicz ( vy ) type for two non - supersymmetric but strongly interacting theories with a dirac fermion either in the two index symmetric or two index antisymmetrix representation of the gauge group .
these theories are planar equivalent , at @xmath3 to sym @xcite . in this limit
the non - supersymmetric effective lagrangians coincide with the bosonic part of the vy lagrangian .
we departed from the supersymmetric limit in two ways .
first , we considered finite values of @xmath4 . then @xmath5 effects break supersymmetry .
we suggested the simplest modification of the vy lagrangian which incorporates these @xmath5 effects , leading to a non - vanishing vacuum energy density .
we analyzed the spectrum of the finite-@xmath4 non - supersymmetric daughters .
for @xmath6 the two - index antisymmetric representation ( one flavor ) _ is one - flavor qcd_. we showed that in this case the scalar quark - antiquark state is heavier than the corresponding pseudoscalar state , the @xmath7 .
second , we added a small fermion mass term which breaks supersymmetry explicitly .
the vacuum degeneracy is lifted , the parity doublets split and we evaluated this splitting . the @xmath8-angle dependence and its implications were also investigated .
this new effective lagrangian provides a number of fundamental results about qcd which can be already tested either experimentally or via lattice simulations .
this new type of expansion in the inverse of number colors in which the quark representation is the two index antisymmetric representation of the gauge group at any given @xmath4 may very well be more convergent then the ordinary @xmath5 expansion . in the ordinary case one
keeps the fermion in the fundamental representation of the gauge group while increasing the number of colors .
indeed recently in @xcite we have studied the dependence on the number of colors ( while keeping the fermions in the fundamental representation of the gauge group ) of the leading pi pi scattering amplitude in chiral dynamics .
we have demonstrated the existence of a critical number of colors for and above which the low energy pi pi scattering amplitude computed from the simple sum of the current algebra and vector meson terms is crossing symmetric and unitary at leading order in a @xmath5 expansion .
the critical number of colors turns out to be @xmath9 and is insensitive to the explicit breaking of chiral symmetry .
this means that the ordinary @xmath5 corrections for the real world are large . at nonzero temperature
the center of the @xmath10 gauge group becomes a relevant symmetry @xcite .
however except for mathematically defined objects such as polyakov loops the physical states of the theory are neutral under the center group symmetry . a new class of effective lagrangians have been constructed to show how the information about the center group symmetry is efficiently transferred to the actual physical states of the theory and will be reviewed in detail elsewhere . via these lagrangians
we were also able to have a deeper understanding of the relation between chiral restoration and deconfinement @xcite for quarks in the fundamental and in the adjoint representation of the gauge group .
i will focus here on the two basic effective lagrangians developed for color superconductivity .
more specifically the lagrangian for the color flavor locked phase ( cfl ) of qcd at high chemical potential and the 2 flavor color superconductive effective lagrangian .
[ uno ] a color superconducting phase is a reasonable candidate for the state of strongly interacting matter for very large quark chemical potential .
many properties of such a state have been investigated for two and three flavor qcd . in some cases these results rely heavily on perturbation theory , which is applicable for very large chemical potentials . some initial applications to supernovae
explosions and gamma ray bursts can be found in @xcite and @xcite respectively , see also @xcite . the interested reader can find a discussion of the effects of color superconductivity on the mass - radius relationship of compact stars in @xcite
for @xmath11 light flavors at very high chemical potential dynamical computations suggest that the preferred phase is a superconductive one and the following ansatz for a quark - quark type of condensate is energetically favored : @xmath12 a similar expression holds for the right transforming fields .
the greek indices represent spin , @xmath13 and @xmath14 denote color while @xmath15 and @xmath16 indicate flavor .
the condensate breaks the gauge group completely while locking the left / right transformations with color .
the final global symmetry group is @xmath17 , and the low energy spectrum consists of @xmath18 goldstone bosons .
here we seek insight regarding the relevant energy scales of various physical states of the color flavor locked phase ( cfl ) , such as the vector mesons and the solitons @xcite .
our results do not support the naive expectation that all massive states are of the order of the color superconductive gap , @xmath19 .
our strategy is based on exploiting the significant information already contained in the low energy effective theory for the massless states .
we transfer this information to the massive states of the theory by making use of the fact that higher derivative operators in the low energy effective theory for the lightest state can also be induced when integrating out heavy fields . for the vector mesons ,
this can be seen by considering a generic theory containing vector mesons and goldstone bosons . after integrating out the vector mesons ,
the induced local effective lagrangian terms for the goldstone bosons must match the local contact terms from operator counting .
we find that each derivative in the ( cfl ) chiral expansion is replaced by a vector field @xmath20 as follows @xmath21 this relation allows us to deduce , among other things , that the energy scale for the vector mesons is @xmath22 where @xmath23 is the vector meson mass .
our result is in agreement with the findings in .
we shall see that this also suggests that the ksrf relation holds in the cfl phase . in the solitonic sector ,
the cfl chiral lagrangian gives us the scaling behavior of the coefficient of the skyrme term and thus shows that the mass of the soliton is of the order of @xmath24 which is contrary to naive expectations .
this is suggestive of a kind of duality between vector mesons and solitons in the same spirit as the duality advocated some years ago by montonen and olive for the @xmath25 georgi - glashow theory @xcite .
this duality becomes more apparent when considering the product @xmath26 which is independent of the scale , @xmath19 . in the present case , if the vector meson self - coupling is @xmath27 , we find that the skyrme coefficient , @xmath28 , can be identified with @xmath27 .
thus , the following relations hold : @xmath29 in this notation the electric - magnetic ( i.e. vector meson - soliton ) duality is transparent . since the topological wess - zumino term in the cfl phase is identical to that in vacuum , we identify the soliton with a physical state having the quantum numbers of the nucleon .
we expect that the product of the nucleon and vector meson masses will scale like @xmath30 for any non - zero chemical potential for three flavors .
interestingly , quark - hadron continuity can be related to duality @xcite . testing this relation
can also be understood as a quantitative check of quark - hadron continuity .
it is important to note that our results are tree level results and that the resulting duality relation can be affected by quantum corrections .
our results have direct phenomenological consequences for the physics of compact stars with a cfl phase .
while vector mesons are expected to play a relevant role , solitons can safely be neglected for large values of the quark chemical potential .
when diquarks condense for the three flavor case , we have the following symmetry breaking : @xmath31 \times su_l(3 ) \times su_r(3 ) \times
u_b(1 ) \rightarrow su_{c+l+r}(3 ) \ . \nonumber\end{aligned}\ ] ] the gauge group undergoes a dynamical higgs mechanism , and nine goldstone bosons emerge . neglecting the goldstone mode associated with the baryon number and quark masses ( which will not be important for our discussion at lowest order ) ,
the derivative expansion of the effective lagrangian describing the octect of goldstone bosons is : @xmath32\equiv \frac{f^2_{\pi}}{2 } { \rm tr}\left[p_{\mu}p^{\mu}\right ] \ , \end{aligned}\ ] ] with @xmath33 , @xmath34 , @xmath35 and @xmath36 is the octet of goldstone bosons .
@xmath37 transforms linearly according to @xmath38 and @xmath39 while @xmath40 transforms non - linearly : @xmath41 this constraint implicitly defines the matrix , @xmath42 . here ,
we wish to examine the cfl spectrum of massive states using the technique of integrating in / out at the level of the effective lagrangian .
@xmath43 is the goldstone boson decay constant .
it is a non - perturbative quantity whose value is determined experimentally or by non - perturbative techniques ( e.g. lattice computation ) . for very large quark chemical potential , @xmath43 can be estimated perturbatively .
it is found to be proportional to the fermi momentum , @xmath44 , with @xmath45 the quark chemical potential @xcite . since a frame
must be fixed in order to introduce a chemical potential , spatial and temporal components of the effective lagrangians split .
this point , however , is not relevant for the validity of our results .
when going beyond the lowest - order term in derivatives , we need a counting scheme . for theories with only one relevant scale ( such as qcd at zero chemical potential ) , each derivative is suppressed by a factor of @xmath43 .
this is not the case for theories with multiple scales . in the cfl phase
, we have both @xmath43 and the gap , @xmath19 , and the general form of the chiral expansion is @xcite : @xmath46 following @xcite , we distinguish between temporal and spatial derivatives .
chiral loops are suppressed by powers of @xmath47 , and higher - order contact terms are suppressed by @xmath48 where @xmath49 is the momentum .
thus , chiral loops are parametrically small compared to contact terms when the chemical potential is large .
there is also a topological term which is essential in order to satisfy the thooft anomaly conditions at the effective lagrangian level .
it is important to note that respecting the thooft anomaly conditions is more than an academic exercise .
in fact , it requires that the form of the wess - zumino term is the same in vacuum and at non - zero chemical potential .
its real importance lies in the fact that it forbids a number of otherwise allowed phases which can not be ruled out given our rudimentary treatment of the non - perturbative physics . as an example , consider a phase with massless protons and neutrons in three - color qcd with three flavors . in this case
chiral symmetry does not break .
this is a reasonable realization of qcd for any chemical potential .
however , it does not satisfy the thooft anomaly conditions and hence can not be considered .
were it not for the thooft anomaly conditions , such a phase could compete with the cfl phase .
gauging the wess - zumino term with to respect the electromagnetic interactions yields the familiar @xmath50 anomalous decay .
this term @xcite can be written compactly using the language of differential forms .
it is useful to introduce the algebra - valued maurer - cartan one form @xmath51 which transforms only under the left @xmath52 flavor group .
the wess - zumino effective action is @xmath53 = c\,\int_{m^{5}}{\rm tr}\left [ \alpha ^{5}\right ] \ .
\label{wz}\end{aligned}\ ] ] the price which must be paid in order to make the action local is that the spatial dimension must be augmented by one .
hence , the integral must be performed over a five - dimensional manifold whose boundary ( @xmath54 ) is ordinary minkowski space . in the constant @xmath55 has been shown to be the same as that at zero density , i.e. @xmath56 where @xmath57 is the number of colors ( three in this case ) . due to the topological nature of the wess - zumino term
its coefficient is a pure number .
it is well known that massive states are relevant for low energy dynamics .
consider , for example , the role played by vector mesons in pion - pion scattering in saturating the unitarity bounds . more specifically
, vector mesons play a relevant role when describing the low energy phenomenology of qcd and may also play a role also in the dynamics of compact stars with a cfl core @xcite . in order to investigate the effects of such states ,
we need to know their in medium properties including their gaps and the strength of their couplings to the cfl goldstone bosons . except for the extra
spontaneously broken @xmath58 symmetry , the symmetry properties of the cfl phase have much in common with those of zero density phase of qcd .
this fact allows us to make some non perturbative but reasonable estimates of vector mesons properties in medium .
we have already presented the general form of the chiral expansion in the cfl phase .
as will soon become clear , we are now interested in the four derivative ( non topological ) terms whose coefficients are proportional to @xmath59 this must be contrasted with the situation at zero chemical potential , where the coefficient of the four derivative term is always a pure number before quantum corrections are taken into account . in vacuum ,
the tree - level lagrangian which simultaneously describes vector mesons , goldstone bosons , and their interactions is : @xmath60+\frac{m^2_{v}}{2 } { \rm tr } \left[\left(\rho_{\mu } + \frac{v_{\mu}}{\widetilde{g}}\right)^2\right ] \nonumber \\ & -&\frac{1}{4}{\rm tr } \left[f_{\mu\nu}(\rho)f^{\mu\nu}(\rho)\right ] \ , \end{aligned}\ ] ] where @xmath61 mev and @xmath62 is the one form @xmath63 with @xmath64 and @xmath65 $ ] . at tree level
this lagrangian agrees with the hidden local symmetry results @xcite .
when the vector mesons are very heavy with respect to relevant momenta , they can be integrated out .
this results in the field constraint : @xmath66 substitution of this relation in the vector meson kinetic term ( i.e. , the replacement of @xmath67 by @xmath68 ) gives the following four derivative operator with two time derivatives and two space derivatives @xcite : @xmath69 ^ 2\right ] \ .\label{4d}\end{aligned}\ ] ] the coefficient is proportional to @xmath70 .
it is also relevant to note that since we are describing physical fields we have considered canonically normalized fields and kinetic terms .
this lagrangian can also be applied to the cfl case . in the vacuum
, @xmath27 is a number of order one independent of the scale at tree level .
this is no longer the case in the cfl phase . here , by comparing the coefficient of the four derivative operator in eq .
( [ 4d ] ) obtained after having integrated out the vector meson with the coefficient of the same operator in the cfl chiral perturbation theory we determine the following scaling behavior of @xmath27 : @xmath71 by expanding the effective lagrangian with the respect to the goldstone boson fields , one sees that @xmath27 is also connected to the vector meson coupling to two pions , @xmath72 , through the relation @xmath73 in vacuum @xmath74 and @xmath75 are quantities of order one . since @xmath62 is essentially a single derivative , the scaling behavior of @xmath27 allows us to conclude that each derivative term is equivalent to @xmath76 with respect to the chiral expansion .
for example , dropping the dimensionless field @xmath37 , the operator with two derivatives becomes a mass operator for the vector meson @xmath77 this demonstrates that the vector meson mass gap is proportional to the color superconducting gap .
this non - perturbative result is relevant for phenomenological applications .
it is interesting to note that our simple counting argument agrees with the underlying qcd perturbative computations of ref .
@xcite and also with recent results of ref .
@xcite . in @xcite , at high chemical potential , vector meson dominance is discussed . however , our approach is more general since it does not rely on any underlying perturbation theory .
it can be applied to theories with multiple scales for which the counting of the goldstone modes is known . since @xmath78 , we find that @xmath79 scales with @xmath27 suggesting that the ksrf relation is a good approximation also in the cfl phase of qcd .
the low energy effective theory supports solitonic excitations which can be identified with the baryonic sector of the theory at non - zero chemical potential . in order to obtain classically stable configurations , it is necessary to include at least a four derivative term ( containing two temporal derivatives ) in addition to the usual two derivative term .
such a term is the skyrme term : @xmath80 ^ 2\right ] \
.\end{aligned}\ ] ] since this is a fourth order term in derivatives not associated with the topological term we have : @xmath81 this term is the same as that which emerges after integrating out the vector mesons ( see eq .
( [ 4d ] ) ) , and one concludes that @xmath82 @xcite . the simplest complete action supporting solitonic excitations is : @xmath83 + l^{\rm skyrme}\right ] + \gamma_{wz } \ .\end{aligned}\ ] ] the wess - zumino term in eq .
( [ wz ] ) guarantees the correct quantization of the soliton as a spin @xmath84 object . here
we neglect the breaking of lorentz symmetries , irrelevant to our discussion .
the euler - lagrangian equations of motion for the classical , time independent , chiral field @xmath85 are highly non - linear partial differential equations . to simplify these equations
skyrme adopted the hedgehog _ ansatz _ which , suitably generalized for the three flavor case , reads @xcite : @xmath86 where @xmath87 represents the pauli matrices and the radial function @xmath88 is called the chiral angle .
ansatz _ is supplemented with the boundary conditions @xmath89 and @xmath90 which guarantee that the configuration posseses unit baryon number . after substituting the _ ansatz _ in the action
one finds that the classical solitonic mass is , up to a numerical factor : @xmath91 and the isoscalar radius , @xmath92 .
interestingly , due to the non perturbative nature of the soliton , its mass turns to be dual to the vector meson mass .
it is also clear that although the vector mesons and the solitons have dual masses , they describe two very distinct types of states .
the present duality is very similar to the one argued in @xcite . indeed , after introducing the collective coordinate quantization , the soliton ( due to the wess - zumino term ) describes baryonic states of half - integer spin while the vectors are spin one mesons . here , the dual nature of the soliton with respect to the vector meson is enhanced by the fact that , in the cfl state , @xmath93 is expected to be substantially reduced with respect to its value in vacuum .
once the soliton is identified with the nucleon ( whose density dependent mass is denoted with @xmath94 ) we predict the following relation to be independent of the matter density : @xmath95 in this way , we can relate duality to quark - hadron continuity .
we considered duality , which is already present at zero chemical potential , between the soliton and the vector mesons a fundamental property of the spectrum of qcd which should persists as we increase the quark chemical potential .
should be noted that differently than in @xcite we have not subtracted the energy cost to excite a soliton from the fermi sea .
since we are already considering the lagrangian written for the excitations near the fermi surface we would expect not to consider such a corrections . in any event
this is of the order @xmath45 @xcite and hence negligible with respect to @xmath96 .
we have shown that the vector mesons in the cfl phase have masses of the order of the color superconductive gap , @xmath19 .
on the other hand the solitons have masses proportional to @xmath97 and hence should play no role for the physics of the cfl phase at large chemical potential .
we have noted that the product of the soliton mass and the vector meson mass is independent of the gap .
this behavior reflects a form of electromagnetic duality in the sense of montonen and olive @xcite .
we have predicted that the nucleon mass times the vector meson mass scales as the square of the pion decay constant at any nonzero chemical potential . in the presence of two or more scales provided by the underlying theory
the spectrum of massive states shows very different behaviors which can not be obtained by assuming a naive dimensional analysis .
qcd with 2 massless flavors has gauge symmetry @xmath98 and global symmetry @xmath99 at very high quark density the ordinary goldstone phase is no longer favored compared with a superconductive one associated to the following type of diquark condensates : @xmath100 if parity is not broken spontaneously , we have @xmath101 , where we choose the condensate to be in the 3rd direction of color .
the order parameters are singlets under the @xmath102 flavor transformations while possessing baryon charge @xmath103 .
the vev leaves invariant the following symmetry group : @xmath104 \times su_{l}(2)\times su_{r}(2)\times \widetilde{u}_{v}(1)\ , \ ] ] where @xmath105 $ ] is the unbroken part of the gauge group .
the @xmath106 generator @xmath107 is the following linear combination of the previous @xmath108 generator @xmath109 and the broken diagonal generator of the @xmath98 gauge group @xmath110 : @xmath111 .
the quarks with color @xmath112 and @xmath113 are neutral under @xmath107 and consequently so is the condensate .
the spectrum in the 2sc state is made of 5 massive gluons with a mass of the order of the gap , 3 massless gluons confined ( at zero temperature ) into light glueballs and gapless up and down quarks in the direction ( say ) 3 of color .
the relevant coset space @xmath114 with @xmath115 is parameterized by : @xmath116 where @xmath117 @xmath118 belong to the coset space @xmath114 and are taken to be @xmath119 for @xmath120 while @xmath121 @xmath122 are the standard generators of @xmath123 .
the coordinates @xmath124 via @xmath125 describe the goldstone bosons which will be absorbed in the longitudinal components of the gluons .
the vevs @xmath126 and @xmath127 are , at asymptotically high densities , proportional to @xmath128 .
@xmath129 transforms non linearly : @xmath130 with @xmath131 it is convenient to define the following differential form : @xmath132 with @xmath133 the gluon fields while @xmath134 is the strong coupling constant
. @xmath135 transforms according to : @xmath136 we decompose @xmath137 into @xmath138 \quad { \rm and}\quad \omega _ { \mu } ^{\perp } = 2x^{i}{\rm tr}\left [ x^{i}\omega _ { \mu } \right ] \ , \end{aligned}\ ] ] @xmath139 are the unbroken generators of @xmath140 , while @xmath141 and @xmath142 .
the most generic two derivative kinetic lagrangian for the goldstone bosons is : @xmath143 + f^{2}a_{2}{\rm tr}\left [ \,\omega _ { \mu } ^{\perp } \,\right ] { \rm tr } \left [ \,\omega ^{\mu \perp } \,\right ] \ . \label{dt}\end{aligned}\ ] ] the double trace term is due to the absence of the condition for the vanishing of the trace for the broken generator @xmath144 .
it emerges naturally in the non linear realization framework at the same order in derivative expansion with respect to the single trace term . in the unitary gauge these two terms
correspond to the five gluon masses @xcite . for the fermions it is convenient to define the dressed fermion fields @xmath145 transforming as @xmath146 .
@xmath147 has the ordinary quark transformations ( i.e. is a dirac spinor ) . -1.5 cm
-.6 cm pictorially @xmath148 can be viewed as a constituent type field or alternatively as the bare quark field @xmath147 immersed in the diquark cloud represented by @xmath129 .
the non linearly realized effective lagrangian describing in medium fermions , gluons and their self interactions , up to two derivatives is : @xmath149 + f^{2}a_{2}{\rm tr}\left [ \,\omega _ { \mu } ^{\perp } \,\right ] { \rm tr}\left [ \,\omega ^{\mu \perp } \,\right ] \nonumber \\ & + & b_{1}\overline{\widetilde{\psi } } i\gamma ^{\mu } ( \partial _ { \mu } -i\omega _ { \mu } ^{\parallel } ) \widetilde{\psi } + b_{2}\overline{\widetilde{\psi } } \gamma ^{\mu } \omega _ { \mu } ^{\perp } \widetilde{\psi } \nonumber \\ & + & m_{m}\overline{\widetilde{\psi } ^{c}}_{i}\gamma ^{5}(it^{2})\widetilde{\psi } _ { j}\varepsilon ^{ij}+{\rm h.c.}\ , \end{aligned}\ ] ] where @xmath150 , @xmath151 are flavor indices and @xmath152 here @xmath153 and @xmath154 are real coefficients while @xmath155 is complex . from the last two terms , representing a majorana mass term for the quarks , we see that the massless degrees of freedom are the @xmath156 .
the latter possesses the correct quantum numbers to match the t hooft anomaly conditions @xcite .
the @xmath157 gauge symmetry does not break spontaneously and confines . calling @xmath140 a mass dimension four composite field describing the scalar glueball we can construct the following lagrangian @xcite : @xmath158 \right .
\nonumber \\ & - & \left .
\frac{b}{2 } h\log\left[\frac{h}{\hat{\lambda}^4}\right ] \right\ } \ .
\label{g - ball}\end{aligned}\ ] ] this lagrangian correctly encodes the underlying @xmath157 trace anomaly .
the glueballs move with the same velocity @xmath159 as the underlying gluons in the 2sc color superconductor .
@xmath160 is related to the intrinsic scale associated with the @xmath157 theory and can be less than or of the order of few mevs @xcite is @xmath161 with @xmath162 the intrinsic scale of @xmath157 after the coordinates have been appropriately rescaled to eliminate the @xmath159 dependence from the action .
] once created , the light @xmath157 glueballs are stable against strong interactions but not with respect to electromagnetic processes @xcite .
indeed , the glueballs couple to two photons via virtual quark loops .
@xmath163 \approx 1.2\times 10^{-2 } \left[\frac{m_h}{1~{\rm mev}}\right]^5~{\rm ev } \ , \end{aligned}\ ] ] where @xmath164 . for illustration purposes
we consider a glueball mass of the order of @xmath112 mev which leads to a decay time @xmath165 .
this completes the effective lagrangian for the 2sc state which corresponds to the wigner - weyl phase .
using this lagrangian one can estimate the @xmath157 glueball melting temperature to be @xcite : @xmath166{\frac{90 { v}^3}{2\,e\ , \pi^2}}\ , { \hat{\lambda } } < t_{csc } \ .\end{aligned}\ ] ] where @xmath167 is the color superconductive transition temperature .
the deconfining / confining @xmath157 phase transition within the color superconductive phase is second order .
the superconductive phase for @xmath168 possesses the same global symmetry group as the confined wigner - weyl phase .
the ungapped fermions have the correct global charges to match the t hooft anomaly conditions as shown in @xcite .
specifically the @xmath169 global anomaly is correctly reproduced in this phase due to the presence of the ungapped fermions .
this is so since a quark in the 2sc case is surrounded by a diquark medium ( i.e. @xmath170 ) and behaves as a baryon . @xmath171
the validity of the thooft anomaly conditions at high matter density have been investigated in .
a delicate part of the proof presented in @xcite is linked necessarily to the infrared behavior of the anomalous three point function . in particular
one has to show the emergence of a singularity ( i.e. a pole structure ) .
this pole is then interpreted as due to a goldstone boson when chiral symmetry is spontaneously broken .
one might be worried that , since the chemical potential explicitly breaks lorentz invariance , the gapless ( goldstone ) pole may disappear modifying the infrared structure of the three point function .
this is not possible .
thanks to the nielsen and chadha theorem @xcite , not used in @xcite , we know that gapless excitations are always present when some symmetries break spontaneously even in the absence of lorentz invariance . since the quark chemical potential is associated with the barionic generator which commutes with all of the non abelian global generators the number of goldstone bosons must be larger or equal to the number of broken generators . besides all of the goldstones must have linear dispersion relations ( i.e. are type i @xcite ) .
this fact not only guarantees the presence of gapless excitations ( justifying the analysis made in @xcite on the infrared behavior of the form factors ) but demonstrates that the pole structure due to the gapless excitations needed to saturate the triangle anomaly is identical to the zero quark chemical potential one in the infrared .
it is also interesting to note that the explicit dependence on the quark chemical potential is communicated to the goldstone excitations via the coefficients of the effective lagrangian ( see @xcite for a review ) .
for example @xmath43 is proportional to @xmath45 in the high chemical potential limit and the low energy effective theory is a good expansion in the number of derivatives which allows to consistently incorporate in the theory the wess - zumino - witten term @xcite and its corrections .
the validity of the anomaly matching conditions have far reaching consequences . indeed , in the three flavor case
, the conditions require the goldstone phase to be present in the hadronic as well as in the color superconductive phase supporting the quark - hadron continuity scenario @xcite . at very high quark chemical potential the effective field theory of low energy modes ( not to be confused with the goldstone excitations ) has positive euclidean path integral measure @xcite . in this limit
the cfl is also shown to be the preferred phase with the aid of the anomaly conditions .
since the fermionic theory has positive measure only at asymptotically high densities one can not use this fact to show that the cfl is the preferred phase for moderate chemical potentials .
this is possible using the anomaly constraints . while the anomaly matching conditions are still in force at nonzero quark chemical potential @xcite the _ persistent mass _ condition @xcite ceases to be valid .
indeed a phase transition , as function of the strange quark mass , between the cfl and the 2sc phases occurs .
we recall that we can saturate the thooft anomaly conditions either with massless fermionic degrees of freedom or with gapless bosonic excitations .
however in absence of lorentz covariance the bosonic excitations are not restricted to be fluctuations related to scalar condensates but may be associated , for example , to vector condensates @xcite .
i thank a.d .
jackson for stimulating collaboration on some of the recent topics presented here and for careful reading of the manuscript .
i also thank r. casalbuoni , p.h .
damgaard , z. duan , m. harada , d.k .
hong , s.d .
hsu , r. marotta , a. mcsy , f. pezzella , j. schechter , m. shifman , k. splittorff and k. tuominen for their valuable collaboration and/or discussions on some of the topics i presented here .
k. tuominen is thanked also for careful reading of the manuscript .
999 j. schechter , phys .
d * 21 * ( 1980 ) 3393 .
c. rosenzweig , j. schechter and g. trahern , phys .
rev . * d21 * , 3388 ( 1980 ) ; p. di vecchia and g. veneziano , nucl .
b171 * , 253 ( 1980 ) ; e. witten , ann . of phys . *
128 * , 363 ( 1980 ) ; p.
nath and a. arnowitt , phys . rev .
* d23 * , 473 ( 1981 ) ; a. aurilia , y. takahashi and d. townsend , phys . lett . *
95b * , 65 ( 1980 ) ; k. kawarabayashi and n. ohta , nucl . phys . *
b175 * , 477 ( 1980 ) .
a. a. migdal and m. a. shifman , phys .
b * 114 * , 445 ( 1982 ) ; j. m. cornwall and a. soni , phys .
d * 29 * , 1424 ( 1984 ) ; phys .
d * 32 * , 764 ( 1985 ) .
a. salomone , j. schechter and t. tudron , phys . rev . *
d23 * , 1143 ( 1981 ) ; j. ellis and j. lanik , phys . lett . *
150b * , 289 ( 1985 ) ; h. gomm and j. schechter , phys . lett . *
158b * , 449 ( 1985 ) ; f. sannino and j. schechter , phys .
d * 60 * , 056004 ( 1999 ) [ hep - ph/9903359 ] .
g. veneziano and s. yankielowicz , phys .
b * 113 * , 231 ( 1982 ) .
r. marotta , f. nicodemi , r. pettorino , f. pezzella and f. sannino , jhep * 0209 * , 010 ( 2002 ) [ arxiv : hep - th/0208153 ] .
r. marotta and f. sannino , phys .
b * 545 * , 162 ( 2002 ) [ arxiv : hep - th/0207163 ] .
m. harada , f. sannino and j. schechter , `` large n(c ) and chiral dynamics , '' arxiv : hep - ph/0309206 . to appear in phys .
d. a. mocsy , f. sannino and k. tuominen , phys .
lett . *
91 * , 092004 ( 2003 ) [ arxiv : hep - ph/0301229 ] .
a. mocsy , f. sannino and k. tuominen , `` induced universal properties and deconfinement , '' arxiv : hep - ph/0306069 .
a. mocsy , f. sannino and k. tuominen , `` confinement versus chiral symmetry , '' arxiv : hep - ph/0308135 .
see t. schafer , `` quark matter , '' arxiv : hep - ph/0304281 and references therein for a concise review on the @xmath45 dependence of @xmath172 . f. sannino , phys .
b * 480 * , 280 ( 2000 ) [ arxiv : hep - ph/0002277 ] .
s. d. hsu , f. sannino and m. schwetz , mod .
a * 16 * , 1871 ( 2001 ) [ arxiv : hep - ph/0006059 ] .
f. sannino , anomaly matching and low energy theories at high matter density arxiv : hep - ph/0301035 .
proceedings for the review talk at the electroweak and strong matter conference , heidelberg 2002 .
f. sannino , phys .
d * 67 * , 054006 ( 2003 ) [ arxiv : hep - ph/0211367 ] .
f. sannino and w. schfer , phys .
b * 527 * , 142 ( 2002 ) hep - ph/0111098 .
f. sannino and w. schfer , hep - ph/0204353.j .
t. lenaghan , f. sannino and k. splittorff , phys .
d * 65 * , 054002 ( 2002 ) [ arxiv : hep - ph/0107099 ] . | i briefly discuss effective lagrangians for strong interactions while concentrating on two specific lagrangians for qcd at large matter density .
i then introduce spectral duality in qcd a la montonen and olive .
the latter is already present in qcd in the hadronic phase .
however it becomes transparent at large chemical potential .
finally i show the relevance of having exact non perturbative constraints such as thooft anomaly conditions at zero and nonzero quark chemical potential on the possible phases of strongly interacting matter .
an important outcome is that for three massless quarks at any chemical potential the only non trivial solution of the constraints is chiral symmetry breaking .
this also shows that for three massless flavors at large quark chemical potential cfl is the ground state . |
convergent expansions such as linked cluster , hopping parameter or high temperature expansions provide an analytic alternative to monte carlo simulations .
originally they have been developed in the infinite volume , meanwhile they have been extended to finite volumes as well @xcite .
in contrast to generic perturbation theory , hopping parameter expansions ( hpes ) are convergent expansions about completely disordered lattice systems .
the expansion parameter @xmath5 is the coefficient of the ( pair ) interaction term .
since we calculate free energies and connected correlations in the hopping parameter expansion , we generate linked cluster expansions ( lces ) .
lces have been generalized to scalar field theories at _ finite _ temperature by @xcite .
compared to the zero temperature results , the finite temperature induces a tiny shift in the critical temperature ( hopping parameter ) .
thus one has to face a similar problem as the improvement of a bad signal to noise ratio in monte carlo simulations . here
the signal corresponds to the finite temperature effect and the noise to the `` background '' of temperature zero graphs .
the price to describe the critical region , in particular critical exponents , with an accurracy of @xmath6 % is to go to the 18th order in the expansion in @xmath5 .
this accurracy has been achieved in @xcite .
thus the question arises , why we are still interested in a generalization of lces to a finite volume .
the reason is twofold .
@xmath7 the first one is to identify 1st order transitions from series expansions in the high temperature phase .
for example a measurement of a vanishing critical exponent @xmath8 is compatible with a gaussian fixed point , but not conclusive for the onset of a first order transition region at a tricritical point .
( the latter interpretation actually applies to our measurements . )
@xmath7 the second reason is to distinguish second order transitions associated with different universality classes .
ideally critical exponents should show a clear gap between plateaus of different universality classes .
practically the gap is smeared out because of the truncation of the series expansions at finite order .
the truncation has a similar rounding effect on this gap as a finite volume on thermodynamic singularities .
in hpes the action is split into a sum of ultralocal parts @xmath9 and a next neighbour part @xmath10 with next neighbour couplings @xmath11 .
a taylor expansion in @xmath5 of the partition function @xmath12 about the ultralocal contribution to @xmath12 finally leads to graphical expansions of n - point susceptibilities @xmath13 [ exa.chi ] _ n ( ) = _ 0 a_^(n ) ^. the coefficients @xmath14 are a sum of graphs each of which consists of a product of the inverse symmetry factor , an internal symmetry factor , a lattice embedding factor , and a polynomial of vertex contributions depending on the couplings involved in @xmath9 .
it is only the embedding factor that depends on the topology of the particular lattice , that will change in passing from an infinite to a finite volume . from our series expansions we measure a _ shift in @xmath15 _ from a fit according to [ abc ] ( | _ c(l ) - _ c ( ) | ) - y_t .
here @xmath16 is a constant , @xmath17 for 2nd order transitions and @xmath18 for 1st order transitions .
the critical couplings @xmath19 and @xmath20 are determined as the radius of convergence of the series expansions in a finite or infinite volume , respectively . since
an extrapolation of the scaling behaviour from the height and width of the critical region is not feasible within the series expansions , we have proposed the following _ monotony criterion_. consider two volumes @xmath21 and @xmath22 such that @xmath23 , @xmath24 , @xmath25 stands for a generic coupling , and @xmath26 for the reduced `` temperature '' .
we define [ fss.rml ] r(v_1,v_2 ) : = 1 - , the monotony criterion says that @xmath27 the different behaviour of @xmath28 comes from the _ singular _ part of @xmath29 .
the underlying observation is a different monotony behaviour of response functions @xmath30 for 1st and 2nd order transitions that is seen for @xmath30s with nonanalytic behaviour in the infinite volume limit .
the @xmath30s are increasing in volume in a certain neighbourhood of @xmath31 for 2nd order transitions and decreasing for 1st order transitions . for a precise specification of the scaling region , i.e. the bounds that @xmath21 , @xmath22,@xmath32 have to satisfy , we refer to @xcite .
the difference reflects the @xmath33-function and power law singularities for 1st or 2nd order transitions , respectively , predicted at @xmath34 in the thermodynamic limit.thus the actual behaviour of @xmath29 we have found in applications to scalar @xmath1 models ( cf .
section 3 ) is neither a pecularity of the model nor an artifact of the series expansions .
@xmath7 it should be noticed that the criterion is not restricted to series expansions , but also applicable to monte carlo calculations ( since both involved volumes may be finite ) if the regular contribution to @xmath30 is really negligible .
@xmath7 it is neither restricted to order parameter susceptibilities as the notation suggests , but similarly applies to other response functions like the specific heat if they diverge in the thermodynamic limit . the effective potential is defined via the effective action @xmath36 evaluated for a constant background field @xmath37 ( @xmath38 ) .
it can be expressed in terms of the 1pi - susceptibilities according to @xmath39 @xmath40 the @xmath41s are obtained from the @xmath42s if the graphical expansion of the @xmath42s is restricted to 1pi graphs .
the ( non)convex shape of the effective potential in the symmetric phase indicates the order of the transition .
its evaluation in a _
finite _ volume leads to an alternative determination of @xmath19 via the coexistence of different minima of the free energy or a vanishing of the coefficient of the quadratic term .
we apply the finite size scaling analysis with hpes to a scalar theory with @xmath45 and @xmath46-terms and @xmath1 symmetry in 3 dimensions , @xmath47 is chosen as @xmath48 and @xmath49 . after a suitable reparametrization the ultralocal part of the action reads [ exa.action ] ( , , ) = ^2 + ( ^2 - 1)^2 + ( ^2 - 1)^3 the hopping parameter term is given as the value for @xmath50 for the first order transitions confirms results of @xcite for @xmath2 and extends results of @xcite for @xmath3 .
it disagrees with the naive expectation of a scaling with @xmath0 rather than @xmath51 .
the figure shows results for the monotony criterion applied to the special case of @xmath52 , @xmath53 , @xmath54 , @xmath3 , @xmath55 .
the additional index @xmath37 refers to the order of truncation .
a similar figure is obtained for @xmath56 as function of @xmath25 for fixed order of the truncation and varying volumes .
thus the intersection of the curves with the @xmath25-axis , which localizes the tricritical point , weakly depends on the order of the truncation and the choice of @xmath21 .
the ultimate determination of the tricritical couplings @xmath57 and @xmath58 needs an extrapolation in both parameters . respecting the mutual relation between the truncation effect and the finite volume dependence of the intersection , the final estimate for @xmath57 leads to @xmath59 . for further details on the involved extrapolations we refer to @xcite .
this result improves the infinite volume estimate of @xmath60 by two orders of magnitude .
it should be noticed that the volume independence of @xmath29 along the tricritical line supports the validity of a mean field analysis of tricritical exponents even in three dimensions as one might have expected from the ginzburg criterion .
the localization of @xmath57 from the change in the shape of the effective potential leads to an upper bound on @xmath57 , we find @xmath61 , since the involved systematic errors are less well under control .
an extension to _ effective _ scalar models for an underlying salam - weinberg theory with hpes in a finite volume will particularly well work in the range of large higgs masses .
it complements the range of higgs masses which has been available in recent monte carlo simulations and allows a determination of the _ critical _ higgs mass above which the electroweak transition turns into a smooth crossover phenomenon @xcite . | linked cluster expansions are generalized from an infinite to a finite volume on a @xmath0-dimensional hypercubic lattice .
they are performed to 20th order in the expansion parameter to investigate the phase structure of scalar @xmath1 models for the cases of @xmath2 and @xmath3 in 3 dimensions . in particular
we propose a new criterion to distinguish first from second order transitions via the volume dependence of response functions for couplings close to but not at the critical value .
the criterion is applicable to monte carlo simulations as well .
here it is used to localize the tricritical line in a @xmath4 theory .
we indicate further applications to the electroweak transition . |
planetary and proto - planetary nebulae ( pne , ppne ) present conspicuous departures from spherical symmetry , including e.g. multiple lobes and jets . to explain their evolution from spherical agb envelopes ,
several models have postulated the presence of dense rings or disks close to the central post - agb stars as the agents of the mechanical collimation of the stellar wind .
existing observations reveal the presence of central disks in several ppne , but their limited spatial resolution can not unveil the very inner regions of the disks that are relevant for the processes mentioned above .
our first vlba observations of sio masers in oh231.8 + 4.2 , carried out at 7 mm ( @xmath0=2 , @xmath1=10 ) in april 2000 , revealed for the first time the structure and kinematics of the close stellar environment in a ppn ( sanchez contreras et al . , 2002 ) .
the sio maser emission arises in several compact , bright spots forming a structure elongated in the direction perpendicular to the symmetry axis of the nebula .
such a distribution is consistent with an equatorial torus with a radius of @xmath26 au around the central star .
a complex velocity gradient was found along the torus , which suggests rotation and infall of material towards the star .
the rotation and infalling velocities deduced are of the same order and range between @xmath27 and @xmath210kms@xmath3 .
such a distribution is remarkably different from that found in other late - type stars , where the masers form a roughly spherical ring - like chain of spots resulting from tangential maser amplification in a thin , spherical shell ( diamond et al . , 1994 , desmurs et al . , 2000 ) .
the new vlba observations at 3 mm were observed on april 2002 .
the data were recorded in dual circular polarization over a bandwidth of 16 mhz , assuming a rest frequency of 86243.442 mhz for the @xmath0=1 @xmath1=21 sio maser emission , for a final spectral resolution of about 0.1 km s@xmath3 .
we frequently observed several calibrators for band - pass and phase calibration ; pointing was checked and corrected every less than half an hour .
the data reduction followed the standard scheme for spectral line calibration data in aips . to improve the phase calibration solutions , we extensively map all the calibrators , and their brightness distribution
were introduced and taken into account in the calibration process .
our vlba maps at 3 mm ( fig .
1 ) show two sio maser spots that are globally aligned with those at 7 mm , i.e. , they are roughly distributed orthogonally to the nebular symmetry axis .
the velocities of the two maser spots observed at 3 mm are in good agreement with velocities observed at 7 mm .
the spatial distribution and velocity structure of the sio maser spots at 3 mm and 7 mm are consistent with a torus with radius @xmath46a.u@xmath5 , perpendicular to the nebular symmetry axis ( inclined @xmath6 with respect to the plane of the sky ) .
this torus - like structure is not totally traced by the maser spots .
such a spot distribution is expected , even for a homogeneous torus , if the maser is tangentially amplified .
this amplification mechanism leads to the most intense maser features at the edges of the torus - like structure .
moreover , given the comparable size of the star and the maser emitting region , occultation by the star of the far side of the torus is very likely , which would partially explain the small number of detected spots . in our opinion ,
the structure traced by the sio maser emission in oh231.8 + 4.2 , compatible with a rotating+infalling torus very close to the central star , is unlikely an accretion disk .
the dynamic is well reproduced by a simple model considering the presence of an inner torus in rotation and infall with a radius of @xmath2 5au ( [email protected] au ) , which means very close to the surface of the star ( sanchez contreras et al . , 2002 ) .
in fact , the dramatic changes with time of the sio @xmath0=2 , @xmath1=10 profile suggest that the masers lie in a region with an unstable structure and/or kinematics , rather than in a stable accretion disk .
moreover , accretion disks are expected around the compact companion in a binary system and less likely around the mass - losing star .
finally , it is worth mentioning that : ( a ) 3 mm maser spots are located further away from the central star than those at 7 mm , similarly to what has been recently found in agb stars ( see soria - ruiz et al . , these proceedings ) ; and ( b ) the rotation velocity measured in the torus of oh231.8 + 4.2 is consistent with the values of the rotation velocity found in inner envelopes of several agbs ( nml cyg : boboltz & marvel , 2000 ; tx cam and irc+10011 : desmurs et al .
, 20001 , r aqr : hollis et al . , 2001 ) .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ boboltz , d. a. & marvel , k. b. , 2000 , , 545 , 149 .
+ desmurs , j. f. , bujarrabal , v. , colomer , f. , & alcolea , j.2000 , , 360 , 189 .
+ desmursj.f . , v. bujarrabal , f. colomer & j. alcolea , 2001 , iau symposium 206 , eds .
v. migenes et al . ,
+ hollis , j. m. , boboltz , d. a. , pedelty , j. a. , white , s. m. & forster , j. r. 2001 , , 559 , l37-l40 .
+ sanchez contreras , c. , desmurs , j. f. , bujarrabal , v. , colomer , f. , alcolea , j. , 2002 , a&a 385 , l1-l4 .
+ soria - ruiz r. , colomer f. , alcolea j. , desmurs j.- f. , bujarrabal v. , marvel k.b . ,
diamond p.j . , & boboltz d. , 2003 , these proceedings .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ | we present milliarsecond - resolution maps of the sio maser emission @xmath0=1 j=21 ( 3 mm ) in the bipolar post - agb nebula oh231.8 + 4.2 , and compare them with our previous observations of the @xmath0=2 j=10 line ( 7 mm ) .
our observations show that the sio masers arise in several bright spots forming a structure elongated in a direction perpendicular to the symmetry axis of the nebula .
this , and the complex velocity gradient observed , is consistent with the presence of an equatorial torus in rotation and with an infall of material towards the star .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
analyzing spike trains from hundreds of neurons to find significant temporal patterns is an important current research problem @xcite . by using experimental techniques such as micro electrode arrays or imaging of neural currents ,
spike data can be recorded simultaneously from many neurons @xcite .
such multi - neuronal spike train data can now be routinely gathered _ in vitro _ from neural cultures or _ in vivo _ from brain slices , awake behaving animals and even humans . such data would be a mixture of stochastic spiking activities of individual neurons as well as that due to correlated activity of groups of neurons due to interconnections , possibly triggered by external inputs . automatically discovering patterns ( regularities ) in these spike trains can lead to better understanding of how interconnected neurons act in a coordinated manner to generate specific functions .
there has been much interest in techniques for analyzing the spike data so as to infer functional connectivity or the functional relationships within the system that produced the spikes @xcite .
in addition to contributing towards our knowledge of brain function , understanding of functional relations embedded in spike trains leads to many applications , e.g. , better brain - machine interfaces .
such an analysis can also ultimately allow us to systematically address the question , `` is there a neural code ? '' . in this paper , we consider the problem of discovering _ statistically significant _ patterns from multi - neuronal spike train data .
the patterns we consider here are ordered firing sequences by a group of neurons with specific time - lags or delays between successive neurons .
such a pattern ( when it repeats many times ) may denote a chain of triggering events and hence unearthing such patterns from spike data can help understand the underlying functional connectivity .
for example , memory traces are probably embedded in such sequential activation of neurons and signals of this form have been found in hippocampal neurons @xcite .
such patterns of ordered firing sequences with fairly constant delays between successive neuronal firings have been observed in many experiments and there is much interest in detecting such patterns and assessing their statistical significance .
( see @xcite and references therein ) . here , we will call patterns of ordered firing sequences as _
sequential patterns_. symbolically , we denote such a pattern as , e.g. , @xmath0 .
this represents the pattern of ordered firing sequence of @xmath1 followed by @xmath2 followed by @xmath3 with a delay of @xmath4 time units between @xmath1 & @xmath2 and @xmath5 time units between @xmath2 & @xmath3 .
( we note here that within any occurrence of such a firing pattern , there could be spikes by other neurons ) .
such a pattern of firings may occur repeatedly in the spike train data if , e.g. , there is an excitatory influence of total delay @xmath4 from @xmath1 to @xmath2 and an excitatory influence of delay @xmath5 between @xmath2 and @xmath3 .
in general , the delays may not be exactly constant because synaptic transmission etc .
could have some random variations .
hence , in our sequential patterns , we will allow the delays to be intervals of small length . at the least
, we can take the length of the interval as the time resolution in our measurements .
in general , such patterns can involve more than three neurons .
the _ size _ of a pattern is the number of neurons in it .
thus , the above example is that of a size 3 pattern or a 3-node pattern .
one of the main computational methods for detecting such patterns that repeat often enough , is due to abeles and gerstein @xcite .
this essentially consists of sliding the spike train of one neuron with respect to another and noting coincidences at specific delays .
there are also some recent variations of this method @xcite .
most of the current methods for detecting such patterns essentially use correlations among time - shifted spike trains ( and some statistics computed from the correlation counts ) , and these are computationally expensive when detecting large - size ( typically greater than 4 ) patterns @xcite . another approach to detecting such ordered firing sequences
is considered in @xcite while analyzing recordings from hippocampal neurons .
given a specific ordering on a set of neurons , they look for longest sequences in the data that respect this order .
this is similar to our sequential patterns which are somewhat more general because we can also specify different delays between consecutive elements of the pattern . in this paper
we use a method based on some temporal datamining techniques that we have recently proposed @xcite .
this method can automatically detect all sequential patterns whose _ frequency _ in the data is above a ( user - specified ) threshold where _ frequency _ of the pattern is maximum number of non - overlapped occurrences of the pattern in the spike data .
the essence of this algorithm is that instead of trying to count all occurrences of the pattern in the data we count only certain well - defined subset of occurrences and this makes the process computationally efficient .
the method is effective in detecting long patterns and it would detect only those patterns that repeat more than a given threshold . also , the method can automatically decide on the most appropriate delays in each detected pattern by choosing from a set of possible delays supplied by the user .
( see @xcite for details ) .
the main contribution of this paper is a method for assessing the statistical significance of such sequential patterns .
the objective is to have a method so that we will detect only those patterns that repeat often enough to be significant ( and thus fix the thresholds for the data mining algorithm automatically ) .
we tackle this issue in a classical hypothesis testing framework .
there have been many approaches for assessing the significance of detected firing patterns @xcite . in the current analytical approaches , one generally employs a null hypothesis that the different spike trains are generated by independent processes . in most cases
one assumes ( possibly inhomogeneous ) bernoulli or poisson processes .
then one can calculate the probability of observing the given number of repetitions of the pattern ( or of any other statistic derived from such counts ) under the null hypothesis of independent processes and hence calculate a minimum number of repetitions needed to conclude that a pattern is significant in the sense of being able to reject the null hypothesis .
there are also some empirical approaches suggested for assessing significance @xcite .
here one creates many surrogate data streams from the experimentally observed data by perturbing the individual spikes while keeping certain statistics same and then assessing significance by noting whether or not the patterns are preserved in the surrogate data .
there are many possibilities for the perturbations to be imposed to generate surrogate data @xcite . in these empirical methods
also , the implicit null hypothesis assumes independence .
the main motivation for the approach presented here is the following .
if a sequential pattern repeats often enough to be significant then one would like to think that there are strong influences among the neurons representing the pattern .
however , different ( detected ) patterns may represent different levels or strengths of influences among their constituent neurons .
hence it would be nice to have a method of significance analysis that can rank order different ( significant ) patterns in terms some ` strength of influence ' among the neurons of the pattern .
for this , here we propose that the strength of influence of @xmath1 on @xmath2 is well represented by the conditional probability that @xmath2 will fire after some prescribed delay given that @xmath1 has fired .
we then employ a composite null hypothesis specified through one parameter that denotes an upper bound on all such pairwise conditional probabilities . using this
we would be able to decide whether or not a given pattern is significant at various values for this parameter in the null hypothesis and thus be able to rank - order different patterns .
there is an additional and important advantage of the above approach that we propose here .
our composite null hypothesis is such that any stochastic model for a set of spiking neurons would be in the null hypothesis if all the relevant pairwise conditional probabilities are below some bound . since
this bound is a parameter that can be chosen by the user , the null hypothesis would include not only independent neuron models but also many models of interdependent neurons where the pair - wise influences among neurons are ` weak ' . hence rejecting such a null hypothesis is more attractive than rejecting a null hypothesis of independence when we want to conclude that a significant pattern indicates ` strong ' interactions among the neurons . in this sense ,
the approach presented here extends the currently available methods for significance analysis .
we analytically derive some bounds on the probability that our counting process would come up with a given number of repetitions of the firing pattern if the data is generated by any model that is contained in our compound null hypothesis .
as mentioned earlier , we use the number of non - overlapped occurrences of a pattern as our test statistic instead of the total number of repetitions and employ a temporal datamining algorithm for counting non - overlapped occurrences of sequential patterns @xcite .
this makes our method attractive for discovering significant patterns involving large number of neurons also .
we show the effectiveness of the method through extensive simulation experiments on synthetic spike train data obtained through a model of inter - dependent non - homogeneous poisson processes . the rest of the paper is organized as follows . in section [ sec
: epi ] we give a brief overview of temporal datamining and explain our algorithm for detecting sequential patterns whose frequency is above some threshold .
the full details of the algorithm are available elsewhere @xcite and we provide only some details which are relevant for understanding the statistical significance analysis which is presented in section [ sec : stat ] . in section [ sec : simu ] , we present some simulation results on synthetic spike train data to show the effectiveness of our method .
we present results to show that our method is capable of ranking different patterns in terms of the synaptic efficacy of the connections .
while we confine our attention in this paper to only sequential patterns , the statistical method we present can be generalized to handle other types of patterns .
we briefly indicate this and conclude the paper with a discussion in section [ sec : dis ] .
temporal datamining is concerned with analyzing symbolic time series data to discover ` interesting ' patterns of temporal dependencies @xcite .
recently we have proposed that some datamining techniques , based on the so called frequent episodes framework , are well suited for analyzing multi - neuronal spike train data @xcite .
many patterns of interest in spike data such as synchronous firings by groups of neurons , the sequential patterns explained in the previous section , and synfire chains which are a combination of synchrony and ordered firings , can be efficiently discovered from the data using these datamining techniques . while the algorithms are seen to be effective through simulations presented in @xcite , no statistical theory was presented there to address the question of whether the detected patterns are significant in any formal sense which is the main issue addressed in this paper . in this section
we first briefly outline the frequent episodes framework and then qualitatively describe this datamining technique for discovering frequently occurring sequential patterns . in the frequent episodes framework of temporal datamining .
the data to be analyzed is a sequence of events denoted by @xmath6 where @xmath7 represents an _ event type _ and @xmath8 the _ time of occurrence _ of the @xmath9 event .
@xmath10 s are drawn from a finite set of event types , @xmath11 .
the sequence is ordered with respect to time of occurrences of the events so that , @xmath12 , @xmath13 .
the following is an example event sequence containing 11 events with 5 event types .
@xmath14 in multi - neuronal spike data , the event type of a spike event is the label of the neuron which generated the spike and the event has the associated time of occurrence .
the neurons in the ensemble under observation fire action potentials at different times , that is , generate spike events .
all these spike events are strung together , in time order , to give a single long data sequence as needed for frequent episode discovery .
it may be noted that there can be more than one event with the same time because two neurons can spike at the same time .
the temporal patterns that we wish to discover in this framework are called episodes . in general , episodes are partially ordered sets of event types . here
we are only interested in _ serial episodes _ which are totally ordered .
a _ serial episode _ is an ordered tuple of event types .
for example , @xmath15 is a _ 3-node _ serial episode .
( we also say that the size of this episode is 3 ) .
the arrows in this notation indicate the order of the events .
such an episode is said to _ occur _ in an event sequence if there are corresponding events in the prescribed order in the data sequence . in sequence ( [ eq : data - seq ] ) , the events \{@xmath16 } constitute an occurrence of the serial episode @xmath17 while the events \{@xmath18 } do not .
we note here that occurrence of an episode does not require the associated event types to occur consecutively ; there can be other intervening events between them . in the multi - neuronal data ,
if neuron @xmath1 makes neuron @xmath2 to fire , then , we expect to see @xmath2 following @xmath1 often
. however , in different occurrences of such a substring , there may be different number of other spikes between @xmath1 and @xmath2 because many other neurons may also be spiking during this time .
thus , the episode structure allows us to unearth patterns in the presence of such noise in spike data .
the objective in frequent episode discovery is to detect _ all _ frequent episodes ( of different lengths ) from the data .
frequent episode _ is one whose _
frequency _ exceeds a ( user specified ) _
frequency threshold_. the frequency of an episode can be defined in many ways .
it is intended to capture some measure of how often an episode occurs in an event sequence .
one chooses a measure of frequency so that frequent episode discovery is computationally efficient and , at the same time , higher frequency would imply that an episode is occurring often . here , we define frequency of an episode as the maximum number of non - overlapped occurrences of the episode in the data stream . two occurrences of an episode are said to be _ non - overlapped _
if no event associated with one occurrence appears in between the events associated with the other .
a set of occurrences is said to be non - overlapped if every pair of occurrences in it are non - overlapped . in our example sequence ( [ eq : data - seq ] ) , there are two non - overlapped occurrences of @xmath19 given by the events : @xmath20 and @xmath21 .
note that there are three distinct occurrences of this episode in the data sequence though we can have only a maximum of two non - overlapped occurrences .
we also note that if we take the occurrence of the episode given by @xmath22 , then there is no other occurrence that is non - overlapped with this occurrence .
that is why we define the frequency to be the maximum number of non - overlapped occurrences .
this definition of frequency results in very efficient counting algorithms with some interesting theoretical properties @xcite .
in addition , in the context of our application , counting non - overlapped occurrences seems natural because we would then be looking at chains that happen at different times again and again . in analyzing neuronal spike data ,
it is useful to consider methods , where , while counting the frequency , we include only those occurrences which satisfy some additional temporal constraints . here
we are interested in what we call inter - event time constraint which is specified by giving an interval of the form @xmath23 $ ] .
the constraint requires that the difference between the times of every pair of successive events in any occurrence of a serial episode should be in this interval . in general
, we may have different time intervals for different pairs of events in each serial episode . as is easy to see , a serial episode with inter - event time constraints corresponds to what we called a
_ sequential pattern _ in the previous section .
these are the temporal patterns of interest in this paper .
the inter - event time constraint allows us to take care of delays involved in the process of one neuron influencing another through a synapse .
suppose neuron @xmath1 is connected to neuron @xmath2 which , in turn , is connected to neuron @xmath3 , through excitatory connections with delays @xmath4 and @xmath5 respectively . then , we should be counting only those occurrences of the episode @xmath24 , where the inter - event times satisfy the delay constraint .
this would be the sequential pattern @xmath25 .
in general , the inter - event constraint could be an interval .
occurrences of such serial episodes with inter - event constraints in spike data are shown schematically in fig .
[ fig : ser - epi - fig ] in the spike trains from neurons @xmath26 . a small interval ( usually 1 ms ) is shown around the second and third spike to indicate possible variation in the delay .
note that within the duration of one occurrence of the pattern there may be other intervening spikes ( from any of the neurons ) . ] in any occurrence of the episode or sequential pattern , we call the difference between the times of the first and last events as its _ span_. the span would be the total of all the delays . if , in the above episode , the span of all occurrences would be @xmath27 and hence we may call it the span of the episode . if the inter - event time constraints are intervals then the span of different occurrences could be different .
there are efficient algorithms for discovering all frequent serial episodes with specified inter - event constraints @xcite .
that is , for discovering all episodes whose frequency ( which is the number of non - overlapped occurrences of the episode ) is above a given threshold .
_ conceptually , the algorithm does the following_. suppose , we are operating at a time resolution of @xmath28 .
( that is , the times of of events or spikes are recorded to a resolution of @xmath28 ) .
then we discretize the time axis into intervals of length @xmath28 .
then , for each episode whose frequency we want to find we do the following .
suppose the episode is the one mentioned above .
we start with time instant 1 .
we check to see whether there is an occurrence of the episode starting from the current instant . for this , we need an @xmath1 at that time instant and then we need a @xmath2 and a @xmath3 within appropriate time windows . if there are such @xmath2 and @xmath3 , then we take the earliest of the @xmath2 and @xmath3 to satisfy the time constraints , increment the counter for the episode and start looking for the occurrence again starting with the next time instant ( after @xmath3 ) . on the other hand ,
if we can not find such an occurrence ( either because @xmath1 does not occur at the current time instant or because there are no @xmath2 or @xmath3 at appropriate times following @xmath1 ) , then we move by one time instant and start the search again
. the actual search process would be very inefficient if implemented as described above .
the algorithm itself does the search in a much more efficient manner .
there are two issues that the algorithm needs to address .
since , a priori , we do not know what patterns to look for , we need to make a reasonable list of candidate patterns and then obtain their frequencies so as to output only those patterns whose frequency exceeds the preset threshold .
the second issue is that in obtaining frequencies , the algorithm is required to count the frequencies of not one but a set of candidates in one pass through the data and we need to do this efficiently . in generating the candidates , we need to tackle the combinatorial explosion because all possible serial episodes of a given size increases exponentially with the size .
this is tackled using an iterative procedure that is popular in datamining . to understand this ,
consider our example 3-node pattern @xmath25 .
this can not be frequent unless certain 2-node _ subepisodes _ of this , namely , @xmath29 and @xmath30 are frequent .
( this is because any two non - overlapped occurrences of the 3-node pattern also gives us two non - overlapped occurrences of the two 2-node patterns mentioned above ) .
thus , we should allow this 3-node episode to be a candidate only if the appropriate 2-node episodes are already found to be frequent . based on this idea
, we have the following structure for the algorithm .
we first get frequent 1-node episodes which are then used to make candidate 2-node episodes . then
, by one more pass over data , we find frequent 2-node episodes which are then used to make candidate 3-node episodes and so on .
such a technique is quite effective in controlling combinatorial explosion and the number of candidates comes down drastically as the size increases .
this is because , as the size increases , many of the combinatorially possible serial episodes of that size would not be frequent .
this allows the algorithm to find large size frequent episodes efficiently . at each stage of this process , we count frequencies of not one but a whole set of candidate episodes ( of a given size ) through one sequential pass over the data
. we do not actually traverse the time axis in time ticks once for each pattern whose occurrences we want to count .
we traverse the time - ordered data stream .
as we traverse the data we remember enough from the data stream to correctly take care of all the occurrence possibilities of all episodes in the candidate set and thus compute all the frequent episodes of a given size through one pass over the data .
the complete details of the algorithm are available in @xcite .
in this section we address the issue of the statistical significance of the sequential patterns discovered by our algorithm .
the question is when are the discovered episodes significant , or , equivalently , what frequency threshold should we choose so that all discovered frequent episodes would be statistically significant . to answer this question
we follow a classical hypothesis testing framework .
intuitively we want significant sequential patterns to represent a chain of strong interactions among those neurons .
so , we have to essentially choose a _ null hypothesis _ that asserts that there is no ` _ structure _ ' or ` _ strong influences _ ' in the system of neurons generating the data .
also , as mentioned earlier , we want the null hypothesis to contain a parameter that allows us to specify what we mean by saying that the influence one neuron has on another is not ` strong ' . for this , we capture the strength of interactions among the neurons in terms of conditional probabilities . let @xmath31 denote the conditional probability that @xmath2 fires in a time interval @xmath32 $ ] given that @xmath1 fired at time zero .
@xmath28 is essentially the time resolution at which we operate .
( for example , @xmath28 = 1ms ) .
thus , @xmath31 is essentially , the conditional probability that @xmath2 fires @xmath33 time units after @xmath1 .
, as a constant
. however , in practice our method can easily take care of the case where the actual delay is uniformly distributed over a small interval with @xmath33 as its expected value . ]
if there is a strong excitatory synapse of delay @xmath33 between @xmath1 and @xmath2 , then this conditional probability would be high . on the other hand
if @xmath1 and @xmath2 are independent , then , this conditional probability is the same as the unconditional probability of @xmath2 firing in an interval of length @xmath28 .
we denote the ( unconditional ) probability that a neuron , @xmath1 , fires in any interval of length @xmath28 by @xmath34 .
( for example , if we take @xmath35 and that the average firing rate of @xmath2 is 20hz , then @xmath36 would be about 0.02 ) .
the main assumption we make is that the conditional probability @xmath37 is not a function of time .
that is , the conditional probability of @xmath2 firing at least once in an interval @xmath38 $ ] given that @xmath1 has fired at @xmath39 is same for all @xmath39 within the time window of the observations ( data stream ) that we are analyzing .
we think this is a reasonable assumption and some recent analysis of spike trains from neural cultures suggests that such an assumption is justified @xcite .
note that this assumption does not mean we are assuming that the firing rates of neurons are not time - varying . as a matter of fact ,
one of the main mechanisms by which this conditional probability is realized is by having a spike from @xmath1 affect the rate of firing by @xmath2 for a short duration of time .
thus , the neurons would be having time - varying firing rates even when the conditional probability is not time - varying .
essentially , the constancy of @xmath40 would only mean that every time @xmath1 spikes , it has the same chance of eliciting a spike from @xmath2 after a delay of @xmath33 .
thus our assumption only means that there is no appreciable change in synaptic efficacies during the period in which the data being analyzed is gathered .
the intuitive idea behind our null hypothesis is that the conditional probability @xmath37 is a good indicator of the ` strength of interaction ' between @xmath1 and @xmath2 . for inferring functional connectivity from repeating sequential patterns
, the constancy of delays ( between spikes by successive neurons ) in multiple repetitions is important .
that is why we defined the conditional probability with respect to a specific delay .
now , an assertion that the interactions among neurons is ` weak ' can be formalized in terms of an upper bound on all such conditional probabilities .
we formulate our composite null hypothesis as follows . _
our composite null hypothesis includes all models of interacting neurons for which we have @xmath41 for all pairs of neurons @xmath42 and for a set of specified delays @xmath33 , where @xmath43 is a fixed user - chosen number in the interval @xmath44 .
_ thus all models of inter - dependent neurons where the probability of @xmath1 causing @xmath2 to fire ( after a delay ) is less that @xmath43 , would be in our null hypothesis .
the actual mechanism by which spikes from @xmath1 affect the firing by @xmath2 is immaterial to us .
whatever may be this mechanism of interaction , if the resulting conditional probability is less than @xmath43 , then that model of interacting neurons would be in our null hypothesis . could be taken as a typical delay involved in the process ; otherwise it can be taken as some integral multiple of such delays . in any case , our interest is in deciding on the significance of sequential patterns with some given values for @xmath33 . ]
the user specified number , @xmath43 , formalizes what we mean by interaction among neurons is strong .
if @xmath1 and @xmath2 are independent then this conditional probability is same as @xmath36 . as mentioned earlier , if @xmath35 and average firing rate for @xmath2 as 20 hz , then @xmath45 .
so , if we choose @xmath46 , it means that we agree to call the influence as strong if the conditional probability is 20 times what it would be if the neurons are independent . by having different values for @xmath43 in the null hypothesis , we can ask what patterns are significant at what value of @xmath43 and thus rank - order patterns . now
if we are able to reject this null hypothesis then it is reasonable to assert that the episode(s ) discovered would indicate ` strong ' interactions among the appropriate neurons .
the ` strength ' of interaction is essentially chosen by us in terms of the bound @xmath43 on the conditional probability in our null hypothesis .
we now present a method for bounding the probability that the frequency ( number of non - overlapped occurrences ) of a given serial episode with inter - event constraints is more than a given threshold under the null hypothesis . to do this ,
we first compute the expectation and variance ( under the null hypothesis ) of the random variable representing the number of non - overlapped occurrences of a serial episode with inter - event constraints by using the following stochastic model .
let @xmath47 be _ iid _ random variables with distribution given by @xmath48 & = & p \nonumber \\
p[x_i = 1 ] & = & 1 - p \label{eq : xi}\end{aligned}\ ] ] where @xmath33 is a fixed constant ( and @xmath49 ) .
let @xmath50 be a random variable defined by @xmath51 where @xmath52 is a fixed constant .
let the random variable @xmath53 denote the number of @xmath54 s out of the first @xmath50 which have value @xmath33 .
define the random variable @xmath55 by @xmath56 all the random variables , @xmath57 depend on the parameters @xmath58 . when it is important to show this dependency we write @xmath59 and so on .
now we will argue that @xmath60 is the random variable representing the number of non - overlapped occurrences of an episode where @xmath33 is the span ( or sum of all delays ) of the episode and @xmath52 is the length of data ( in terms of time duration ) .
we would fix @xmath61 based on the bound @xmath43 in our null hypothesis as explained below .
superimposed on the spike trains from neurons @xmath1 and @xmath2 . in the yellow region
there are no occurrences of the pattern starting with that time instant and the counting scheme moves forward by one time step . in the blue region
there is an occurrence and the counting process moves by @xmath33 time steps .
the random variables @xmath54 , defined by eq .
( [ eq : xi ] ) , capture the evolution of the counting process ] consider an episode @xmath62 with an inter - event time constraint ( or delay ) of @xmath33 .
now , the sequence @xmath54 essentially captures the counting process of our algorithm .
a schematic of the counting process ( as relevant for this discussion ) is shown in fig .
[ fig : counting ] .
as explained earlier , the algorithm can be viewed as traversing a _ discretized _ time axis in steps of @xmath28 , looking for an occurrence of the episode starting at each time instant . at each time instant ( which , on the discretized time axis corresponds to an interval of length @xmath28 ) , let @xmath63 denote the probability of spiking by @xmath1 and let @xmath64 denote the conditional probability that @xmath2 generates a spike @xmath33 instants later given that @xmath1 has spiked now . in terms of our earlier notation , @xmath65 , @xmath66 .
thus , at any instant , @xmath67 denotes the probability of occurrence of the episode starting at that instant .
now , in eq.([eq : xi ] ) let @xmath68 .
then @xmath61 represents the probability that this episode occurs starting with any given time instant . because we do not know the exact value for @xmath37 .
but finally we would bound the relevant probability by using @xmath43 to bound @xmath37 . ]
let @xmath52 in eq.([eq : n ] ) denote the data length ( in time units ) .
then the sequence , @xmath69 , represents our counting process .
if , at the first instant there is an occurrence of the episode starting at that instant then we advance by @xmath33 units on the time - axis and then look for another occurrence ( since we are counting non - overlapped occurrences ) ; if there is no occurrence starting at the first instant then we advance by one unit and look for an occurrence . also , whether or not there is an occurrence starting from the current instant is independent of how many occurrences are completed before the current instant ( because we are counting only non - overlapped occurrences )
so , the counting process is well captured by accumulating the @xmath54 s defined above till we reach the end of data .
hence @xmath50 captures the number of such @xmath54 that we accumulate because @xmath52 is the data length in terms of time . since @xmath54 take values @xmath70 or @xmath33 , the only way @xmath71 exceeds @xmath52 is if the last @xmath54 takes value @xmath33 which in turn implies that when we reached end of data we have a partial occurrence of the episode . in this case
the total number of completed occurrences is one less than the number of @xmath54 ( out of @xmath50 ) that take value @xmath33 .
if the last @xmath54 has taken value 1 ( and hence the sum is equal to @xmath52 ) then the number of completed occurrences is equal to the number of @xmath54 that take value @xmath33 .
now , it is clear that @xmath55 is the number of non - overlapped occurrences counted .
it is easy to see that the model captures counting of episodes of arbitrary length also .
for example , if our episode is @xmath0 then @xmath33 is eq.([eq : xi ] ) would be @xmath72 and @xmath61 would be @xmath73 . after a delay of @xmath5 from @xmath2 , conditioned on firing of @xmath2 , is conditionally independent of earlier firing of @xmath1 . since our objective is to unearth significant triggering chains , this is a reasonable assumption .
also , this allows us to capture the null hypothesis with a single parameter @xmath43 .
we discuss this further in section [ sec : dis ] . ]
suppose in a @xmath74-node episode the conditional probability of @xmath75 neuron firing ( after the prescribed delay ) given that the previous one has fired , is equal to @xmath76 .
let the successive delays be @xmath77 .
let the ( unconditional ) probability of the first neuron ( of the episode ) firing at any instant ( that is , in any interval of length @xmath28 ) is @xmath78
. then we will take ( for the @xmath74-node episode ) @xmath79 and @xmath80 .
now , we first derive some recurrence relations to calculate the mean and variance of @xmath59 for a given episode . fixing an episode fixes the value of @xmath61 and @xmath33 .
let @xmath81 where @xmath82 denotes expectation .
we can derive a recurrence relation for @xmath83 as follows .
@xmath84 \ ; \right ] \nonumber \\ & = & e\ : [ m(l , t , p ) \ ; | \ ; x_1 = 1 ] ( 1-p ) \
; + \ : e[m(l , t , p ) \ ; | \ ; x_1 \neq 1 ] p \nonumber \\ & = & ( 1-p ) e\:[m(l-1 , t , p ) ] \ : + \ : p ( 1 \ : + \ :
e[m(l - t , t , p ) ] ) \nonumber \\ & & \end{aligned}\ ] ] in words what this says is : if the first @xmath54 is @xmath70 ( which happens with probability @xmath85 ) , then the expected number of occurrences is same as those in data of length @xmath86 ; on the other hand , if first @xmath54 is not @xmath70 ( which happens with probability @xmath61 ) then the expected number of occurrences are 1 plus the expected number of occurrences in data of length @xmath87 . hence our recurrence relation is : @xmath88 the boundary conditions for this recurrence are : @xmath89 let @xmath90 $ ] .
that is @xmath91 is the second moment of @xmath60 . using the same idea as in case of @xmath83 we can derive recurrence relation for @xmath92 as follows .
@xmath93 & = & e\ : \left [ e\ : [ m^2(l , t , p ) \ ; | \ ; x_1 ] \right ] \nonumber \\ & = & e\ : [ m^2(l , t , p ) \ ; | \ ; x_1 = 1 ] ( 1-p ) \ ; \nonumber \\ & & \ \ \ \ \ \ \ \ \ \ + \ : e[m^2(l , t , p ) \ ; | \ ; x_1 \neq 1 ] p \nonumber \\ & = & ( 1-p ) e\:[m^2(l-1 , t , p ) ] \ : + \ : p e ( 1 \ : + \ : m(l - t , t , p ) ) ^2 \nonumber \\ & = & ( 1-p ) e\:[m^2(l-1 , t , p ) ] \ : + \nonumber \\ & & \ \ \ \ \ \ \ \ \ \ \ : p e ( 1 \ : + \ : m^2(l - t , t , p ) \ : +
\ : 2 m(l - t , t , p ) ) \nonumber \\ & & \end{aligned}\ ] ] thus we get @xmath94 solving the above , we get the second moment of @xmath55 .
let , @xmath95 be the variance of @xmath96 .
then we have @xmath97 once we have the mean and variance we can bound the probability that the number of non - overlapped occurrences is beyond something .
for example , we can use chebyshev inequality as @xmath98 \leq \frac{1}{k^2 } \label{eq : cheb}\ ] ] for any positive @xmath99 .. such bounds can be used for test of statistical significance as explained below .
suppose we are considering @xmath74-node episodes .
let the allowable type i error for the test be @xmath100 .
then what we need is a threshold , say , @xmath101 for which we have @xmath102 \leq \epsilon , \label{eq : test1}\ ] ] where @xmath103 is the frequency of any @xmath74-node episode and @xmath104 denotes probability under the null hypothesis models .
this would imply that if we find a @xmath74-node episode with frequency greater than @xmath101 then , with @xmath105 confidence we can reject our null hypothesis and hence assert that the discovered episode represents ` strong ' interactions among those neurons .
now the above can be used for assessing statistical significance of any episode as follows .
suppose we are considering an @xmath74-node ( serial ) episode .
let the first node of this episode have event type @xmath1 .
( that is , it corresponds to neuron @xmath1 ) .
let @xmath34 be the probability that @xmath1 will spike in any interval of length @xmath28 .
( we will fix @xmath28 by the time resolution being considered ) .
let @xmath100 be the prescribed confidence level .
let @xmath99 be such that @xmath106 .
fix @xmath107 .
let @xmath33 be the sum of all inter - event delay times in the episode .
let @xmath52 be the total length of data ( as time span in units of @xmath28 ) .
our null hypothesis is that the conditional probability for any pair of neurons is less than @xmath43 .
further , our random variable @xmath55 is such that its probability of taking higher values increases monotonically with @xmath61 . hence , with the above @xmath61 , the probability of @xmath60 being greater than any value is an upper bound on the probability of the episode frequency being greater than that value under any of the models in our null hypothesis . thus , a threshold for significance is @xmath108 because , from eq .
( [ eq : cheb ] ) we have @xmath109 \leq \frac{1}{k^2 } \leq \epsilon .
\label{eq : cheb1}\ ] ] though we do not have closed form expressions for @xmath83 and @xmath110 , using our recurrence relations , we can calculate @xmath111 and @xmath95 for any given values of @xmath112 and hence can calculate the above threshold .
the only thing unspecified for this calculation is how do we get @xmath34 .
we can obtain @xmath34 by either estimating the average rate of firing for this neuron from the data or from other prior knowledge .
thus , we can use eq .
( [ eq : cheb1 ] ) either for assessing the significance of a specific @xmath74-node episode or for fixing a threshold of any @xmath74-node episode in our datamining algorithm . in either case , this allows us to deduce the ` strong connections ' ( if any ) in the neural system being analyzed by using our datamining method .
we can summarize the the test of significance as follows .
suppose the allowed type - i error is @xmath100 .
we choose integer @xmath99 such that @xmath113 .
suppose we want to assess the significance of a n - node sequential pattern with the total delay being @xmath33 based on its count .
suppose @xmath43 is the bound we use in our null hypothesis .
let @xmath52 be the total data length in time units .
let @xmath78 be the average firing rate of the first neuron in the data .
let @xmath114 .
we calculate @xmath111 and @xmath95 using ( [ eq : rec1 ] ) , ( [ eq : var - rec1 ] ) and ( [ eq : var ] )
. then the pattern is declared significant if its count exceeds @xmath115 .
_ we like to emphasize that the threshold frequency ( count ) given above for an episode to be significant ( and hence represent strong interactions ) is likely to be larger than that needed .
this is because it is obtained through a chebyshev bound which is often loose . _
thus , for example , if we choose @xmath46 then some strong connections which may result in the effective conditional probability value of up to 0.5 may not satisfy the test of significance at a particular significance level .
this , in general , is usual in any hypothesis testing framework . in practice
, we found that we can very accurately discover all connections whose strengths in terms of the conditional probabilities are about 0.2 more than @xmath43 at 5% confidence level . at @xmath116 ,
the threshold is about 4.5 standard deviations above the mean . in a specific application , for example ,
if we feel that three standard deviations above the mean is a good enough threshold , then correspondingly we will be able to discover even those connections whose effective conditional probability is only a little above @xmath43 .
this test of significance allows us to rank order the discovered patterns . for this
, we run our datamining method with different thresholds corresponding to different @xmath43 values .
then , by looking at the sets of episodes found at different @xmath43 values , we can essentially rank order the strengths of different connections in the underlying system . since any manner of assigning numerical values to strengths of connections
is bound to be somewhat arbitrary , this method of rank ordering different connections in terms of strengths can be much more useful in analyzing microcircuits .
we illustrate all these through our simulation experiments in section [ sec : simu ] . so far in this section
we have assumed that the individual delays and hence the span of an episode , @xmath33 , to be constant . in practice ,
even if delay is random and varies over a small interval around @xmath33 , the threshold we calculated earlier would be adequate .
in addition to this , it is possible to extend our model to take care of some random variations in such delays . since we have assumed that @xmath28 is the time resolution at which we are working , it is reasonable to assume that the delay @xmath33 is actually specified in units of @xmath28 .
then we can think of the delay as a random variable taking values in a set @xmath117 where @xmath118 is a small ( relative to t ) integer .
for example , suppose the delay is uniformly distributed over @xmath119 .
now we can change our model as follows : the @xmath47 will now be _ iid _ random variables with distribution @xmath120 & = & 1 - p \nonumber \\
\mbox{prob}[x_i = t-1 ] = \mbox{prob}[x_i = t ] = \mbox{prob}[x_i = t+1 ] & = & \frac{p}{3 } \nonumber \label{eq : xi - new}\end{aligned}\ ] ] where we now assume that @xmath121 .
we will define @xmath50 as earlier by eq .
( [ eq : n ] ) .
we will now define @xmath53 as the number of @xmath54 out of first @xmath50 that * do not * take value 1 . in terms of this @xmath53 , we will define @xmath55 as earlier by eq .
( [ eq : m ] ) .
now it is easy to see that our @xmath60 would again be the random variable corresponding to number of non - overlapped occurrences in this new scenario where there are random variations in the delays . now
the recurrence relation for @xmath111 would become @xmath122 the recurrence relation for variance of @xmath60 can also be similarly derived .
now , we can easily implement the significance test as derived earlier .
while the recurrence relations are a little more complicated , it makes no difference to our method of significance analysis because these recurrence relations are anyway to be solved numerically .
it is easy to see that this method can , in principle , take care of any distribution of the total delay ( viewed as a random variable taking values in a finite set ) by modifying the recurrence relation suitably .
in this section we describe some simulation experiments to show the effectiveness of our method of statistical significance analysis .
we show that our stochastic model properly captures our counting process and that the frequency threshold we calculate is effective for separating connections that are ` strong ' ( in the sense of conditional probabilities ) .
we also show that our frequency can properly rank order the strengths of connections in terms of conditional probabilities . as a matter of fact ,
our results provide good justification for saying that conditional probabilities provide a very good scale for denoting connection strengths . for all our experiments we choose
synthetically generated spike trains .
this is because then we know the ground truth about connection strengths and hence can test the validity of our statistical theory . for the simulations we use a data generation scheme where we model the spiking of each neuron as an inhomogeneous poisson process
on which is imposed an additional constraint of refractory period .
( thus the actual spike trains are not truly poisson even if we keep the rate fixed ) .
the inhomogeneity in the poisson process are due to the instantaneous firing rates being modified based on total input spikes received by a neuron through its synapses .
we have shown elsewhere @xcite that our datamining algorithms are very efficient in discovering interesting patterns of firings from spike trains and that we can discover patterns of more than ten neurons also . since in this paper
the focus is on statistical significance of the discovered patterns , we would not be presenting any results for showing the computational efficiency of the method .
we use a simulator for generating the spike data from a network of interconnected neurons .
let @xmath50 denote the number of neurons in the network .
the spiking of each neuron is modelled as an inhomogeneous poisson process whose rate of firing is updated at time intervals of @xmath28 .
( we normally take @xmath28 to be 1ms ) .
the neurons are interconnected by synapses and each synapse is characterized by a delay ( which is in integral multiples of @xmath28 ) and a weight which is a real number .
all neurons also have a refractory period .
the rate of the poisson process is varied with time as follows .
@xmath123 where @xmath124 is the firing rate of @xmath125 neuron at time @xmath126 , and @xmath127 are two parameters .
@xmath128 is the total input into @xmath125 neuron at time @xmath126 and it is given by @xmath129 where @xmath130 is the output of @xmath9 neuron ( as seen by the @xmath125 neuron ) at time @xmath126 and @xmath131 is the weight of synapse from @xmath9 to @xmath125 neuron .
@xmath130 is taken to be the number of spikes by the @xmath9 neuron in the time interval @xmath132 $ ] where @xmath133 represents the delay ( in units of @xmath28 ) for the synapse from @xmath134 to @xmath135 .
the parameter @xmath136 is chosen based on the dynamic range of firing rates that we need to span .
the parameter @xmath137 determines the ` background ' spiking rate , say , @xmath138 .
this is the firing rate of the @xmath125 neuron under zero input . after choosing a suitable value for @xmath136
, we fix the value of @xmath137 based on this background firing rate specified for each neuron .
we first build a network that has many random interconnections with low weight values and a few strong interconnections with large weight values .
we then generate spike data from the network and show how our method can detect all strong connections . to build the network
we specify the background firing rate ( which we normally keep same for all neurons ) which then fixes the value of @xmath137 in ( [ eq : lambda - update ] ) .
we specify all weights in terms of conditional probabilities . given a conditional probability we first calculate the needed instantaneous firing rate so that probability of at least one spike in the @xmath28 interval is equal to the specified conditional probability .
then , using ( [ eq : lambda - update ] ) and ( [ eq : input ] ) , we calculate the value of @xmath131 needed so that the receiving neuron ( @xmath135 ) reaches this instantaneous rate given that the sending neuron ( @xmath134 ) spikes once in the appropriate interval and assuming that input into the receiving neurons from all other neurons is zero .
we note here that the background firing rate as well as the effective conditional probabilities in our system would have some small random variations .
as said above , we fix @xmath137 so that on zero input the neuron would have the background firing rate .
however , all neurons would have synapses with randomly selected other neurons and the weights of these synapses are also random .
hence , even in the absence of any strong connections , the firing rates of different neurons keep fluctuating around the background rate that is specified .
since we choose random weights from a zero mean distribution , in an expected sense we can assume the input into a neuron to be zero and hence the average rate of spiking would be the background rate specified .
we also note that the way we calculate the effective weight for a given conditional probability is also approximate and we chose it for simplicity . if we specify a conditional probability for the connection from @xmath1 to @xmath2 , then , the method stated in the previous paragraph fixes the weight of connection so that the probability of @xmath2 firing at least once in an appropriate interval given that @xmath1 has fired is equal to this conditional probability _ when all other input into @xmath2 is zero_. but since @xmath2 would be getting small random input from other neurons also , the effective conditional probability would also be fluctuating around the nominal value specified . further , even if the random weights have zero mean , the fluctuations in the conditional probability may not have zero mean due to the nonlinear sigmoidal relationship in ( [ eq : lambda - update ] ) .
the nominal conditional probability value determines where we operate on this sigmoid curve and that determines the bias in the excursions in conditional probability for equal fluctuations in either directions in the random input into the neurons .
we consider this as a noise in the system and show that our method of significance analysis is still effective .
the simulator is run as follows .
first , for any neuron we fix a fraction ( e.g. , 25% ) of all other neurons that it is connected to .
the actual neurons that are connected to any neuron are then selected at random using a uniform distribution .
we fix the delays and background firing rates for all neurons .
we then assign random weights to connections by choosing uniformly from an interval . in our simulation experiments
we specify this range in terms of conditional probabilities .
for example suppose the background firing rate is 20 hz .
then with @xmath35 , the probability of firing in any interval of length @xmath28 is ( approximately ) 0.02 .
hence a conditional probability of 0.02 would correspond to a weight value of zero .
then a range of conditional probabilities such as @xmath139 $ ] ( increase or decrease by a factor of 2 in either direction ) would correspond to a weight range around zero .
after fixing these random weights , we incorporate a few strong connections which vary in different simulation experiments .
these weight values are also specified in terms of conditional probabilities .
we then generate a spike train by simulating all the inhomogeneous poisson processes where rates are updated every @xmath28 time instants .
we also fix refractory period for neurons ( which is same for all neurons ) .
once a neuron is fired , we will not let it fire till the refractory period is over . for the results reported here we used a network of 100 neurons with the nominal firing rate being 20 hz .
each neuron is connected to 25 randomly selected neurons with the effective conditional probability of the connection strength ranging over @xmath139 $ ] . with 20hz firing rate and 1ms time resolution , the effective conditional probability when two neurons are independent is 0.02 .
thus the random connections have conditional probabilities that vary by a factor of two on either side as compared to the independent case .
we then incorporated some strong connections among some neurons .
for this we put in one 3-node episode , three 4-node episodes , three 5-node episodes and one 6-node episode with different strengths for the connections .
the connection strengths are so chosen so that we have enough number of 3-node and 4-node episodes ( as possibly subepisodes of the embedded episodes ) spanning the range of conditional probabilities from 0.1 to 0.8 .
all synaptic connections have a delay of 5ms . using our simulator described earlier ,
we generated spike trains for 20 sec of time duration ( during which there are about 50,000 spikes typically ) , and obtained the counts of non - overlapped occurrences of episodes of all sizes using our datamining algorithms . in all results presented below , all statistics are calculated using 1000 repetitions of this simulation . typically , on a data sequence for 20 sec duration ,
the mining algorithms ( run on a dual - core pentium machine ) take about a couple of minutes . as explained earlier , in our simulator
, the rate of the poisson process ( representing the spiking of a neuron ) is updated every 1ms based on the actual spike inputs received by that neuron .
this would , in general , imply that many pairs of neurons ( especially those with strong connections ) are not spiking as independent processes .
[ fig : corr ] shows this for a few pairs of neurons .
the figure shows the cross correlograms ( with bin size of 1 ms and obtained using 1000 replications ) for pairs of neurons that have weak connections and for pairs of neurons that have strong connections .
there is a marked peakiness in the cross correlogram for neurons with strong interconnections , as expected .
[ fig : acc-3 - 4 - 5 ] shows that our theoretical model for calculating the mean and variance of of the non - overlapped count ( given by @xmath83 and @xmath110 determined through eqns .
( [ eq : rec1 ] ) and ( [ eq : var ] ) ) are accurate .
the figure shows plot of the mean ( @xmath83 ) and mean plus three times standard deviation ( @xmath140 ) for different values of the connection strength in terms of conditional probabilities ( @xmath43 ) , for the different episode sizes .
also shown are the actual counts obtained for episodes of that size with different @xmath43 values . as is easily seen ,
the theoretically calculated mean and standard deviations are very accurate .
notice that most of the observed counts are below the @xmath141 threshold for @xmath142 even though this corresponds to a type - i error of just over 10% .
thus our statistical test with @xmath142 or @xmath143 should be quite effective . [ cols="^,^ " , ]
in this paper we addressed the problem of detecting statistically significant sequential patterns in multi - neuronal spike train data .
we employed an efficient datamining algorithm that detects all frequently occurring sequential patterns with prescribed inter - neuron delays .
a pattern is frequent if the number of non - overlapping occurrences of the pattern is above a threshold .
the strategy of counting only the non - overlapped occurrences rather than all occurrences makes the method computationally attractive .
the main contribution of the paper is a new statistical significance test to determine when the count obtained by our algorithm is statistically significant . or , equivalently
, the method gives a threshold for different patterns so that the algorithm can detect only the significant patterns .
the novelty in assessing the significance in our approach is in the structure of the null hypothesis .
the idea is to use conditional probability as a mechanism to capture strength of influence of one neuron on another .
our null hypothesis is specified in terms of a ( user - chosen ) bound on the conditional probability that @xmath2 will fire after a specified delay given that @xmath1 has fired , for any pair of neurons @xmath1 and @xmath2 .
thus this compound null hypothesis includes many models of inter - dependent neurons where the influences among neurons are ` weak ' in the sense that all such pairwise conditional probabilities are below the bound .
being able to reject such a null hypothesis makes a stronger case for concluding that the detected patterns represent significant functional connectivity .
equally interestingly , such a null hypothesis allows us to rank order the different patterns in terms of their strengths of influence .
if we chose this bound @xmath43 to be the value of the conditional probability when the different neurons are independent , then we get the usual null hypothesis of independent neuron model .
but since we can choose the @xmath43 to be much higher , we can decide which patterns are significant at different levels of @xmath43 and hence get an idea of the strength of interaction they represent .
thus , the method presented here extends the current techniques of significance analysis .
while we specify our null hypothesis in terms of a bound on the conditional probability , note that we are not in any way estimating such conditional probabilities . estimating all relevant conditional probabilities would be computationally intensive .
since our algorithm counts only non - overlapped occurrences and also uses the datamining idea of counting frequencies for only the relevant candidate patterns , our counts do not give us all the pair - wise conditional probabilities .
however , the statistical analysis presented here allows us to obtain thresholds on the non - overlapped occurrences possible ( at the given confidence level ) if all the conditional probabilities are below our bound .
this is what gives us the test of significance .
we presented a method for bounding the probability that , under the null hypothesis , a pattern would have more than some number of non - overlapped occurrences . because we are counting non - overlapped occurrences , we are able to capture our counting process in an interesting model specified in terms of sums of independent random variables . this model allowed us to get recurrence relations for mean and variance of the random variable representing our count under the null hypothesis which allowed us to get the required threshold using chebyshev inequality .
while this may be a loose bound , as shown through our simulation results , the bound we calculate is very effective .
our method of analysis is quite general and it can be used in situations other than what we considered here . by choosing the value of @xmath61 in eq.([eq : xi ] )
appropriately we can realize this generality in the model . as an illustration of this
we will briefly describe one extension of the model . in the method presented , while analyzing significance of a pattern @xmath25 , we are assuming that firing of @xmath3 after @xmath5 given that @xmath2 has fired is independent of @xmath1 having fired earlier .
that is why we have used @xmath144 while calculating our threshold . but
suppose we do not want to assume this .
then we can have a null hypothesis that is specified by bounds on different conditional probabilities .
suppose @xmath145 is the conditional probability that @xmath146 fires after @xmath33 given @xmath147 has fired and suppose @xmath148 be the probability that @xmath146 fires after @xmath4 and @xmath149 fires after another @xmath5 given @xmath147 has fired .
now we specify the null hypothesis in terms of two parameters as : @xmath150 and @xmath151 . now for assessing significance of 3-node episodes we can use @xmath152 .
our method of analysis is still applicable without any modifications .
of course , now the user has to specify two bounds on different conditional probabilities and he has to have some reasons for distinguishing between the two conditional probabilities .
but the main point here is that the model is fairly general and can accommodate many such extensions .
there are many other ways in which the idea presented here can be extended .
suppose we want to assess significance of synchronous firing patterns rather than sequential patterns based on the count of number of non - overlapped occurrences of the synchronous firing pattern .
one possibility would be to use conditional probabilities of @xmath1 firing within an appropriate short time interval from @xmath2 in our null hypothesis and then use an appropriate expression for @xmath61 in our model .
another example could be that of analyzing occurrences of neuronal firing sequences that respect a pre - set order on the neurons as discussed in @xcite .
suppose we want to assess the significance of count of such patterns of a fixed length .
if we use our type of non - overlapped occurrences count as the statistic , then the model presented here can be used to assess the significance .
now the parameter @xmath61 would be the probability of occurrence of a sequence of that length ( which respects the global order on the neurons ) starting from any time instant . for a given null hypothesis , e.g. , of independence
, this would be a combinatorial problem similar to the one tackled in @xcite .
once we can derive an expression for @xmath61 we can use our method for assessing significance . though we did not discuss the computational issues in this paper , the data mining algorithms used for discovering sequential patterns are computationally efficient ( see @xcite for details ) .
one computational issue that may be relevant for this paper may be that of data sufficiency .
all the results reported here are on spike data of 20 sec duration with background spiking rate of 20 hz .
( that works out to about 400 spikes per neuron on the average in the data ) . from fig .
[ fig : acc-3 - 4 - 5 ] we can see that , with this much of data , we can certainly distinguish between connection strengths that differ by about 0.2 on the conditional probability scale .
( notice that , in the figure , the mean plus three sigma range of the count distribution at a connection strength is below our threshold ( with @xmath142 ) at a connection strength 0.2 more ) . in fig .
[ fig : rankorder-3 - 4 - 5 ] we showed that we can reliably rank order connection strengths with about the same resolution .
thus we can say that 20 sec of data is good enough for this level of discrimination .
obviously , if we need to distinguish between only widely different strengths , much less data would suffice . in terms of computational issues
, we feel that one of the important conclusions from this paper is that temporal data mining may be an attractive approach for tackling the problem of discovering firing patterns ( or microcircuits ) in multi - neuronal spike trains . in temporal data mining literature , episodes are , in general , partially ordered sets of event types . here
we used the methods for discovery of serial episodes which correspond to our sequential patterns .
a general episode would correspond to a graph of interconnections among neurons .
however , at present , there are no efficient algorithms for discovering frequently occurring graph patterns from a data stream of events .
extending our data mining algorithm and our analysis technique to tackle such graph patterns is another interesting open problem .
this would allow for discovery of more general microcircuits from spike trains . in summary
, we feel that the general approach presented here has a lot of potential and it can be specialized to handle many of the data analysis needs in multi - neuronal spike train data .
we would be exploring many of these issues in our future work .
we wish to thank mr .
debprakash patnaik and mr .
casey diekman for their help in preparing this paper .
the simulator described here as well as the data mining package for analyzing data streams is written by mr .
patnaik @xcite and he has helped in running the simulator .
diekman has helped in generating all the figures .
the work reported here is partially supported by a project funded by general motors r&d center , warren through sid , indian institute of science , bangalore .
r. agrawal and r. srikant .
mining sequential patterns . in philip
s. yu and arbee s. p. chen , editors , _ eleventh international conference on data engineering _ , pages 314 , taipei , taiwan , 1995 .
ieee computer society press . j. le feber , w. l. c. rutten , j. stegenga , p.
s. wolters , g. j. a. ramakers and j. van pelt .
conditional firing probabilities in cultured neuronal networks : a stable underlying structure in widely varying spontaneous activity patterns . , 4:5467 , 2007 .
s. laxman , p. s. sastry , and k. p. unnikrishnan . a fast algorithm for finding frequent episodes in event streams . in _ proc .
acm sigkdd int .
conf . knowledge discovery and datamining _ , san jose , usa , august 2007 .
k. p. unnikrishnan , d. patnaik , and p. s. sastry .
discovering patterns in multi - neuronal spike trains using the frequent episode method .
technical report , 2007 .
available at http://www.citebase.org/abstract?id=oai:arxiv.org:0709.0566 . | in this paper we consider the problem of detecting statistically significant sequential patterns in multi - neuronal spike trains .
these patterns are characterized by an ordered sequences of spikes from different neurons with specific delays between spikes .
we have previously proposed a datamining scheme @xcite to efficiently discover such patterns which are frequent in the sense that the count of non - overlapping occurrences of the pattern in the data stream is above a threshold . here
we propose a method to determine the statistical significance of these repeating patterns and to set the thresholds automatically .
the novelty of our approach is that we use a compound null hypothesis that includes not only models of independent neurons but also models where neurons have weak dependencies .
the strength of interaction among the neurons is represented in terms of certain pair - wise conditional probabilities .
we specify our null hypothesis by putting an upper bound on all such conditional probabilities .
we construct a probabilistic model that captures the counting process and use this to calculate the mean and variance of the count for any pattern . using this
we derive a test of significance for rejecting such a null hypothesis .
this also allows us to rank - order different significant patterns .
we illustrate the effectiveness of our approach using spike trains generated from a non - homogeneous poisson model with embedded dependencies . |
any discussion of the role of carbon stars ( c stars ) in the chemical evolution of galaxies raises a number of important questions such as : for which metal abundances and initial stellar masses do stars become c stars ? how much mass does a star lose during its c star phase ?
what are the elemental compositions of these ejecta ?
c stars are brilliant and show easily recognizable spectral characteristics .
thus , their occurrence and frequency in different galaxies may be interpreted in terms of the properties ( metallicity and age ) of the stellar populations , if we know which stars become c stars .
current answers to this question are , however , not very precise , although stars with masses somewhere in the interval 1.2 to 4m@xmath3 may be a realistic answer .
studies of the distribution of c stars perpendicular to the galactic plane suggest typical masses in the interval 1.2 1.6m@xmath3 ( claussen et al .
1987 ; cf . also groenewegen et al .
1995 ) . for stellar masses greater than about 4m@xmath3 hot - bottom burning ( hbb ) is thought to prevent c star formation ( boothroyd et al .
1993 ) . in metal - poor populations , stars with even smaller masses may become sufficiently carbon enriched to show c star spectra , while hbb may also be effective in inhibiting c star formation at lower masses than for pop .
an important issue in this respect is the frequency of c - rich planetary nebulae ( pne ) in the galaxy , which is so high ( cf .
zuckerman & aller 1986 ; rola & strasiska 1994 ) that a considerable fraction of all stars must become c stars in the end this is a strong reason for not increasing the lower limit of the mass range discussed here too much above 1m@xmath3 .
there is consensus that mass loss probably both sets the maximum luminosity that a star achieves and determines the time it spends in its evolution up along the agb .
the total mass returned to the interstellar medium is obviously the difference between the initial stellar mass and the final remnant mass ( typically about 0.65m@xmath3 for the stars discussed here ) . in order to determine the total mass lost from the c
stars we first need to know how much mass these stars have lost before becoming c stars .
most authors assume that the total mass loss on the rgb is on the order of 0.2m@xmath3 for solar - mass stars , following arguments by renzini & fusi pecci ( 1988 ) , while more massive stars being more compact and not undergoing the he core flash presumably lose less . many studies also assume some continuous mass loss , specified by the reimers relation ( with a suitable fitting parameter ) , followed at the end of the agb by a
superwind " with a steeper dependence of mass loss on stellar parameters .
detailed models of pulsating stars ( bowen & willson 1991 ; willson et al .
1996 ) demonstrate the inadequacy of simple parametrizations the mass loss rate turns out to be quite sensitive to stellar parameters , including mass and composition , and this can not be determined empirically from available observed mass loss rates because the stellar parameters are not well known . since the atmospheres of agb stars , and especially c stars ,
are enriched in many elements by interior nucleosynthesis and various mixing mechanisms , they contribute substantial amounts of heavy elements to the interstellar medium ( ism ) .
( also , the physical and chemical properties of the ism are affected by the grains produced by c stars . )
the main yield to the ism from the c stars depends , however , on the interplay between mass loss and dredge - up of processed material from the inner stellar layers .
neither of these processes is very well understood . in recent model calculations dredge - up , as well as mass loss ,
is incorporated with free parameters adjusted to fit some selected observational constraints , such as the luminosity distribution of c stars in the large magellanic cloud ( lmc ) .
major recent contributions of this type are due to groenewegen & de jong ( 1993 ) , groenewegen et al .
( 1995 ) , blcker ( 1995 ) and marigo et al .
this may be a questionable procedure in view of the assumptions made concerning the dependence of the adjustable parameters on stellar properties .
key issues in the present discussion are when , during the agb evolution with its gradual dredge - up of enriched material , the most significant mass loss occurs and how this depends on the metallicity of the star .
we are still far from definitive answers to these questions . here , a rather empirical discussion will be given , based on the observed chemical compositions of stars and pne . for a detailed and more theoretical approach , based on calculations of nuclear processing and mixing during the agb with adjustable parameters describing the dredge - up as well as with semiempirical recipes for mass loss rate as a function of luminosity and pulsation period , see marigo et al .
these authors also contribute tables of predicted yields , partially superseding those of renzini & voli ( 1981 ) .
another , more indirect but very significant basis for estimating yields will also be used below : the trends in relative abundances of solar - type stars of different ages in the solar neighborhood . from such results ( e.g. those of edvardsson et al . 1993 )
one can estimate relative time scales of the build - up of different elements , and thus separate the contribution of elements from , say , slowly evolving low mass stars as compared with contributions from rapidly evolving high mass stars through supernova ( sn ) explosions .
we shall here review the current knowledge , and lack of knowledge , as regards the role of c star production of the nuclei of li , c ,
n , o , f , ne , na , and al as well as of the @xmath0-process elements . our discussion will be based on what is known from stars in the galactic disk and in the lmc , and it is important to stress that the understanding concerning the mechanisms of the formation , nucleosynthesis and mass loss of c stars is so weak that extrapolations to stellar populations with other properties are highly uncertain .
in fact , the mixture of arguments , partly based on galactic , partly on lmc c stars , in contemporary discussions is in itself dangerous , but necessary .
as has been demonstrated by reeves et al .
( 1990 ) , spallation of cno nuclei in the interstellar medium , although contributing @xmath4li , be and b , can not be the main process for raising the @xmath5li abundance from the pop .
ii plateau value found by spite & spite ( 1982 ) to the more than 10 times higher value found in the solar system and in young stars .
the processes by which li has been built up are as yet not determined , but several processes have been proposed : li - production in agb stars , the @xmath6-process in sne of type ii ( sneii ) , and production in novae . in a study based on models of galactic chemical evolution and observed abundances of light elements , brown ( 1992 ) suggested intermediate - mass stars to be main contributors .
some c stars show very strong lii resonance lines ( mckellar 1940 ) and these stars have been proposed to be a significant source of li to the ism ( scalo 1976 ) .
these rare stars , having logarithmic li abundances , log @xmath7(li ) , of about 10 times the solar system value or even greater ( @xmath84 on a scale with log @xmath7(h ) set to 12 ) , are still not fully understood ; attempts to explain them include hbb and the be transport mechanism of cameron & fowler ( 1971 ) ( sackmann & boothroyd 1992 ; abia et al . 1993b ) .
abia & isern ( 1996 ) found that the @xmath1c abundances of galactic c stars correlate with their li abundances ; this may give a further clue to the origin of the super li - rich phenomenon .
these authors suggest that hbb might produce the abundances found if one invokes a more efficient convection than is usually assumed . this could decrease the lower mass limit of hbb to below 2m@xmath3 .
alternatively , the plume mixing model of scalo & ulrich ( 1973 ) might operate in these stars .
the significance of c stars in general for galactic li production is entirely dependent on the role of the super li - rich stars ; the more normal n stars have log @xmath7(li ) ranging from 2 to 2 ( abia et al .
1993b ) which is far too low for the stars to be of significance in this respect .
key questions concerning the super li - rich stars are : ( 1 ) how common are they ? ( note that the conclusion of abia et al . that these stars may provide a significant fraction of the galactic li is based on only one star ! ) ( 2 ) how does their frequency depend on mass , metallicity and other parameters ?
( 3 ) what are their true li abundances ?
( we note that the abundance estimates are severely dependent on model atmospheres , spectral synthesis and adopted c / o ratios for the models . thus , abia et al . ( 1993a ) derive an abundance for wz cas of log @xmath7(li ) = 5 , while boesgaard et al .
( 1996 ) find a 3 times lower abundance with the same grid of model atmospheres . )
( 4 ) are the super li - rich stars overrepresented , or represented at all , among the dust - enshrouded stars with mass loss rates @xmath9m@xmath3/year ? ( 5 ) do the super li - rich stars represent relatively short episodes , followed by li burning , and certain limited mass intervals as is suggested by the model calculations by sackmann & boothroyd ( 1992 ) and frost & lattanzio ( 1995 ) or could they possibly mark the ending stage of agb evolution in general ?
a conclusion in several studies is that the c stars probably do not contribute significant amounts of li ( cf .
matteucci et al .
1995 ) , even though contributions as high as 30% or more of the interstellar li may be possible if some super li - rich stars also have high mass - loss rates ( cf .
abia et al .
1993b ) .
an interesting issue is whether this situation is different for low metallicities i.e. , in early phases of galactic evolution or in the present evolution of dwarf galaxies .
a low metal abundance seems to lead to hbb at smaller stellar masses ( cf . sackmann & boothroyd 1992 ) . on the other hand , the low metallicity stars may lose their mass at lower rates , such that the li produced may be burnt away before most of the mass loss occurs ( matteucci et al .
empirical support for this may be the li abundance of log @xmath7(li ) @xmath10 derived for luminous agb ( s ) stars in the smc ( with a metallicity 1/3 of solar ) found by plez et al .
( 1993 ) , which , in spite of being significantly higher than expected from standard agb evolution is still lower than values for the most li - rich agb stars in the galaxy . in conclusion
it seems probable that another process , such as the @xmath6-process in sneii ( woosley et al . 1990 ) , is the main mechanism responsible for li formation .
however , until we understand the nature of the super - li rich c stars better , more definitive statements must be avoided . for some time ,
intermediate or low mass stars have been considered as probably important for the production of carbon in the galaxy ( see , e.g. , tinsley 1978 ) .
sarmiento & peimbert ( 1985 ) estimated the carbon contribution from agb stars to be on the order of 6080% of all galactic carbon .
their argument was essentially based on the fact that models of sne and novae suggested that these objects could not produce more than a minor part of the carbon in the interstellar medium .
sackmann & boothroyd ( 1991 ) claimed , on the basis of calculations of dredge - up of carbon in agb model sequences , that low and intermediate mass agb stars are the dominant sources of carbon in the universe .
the more recent supernova models of woosley & weaver ( 1995 ) and thielemann et al .
( 1996 ) predict yields that are mutually different by about a factor of two , the thielemann et al .
yields being lower due to a high rate for the @xmath11c(@xmath12,@xmath13)@xmath14o reaction . adopting the higher yields of woosley & weaver ( 1995 ) ,
timmes et al .
( 1995 ) still find in their models of galactic evolution that sne and novae can not produce enough , so that the contributions from stars with a mass less than 11m@xmath3 must be dominant .
timmes et al . also adopt the ( now partly obsolete ) yields from intermediate and low mass stars of renzini & voli ( 1981 ) and find that the carbon abundance relative to iron , [ c / fe ] , was lower by about a factor of two than its present value in the early evolution of our galactic disk ; then it increased above its present value and finally decreased again as a result of iron production in less massive sneia .
this particular behavior of [ c / fe ] relative to [ fe / h ] is not seen in the composition of solar - type disk stars of different ages , according to tomkin et al .
( 1995 ) . instead
, these authors find a steady decrease in [ c / fe ] _
vs. _ [ fe / h ] , suggesting a less dramatic variation of contributing sites during the history of the galaxy .
in fact , the ratios of carbon relative to oxygen and @xmath12-elements ( produced in sneii ) only vary slowly and smoothly with [ fe / h ] .
this suggests that the carbon may well have been produced in high - mass stars , although carbon does not follow oxygen in the halo ( tomkin et al .
1992 ) , nor in dwarf galaxies ( garnett et al .
the decrease of [ c / fe ] with increasing [ fe / h ] in the galactic disk is very different from the variation of the @xmath0-element abundances with [ fe / h ] ( see also edvardsson et al .
this suggests that the @xmath0-elements were formed in stars with considerably longer characteristic lifetimes , i.e. smaller masses .
prantzos et al .
( 1994 ) studied the formation of oxygen and carbon in the galaxy and found that , if the duration of the halo phase was on the order of 12 gyears as is currently believed , intermediate or low mass stars should not have been the main carbon sources . instead
, these authors suggest that massive stars contribute with metal - dependent yields ( maeder 1992 ) .
we note that the model calculations of prantzos et al .
, with contributions added also from the intermediate and low mass stars according to renzini & voli ( 1981 ) , fit the results of tomkin et al .
( 1995 ) rather well , although the c / o ratios of comparatively metal - rich halo stars become too high . in that model ,
about 40% of the carbon in the disk stars was contributed by intermediate and low mass stars .
adoption of the yields of marigo et al .
( 1996 ) would diminish the discrepancy for the metal - rich halo stars but may produce too much carbon at late stages .
other data suggest that the intermediate and low mass stars may play a significant role in carbon synthesis . adopting the findings by zuckerman & aller ( 1986 ) and rola & strasiska ( 1994 ) that about half of the pne have c / o @xmath15 1 , we find that these stars should contribute at least about 0.001m@xmath3/year of carbon to the galaxy .
( here , we have used the pne birth - rate of about 1 pn per year in the galaxy estimated by pottasch ( 1992 ) , and a characteristic pn mass of 0.3m@xmath3 . ) assuming the yields recently derived from a semi - empirical modelling of the agb phase by marigo et al .
( 1996 ) would presently give contributions on the order of 0.003m@xmath16 , depending on the star formation rate adopted .
this contribution is of the same order of magnitude as the total present contribution by sneii , assuming yields from woosley & weaver ( 1995 ) and a sn rate of 3 per century . in making this estimate
we have assumed a closed box model ; this may overestimate the significance of sne relative to pne , in view of the much higher expansion velocities of material ejected in sne .
one should note that the number of c - rich pne is so high that they can not only be produced by intermediate - mass stars ; a significant fraction of the c - rich pne must be formed by stars in the mass interval between 1 and 2m@xmath3 . if low - mass stars are assumed to produce significant amounts of carbon , further discrepancies between models of galactic chemical evolution and observed abundances may occur in the relative - abundance diagrams , such as those of tomkin et al .
possibly , this conflict could be resolved by advocating more efficient carbon production by intermediate and low mass metal - poor stars , following the discussion of boothroyd & sackmann ( 1988 ) that suggests a more efficient dredge - up of carbon for models with low abundances of heavy elements .
we note that the carbon yields by marigo et al .
( 1996 ) are several times greater at a given initial stellar mass for their metal - poor ( z=0.008 ) models than for their pop .
i ( z=0.02 ) models .
( this is , however , dependent on the assumptions made in these models that the two adjustable third - dredge - up parameters , the minimum core mass for dredge - up and the dredge - up efficiency , are independent of metallicity , and that the mass loss rate is dependent only on the metallicity through its effects on evolutionary tracks and pulsation period . ) another possibility , which should be further investigated , is that the spectrum analysis of the high - excitation ci lines in solar - type stars may give systematic errors , different for stars of different metallicities , e.g. due to errors in the effective temperature scale , or errors due to inhomogeneities and departures from lte ( we note , however , the small non - lte effects found for carbon in the sun by strenburg & holweger 1990 ) .
the study of andersson & edvardsson ( 1994 ) , using the low excitation [ ci ] line at @xmath17 , seems to verify the results of tomkin et al .
but the weakness of that line only admitted upper limits for most of the metal - poor disk stars in their sample .
further studies based on this line with higher s / n are underway .
do the c stars contribute significantly to galactic carbon ?
number densities and mass - loss observations of bright , as well as dust - enshrouded , c stars in the galaxy ( olofsson et al . 1993a , 1996 ; jura & kleinmann 1989 ) suggest a total contribution of about @xmath18m@xmath16/year .
the indication , from the statistics of pne , that c stars may contribute one order of magnitude more is the result of the fact that about every second pn is carbon rich .
this illustrates the key significance , mentioned in the introduction , of the question of the chemical composition of the stellar envelope just prior to the extensive mass loss that ends the agb evolution .
we conclude that the significance of low and intermediate mass stars for carbon synthesis is still unclear , and deserves further exploration .
prantzos et al .
( 1996 ) have discussed the observations of carbon isotope ratios in the galactic disk and conclude that @xmath1c has probably a mixed origin , being produced both as a primary element presumably in intermediate - mass stars by hbb or by other processes where protons are mixed with @xmath11c produced in the star by he burning and as a secondary element by cn burning , occurring in stars of all masses .
the @xmath11c/@xmath1c ratios observed in c stars range in the interval 30100 , with a pronouced peak around 50 ( lambert et al .
1986 ; a lower range , 1540 , has , however , been obtained by ohnaka & tsuji 1996 , and these differences need further exploration ) . excluded from this
is a minority of j - type stars with considerably lower ratios .
values around 50 are consistent with the c / o ratios derived by lambert et al . ;
i.e. , the result may be explained as the consequence of mixing @xmath11c to the surface layers if @xmath1c is left from the first dredge - up .
these values agree reasonably well with those ( 3040 ) derived for dusty c stars from mm - line observations by kahane et al .
these authors have also observed the c - rich pn ngc 7027 and find a @xmath11c/@xmath1c lower limit of 65 .
in fact , the difference in c / o ratio between bright n stars and c - rich pne ( see lambert et al .
1986 ) suggests that further @xmath11c enrichment has taken place in the latter , which would correspond to an increase from 50 to about 80 in @xmath11c/@xmath1c .
although this agrees with the local ism value , that may well be fortuitous little is known about the @xmath11c/@xmath1c ratios in pne .
one should also note that the @xmath1c produced " by normal c stars is , as presumed above , just the result of the cno burning and the first dredge - up .
that is , however , not necessarily the case ; we note that plez et al .
( 1993 ) find low @xmath11c/@xmath1c ratios for s stars in the small magellanic cloud indicating h - burning in the envelope in late evolutionary phases .
one might ask what the role of the extremely @xmath1c - rich j stars may be in providing @xmath1c .
the evolutionary history of these stars is still not understood ( cf .
lambert et al .
1986 for some suggestions ) .
they constitute about 1/5 of the bright n - type stars in the magnitude - limited lambert et al . sample and have a @xmath11c/@xmath1c ratio of typically 5 .
they tend , however , to be systematically oxygen - poor ; therefore , the @xmath1c abundances relative to hydrogen are not more than typically a factor of 5 higher than those of other c stars .
the j - type stars do not show any tendency of having higher present mass - loss rates than the rest of the n stars ( olofsson et al .
1993a ) . from this one
would only conclude that the j stars at present contribute approximately the same total amount of @xmath1c as the rest of the visible n stars .
similarily , if the j - type stars lose their envelopes without first burning their @xmath1c , and if the fraction of j stars among c stars of 20% really is representative also of the population of immediate progenitors of c - rich pne , which is highly questionable , the j stars could then deliver as much @xmath1c as the sum of the rest of the c stars .
there is at least one additional circumstance that might suggest that the j - type stars play a significant role in this respect : a great fraction of the most luminous red agb stars in the lmc are @xmath1c - rich c stars ( richer et al .
1979 ) , possibly indicating that a greater fraction than 20% of the more massive c stars reach this stage close to the end of their agb evolution , at least in metal - poor populations . however , there is also a population of low - luminosity j - type stars present ( cf .
richer 1981 ; bessell et al .
1983 ) which does not necessarily correspond to the latest phases of agb evolution .
brewer ( 1996 ) recently discovered 7 j - type stars among 48 c stars in two m31 fields and found the j stars relatively faint .
@xmath19n is thought to be produced by equilibrium cno - burning in the hydrogen burning shell , brought to the surface layers at the first dredge - up and even more at the second dredge - up , occurring in stars of 48 m@xmath16 and expelled during later stellar evolution .
timmes et al .
( 1995 ) , however , conclude , on the basis of abundance trends among galactic stars and radial gradients of cno abundances in hii regions in galaxies , that @xmath19n has a strong primary component which they ascribe to low metallicity massive stars .
woosley & weaver ( 1995 ) show that it is possible to create this isotope in massive stars and claim that approximatively 25% of the solar abundance can be ascribed to these sites .
vila - costas & edmunds ( 1993 ) show evidence for a delayed primary and a secondary component relative to oxygen , the delay indicating that intermediate mass stars are responsible .
the low n / o ratio observed for a high redshift gas cloud by pettini et al . (
1995 ) also supports an intermediate - mass primary component .
the secondary component is dominant at high metallicities .
the bulk of the solar abundance of @xmath19n is attributed to low and intermediate mass stars ( see e.g. timmes et al .
1995 ) .
empirically , the evidence that c stars contribute a significant extra amount of n is weak .
lambert et al .
( 1986 ) find , from their analyses of cn lines in n - type star spectra , that these stars are not very n - rich , and this result is strengthened by recent values of the dissociation energy of cn .
olofsson et al .
( 1993b ) find an n abundance higher by about a factor of 5 or so from their analysis of hcn mm lines from the circumstellar envelopes of the same sample of n stars , but this may be due to an underestimated mass loss rate ( cf . also olofsson et al .
the c - rich pne are , according to zuckerman & aller ( 1986 ) , generally not very enriched in n only very few in their list have n abundances in excess of twice the solar value . pasquali & perinotto ( 1993 ) list a mean nitrogen abundance of twice solar for their c - rich pne of type ii and iii , which represent disk stars of intermediate and low mass , while the mean n abundance given for pne of type i , that are more carbon - poor and represent younger , more massive objects , is three times greater .
kingsburgh & barlow ( 1994 ) give a mean nitrogen abundance of only 1.4 times solar for their 36 non - type i pne .
similar tendencies may be traced in the yields calculated by marigo et al .
it seems probable that intermediate - mass stars that generally are prevented from becoming c stars by hbb may provide most of the nitrogen .
we note that brett ( 1991 ) , in his analysis of luminous agb stars in the smc , finds strong cn bands , indicating c to n conversion by hbb .
the galactic production of the single stable isotope of flourine , @xmath20f , has been a matter of discussion and is still rather unclear .
jorissen et al . ( 1992 ) identified lines of hf in the 2 micron region for k , m , s and n giants and derived f abundances .
they found that the f / o ratio in agb stars increases with c / o , with maximum values for stars with c / o ratios not too much in excess of 1 ( sc stars ) . their result seems to imply that the thermal pulses produce fluorine , and they ( cf . also forestini et al .
1992 ) suggest as the most probable scenario that @xmath1c(@xmath12,n)@xmath14o reactions produce neutrons , some of which are captured by @xmath19n to produce @xmath19c and protons , which in turn are captured by @xmath21o : @xmath21o(p,@xmath12)@xmath22n(@xmath12,@xmath13)@xmath20f .
the flourine is then brought to the surface by convection following the thermal pulse .
this scenario is complicated by several circumstances , e.g. the new higher rate for the reaction @xmath21o(@xmath12,@xmath13)@xmath2ne which may deplete @xmath21o enough to decrease the significance of the second part of the reaction chain ( cf . frost & lattanzio 1995 ) . in any case , the over - abundances of f found in sc and n stars by jorissen et al .
( 1992 ) , ranging from 3 to 30 times the solar values , indicate that these stars may well be major contributors to the galactic flourine ; an alternative is production in snii ( woosley et al .
1990 ; timmes et al .
1995 ) . an interesting way of clarifying this issue further may be to determine the flourine abundances in less evolved m stars with different metal abundances , since different scalings of , say , [ f / o ] with [ fe / h ] may be expected for the different production sites .
the dominant neon isotope in the solar system , @xmath23ne , is probably mainly produced by sneii ( cf .
woosley & weaver 1995 ; timmes et al .
the sne yields of @xmath2ne are about one order of magnitude lower than those of @xmath23ne which is consistent with the isotopic ratios observed in the solar system .
however , @xmath2ne may be formed at thermal pulses in agb stars in the convective intershell through @xmath12-capture on @xmath19n , which after @xmath24-decay leads to @xmath21o(@xmath12,@xmath13)@xmath2ne ( boothroyd & sackman 1988 ; gallino et al .
1990 ) . in the agb model calculations of marigo
et al . ( 1996 ) , considerable amounts of @xmath2ne are produced and ejected , in particular for models with masses around 2.5m@xmath16 which get most carbon rich .
the predicted @xmath2ne yields are fairly independent of metallicity ( though higher for z=0.008 than for z=0.02 ) , and so high that the agb stars are suggested to be a major contributor of galactic @xmath2ne .
in particular , the envelopes of c stars are predicted to have a @xmath23ne/@xmath2ne ratio considerably less than 1 , even close to 0.1 . the ne enrichment expected in c - rich pne according to these results
was not traced in the sample of southern pne studied by kingsburgh & barlow ( 1994 ) , nor can it be seen in the sample of pne from the magellanic clouds studied by leisy & dennefeld ( 1996 ) .
however , corrardi & schwarz ( 1995 ) found a sample of bipolar pne to be enriched in ne and marigo et al .
( 1996 ) suggest that the build - up of @xmath2ne could be the reason for this .
we note that lewis et al .
( 1990 ) and nichols et al .
( 1993 ) find consistently low isotopic ratios in gas - rich meteoritic sic grains .
na and al are , as well as mg , generally identified as products of ne and c burning in massive stars while the production in sneia is probably small ( nomoto et al . 1992 ) .
the synthesis of na and al is controlled by the neutron flux during ne and c burning which in turn is dependent on the initial metallicity and primarily on the initial o abundance ( @xmath14o being converted to @xmath2ne , via @xmath19n , in he - burning and the extra neutrons in @xmath2ne , when being liberated , are essential to the formation of na and al ) . therefore , one expects a rapid increase of na / mg ratios with fe / h , and such a tendency is also apparent in the calculations of timmes et al .
for the most metal - rich solar - type stars the positive correlation between na / mg and fe / h is well established ( feltzing & gustafsson 1996 ) . however , the observations of disk stars by edvardsson et al .
( 1993 ) , as well as observations of halo stars , do not show this tendency very clearly ( timmes et al .
edvardsson et al .
find a tendency for the abundance ratio of na / mg to be smaller in the inner galaxy than in the outer , at a given mg / h , for disk - stars more metal - poor than the sun .
this suggests that na and mg , respectively , were formed in stars of different types .
thus , there might be additional sources of sodium and possibly of aluminum .
there is some additional evidence that intermediate - mass stars or even low - mass stars might contribute in these respects . in intermediate - mass stars sodium
may be synthesized in the hydrogen - burning shell through the neon - sodium cycle , and by proton captures on @xmath2ne ( denisenkov 1989 ; denisenkov & denisenkova 1990 ; langer et al . 1993 ) . on longer time - scales , i.e. smaller masses ,
proton capture on @xmath23ne may also be significant , and the results may be brought to the surface if efficient mixing is at hand .
observationally , yellow field supergiants appear sodium rich ( boyarchuk & lyubimkov 1985 ) but they hardly contribute significantly to the galactic na . a fraction of low mass red giants in globular clusters show enhanced na and al and are correspondingly poor in o ( norris & da costa 1995 and references given therein ) .
recently cavallo et al .
( 1996 ) have attempted to model this , by following the nucleosynthesis in low - mass red giant model sequences , and they find substantial na - enrichments in the region above the hydrogen - burning shell throughout the giant branch , independent of metallicity .
for the lowest metallicities al is also produced . with suitable mixing processes
these elements will be visible on the surface and may even , after mass loss , contribute to the global enrichment of the galaxy .
there is , however , no strong argument for this low - mass star production of na and al to be linked to the existence of c stars .
conversely , the meager abundance analyses that exist do not support such a hypothesis ; e.g. lambert et al .
( 1986 ) find normal na / ca ratios on the basis of 1 line for each element in the 2 micron region .
there is an interesting possibility that agb stars contribute radioactive @xmath25al ( kudryashov & tutukov 1988 ) , decaying to @xmath25 mg with emission of 1809 kev photons , detected in space .
guelin et al . (
1995 ) have recently found some evidence for @xmath25alf millimetre emission from the carbon - rich envelope of irc+10216 .
huss & wasserburg ( 1996 ) comment on the finding of high @xmath25al/@xmath26al ratios in some sic grains .
the high temperatures needed to produce @xmath25al suggest rather high mass stars as sites if hbb is invoked .
this is , however , not easily reconciled with other isotopic ratios .
the early discovery by merrill ( 1952 ) of the unstable element technetium in s stars was a clear indication that these stars could be a source of the @xmath0-process elements in general ( see also cameron 1955 ) .
agb stars are now generally found to be more or less @xmath0-element rich .
c stars are believed to be the most enriched ( see below ) .
lambert ( 1992 ) gives several arguments in favor of the idea that @xmath0-elements are produced in agb stars of low mass ( @xmath27m@xmath3 ) rather than in stars of intermediate mass . in intermediate - mass agb stars ,
the temperature in the he - burning shell is high enough to ignite the @xmath2ne(@xmath12,n)@xmath28 mg neutron source .
that would lead to an excess abundance of @xmath28 mg ( cf .
lambert 1991 , and references therein ) .
smith & lambert ( 1986 ) found no @xmath28 mg enhancement in their seven randomly selected @xmath0-element enriched stars , which suggests that these were all low - mass agb stars . in these stars
the @xmath1c(@xmath12,n)@xmath14o reaction is the probable neutron source .
the neutron density at the @xmath0-process site can be determined from observations of isotope abundances at certain branching points of the @xmath0-path in the chart of nuclides .
two useful branching points are the ones at @xmath29zr and @xmath30kr .
lambert et al . ( 1995 ) found no evidence of @xmath31zr in zro bands in s star spectra , implying a neutron density of less than @xmath32 @xmath33 , much less than what is expected in regions where the @xmath2ne neutron source operates .
this again suggests that intermediate - mass agb stars are not the major contributors of @xmath0-processed elements in our galaxy . until recently
, it was believed that the @xmath1c neutron source was in operation during the thermal pulses of low - mass stars .
straniero et al .
( 1995 ) have , however , found from an evolutionary model sequence ( for 3m@xmath3 ) that the @xmath1c neutron source is activated during the intervals _ between _ the thermal pulses .
this suggests a lower neutron density ( @xmath34 @xmath33 ) , and that the @xmath1c is consumed during the interpulse period .
semiconvection is essential for the @xmath0-process ( see sackmann & boothroyd 1991 , and references therein ) ; this will allow mixing of protons into the carbon pocket of the thermal pulses leading to fresh @xmath1c .
predictions from the models of straniero et al .
regarding one of the signatures of the low neutron density , the rb / sr ratio which depends on the @xmath0-path at the @xmath30kr branch , are in agreement with recent observations of ms and s stars by lambert et al .
( 1995 ) .
mean neutron exposures deduced from observations of 34 ms and s stars by smith et al .
( 1987 ) suggest that the solar system @xmath0-element distribution could arise from a mixture of ms , s and c stars which return @xmath0-element enriched material to the ism .
their @xmath0-element excesses are for ms stars a factor of 23 , for s stars 46 and for c stars
note that the latter abundances are based on the uncertain abundance analysis of the crowded visual spectra by utsumi ( 1985 ) .
these figures , in combination with estimated total mass loss , are high enough to explain the @xmath0-element abundance in the solar system .
also , parthasarathy ( 1996 ) found @xmath0-element enhancement relative to the sun of 240 for some post - agb stars .
there is also other more indirect , though strong , empirical evidence for the @xmath0-elements being formed mainly in low - mass agb stars .
edvardsson et al .
( 1993 ) showed , from studies of the composition of solar - type disk stars of different ages , that the major contributors to the enrichment of @xmath0-elements in the disk are stars with characteristic evolutionary times @xmath35 10@xmath36 years , i.e. low - mass stars .
this is significantly longer than the time scale of the iron production from sneia .
the low - mass contribution to the @xmath0-elements in the galactic disk is also verified by pagel & tautvaiien@xmath37 ( 1996 ) . moreover , from abundances in the halo they also trace a contribution of unknown origin but with a time scale characteristic of stars of @xmath38m@xmath3 .
the third dredge - up may be even more significant for populations with lower metallicity than solar ( cf .
wood 1981 ) and provide strong @xmath0-element enrichment .
we note that kipper et al .
( 1996 ) found three presumably intrinsic c stars of pop .
ii to have @xmath0-element enhancements of 13 dex .
we conclude that mass loss from precursors of c - rich pne are most probably a main source of the @xmath0-elements in galaxies .
it seems most probable that the agb stars in general , and c stars in particular , are the main contributors of @xmath0-elements ; probable that their contributions of flourine and carbon are quite significant ; and possible that their contributions of lithium ( from the super li - rich c stars ) , of @xmath1c ( from j stars ) and of ne ( i.e. @xmath2ne ) are of some importance .
however , all of these statements are more or less uncertain , and the uncertainty is also great concerning the role of intermediate and low mass stars in the production of n , na and al .
this is because a number of relatively fundamental questions concerning c stars still remain unanswered . in order to improve the situation new and more detailed and reliable abundance analyses for c stars and pne
are needed ( also in nearby galaxies with different metallicities ) , as well as better models of nucleosynthesis and dredge - up in agb stars .
we also need to know more about mass loss how much matter the stars eject , as a function of initial mass and composition , as well as when the stars lose mass relative to the time when their envelope enrichment occurs . further studies of abundances in less evolved stars of different stellar populations are also of great significance . as regards the nucleosynthesis of two of the most important elements , carbon and nitrogen , more work on stars more massive than normal c stars seems also important .
the situation reminds us of one of the stories about the famous nasrettin hoca of konya , a place not very far from antalya .
nasrettin saw ducks on the lake shore and , thinking of his dinner , tried to catch one but it jumped into the lake and swam out of reach .
the next one did the same .
after several attempts he took a spoon and sat down at the shore and started eating water .
some friends passed and asked him what he was doing .
he answered :
i am having duck soup . "
we have tried to catch " the evasive secrets of the c stars with different , more or less sophisticated methods . however , in spite of much progress presented at this meeting , the solutions of the problems seem to be at some distance .
this partly reflects the experience , so common in science , that phenomena tend to become more complex when studied . in any case , more systematic investigations are now needed .
if we undertake them , it seems that we shall get the most important answers long before nasrettin hoca has been able to walk to the center of his lake and catch his ducks .
abia , c. , boffin , h. m. j. and rebolo , r. 1993a , _ astron .
_ , 272 , 455 abia , c. , isern , j. and canal , r. 1993b , _ astron .
_ , 275 , 96 abia , c. and isern , j. 1996 , _ ap.j . _ , 460 , 443 andersson , h. and edvardsson , b. 1994 , _ astron .
_ , 290 , 590 bessell , m. s. , wood , p. r. and lloyd evans , t. 1983 , _ mon . not
_ , 202 , 59 blcker , t. 1995 , _ astron .
_ , 297 , 727 boesgaard , a. , eriksson , k. and gustafsson , b. 1996 , in prep .
boothroyd , a. i. and sackmann , i .- j .
1988 , _ ap.j .
_ , 328 , 653 boothroyd , a. i. , sackmann , i .-
j . and ahern , s. c. 1993 , _ ap.j .
_ , 416 , 762 bowen , g. h. and willson , l. a. 1991 , _
_ , 375 , l53 boyarchuk , a. a. and lyubimkov , l. s. 1985 , _ bull .
crimean astrophys .
_ , 66 , 119 brett , j. m. 1991 , _ mon . not .
_ , 249 , 538 brewer , j. p. 1996 , _ publ .
soc . pacific _ , 108 , 379 brown , l. e. 1992 , _ ap.j . _ , 389 , 251 cameron , a. g. w. 1955 , _ ap.j . _ , 121 , 144 cameron , a. g. w. and fowler , w. a. 1971 , _ ap.j . _ , 164 , 111 cavallo , r. m. , sweigart , a. v. and bell , r. a. 1996 , _ ap.j .
_ , 464 , l79 claussen , m. j. , kleinmann , s. c. , joyce , r. r. and jura , m. 1987 , _ ap.j .
_ , 65 , 385 corrardi , r. l. m. and schwarz , h. e. 1995 , _ astron . astrophys .
_ , 293 , 871 denisenkov , p. a. 1989 , _
soviet astr .
_ , 14 , 435 denisenkov , p. a. and denisenkova , s. n. 1990 , _
soviet astr .
_ , 16 , 275 edvardsson , b. , andersen , j. , gustafsson , b. , et al . 1993 , _ astron .
_ , 275 , 101 feltzing , s. and gustafsson , b. 1996 , in prep .
forestini , m. , goriely , s. , jorissen , a. and arnould , m. 1992 , _ astron .
_ , 261 , 157 frost , c. a. and lattanzio , j. c. 1995 , in the 32@xmath39 liege inst .
stellar evolution : what should be done ? _ garnett , d. r. , skillman , e. d. , dufour , r. j. , et al .
1995 , _ ap.j .
_ , 443 , 64 gallino , r. , busso , m. , picchio , g. and raiteri , c. m. 1990 , _ nature _ , 348 , 298 groenewegen , m. a. t. and de jong , t. 1993 , _ astron .
_ , 267 , 410 groenewegen , m. a. t. , van den hoek , l. b. and de jong , t. 1995 , _ astron .
_ , 293 , 381 guelin , m. , forestini , m. , valarion , p. , ziurys , l. m. , anderson , m. , chernicharo , j. and kahane , c. 1995 , _ astron .
_ , 297 , 183 huss , g. r. and wasserburg , g. j. 1996 , _ lun .
planetary sci .
_ , 27 , 573 jorissen , a. , smith , v. v. and lambert , d. l. 1992 , _ astron .
_ , 261 , 164 jura , m. and kleinmann , s. 1989 , _ ap.j . _ , 341 , 359 kahane , c. , cernicharo , j. , gmez - gonzlez , j. and gulin , m. 1992 , _ astron . astrophys . _ , 256 , 235 kingsburgh , r. l. and barlow , m. j. 1994 , _ mon . not .
_ , 271 , 257 kipper , t. , jrgensen , u. g. , kochkova , v. g. and pauchuk , v. e. 1996 , _ astron .
_ , 306 , 489 kudryashov , a. s. and tutukov , a. 1988 , _ astron .
_ , 1525 , 11 lambert , d. l. 1991 , in iau symp . 145 : _ evolution of stars : the photospheric abundance connection _ , ed
. g. michaud and a. tutukov ( kluwer ) , p.299 lambert , d. l. 1992 , in the 31@xmath40 herstmonceux conf . : _
elements and the cosmos _
m. g. edmunds and r. terlevich ( cambridge university press ) , p. 92
lambert , d. l. , gustafsson , b. , eriksson , k. and hinkle , k. h. 1986 , _
_ , 62 , 373 lambert , d. l. , smith , v. v. , busso , m. , gallino , r. and straniero , o. 1995 , _ ap.j . _ , 450 , 302 langer , g. e. , hoffman , r. and sneden , c. 1993 , _ publ .
soc . pacific _ , 104 , 301 leisy , p. and dennefeld , m. 1996 , _ astron .
_ , 16 , 95 lewis , r. s. , amari , s. and anders , e. 1990 , _ nature _ , 348 , 293 maeder , a. 1992 , _ astron .
_ , 264 , 105 marigo , p. , bressan , a. and chiosi , c. 1996 , _ astron .
_ , in press matteucci , f. , dantona , f. and timmes , f.x .
1995 , _ astron .
_ , 303 , 460 mckellar , a. 1940 , _ publ .
soc . pacific _ , 52 , 407 merrill , p. w. 1952 , _
_ , 116 , 21 nichols , r. h. jr . ,
amari , s. , hohenberg , c. m. , et al .
1993 , _ meteoritics _ , 28 , 410 nomoto , k. , tsujimoto , t. , yamaoka , h. , et al .
1992 , in the 31@xmath41 herstmonceux conf . : _
elements and the cosmos _ , ed . m. g. edmunds and r. terlevich ( cambridge university press ) , p. 55
norris , j. e. and da costa , g. s. 1995 , _ ap.j .
_ , 441 , l81 ohnaka , k. and tsuji , t. 1996 , _ astron .
_ , 310 , 933 olofsson , h. , eriksson , k. , gustafsson , b. and carlstrm , u. 1993a , _ ap.j .
_ , 87 , 267 olofsson , h. , eriksson , k. , gustafsson , b. and carlstrm , u. 1993b , _
_ , 87 , 305 olofsson , h. , bergman , p. , eriksson , k. and gustafsson , b. 1996 , _ astron .
_ , 311 , 587 pagel , b. e. j. and tautvaiien@xmath37 , g. 1996 , _ mon . not .
_ , in press parthasarathy , m. 1996 , private communication pasquali , a. and perinotto , m. 1993 , _ astron .
_ , 280 , 581 pettini , m. , lipman , k. and hunstead , r. w. 1995 , _ ap.j . _ ,
451 , 100 plez , b. , smith , v. v. and lambert , d. l. 1993 , _ ap.j .
_ , 418 , 812 pottasch , s. r. 1992 , _ astron .
_ , 4 , 215 prantzos , n. , vangioni - flam , e. and chauveau , s. 1994 , _ astron .
_ , 285 , 132 prantzos , n. , aubert , o. and audouze , j. 1996 , _ astron .
_ , 309 , 760 renzini , a. and voli , m. 1981 , _ astron .
_ , 94 , 175 renzini , a. and fusi pecci , f. 1988 , _ ann .
_ , 26 , 199 reeves , h. , richer , j. , sato , k. and teresawa , n. 1990 , _ ap.j .
_ , 355 , 18 richer , h. b. 1981 , _ ap.j . _ , 243 , 744 richer , h. b. , olander , n. and westerlund , b. e. 1979 , _ ap.j . _ , 230 , 724 rola , c. and strasiska , g. 1994 , _ astron .
_ , 282 , 199 sackmann , i .-
j . and boothroyd , a. i. 1991 , in iau symp .
145 : _ evolution of stars : the photospheric abundance connection _ , ed .
g. michaud and a. tutukov ( kluwer ) , p. 275
sackmann , i .-
j . and boothroyd , a. i. 1992 , _ ap.j .
_ , 392 , l71 sarmiento , a. and peimbert , m. 1985 , _ rev .
_ , 11 , 73 scalo , j. m. 1976 , _ ap.j . _ ,
206 , 795 scalo , j. m. and ulrich , r. g. 1973 , _ ap.j .
_ , 183 , 151 smith , v. v. and lambert , d. l. 1986 , _ ap.j .
_ , 311 , 843 smith , v. v. , lambert , d. l. and mcwilliam , a. 1987 , _ ap.j . _ , 320 , 862 spite , f. and spite , m. 1982 , _ astron .
_ , 115 , 357 straniero , o. , gallino , r. , busso , m. , et al . 1995 , _ ap.j . _ , 440 , l85 strenburg , s. and holweger , h. 1990 , _ astron .
_ , 237 , 125 thielemann , f .- k . , nomoto , k. and hashimoto , m .- a .
1996 , _ ap.j . _ , 460 , 408 timmes , f. x. , woosley , s. e. and weaver , t. a. 1995 , _ ap.j .
_ , 98 , 617 tinsley , b. m. 1978 , in iau symp . 76 : _ planetary nebulae _ , ed . y. terzian ( reidel ) , p. 341 tomkin , j. , lemke , m. , lambert , d. l. and sneden , c. 1992 , _
_ , 104 , 1568 tomkin , j. , woolf , v. m. , lambert , d. l. and lemke , m. 1995 , _ a.j . _ , 109 , 2204 utsumi , k. 1985 , in _ cool stars with excesses of heavy elements _ , ed . c. jaschek and p. c. keenan ( reidel ) , p. 243
vila - costas , m. b. and edmunds , m. g. 1993 , _ mon . not .
_ , 265 , 199 willson , l. a. , bowen , g. h. and struck , c. 1996 , in asp conf .
98 : _ from stars to galaxies _ , ed . c. leitherer , u. fritze - v .
alvensleben and j. huchra , p. 197
wood , p. r. 1981 , in _ physical processes in red giants _ , ed .
i. iben jr . and a. renzini ( reidel ) , p. 135
woosley , s. e. , hartmann , d. h. , hoffman , r. d. and haxton , w. c. 1990 , _
_ , 356 , 272 woosley , s. e. and weaver , t. a. 1995 , _ ap.j
_ , 101 , 181 zuckerman , b. and aller , l. h. 1986 , _ ap.j . _ , 301 , 772 | the role of carbon stars in the build - up of chemical elements in galaxies is discussed on the basis of stellar evolution calculations and estimated stellar yields , abundance analyses of agb stars , galactic - evolution models and abundance trends among solar - type disk stars .
we conclude that the agb stars in general , and carbon stars in particular , probably are main contributors of @xmath0-elements , that their contributions of flourine and carbon are quite significant , and that possibly their contributions of lithium , @xmath1c and @xmath2ne are of some importance
. also contributions of n , na and al are discussed .
the major uncertainties that characterize almost any statement concerning these issues are underlined . |
quantum dynamics of bosonic systems @xcite is a highly active branch of modern physics .
it plays a crucial role in experiments with trapped ultra cold atomic or molecular clouds especially when external trap potentials are varied in time .
theoretical predictions on the dynamics of quantum systems in these cases are highly desirable .
quantum dynamics is usually governed by the time - dependent schrdinger equation @xcite . to solve it means to define an initial condition , i.e.
, initial wave packet , and to find the evolution of this initial state in time , i.e. , to propagate it . in this work
we are interested in the many - body dynamics of quantum systems made of indistinguishable bosons . specifically , to solve the time - dependend multi - boson schrdinger equation we use a the multiconfigurational time - dependent hartree method for bosons ( mctdhb ) , developed recently in refs .
the mctdhb(@xmath2 ) is a very efficient and successful propagation tool , because it utilizes a very powerful ansatz for @xmath3-boson wave function , defined in a complete fock subspace , and spanned by all possible many - body basis functions ( permanents ) appearing as permutation of @xmath3 bosons over @xmath2 time - dependent self - consistent orbitals .
the key idea of the original mctdh technique ref .
@xcite was to consider as many - body basis functions the hartree products build of @xmath3 time - dependent orbitals , which can be , in principle , identical .
this method has been widely and successfully applied ( see ref .
@xcite ) to study multi - dimensional quantum dynamics of polyatomic molecules , where each vibrational mode has been assigned to a distinguishable quasi - particle .
lately , it was realized that by applying anti - symmetrization operation to the hartree products one can study dynamics of fermionic systems , so the original mctdh method has been expanded in ref .
@xcite to the mctdhf , here the letter `` f '' means fermions .
analogously , we took symmetrized hartree products ( permanents ) as many - body basis functions in ref .
@xcite and showed their applicability for stationary multi - boson problems and , for the first time in this context , reformulated the representation of the working equations in terms of elements of the reduced one- and two - body density matrices .
this reformulation has provided a deeper insight onto the underlying structure of the mctdh `` ideology '' and allowed us to derive in refs .
@xcite closed - from mctdhb equations for time - dependent bosonic problems and extend the mctdh ideas further in refs .
@xcite to mixed systems of bosons , fermions and spinors , as well as to systems with particle conversion ref . @xcite . in the latter case particles of one kind , e.g. , atoms , are allowed to undergo transformation and build up particles of another kind , i.e. , molecules , and vice - versa .
the developed mctdhb theory has been successfully applied to study a splitting of a repulsive bosonic cloud by ramping - up a potential barrier in ref .
an optimal control of this process has been addressed in ref .
the first- and second - order correlation functions of the split sub - clouds have been analyzed in ref .
@xcite . in ultra - cold atomic clouds with attractive inter - boson interactions
the split phenomena take even more interesting forms . in ref .
@xcite we have shown that by pumping energy into an initially - coherent localized attractive bosonic cloud one can fragment it into two dynamically stable sub - clouds , which are still strongly coupled . moreover , attractive bosonic systems also support another kind of dynamically stable fragmented states the famous quantum superposition states . these can be formed by scattering an attractive cloud from a potential barrier , as was predicted independently by two groups using absolutely different many - body methods @xcite . in ref .
@xcite within the framework of the mctdhb method we have shown that there is a much more efficient way to generate the quantum superposition states instead of scattering an initially - coherent attractive cloud from the potential barrier one can thread it by a barrier .
the mctdh method is so numerically stable and powerfull that one can use the many - body basis sets made of non - symmetrized hartree products , originally proposed to describe distinguishable particles , to study systems of indistinguishable particles .
the configurational spaces in these cases , however , are much larger than actually needed restricting the applicability of such a scheme to few - particle problems .
nevertheless , numerically - exact studies of few - boson problems using this scheme have been successfully done in refs .
@xcite . to specify the many - body wave function in mctdhb means to provide two sets of complex numbers a set of expansion coefficients , i.e. , a vector - array of the size of the respective fock subspace , and a set of @xmath2 one - particle functions usually defined on a real space grid .
both these sets are time - dependent , i.e. , they are changing in time .
the time - evolution of the expansion coefficients in the fock space and the orbitals in space are governed by the respective mctdhb equations @xcite . in ref .
@xcite we have developed a novel , effective and general algorithm to operate with state vectors in fock space , without resorting to matrix representation of operators and even to its elements .
now our goal is to utilize , implement , and apply these ideas to the mctdhb method .
in particular , we expect that a parallel implementation of the mctdhb method augmented by the proposed algorithm allows us to propagate the time - dependent schrdinger equation faster and more efficiently for larger configurational spaces , and thus be able to attack bigger systems , and to propagate for longer times . in this paper
we apply the mctdhb method augmented by the invented algorithm @xcite to study real - space dynamics of repulsive bosonic systems made of @xmath0 , 51 and 3003 bosons in triple - well traps with periodic boundaries . as initial wave packets
we take fully threefold fragmented states and displace them suddenly and strongly out of equilibrium .
the periodic trap used for the propagations supports at least six bound eigenstates per well , i.e. , six bands are available for the quantum dynamics . as a consequence of the displacement applied , the energies per particle of the evolving wave packets are located between the fourth and fifth bands which implies strongly non - equilibrium dynamics . on the one hand
, we would like to see how the properties of these highly non - equilibrium systems change in time and what are the most relevant physical phenomena behind .
for example , to clarify the role of finite @xmath3 effects we analyze and compare the quantum dynamics of the systems made of different number of bosons but characterized by very similar interparticle interaction energies . on the other hand
, we also plan to demonstrate the convergence of the mctdhb(@xmath2 ) method , for short , intermediate and long propagation times .
more specifically , we compare the computations done at different levels @xmath2 of the mctdhb(@xmath2 ) theory , i.e. , by systematically increasing the number @xmath2 of time - dependent self - consistent orbitals .
the key methodological consequence of this study is to show that due to the algorithm used one can tackle large configurational spaces thus opening many new opportunities for the mctdhb method .
the structure of the paper is as follows . in section [ secii ]
we consider a general many - boson system in real space , define its hamiltonian and the @xmath3-boson state vector ( wave function ) .
the basics of the multiconfigurational time - dependent hartree method for bosons ( mctdhb ) are given in subsection [ secii.1 ] .
the implementation of the novel idea how to operate by a bosonic hamiltonian on a state vector within the mctdhb method is discussed in subsection [ secii.2 ] . in particular , we describe how to map and address permanents ( configurations ) , how to operate with one- and two - body operators as well as with total hamiltonian , and how to construct the respective expectation values .
application of the mctdhb method to many - boson quantum dynamics in triple - well traps is illustrated in section [ seciii ] . in subsection [ seciii.1 ]
we provide the specification of the systems under consideration initial and final trap potentials , initial wave packets , interparticle interactions , etc .
the detailed analysis of the evolving wave packets for the systems made of @xmath0 , 51 and 3003 bosons are reported in the subsections [ seciii.2 ] , [ seciii.3 ] , and [ seciii.4 ] respectively .
section [ seciv ] summarizes and concludes .
the quantum dynamics of a system made of @xmath3 bosons is governed by the many - body time - dependent schrdinger equation .
this equation describes the evolution of a many - boson wave function @xmath4 in time : h ( t ) & = & i , [ mbse ] + h(_1,_2,
,_n ) & = & _ j=1^n
h(_j ) + _ k > j=1^n
w(_j-_k ) . [ mbse1 ]
were @xmath5 , @xmath6 is the coordinate of the @xmath7-th boson , @xmath8 is the one - body hamiltonian containing kinetic and potential energy terms , and @xmath9 describes the pairwise interaction between the @xmath7-th and @xmath10-th bosons . in the most general case , the one - body potential @xmath11 and the two - body interaction @xmath9 and , hence , the many - boson hamiltonian @xmath12 itself may be time - dependent . to solve eq .
( [ mbse ] ) means to specify initial conditions at @xmath13 and propagate the ansatz in time .
more precisely , we have to specify the many - body wave function @xmath14 , i.e. , an object which depends on coordinates @xmath15 of each of @xmath3 particles and by using eq .
( [ mbse ] ) find the many - boson wave function at desired time @xmath4 .
general analytical solutions of the time ( in)dependent schrdinger are unknown , therefore , different numerical approaches are in use @xcite . in this section ,
we consider one of them the multiconfigurational time - dependent hartree method for bosons ( mctdhb ) .
the general formulation and derivation of the mctdhb method have been given elsewhere , see refs .
@xcite for the details . here , we present the final equations and give some remarks on the numerical realization of the novel ideas in @xcite how to operate in a configurational ( fock ) space without resorting to the matrix representation of the respective operators .
the mctdhb ansatz , i.e. , the proposed structure of the many - boson wave function is a linear combination of time - dependent many - body basis functions @xmath16 : [ mctdhb_ansatz ] |(t ) > = ^n__c_(t)|;t>. for a system of indistinguishable bosons we take permanents as many - body basis functions .
every permanent is constructed as a symmetrized hartree product of @xmath3 one - particle functions ( orbitals ) : [ basic_permanents_x ] _
1,_2,_3,,_n @xmath17 bosons reside in the first orbital , @xmath18 bosons reside in second one , etc . , and
@xmath19 bosons reside in the @xmath2-th orbital .
the possible occupation patterns @xmath20 , also called configurations , are constrainted only by the condition that the total number of particles is conserved @xmath21 .
so , the complete configurational space of the standard mctdhb(@xmath2 ) method is formed by all possible permanents obtained by distributing @xmath3 bosons over @xmath2 orbitals @xcite : [ nconf ] n_= , where @xmath22 .
the specification mctdhb(@xmath2 ) means that @xmath2 self - consistent time - adaptive orbitals have been used in the computation and the size of the respective fock space , i.e. , the total number of the configurations is determined by eq .
( [ nconf ] ) . at
any given time the many - boson mctdhb(@xmath2 ) wave function @xmath23 is known if the set of expansion coefficients @xmath24 , i.e. , a complex array of @xmath25 elements and the set of @xmath2 one - particle functions @xmath26 are specified . here , we do not discuss how one - particle functions are represented because this depends on the dimensionality of the problem and on the particular realization of the code .
for example , in 1d cases , one can represent each of these functions on a @xmath27-dimensional grid as a vector - array of complex numbers .
the mctdhb(@xmath2 ) ansatz eq .
( [ mctdhb_ansatz ] ) possesses a standard feature of any many - body time - dependent expansion its expansion coefficients @xmath28 depend on time .
it reflects the basic idea of quantum propagation : the many - body wave packet is composed of contributions of very many states ( configurations ) and these contributions are changing in time . the ansatz eq .
( [ mctdhb_ansatz ] ) has an additional , the mctdhb- specific feature dependency of the orbitals , i.e. , one - particle functions @xmath29 used to construct the permanents eq .
( [ basic_permanents_x ] ) on time .
the orbitals @xmath29 of the evolving many - boson wave packet are optimized at every point in time according to the dirac - frenkel variational principle @xcite , changing , thereby , the many - body basis permanents .
we shall see later how this additional mctdhb - specific feature allows one to span effectively much larger configurational spaces in comparison to methods using fixed , i.e. , time - independent orbital basis sets
. in the mctdhb theory the expansion coefficients @xmath28 and the shapes of one - particle functions @xmath26 are considered as independent variational parameters , see refs .
we substitute the mctdhb(@xmath2 ) ansatz eq .
( [ mctdhb_ansatz ] ) into the hamiltonian and introduce lagrange multipliers to guarantee the normalization of the total wave function as well as orthonormalization of the orbitals .
then , by utilizing the dirac - frenkel time - dependent variational principle @xcite we get a coupled system of integro - differential equations which have to be solved ( integrated ) simultaneously . according to the derivation
, the coupled system is composed of two sets of the variational equations , one for the expansion coefficients @xmath28 and another one for the one - particle functions @xmath29 . in numerical implementations of these equations , it is usually assumed that during the elementary propagation of the first set of variational parameters , say expansion coefficients from @xmath28 to @xmath30 , the second one , i.e. , the orbital set @xmath29 , is kept fixed , and vice - versa , during the propagation of the orbitals from @xmath29 to @xmath31 the @xmath28-set is kept fixed .
of course , the @xmath32 has to be sufficiently small .
the first set of the mctdhb equations describes the evolution of @xmath25 expansion coefficients and can be rewritten as : [ mctdhb_ci ] i= ( t ) , where @xmath33 is the matrix representation of the hamiltonian . in this equation
it is assumed that at each point in time , the shapes of the orbitals @xmath29 are known and , therefore , during an elementary numerical propagation time step they remain unchanged , i.e. , fixed . hence , the original real - space hamiltonian operator [ see eq .
( [ mbse1 ] ) ] is represented now in the second quantized form : [ mb_ham ] h = _ k , q h_kq b_k^b_q + _ k , s , q , l w_ksql
b_k^b_s^b_l b_q , where the elements @xmath34 of the one- and @xmath35 of the two - body operators are directly available as integrals over the discussed above orbitals : [ matrix_elements ] h_kq & = & _ k^(,t ) h ( ) _ q(,t ) d , + w_ksql & = & _ k^(,t ) _
s^(,t ) w(- ) _ q(,t ) _
l(,t ) dd. the second set of the mctdhb equations , also called equations - of - motion for orbitals , describes the evolution of @xmath2 orbitals @xmath36 and can be written as : [ mctdhb_psi ] i|_j > & = & , j=1,,m , + & = & 1-_j=1^m|_j > , + _ kslq&=&. since the orbitals @xmath37 are time - dependent functions , the integrals @xmath38 and @xmath35 , computed according to eq .
( [ matrix_elements ] ) become time - dependent functions even for time independent hamiltonians . hence
, the price of using the self - consistent time - adaptive orbitals is that at every elementary numerical propagation time step one has to update these integrals , i.e. , the second quantized form of the hamiltonian , eq .
( [ mb_ham ] ) . having the mctdhb(@xmath2 )
wave function @xmath23 at every given time , i.e. , the @xmath25 complex numbers @xmath28 and @xmath2 one - particle functions @xmath29 , we have an access to the elements of the reduced one- , two- or higher - order density matrices and can construct , thereby , respective correlation functions @xcite or compute any desired properties .
specifically , in this work we construct and diagonalize the reduced one - body density matrix : [ no ] ( , ;t)&= & = + & = & _ k , q _ kq _ k^(,t ) _ q(,t)=n_k n_k(t ) _
k^no(,t ) _
k^no(,t ) , the obtained eigenvalues @xmath39 and eigenvectors @xmath40 are called natural occupation numbers and natural occupation orbitals , respectively .
hence , this diagonalization procedure is often called natural analysis .
a simple visualization of the reduced one - body density is obtained by plotting only its diagonal part @xmath41 , also referred to as density .
summarizing , there are @xmath25 mctdhb equations ( [ mctdhb_ci ] ) describing the evolution of the expansion coefficients , and @xmath2 non - linear integro - differential equations ( [ mctdhb_psi ] ) describing the evolution of the one - particle functions . to propagate the initial state , we divide time onto small elementary slices or steps . during every elementary time step either set of expansion coefficients or one - particle functions
is assumed to be unchanged .
this allows within a given time step to integrate the mctdhb eqs .
( [ mctdhb_ci ] ) and eqs .
( [ mctdhb_psi ] ) independently , see refs .
@xcite for more details .
the evolution of the one - particle functions , i.e. , propagation of the respective eqs .
( [ mctdhb_psi ] ) can be done by using general variable - order integrators , such as the adams - bashforth - moulton ( abm ) predictor - corrector integrator @xcite . to propagate eqs .
( [ mctdhb_ci ] ) for expansion coefficients we can use the short iterative lanczos ( sil ) scheme @xcite , because in the present work we deal with hermitian hamiltonians , i.e. , with real traps and real interparticle interaction potentials .
the key ingredient of the sil technique is a @xmath42-dimensional krylov subspace , formed by repeated operations of the hamiltonian : @xmath43 . to integrate the configurational part of the mctdhb(@xmath2 ) equations
[ see eqs .
( [ mctdhb_ci ] ) ] at a given point in time we need these vectors only . in a standard computational scheme to construct krylov vectors at every integration step one
first recalculates the hamiltonian matrix elements @xmath44 which is a quite tedious task that even for a moderate - sized systems .
and then by multiplying this matrix with the complex vector - array of the expansion coefficients repeatedly one builds up the krylov subspace of the desired size . in ref.@xcite
we invented an alternative technique to compute the action of the hamiltonian on a state vector .
now , we discuss how to implement the proposed ideas to the mctdhb method to obtain quickly and efficiently all the required vectors without ( re)construction of the respective hamiltonian matrix at every point in time .
this allows , as we shall demonstrate below , for faster operations in larger configurational spaces .
the propagation of the mctdhb wave function within a given time interval consists of two steps : propagation of the expansion coefficients keeping the orbitals fixed , eqs .
( [ mctdhb_ci ] ) , and propagation of the orbitals keeping the expansion coefficients unchanged , see eqs .
( [ mctdhb_psi ] ) .
the key operation at the first step is the action of the hamiltonian eq .
( [ mb_ham ] ) on a state vector @xmath45 , having at hands the integrals @xmath34 and @xmath35 .
at the second step we need to know the elements @xmath46 and @xmath47 of the reduced one- and two - body density matrices .
this subsection shows how to get these quantities by implementing and utilizing the ideas reported in ref .
@xcite . as we have seen above , the set of the one - particle functions @xmath29 and the set of expansion coefficients @xmath28 , i.e. , the complex array of @xmath25 elements , specify the mctdhb wave function @xmath23 at any given time . we first introduce a simple and compact scheme to enumerate elements of the complex array @xmath48 .
more strictly , we utilize the explicit scheme of ref .
@xcite to map @xmath2 integers @xmath49 characterizing each configuration to one integer @xmath50 addressing it as a coordinate of the state vector .
the address @xmath50 of the bosonic configuration @xmath51 in the state vector is computed as : @xmath52 technically , it means that for any given set of @xmath2 integer occupation numbers @xmath53 we can immediately compute one integer
its address @xmath50 . consequently , using this mapping we can access any desired element of the complex array @xmath48 by providing the respective set of the occupation numbers .
the next question is how to access the elements of the @xmath48 array .
let us take the definition of the mctdhb(@xmath2 ) ansatz ( [ mctdhb_ansatz ] ) and rewrite it : [ j_psi ] | > = ^n__j=1c_j|j ( ) > , where the summation index @xmath50 running over all @xmath25 configurations is associated with the above defined mapping index eq .
( [ j - numbering ] ) .
to access every @xmath54 element by starting one _ do-_loop over @xmath50 is inconvenient , because one has to , first , restore @xmath2 occupation numbers @xmath55 corresponding to every given address @xmath50 at every step , i.e. , to solve the respective inverse problem .
that is in principle possible but not necessary . to avoid it ,
we start , instead , @xmath2 nested _
do-_loops running over occupation numbers @xmath56 with the condition that @xmath57 , the index @xmath50 is immediately available by applying eq .
( [ j - numbering ] ) .
hence , in the case of an @xmath2-loop implementation there is no need to solve the inverse problem at all .
from now on , by writting a sum over index @xmath50 we actually mean that in practical implementations @xmath2 nested loops over @xmath56 are running instead .
the repeated recomputations of the binomial coefficients needed by eq .
( [ j - numbering ] ) can be avoided by introducing a small static two - dimensional array and filling it only once by the respective binomial integers .
hence , to find the address @xmath50 corresponding to a given set @xmath55 of the occupation numbers one needs to sum up @xmath58 elements of this integer array .
now , we approach the key issue how to find the result of the action of the hamiltonian eq .
( [ mb_ham ] ) on a bosonic state vector . within a standard quantum mechanical text - book approach , e.g. , ref .
@xcite , to get @xmath59 one needs , first , to construct the respective hamiltonian matrix and then to multiply it by the complex vector of expansion coefficients @xmath48 . in ref .
@xcite we propose a simple alternative to it .
the hamiltonian is a sum of one - body @xmath60 and two - body @xmath61 terms . if the result of action of every basic one - body @xmath62 and two - body @xmath63 operator on a state vector is known , the total action of the hamiltonian is obtained by summing them up . in @xcite
we discussed these operations in great details , here we give only the final formulations . in ref .
@xcite we have seen that the action of the @xmath62 combination of creation and annihilation operators on every configuration @xmath64 of the incoming state vector can be interpreted as a re - addressing of this configuration to another one @xmath65 scaled by a statistics - dependent weight .
therefore , using the introduced above mapping we can get the result of the action of the @xmath62 on every component of the bosonic state vector @xmath45 , for a given pair @xmath10 and @xmath66 : [ one - body_oper ] | ^kq >
b^_kb_q | > = ^n__jc^kq_j|j ( ) > , + c^kq_j = c_j^kq . as we have discussed above , instead of running one _ do-_loop over index @xmath50
, we start here @xmath2 nested _ do-_loops over @xmath56 . the occupation pattern of the current configuration @xmath50 is explicitly known @xmath67 .
the occupation pattern of the configuration that we reffer to as @xmath68 is obtained from the current one by reducing the occupation number of the @xmath10-th orbital by one , i.e. , now it becomes @xmath69 and by increasing the occupation number of the @xmath66-th orbitals by one @xmath70 : @xmath71 .
the change between @xmath66 and @xmath10 indices appears due to the change of the summation index , see ref .
@xcite for more details .
using eq .
( [ j - numbering ] ) we compute the addresses of the current @xmath72 and referred @xmath73 configurations .
the element @xmath74 with address @xmath50 of the resulting state vector is obtained as a product of the @xmath75 component of the incoming state vector , having address @xmath68 , and the @xmath76 prefactor .
this simple and straightforward methodology is ideally suitable for programming .
moreover , within the framework of this method we also get a direct access to the elements @xmath46 of the reduced one - body density matrix . indeed , according to the standard definition , for a given state vector @xmath45 , the one - body density matrix element reads : [ rho_ij ] _
kq= ] = = ^n__jc^*_jc^kq_j .
thus , the matrix elements @xmath46 can be immediately computed as a dot - product of the incoming @xmath45 and resulting @xmath77 state vectors .
similarly , we get the action of the general ( @xmath78 ) two - body @xmath63 operator on the incoming state vector @xmath45 : [ two - body_oper ] |^kslq > b^_kb^_sb_lb_q| > = ^n__jc^kslq_j|j ( ) > , + c^kslq_j = c_j^kslq , where the current index @xmath79 of the resulting vector as well as the address of the incoming vector , which we refer to @xmath80 are computed using eq .
( [ j - numbering ] ) .
the respective element of the two - body density is obtained as a dot - product of the incoming @xmath45 and resulting @xmath81 state vectors : [ rho_ijkl ] _
kslq= ] = = ^n__jc^*_jc^kslq_j .
obviously , in the discussed scheme the elements of the one- and two - body density matrices , needed by the second part of the mctdhb(@xmath2 ) equations ( [ mctdhb_psi ] ) are naturally available .
clearly , the elements of three- and higher - body density matrices can be obtained in a very similar way .
the following strategy is implemented in the mctdhb(@xmath2 ) code to obtain the result of the action of the hamiltonian on a state vector .
first , for a given vector @xmath45 , i.e. , a given set of the expansion coefficients @xmath48 , the actions of every one- and two - body basic operators are evaluated using eqs .
( [ one - body_oper ] ) or ( [ two - body_oper ] ) . each of these actions can be computed independently , implying effective parallelization .
namely , the @xmath48 array , corresponding to the incoming vector is copied over all the available nodes .
then , on every node , independently , we start @xmath2 nested loops to get the resulting @xmath77 and @xmath81 vectors , corresponding to every pair @xmath82 and quartet @xmath83 .
second , by taking the dot - product of the incoming and the resulting state vectors , i.e. , by using eqs .
( [ rho_ij ] ) or ( [ rho_ijkl ] ) we obtain the respective elements @xmath46 and @xmath84 of the reduced one- and two - body density matrices .
third , by multiplying all the elements of the resulting vectors by the corresponding integrals we get the desired actions of the respective @xmath85 and @xmath86 terms of the hamiltonian on the incoming state vector .
fourth , by summing up the resulting vectors from all the terms we obtain the total action of the hamiltonian on the incoming state vector .
finally , we easily compute the expectation value of the hamiltonian @xmath87 as a dot - product of the total resulting and incoming state vectors . in principle
, the desired vector @xmath59 can be obtained by collecting actions from one- and two- body operators into two separate groups and summing only two resulting vectors afterwards . using the results of the previous subsections we determine the total action of all the one - body terms on an initial state vector @xmath45 as : [ h - psi_one - body ] h | > & = & _ k , q h_kq = _ k , q h_kq |^kq>=^n__jc^h_j|j ( ) > , + c^h_j&= & _ k , q h_kq c^kq_j , where every component @xmath88 of the resulting vector having address @xmath72 , is obtained as a sum of the @xmath89 numbers evaluated for every given @xmath10 and @xmath66 according to eq .
( [ one - body_oper ] ) .
analogously , we group the contribution from all the two - body terms : [ h - psi_two - body ] w | > & = & _ k , s , q , l w_ksql = _ k , s , q , l w_ksql c^w_j&=&_k , s , q , l w_ksql c^kslq_j , where @xmath90 are computed using eq .
( [ two - body_oper ] ) for every given quartet @xmath83 of the summation indices .
finally , the desired result @xmath59 is obtained as a sum of the two computed above vectors eqs .
( [ h - psi_one - body ] ) and ( [ h - psi_two - body ] ) : [ hpsi ] h | & = & h |>+w c^h_j&=&c^h_j + c^w_j , and the respective expectation value of the hamiltonian is given as a dot product : [ psihpsi ] ] + ] = ^n__jc^*_j ( c^h_j+c^w_j ) .
these closed form results can be considered as a formal proof that one can operate with state vectors without construction of the corresponding hamiltonian matrix .
it also justifies the used representation of eq .
( [ mctdhb_ci ] ) .
now , instead of the standard matrix representation of the time-(in)dependent schrdinger equations we can rewrite them in terms of the enumerated configurations , i.e. , in a formally `` vectorized '' form : [ newform ] i & = & c^h_j , j , + c^h_j & = & e c_j , j , where every @xmath91 element is computed according to eq .
( [ hpsi ] ) as a sum of the respective one- ( [ one - body_oper],[h - psi_one - body ] ) and two - body ( [ two - body_oper],[h - psi_two - body ] ) contributions .
let us conclude .
we implement in the mctdhb(@xmath2 ) method the ideas proposed in ref .
@xcite allowing to determine the result of the action of the hamiltonian written in second quantized form on a state vector without building up the respective hamiltonian matrix .
the implemented computational scheme has several advantages comparing to the standard one .
first of all it does not require the ( re)evaluation of the hamiltonian matrix elements in the given many - body basis set at each time step .
consequently , there is no need to construct , store and address the hamiltonian matrix or its elements at all .
second , the elements of the reduced one- and two - body matrices , needed by the mctdhb(@xmath2 ) method are immediately available as scalar products of the incoming and resulting state vectors , obtained by acting the respective combination of creation and annihilation operators .
third , the proposed ideas can be easily extended to three - body and higher - order interactions potentials .
last but not least , the proposed ideas are implemented using very effective parallelization schemes , which are of high demands in modern computational physics .
condensation is one of the key physical phenomena taking place in quantum many - boson systems @xcite . to characterize a many - boson quantum state one constructs and diagonalizes the reduced one - body density matrix ,
( [ no ] ) .
according to penrose and onsager ref .
@xcite the bosonic state is condensed if only one natural orbital has macroscopic occupation . on the other hand ,
when several natural orbitals are macroscopically occupied the system becomes several - fold fragmented @xcite .
fragmentation usually takes place in repulsive condensates due to multi - well structure of the external potentials trapping bosonic clouds .
increasing , for example , the height of the barrier separating two wells of the double - well trap one can change the ground state property of a weakly - interacting repulsive many - boson system from almost pure condensed to fully twofold fragmented refs .
analogously , ground state properties of repulsive bosons in optical lattices depend on the intensity of the laser light forming the lattice .
the ground state of weakly interacting bosons in shallow optical lattices is superfluid , i.e. , condensed @xcite . increasing the intensity of the optical potential ,
i.e. , its depth , one can drive the system from the superfluid state to a completely different , so called mott insulating state @xcite . considering an optical lattice as an extreme case of a multi - well trap , one can associate the fragmented states in multi - well systems with mott insulating states in optical lattices .
of course , occupations of the fragments in fragmented states of multi - well systems are macroscopic , while in optical lattices the mott insulating states are usually characterized by population of a few bosons per site . for trap potentials manipulated in time , the quantum dynamics starts to play a very crucial role .
for example , when a repulsive bosonic cloud is split by ramping - up a potential barrier the final state reveals long - lived oscillatory dynamics between condensed and twofold fragmented states @xcite .
similarly , when the intensity of the optical lattice is driven non - adiabatically , one observes dynamical oscillations between superfluid and mott insulating states @xcite .
these oscillations are damped . in the numerical examples of the present section we consider many - boson systems trapped in a real - space one - dimensional triple - well potential and
study their evolutions by manipulating the shape of the trap , and varying the number of bosons and their interaction strength . compared to numerous theoretical and experimental studies of bosonic systems trapped in double - well traps ,
studies of bosonic systems in triple - wells are quite scarce .
however , they can provide a deeper insight into the physics of the additional interactions appearing between neighboring wells due to the presence of the third one .
this makes one step forward to more detailed understandings of the dynamical oscillations between superfluid and mott insulating states taking place in optical lattices . in this work we use dimensionless units . to translate the time - dependent many - boson schrdinger equation eq .
( [ mbse ] ) to dimensionless units we divide the hamiltonian by the energy unit @xmath92 , where @xmath93 is the mass of a boson and @xmath94 is a typical length scale of the system .
the dimensionless one - body hamiltonian then reads @xmath95 .
the bosons interact via the popular short - range contact repulsive potential @xmath96 of strength @xmath97 , see refs .
@xcite and references therein .
let us consider systems made of @xmath0 , 51 and 3003 repulsive bosons confined in a periodic triple - well potential .
we use periodic boundary conditions to ensure that all three wells are equivalent . for weakly interacting bosons and sufficiently large barrier heights
one expects that the ground state of the system becomes fully threefold fragmented , i.e. , according to the standard definition @xcite three natural orbitals in eq .
( [ no ] ) acquire macroscopic occupations . from a perspective of optical lattice physics such a state
can be associated with a perfect mott insulating state having occupation of @xmath98 bosons per site .
we use these fully threefold fragmented states in our dynamical studies as initial wave packets for propagation . at the beginning of the propagation
the optical potential has been suddenly shifted and ramped down to some final value @xmath99 .
we summarize all the trap manipulations as follows : [ trap ] v(x , t)=v ( x/2+)^2 , x [ -3,3 ) \ { t<0 , v&=&v_0 , = 0 + t0 , v&=&v_final , = /4 .. here , @xmath100 is the barrier height of the initial trap where fully treefold fragmented states have been prepeared at @xmath101 .
@xmath102 is the barrier height of the final trap where we study the dynamics for @xmath103 , and @xmath104 is the shift applied to this final potential .
we propagate the time - dependent schrdinger equation for different numbers @xmath3 of bosons in real space within framework of the mctdhb(@xmath2 ) method , as described in the previous section [ secii ] .
the interparticle interaction is chosen to be @xmath105 . in this case
the systems made of different numbers @xmath3 of bosons become equivalent in the sense that the fraction of interparticle interaction energy , also associated with the total non - linearity strength @xmath106 , remains the same for every system .
consequently , the total energies per particle in these systems are expected to be very close to each other .
that allows us to compare their evolutions . in fig .
[ fig.0 ] we schematically depict the initial and final shifted trap potentials as well as the initial many - body density .
the topology of the used triple - well trap implies triple degeneracy of the respective one - particle energy levels @xmath107 of @xmath108 , i.e. , @xmath109 , @xmath110 , etc . again , using the analogy between optical lattices and multi - well potentials we can associate each of these triples of quasi - degenerate levels with a band .
the split between the levels forming a band is smaller for deeper bands . in fig .
[ fig.0 ] we show the lowest - energy levels of each band .
the energy per particle @xmath111 of the evolving systems is also depicted by a red solid bold line . generally speaking for different @xmath3 these energies
are slightly different , see table .
[ tab.1 ] , but at the plotted scale these differences are indistinguishable .
it is important to stress that here we study the dynamics of the interacting bosonic systems in the periodic triple - well potential where at least six bands are available , see fig .
[ fig.0 ] .
the shift @xmath104 applied to the final potential means that every initial wave packet is pushed far away from equilibrium . due to this shift
a lot of energy has been pumped into the system , such that the energy per particle of the initial wave packet is located around the fourth band .
hence , the system is in a non - equilibrium highly excited state .
after the sudden shift the total energy of the system is conserved because during the wave packet propagation the final trap potential remains unchanged and the respective hamiltonian is time - independent and hermitian .
we first study the dynamics of the system made of @xmath0 bosons .
the ground state of such a system becomes fully threefold fragmented already at the barrier height @xmath112 .
however , to be able to relate later the results to those obtained for systems made of different numbers of bosons we choose as an initial state the ground state at a much higher barrier @xmath100 . to get the initial state we propagate the mctdhb(@xmath2 ) equations eqs .
( [ mctdhb_ci],[mctdhb_psi ] ) in imaginary time until the total energy is converged to the desired precision of @xmath113 .
we recall that throughout this work we use dimensionless units . to simplify the differentiation and integration operations appearing in eq .
( [ mctdhb_psi ] ) we use a discrete variable representation ( dvr ) , see ref .
@xcite , for the self - consistent time - adaptive orbitals .
more specifically , each time - dependent orbital is represented in an exponential dvr grid with @xmath114 or @xmath115 points .
so , in the present computations the linear combination of the @xmath116 primitive exponential functions define the shape of each orbital at every point in time .
the quality of the used dvr grid can be seen in fig .
[ fig.0 ] where the density of the initial state is plotted as a dotted line : every depicted point corresponds to the actual dvr point in coordinate space ( @xmath114 ) .
we computed the ground state many - boson wave function at different levels of the mctdhb(@xmath2 ) theory , namely with @xmath1 , 6 , 9 and 12 self - consistent time - adaptive orbitals .
the obtained ground state energies per particle @xmath117 of the initial fully threefold fragmented states are presented in table .
[ tab.1 ] . from this table
we clearly see that the ground state energy is converged with increasing size @xmath25 of the configurational subspace , i.e. , with increasing level @xmath2 of the mctdhb(@xmath2 ) theory .
now , it is worthwhile to clarify the role of self - consistency in the mctdhb method . according to eq .
( [ nconf ] ) , for @xmath0 bosons and three ( @xmath1 ) self - consistent orbitals the total number of configurations used to describe the many - boson wave function eq .
( [ j_psi ] ) is @xmath118 .
however , since in the numerical computations each orbital is represented as a sum of @xmath116 exponential dvr functions we can formally expand these sums and compute the total size of the fock subspace spanned by these _ fixed _ primitive dvr functions .
for example , for @xmath114 exponential functions used in the present computation the formal dimension of the respective fock space is huge
@xmath119 , i.e. , it is larger than @xmath120 ! clearly , this is beyond reach and one can not carry out even approximately converged computations using the standard method of fixed orbitals .
the mathematical role of the self - consistent orbitals in mctdhb is to contract the fixed primitive dvr basis functions ( exponential functions in the present example ) , according to the variational principle , and to project , thereby , the huge configurational subspace to the optimal and compact one . from a physical point of view
this means that every excitation constructed using self - consistent orbitals actually utilizes a huge number of the primitive excitations built on the fixed dvr functions . as we see from table .
[ tab.1 ] even a modest number of the self - consistent orbitals can provide a very accurate description of the quantum system . from table .
[ tab.1 ] we see that the ground state energy obtained at the three - orbital level of the mctdhb(@xmath1 ) theory is very close to those computed using @xmath121 , 9 , 12 self - consistent orbitals .
we analyze the properties of the respective solutions by constructing and diagonalizing the reduced one body density [ see eq .
( [ no ] ) ] and find that , indeed , only three natural orbitals are macroscopically occupied : @xmath122 .
we conclude that a physically correct description of the ground state is obtained already at the mctdhb(@xmath1 ) level
. further increase of the hilbert subspace does not lead to significant changes , but costs , of course , considerable computational efforts ( see @xmath25 in table . [ tab.1 ] ) .
the ground state density obtained for @xmath123 is depicted in fig .
[ fig.0 ] .
the ground state density profiles , obtained using @xmath1 , 6 , 9 self - consistent orbitals can not be distinguished from this one and , therefore , are not shown . .
the energies of the ground and evolving states , and the number of configurations @xmath25 obtained at different levels @xmath2 of mctdhb(@xmath2 ) theory , i.e. , by using different numbers @xmath2 of self - consistent time - adaptive orbitals .
results are shown for systems made of @xmath0 , 51 and 3003 bosons . [ cols="<,^,^,<,<",options="header " , ] we now turn to the propagation of the initial many - boson wave functions obtained above .
let us first define the natural time - scale of the problem .
the energy difference between the second and first bands of the periodic potential @xmath124 is @xmath125 , where @xmath126 and @xmath127 are the respective lowest - energy levels forming the first and second bands .
this energy difference relates to oscillations of the density in each well ( on each site ) and can be used to define the time - scale of the problem @xmath128 .
we present here the results of the propagation until @xmath129 which approximately corresponds to @xmath130 periods of oscillations of the one - site densities . having obtained the initial - state wave function , i.e. , respective expansion coefficients and one - particle functions ( orbitals ) , we use them as initial conditions for integration of the mctdhb(@xmath2 ) equations ( [ mctdhb_ci ] ) and ( [ mctdhb_psi ] ) with @xmath1 , 6 , 9 , 12 orbital sets .
we construct and diagonalize the reduced one - body density eq .
( [ no ] ) at each integration time for every set of the orbitals .
the diagonal parts of the reduced one - body densities obtained for a short @xmath131 , intermediate @xmath132 and long @xmath133 propagation times are depicted in the bottom , middle and upper panels of fig .
[ fig.i_a ] respectively .
it is clearly seen that the densities obtained at the mctdhb(@xmath1 ) level have only marginal deviations from the @xmath121 , 9 , 12 ones at all the presented times .
so , the first conclusion is that three time - dependent ( self - consistent ) orbitals provide the physically correct description of the density of the evolving many - boson wave packets for quite long propagation times . the convergence with @xmath2 is clearly seen in fig . [ fig.i_a ] .
now , we compare more involved properties of the many - body wave functions
. the most important characteristics of bosonic systems are the degrees of condensation and fragmentation .
these quantities correspond to the values of the natural occupation numbers , computed according to eq .
( [ no ] ) . in fig .
[ fig.i_b ] we collect the evolutions of the occupation numbers as functions of propagation time computed at different levels @xmath2 of the mctdhb(@xmath2 ) theory , i.e. , within different numbers of the time - adaptive orbitals used .
the upper left panel depicts evolutions of the twelve natural occupation numbers obtained at the @xmath123 level of computation the best converged variational results which we use as a reference .
we recall first that at @xmath134 the initial state is fully threefold fragmented or using the terminology of optical - lattice physics it is a pure mott insulating state with @xmath135 atoms per site .
second , the evolution takes place in real space and the shapes of the wave packets are drastically changing in time , thereby demonstrating the importance of time - dependent orbitals .
third , the applied shift of the trap potential results in considerable pumping of energy into the system compare the energies listed in table .
[ tab.1 ] for the initial ground @xmath136 ( @xmath101 ) and evolving @xmath137 ( @xmath138 ) wave packets .
the evolving wave packets are located energetically between the fourth and fifth bands as schematically depicted in fig .
[ fig.0 ] . nevertheless , despite such a highly non - equilibrium problem
, we see that only three natural orbitals and the respective occupation numbers provide dominant contributions to the physics of the evolving wave packets , up to rather long propagation times . in other words ,
three time - adaptive orbitals are capable to describe the many - boson quantum dynamics of the system .
the evolutions of the three largest occupation numbers computed at different levels @xmath2 of the mctdhb(@xmath2 ) theory , are compared in the three other panels of fig .
[ fig.i_b ] : @xmath17 in the right upper , @xmath18 in the left bottom , and @xmath139 in the right bottom panel .
the plotted curves corresponding to the evolutions of @xmath140 and @xmath139 are not smooth because the natural occupation numbers obtained by numerical diagonalizing the density matrix eq .
( [ no ] ) are sorted according to their values .
therefore , instead of nicely separated crossing curves we see their connected parts . in fig .
[ fig.i_b ] one can see that until @xmath141 all the mctdhb(@xmath2 ) curves for @xmath1 , 6 , 9 , 12 are indistinguishable on the resolution of the scale shown .
in other words , for this time interval which is about @xmath142 on - site oscillations @xmath143 the three orbital ansatz provides essentially exact description of the many - body dynamics . another interesting feature observed from figure fig .
[ fig.i_b ] is that until @xmath144 the evolving system stays in a quite pure insulating state @xmath145 , despite the fact that the on - site densities undergo drastic changes they are split onto three big humps separated by very deep dips , see lower panel of fig .
[ fig.i_a ] . with time
the many - body dynamics destroys the perfect threefold fragmentation of the initial state . during the propagation two of the three natural orbitals with largest occupancies are evolving in time in a very similar manner forming a pair , while the third orbital is split off and demonstrates a quite different behavior ,
see upper left panel of fig .
[ fig.i_b ] .
we might associate these evolutions with the structure of the one - particle states of the triple - well trap used .
we recall that each band of the triple - well periodic potential eq .
( [ trap ] ) is formed by three discrete eigenstates . in the lowest bands
all of them are almost degenerate , in the higher ones two of the three one - particle eigenstates are degenerate while the third one is split off .
we thus conclude that the topology of the external trap might be responsible for the observed behavior of the natural occupation numbers of the evolving many - body wave packets .
the structure of the evolving natural occupation numbers presented in fig . [ fig.i_b ] has a very complicated character .
we can consider two main scales : the general leading trend
gradual long - time oscillation with a period of @xmath146 , and small modulating short - time oscillations on top of it . comparing , first
, these small modulating oscillations for different @xmath2 we conclude that they are nicely reproduced at all the presented levels of the mctdhb(@xmath2 ) computations , starting already at @xmath1 .
the long - scale dynamics might be viewed as a damped oscillations around perfect threefold fragmented state .
these oscillations may be attributed to periodic attempts to develop coherence in the system .
alternatively , using optical - lattice terminology one may say that this behavior mimics the oscillations between mott insulating and superfluid states . till @xmath141 all
the results practically coincide one can not see the deviations at the presented scales of fig . [ fig.i_b ] . for @xmath147
the discrepancy in the long time behavior between @xmath1 and @xmath121 , 9 , 12 computations starts to develop , remaining nevertheless quite modest .
this , again , indicates that the essential part of the evolution is nicely described at the mctdhb(@xmath1 ) level .
let us now analyze the physical origin of the observed imperfection ( discrepancy ) in the long time behavior between the @xmath1 and @xmath121 , 9 , 12 results for the evolving natural occupation numbers depicted in fig . [ fig.i_b ] .
we recall first that in all the presented computations the total energy of the evolving systems is preserved and , second , temperature does not enter the computations at all .
the key difference between @xmath1 and @xmath148 mctdhb(@xmath2 ) computations is the number of time - adaptive orbitals available for dynamics , i.e. , the number of excited states taken into account ( see respective dimensions @xmath25 listed in table . [ tab.1 ] ) .
we can suppose that excitations to higher orbitals , not available at the mctdhb(@xmath1 ) level , are responsible for the discrepancy .
on the other hand , these quantum excitations deplete the occupations of the three leading fragments and , hence , we can call this process quantum depletion . the proposed explanation can be directly verified by comparing the details of the three - orbital and @xmath148 computations . in the @xmath148 cases the depletion
is naturally described by populating higher time - adaptive orbitals .
indeed , the mctdhb(@xmath2 ) results using @xmath121 , 9 and 12 time - adaptive orbitals are very close to each other and show that the populations of the higher @xmath148 natural orbitals increase with time .
this can be seen from the left upper panel of fig .
[ fig.i_b ] where all the natural occupation numbers for @xmath123 case are plotted .
now , we elaborate more on the role of depletions in the three - orbital mctdhb(@xmath1 ) description .
we recall that every time - adaptive orbital is a sum of primitive dvr functions ( plane waves in this study ) .
hence , an excitation involving only a few time - dependent optimized orbitals actually contracts a huge number of excitations between the primitive plane waves .
so , due to this mutual self - consistency of the mctdhb method large fraction of quantum depletion is simulated by readjustments of the orbitals shapes already at the three - orbitals level .
since the total number of particles and self - consistent orbitals are kept fixed , the `` depleted '' populations can only be redistributed between available self - consistent orbitals , which is reflected as changes of the natural occupation numbers . of course ,
at longer propagation times contributions from higher excited states become more and more relevant and a larger number of the time - adaptive orbitals are needed to provide a more accurate description of these excitations .
this explains the small differences in the densities between @xmath1 and @xmath148 cases seen in fig .
[ fig.i_a ] at long propagation times ( @xmath149 ) . in the scenario studied here the processes of quantum depletion
play a very important role
they lead to a considerable damping of the long - scale oscillation around the perfect threefold fragmented state .
the general conclusion is that at the mctdhb(@xmath1 ) level of description almost all key physical features of the evolution are accurately reproduced , while numerically converged results are obtained using more time - adaptive orbitals .
now we discuss the second numerical example with @xmath150 repulsive bosons trapped in the same triple - well potential as before , eq .
( [ trap ] ) . as an initial state
we take , again , the ground state of the system at @xmath100 and @xmath105 . similarly to the @xmath0 case , the ground state becomes fully treefold fragmented @xmath151 at smaller barrier heights starting at @xmath152 .
we propagate the mctdhb(@xmath2 ) equations ( [ mctdhb_ci ] ) and ( [ mctdhb_psi ] ) using @xmath1 and @xmath121 time - adaptive orbitals in imaginary time to get the ground state expansion coefficients and respective one - particle functions .
the obtained ground state energies per particle are listed in table .
[ tab.1 ] .
first of all , we see a convergence of the ground state energy with @xmath2 .
second , comparing the results for @xmath0 and @xmath150 we conclude that these systems are indeed very similar from the point of view of energy per particle . the next step is dynamics
we ramped down the periodic potential suddenly to the depth @xmath102 of the triple - well and shift it to @xmath104 , see eq .
( [ trap ] ) , and then perform the real - time propagation .
we again follow the evolution of the many - boson wave function and construct and diagonalize the reduced one - body density matrix according to eq .
( [ no ] ) at each point in time .
the representative densities obtained at @xmath131 , @xmath132 and @xmath133 , i.e. , at the short , intermediate and long evolution times are depicted in fig .
[ fig.ii_a ] .
the densities computed at the @xmath1 and @xmath121 levels of the mctdhb(@xmath2 ) theory are very similar at all the propagation times , indicating that three time - adaptive mctdhb(@xmath1 ) orbitals provide an adequate description of the evolution in this case as well , as far as the density is considered .
the one - particle densities obtained at the @xmath1 and @xmath121 levels are plotted in fig .
[ fig.ii_a ] using red and green colors . in fig .
[ fig.ii_b ] we plot the evolutions of the corresponding natural occupation numbers as functions of time using the same color scheme red and green curves represent mctdhb(@xmath2 ) results for the @xmath1 and @xmath121 computations .
the general behavior of the system with @xmath150 bosons is quite similar to that discussed above for @xmath0 . until @xmath153 the system continues to stay in a quite pure threefold fragmented state . during this time the density , however , drastically changes , see the density profile at @xmath154 depicted in the lower panel of fig .
[ fig.ii_a ] .
afterwards , the occupations of two of the three initially equally occupied natural orbitals are reduced .
the largest occupation number reveals an opposite tendency it borrows particles from the other two natural orbitals , increasing thereby its own natural occupation .
this picture described at the @xmath1 level is confirmed by the computation with @xmath121 orbitals .
thus , short and intermediate time evolutions ( @xmath155 ) are identically described at the @xmath1 and @xmath121 levels of mctdhb theory . quantum depletions
excitations from three macroscopically - populated self - consistent time - adaptive orbitals to higher ones also become relevant for this system as time proceeds and appear as a gradual decrease of the occupations of all the three fragments .
as we have already mentioned above , the considered systems of @xmath0 and @xmath150 bosons are similar because we took the same trap potentials , the `` same '' initial state and very similar interparticle energies defined by the same @xmath156 .
we can compare the behavior of the depletions as a function of time for these systems by comparing the occupations of higher @xmath148 natural orbitals .
these are presented in the left upper panel of fig .
[ fig.i_b ] for @xmath0 and in fig .
[ fig.ii_b ] for @xmath150 .
we conclude that quantum depletion develops slower for the system made of a larger number of bosons .
let us now discuss the last numerical example , where we consider dynamics of @xmath157 repulsive bosons with contact interaction strength @xmath105 trapped in the same triple - well potential as before . to obtain a perfect threefold fragmented , i.e. , mott insulating state with 1001 bosons per site , the optical depth of the periodic potentials has to be very large , @xmath100 .
we first propagate the mctdhb(@xmath2 ) equations with @xmath1 time - adaptive orbitals in imaginary time to get the ground state solution of this system . using eq .
( [ no ] ) we verify that the obtained state is indeed threefold fragmented the respective natural occupation numbers are @xmath151 .
then , we suddenly shift the potential to @xmath104 and reduce its amplitude to @xmath102 as before , see eq .
( [ trap ] ) , and apply the real - time propagation scenario .
the computed energies of the initial system @xmath136 as well as of the evolving wave packet @xmath137 are presented in table .
[ tab.1 ] .
now , we analyze the results of the real - time propagation . in fig . [ fig.iii_a ]
we plot the densities at the short @xmath131 , intermediate @xmath132 and long @xmath133 propagation times . comparing these results with the respective densities depicted in fig .
[ fig.i_a ] and fig .
[ fig.ii_a ] for the bosonic systems made of @xmath0 and @xmath150 particles we conclude that all three systems exhibit very similar behaviors .
we apply eq .
( [ no ] ) to analyze the properties of the propagating wave packet and depict in fig . [ fig.iii_b ] the obtained evolutions of the three available natural occupation numbers using blue lines .
two of the three , initially equally occupied natural orbitals start to donate the bosons , i.e. , decrease their occupancies , while the third one accepts the particles , increasing its own occupation . then , around @xmath158 , the tendency changes and the most - occupied natural orbital starts to donate bosons , transferring them back to the two other orbitals .
a signature of this imperfect oscillatory behavior can be observed during all the computed propagation times depicted in fig .
[ fig.iii_b ] . for comparison in this figure
we also plot the natural occupation numbers obtained at the @xmath1 level of the mctdhb(@xmath2 ) theory for the systems made of @xmath0 and @xmath150 bosons with red and green colors , respectively .
this allows us to analyze the @xmath3-dependence of the evolutions .
the key observation is that evolutions of all the three systems are qualitatively the same .
moreover , the evolutions of the systems made of @xmath150 and @xmath157 bosons are much closer to each other than to the @xmath0 ones .
therefore , we might speculate that for the present studies the many - body properties are saturated to some average values already at @xmath150
. hence one can expect that key aspects of the dynamics of the initially threefold fragmented systems with larger @xmath3 will be very similar to those obtained for @xmath150 .
let us now summarize all three studies .
we have considered systems made of @xmath0 , @xmath150 and @xmath157 particles choosing the strength of the contact interparticle interaction potential such that the non - linearity parameter @xmath106 is the same for all systems .
we have solved the many - particle time - dependent schrdinger equation using the scenario of eq .
( [ trap ] ) ( see also fig .
[ fig.0 ] ) for all the considered systems . comparing the diagonal parts of the reduced one - particle density matrices , i.e. , the densities computed at different propagation times and depicted for @xmath0 , @xmath150 , and @xmath157 bosons in figs .
[ fig.i_a ] , [ fig.ii_a ] and [ fig.iii_a ] , respectively , we observe that the presented densities look very similar at the short , intermediate and long propagation times .
we also verify that the energy per particle of the considered systems are very close to each other , as one can see by comparing the data collected in table .
[ tab.1 ] .
hence , we can conclude that the non - linearity parameter @xmath106 is useful to characterize the global behavior of the densities in the systems made of different numbers of particles , at least for the dynamical studies of the presented type .
the compared results for the time - dependent densities are quite similar , but more involved properties reveal differences . indeed , the comparison of the properties of the evolving wave packets , e.g. , evolutions of the natural occupation numbers presented in figs .
[ fig.i_b ] , [ fig.ii_b ] and [ fig.iii_b ] for @xmath0 , @xmath150 , and @xmath157 , respectively , shows that global tendencies are @xmath156 dependent , while the depletions , i.e. , excitations from the three most - occupied orbitals to higher ones depend in addition on total number @xmath3 of the bosons in the system as well as on their interaction strength @xmath97 .
we conclude that in the presented study the dependence of the properties on the number of particles @xmath3 explicitly appears at longer propagation times and is related to the quantum depletion which is also found to grow with time .
recently , in ref .
@xcite we proposed a novel , effective and general theory to construct the result of action of any many - particle operator represented in the second quantized form on a state vector , without resorting to matrix representation of operators and even to its elements . in the present work we adopt the bosonic version of this theory to the mctdhb method .
we first discuss the practical realization of the proposed ideas in the mctdhb(@xmath2 ) computational scheme and address its parallel implementation in some details . using this new tool we study highly non - equilibrium dynamics of repulsive systems made of @xmath0 , 51 , 3003 particles in real - space triple - well traps . as an initial wave packet
we take the fully threefold fragmented quantum state , strongly displace it out of equilibrium and then , suddenly , reduce the depth of the potential wells .
from the perspective of optical - lattice physics this scenario can be seen as a propagation of the initially pure mott insulating state , strongly displaced out of the potential minima . as a result of such a sequence of trap manipulations at least four bands of the periodic trap potential
become energetically open , i.e. , available for excitations .
we have shown that quantum dynamics for this scenario is very complicated .
first , we have demonstrated the convergence of the mctdhb(@xmath2 ) approach with @xmath1 , 6 , 9 , 12 self - consistent time - adaptive orbitals by applying the described above scenario to study long - time dynamics of @xmath0 bosons .
next , we see that a quite accurate description of the evolution for short and even intermediate propagation times is obtained already at the mctdhb(@xmath1 ) level of theory
. this result provides , thereby , a confirmation of an intuitive supposition that many - body dynamics in three - well traps can qualitatively and sometimes quantitatively be described by using three self - consistent time - adaptive orbitals .
we stress here the last point
each used self - consistent time - adaptive orbital actually contracts very many fixed ( primitive ) functions ( @xmath114 or 555 plane waves in our work ) , and , therefore , the mctdhb many - body wave function constructed by using these few self - consistent time - adaptive orbitals spans a huge effective fock space . as a next step
we compute the dynamics for the systems made of @xmath150 and @xmath157 bosons and characterized by the same total non - linearity parameter @xmath159 . in the region of the trap / state manipulations considered here
, the total non - linearity @xmath156 appears to be quite a useful parameter . because the systems made of different numbers of atoms but with the same @xmath156 demonstrate very similar dynamical behavior at least for the evolving density profiles of the initially three - fold fragmented ( mott insulating ) states .
more involved properties , like natural occupation numbers are more sensitive to the individual @xmath97 and @xmath3 parameters , especially for a small number of particles and at longer propagation times .
the evolutions of the natural occupation numbers show up a tendency to a long - time oscillatory behavior around the initially perfect threefold fragmented state .
however , quantum depletions , i.e. , excitations to higher orbitals lead to suppression or damping of these oscillations . for the short and intermediate propagation times this process is adequately reproduced already at the mctdhb(@xmath1 ) level due to mutual self - consistency of the time - adaptive orbitals used .
however , for longer propagation times more self - consistent orbitals are needed for the correct description of the depletions and many - body dynamics . at higher levels of the mctdhb(@xmath148 ) description , e.g. , within @xmath121 self - consistent orbitals , the depletions are naturally included and with @xmath160 , 12 orbitals they are accurately described .
the mctdhb theory augmented by the proposed , developed and implemented ideas for effective operations in the bosonic fock space allows for a faster solution of the time - dependent many - boson schrdinger equation in larger configurational spaces and for longer propagation times .
this opens new perspectives for quantitative descriptions of the quantum many - body dynamics of larger systems , as recently announced through the numerically exact studies of the one - dimensional bosonic josephson junction problem @xcite
. j. zanghellini , m. kitzler , c. fabian , t. brabec , and a. scrinzi , laser phys .
* 13 * , 1064 ( 2003 ) ; t. kato and h. kono , chem .
phys . lett . *
392 * , 533 ( 2004 ) ; m. nest , t. klamroth , and p. saalfrank , j. chem . phys . * 122 * , 124102 ( 2005 ) .
a. i. streltsov , o. e. alon and l. s. cederbaum , j. phys .
b , * 42 * , 091004 ( 2009 ) s. zllner , h .- d
. meyer , and p. schmelcher , phys .
a * 74 * , 053612 ( 2006 ) ; s. zllner , h .- d .
meyer , and p. schmelcher , phys .
a * 74 * , 063611 ( 2006 ) ; c. matthies , s. zllner , h .- d .
meyer , and p. schmelcher , phys .
a * 76 * , 023602 ( 2007 ) ; s. zllner , h .- d .
meyer , and p. schmelcher , phys .
* 100 * , 040401 ( 2008 ) ; s. zllner , h .-
meyer , and p. schmelcher , phys .
a * 78 * , 013621 ( 2008 ) .
a. u. j. lode , a. i. streltsov , o. e. alon , h .-
meyer , and l. s. cederbaum , j. phys .
b , * 42 * , 044018 ( 2009 ) .
a. i. streltsov , o. e. alon , and l. s. cederbaum , _
`` a novel general mapping for bosonic and fermionic operators in fock space '' _ , arxiv:0910.2577v1 . p. nozires and d. saint james , j. phys .
( paris ) * 43 * , 1133 ( 1982 ) ; p. nozires , in bose - einstein condensation , edited by a. griffin , d.w .
snoke , and s. stringari ( cambridge university press , cambridge , england , 1996 ) .
a. fisher , p. b. weichman , g. grinstein , and d. s. fisher , phys .
b * 40 * , 546 ( 1989 ) ; d. jaksch , c. bruder , j. i. cirac , c. w. gardiner , and p. zoller , phys .
. lett . * 81 * , 3108 ( 1998 ) ; m. greiner , o. mandel , t. esslinger , t. w. hnsch , and i. bloch , nature ( london ) * 415 * , 39 ( 2002 ) ; t. stferle , h. moritz , c. schori , m. khl , and t. esslinger , phys .
lett . * 92 * , 130403 ( 2004 ) . | to solve the many - boson schrdinger equation we utilize the multiconfigurational time - dependent hartree method for bosons ( mctdhb ) . to be able to attack larger systems and/or to propagate the solution for longer times , we implement a parallel version of the mctdhb method thereby realizing the recently proposed [ streltsov _ et al .
_ arxiv:0910.2577v1 ] novel idea how to construct efficiently the result of the action of the hamiltonian on a bosonic state vector .
we study the real - space dynamics of repulsive bosonic systems made of @xmath0 , 51 and 3003 bosons in triple - well periodic potentials .
the ground state of this system is three - fold fragmented . by suddenly strongly distorting the trap potential , the system performs complex many - body quantum dynamics . at long times
it reveals a tendency to an oscillatory behavior around a threefold fragmented state .
these oscillations are strongly suppressed and damped by quantum depletions . in spite of the richness of the observed dynamics , the three time - adaptive orbitals of mctdhb(@xmath1 ) are capable to describe the many - boson quantum dynamics of the system for short and intermediate times . for longer times , however , more self - consistent time - adaptive orbitals are needed to correctly describe the non - equilibrium many - body physics .
the convergence of the mctdhb(@xmath2 ) method with the number @xmath2 of self - consistent time - dependent orbitals used is demonstrated . |
over the past decade , the analysis of particle multiplicities in heavy ion collions has shown overwhelming evidence for chemical equilibrium in the final state except for particles carrying strangeness which are mildly suppressed ; however , their relative yields fulfill statistical equilibrium . a summary as of 2006 , combining the results from many different groups @xcite , is shown in fig . [ eovern ] . us determined from heavy ion collisions at different beam energies .
the lower ags points are based on a preliminary analysis of @xmath2 data.,width=226 ] except for particle multiplicities at rhic energies , all data in fig .
[ eovern ] use integrated particle yields , the very systematic change of thermal parameters over the full range of beam energies is one of the most impressive features of relativistic ion collisions to date .
it is now possible to use the thermal model to make solid predictions for particle multiplicities at lhc energies @xcite and to determine which beam energy will lead to the highest baryon density at freeze - out @xcite . with chemical equilibrium for integrated yields
thus firmly established , we focus on other properties , in particular , since the rapidity distributions of identified particles is now becoming available also at rhic energies @xcite , it is now possible to determine the rapidity dependence of thermal parameters .
a first analysis was done by stiles and murray @xcite for the data obtained by the brahms collaboration at 200 gev @xcite .
this shows a clear dependence of the baryon chemical potential on rapidity in particular due to the changing @xmath3 ratio .
subsequently it was shown by rhrich @xcite that the particle ratios at large rapidities are consistent with those measured at the sps energies .
this opens the possibility to compare measurements for e.g. the @xmath4 ratio at high rapidities and check them with the corresponding values measured in the energy scan at the sps , thus complementing the rapid variation of this ratio as a function of beam energy .
a further analysis of the rapidity dependence was recently done in ref .
a detailed report of our results will be presented elsewhere @xcite .
the general procedure is as follows : the rapidity axis is populated with fireballs following a gaussian distribution function given by @xmath5 where @xmath6 is the rapidity of the fireball .
particles will appear when the fireball freezes out and will follow a thermal distribution centered around the position of the fireball @xmath7 the momentum distribution of hadron @xmath8 is then calculated from the distribution of fireballs as given by eq .
[ [ eqrho ] ] along the rapidity axis as follows @xmath9 where @xmath10 is the the distribution of hadrons from a single fireball .
the temperature @xmath0 and the baryon chemical potential @xmath1 will depend on the rapidity of the fireball and are not assumed to be constant .
an important parameter is the width of the distribution . for the rhic data at 200 gev
this was determined from the @xmath11 s as these are very sensitive to the value of @xmath12 and less to variations in the baryon chemical potential .
the width of the distribution @xmath13 is compatible with the values quoted by the brahms collaborationi @xcite , e.g. @xmath14 and @xmath15 .
the hadrons described by eq . [ [ eqdist ] ] are mainly resonances
. only a fraction of these are stable under strong interactions .
the majority of them will decay into stable hadrons at chemical freeze - out , hence the need to implement multi - particle decays .
we assume that the temperature @xmath0 and the chemical potential @xmath1 are always related via the freeze - out curve as given in fig . [ [ eovern ] ] ; if the temperature varies along the rapidity axis , then also the chemical potential will vary .
thus a decrease in the temperature of the fireball will be accompanied by an increase in the baryon chemical potential . in other words , we assue a universality of the chemical freeze - out condition .
this relationship between temperature and baryon chemical potential is very reasonable since all particle abundancies measured so far follow it .
once the width of the distribution of fireballs has been fixed , we can go on with the dependence of the baryon chemical potential on the rapidity of the fireball ( units are in gev ) . @xmath16
the dependence of the @xmath3 ratio on the parameter @xmath17 is shown in figure [ [ fig : mub_y ] ] .
the best value is @xmath18 gev .
ratio as a function of rapidity .
the curves show the dependence on the baryon chemical potential on the rapidity as discussed in the text.,width=226 ] the variation of the temperature along the rapidity axis is shown in fig .
[ fig : tmuy ]
. the temperature is maximal at mid - rapidity and gradually decreases towards higher ( absolute ) values of the rapidity .
the baryon chemical potential follows a different pattern : it has a minimum at mid - rapidity and increases quite substantially towards higher values of the rapidity .
of particular interest are the largest values of the rapidity ( say @xmath19 ) .
it can be seen that there is a region of overlap with the data obtained at the na49 experiment at cern - sps .
particle yields measured in heavy ion collisions show an overwhelming evidence for chemical equilibrium at all beam energies . the rapidity dependence of the thermal parameters @xmath0 and @xmath1 can now be determined over a wide range of rapidities and
show a systematic behavior towards an increase in @xmath1 as the rapidity is increased .
this work was supported in part by the scientific and technological co - operation programme between italy and south africa , project number 16 .
the help of d. cebra with the preliminary 4@xmath20 data at the lower ags beam energies is gratefully acknowledged , r. adams has helped in the analysis of these data in fig .
[ 1 ] is acknowledged .
99 for references see e.g. j. cleymans , h. oeschler , k. redlich , s. wheaton , phys .
c73 ( 2006 ) 034905 .
j. cleymans , i. kraus , h. oeschler , k. redlich , s. wheaton , phys .
c74 ( 2006 ) 034903 .
j. randrup and j. cleymans , phys .
c74 ( 2006 ) 047901 .
stiles and m. murray , nucl - ex/0601039 , unpublished .
d. rhrich , proceedings of the talk given at the workshop on `` critical point and the onset of deconfinement '' , florence , italy , july 2006 .
i. arsene at al . ,
brahms collaboration , phys .
94 ( 2005 ) 162301 .
bearden , brahms collaboration , contribution to the qm2006 conference , shanghai , china , november 2006 .
b. biedron , w. broniowski , nucl - th/0610083 . f. becattini and j. cleymans ( in preparation ) .
w. broniowski and f. florkowski , phys .
c65 ( 2002 ) 064905 ; j. phys .
g31 ( 2005 ) s1087 . | particle yields in heavy ion collisions show an overwhelming evidence for chemical or relative chemical equilibrium at all beam energies . the rapidity dependence of the thermal parameters @xmath0 and @xmath1 can now be determined over a wide range of rapidities and
show a systematic behavior towards an increase in @xmath1 away from mid - rapidity . |
since the concept of stellar populations was invented @xcite , stellar population synthesis ( sps ) has been proven to be a useful tool for revealing clues about galaxy formation @xcite .
research on several crucial factors of stellar modelling , like convection , opacity , heavy - element mixing , helium content , and mass loss @xcite , resulted in increasingly accurate sps models @xcite . in this scheme ,
estimation of single burst equivalent age and metallicity , was accomplished by comparing integrated light of observed galaxy with that of sps models . however , most of the colours and absorption feature strength appear identical if the changes of age and metallicity satisfy @xmath2 .
this so - called age - metallicity degeneracy @xcite blocks the way to estimating galaxy accurate ages from optical broad band colours . at spectroscopically narrow bands , however , several age sensitive or metallicity sensitive lick indices
@xcite , taken in pairs , are effective at breaking this degeneracy .
balmer indices vs. fe - peak indices are widely used , for example , because they are relatively easy to observe ( all features are in the optical so one spectrograph can cover them all ) and the model grids in the observed index - index space open up to nearly orthogonal grids rather than collapsed into linear , degenerate overlapping segments .
spectroscopy is effective , but photometry also has its strong points .
it is much more efficient for faint objects and low surface brightness galaxies . in the epoch of large sky surveys , like _ sloan digital sky survey ( sdss ) , two micron all sky survey ( 2mass ) , spitzer space telescope ( sst ) , large synoptic survey telescope ( lsst ) _ , and _ james webb space telescope ( jwst ) _ , photometry is undoubtedly more feasible than spectrometry .
in fact , optical near - infrared ( near - ir ) colours have been exploited to possibly constrain age and metallicity by several studies @xcite .
@xcite suggested ( @xmath3 ) colour shows good balance between parameter degeneracy and sensitivity .
@xcite found that four age intervals ( @xmath410 gyr , 2@xmath59 gyr , 1@xmath52 gyr , and 0.2@xmath51 gyr ) are clearly separated in the optical near - ir colour - colour plot .
these results hinge upon the models , and specifically upon the number of thermally pulsing asymptotic giant branch ( tp - agb ) stars , which dominate near - ir light between 0.3 and 2 gyr of age @xcite .
the present work looks for age - sensitive or metallicity - sensitive indices in three photometric systems : david dunlap observatory ( ddo , @xcite ) , beijing - arizona - taiwan - connecticut ( batc , @xcite ) , and strmgren @xcite systems .
these filters are narrower in wavelength than johnson - cousins broad band filters , but resemble the _ nircam _ narrow filters on board _
jwst_. can we find indices which are practical to break the age - metallicity degeneracy ?
the tables and graphs in @xmath6 [ sub : three ] explore this issue . in addition , equipped with the latest stellar evolutionary isochrones of the padova group ( @xcite , hereafter b09 ) , we plot the model grids in ( @xmath7 ) vs. ( @xmath3 ) space , and compare with previous models in @xmath6 [ sub : bvk ] . observables from three samples are chosen to verify the credibility of parameters given by the model grids .
discrepancies between two model grids are briefly discussed in @xmath6 [ sect : disc ] , and a brief summary of the results and outlook for the future is given in @xmath6 [ sect : sum ] .
in order to find the age - sensitive or metallicity - sensitive index , we calculate the the metallicity - to - age sensitivity parameter ( @xmath8 ) using the evolving ( * ? ? ? *
hereafter w94 ) models , following @xcite and @xcite .
the spectral library is synthetic and is sensitive to individual elemental abundances in the 300 to 1000 nm wavelength range .
the metallicity sensitivity parameter is : @xmath9 where @xmath10 is the metallicity partial derivative of the index at 8 gyr , and @xmath11 is the age partial derivative of the index at solar metallicity .
thus , @xmath12 is the metallicity sensitivity at 8 gyr , solar metallicity .
@xmath12 is surveyed for indices in three photometric systems .
we refer the readers to related papers for detailed description of filters and indices ( see @xmath6[sect : intro ] ) .
as shown in @xcite , indices with @xmath12 close to 1.5 are less likely to break the age - metallicity degeneracy .
hence , indices with extreme @xmath12 values are selected as candidates and listed in table [ tab : ss]k , batc6075@xmath5k , were also examined . however , no promising index is found . ] . what is not quantified in table [ tab : ss ]
is the dynamic range ( compared to observational error ) of a given index , and the dynamic range is low in ddo c(41 - 42 ) , alas .
.metallicity sensitivities at 8 gyr , solar metallicity , for our best candidate pairs of colors [ cols=">,>",options="header " , ] in a non - solar scaled environment , indices may reflect those abundance changes . to trace the effects of different elements
, we increase the stellar atmosphere model abundances of six major elements ( c , n , o , na , mg , fe ) by 0.3 dex , one single element at a time .
the sensitivity is modeled only in the stellar library ; the isochrones are kept as published .
table [ tab : eh ] outlines the index changes at 8 gyr , solar metallicity . * the prominent carbon and nitrogen effect on ddo c(41 - 42 ) colour is not surprising , since the ddo 41 band overlaps cn@xmath13 and cn@xmath14 lick indices in wavelength .
fe and cn absorption lines are the main features under 4000 , which explains the ddo c(35 - 48 ) colour changes at super - solar c , n , and fe abundances . the behavior of increased o is opposite due to molecular balancing : the co molecule has the strongest binding energy and so adding o decreases the c supply , weakening c@xmath14 and cn features .
thus , adding more o causes bluer ddo c(41 - 42 ) and c(35 - 38 ) colours .
* the batc 6075 - 6660 and 5795 - 6075 colours are relatively unaffected by six major elements , which makes them good colours to study systems with unknown element enhancements .
what we find matches the basic philosophy of creating this photometric system : avoiding known bright feature lines and sky lines , in order to study the continuum of the spectrum @xcite . *
strmgren @xmath1 $ ] and @xmath0 $ ] are different from the four colours above , because they are the subtractions of two colours . @xmath1 $ ] is found to be metallicity sensitive in @xmath6 [ sub : three ] , which is consistent with its behavior in table [ tab : eh ] : five elements have substantial influences on @xmath1 $ ] .
the trend of the @xmath0 $ ] index change due to c and o abundance variations is reverse compared to other indices .
since @xmath15 and @xmath16 bands are both blanketed by cn absorption lines , our models predict that the impact of cn on the @xmath16 band is greater than that on the @xmath15 band .
therefore , higher cn abundance means a larger magnitude increase in the @xmath16 band than that in the @xmath15 band , leading to a smaller @xmath0 $ ] value . the strong mg effect on @xmath0 $ ] is echoed by the mg3835 ( @xmath15 band ) and mg4780 ( @xmath17 band ) indices defined in @xcite .
the possibility of breaking age - metallicity degeneracy with optical plus near - ir colours @xcite , and the upcoming era of near - ir telescopes ( eg : _ jwst _ ) inspire us to delve the optical near - ir colour@xmath5colour plot . prior to this work
, @xcite inspected the ( @xmath7 ) vs. ( @xmath3 ) plot using ( * ? ? ?
* hereafter bc03 ) , @xcite , and basti @xcite models .
they suggested different treatments of convective core overshooting and differences in the tp - agb phase cause the inconsistency among different model grids ( also see @xcite ) . to further discuss the robustness of the models , we investigate the isochrones that looked so promising ( b94 ) and a later , even more complete version ( b09 ) from the same group .
* b94 models + perhaps partly because of the inclusion of all stellar evolutionary phases , b94 models were widely adopted as basic stellar evolution isochrones for evolutionary synthesis models , such as : bc03 ; pegase @xcite ; and starburst99 @xcite models .
figure [ fig : pad94 ] shows the ( @xmath7 ) vs. ( @xmath3 ) plot of b94 models .
three observational samples are selected for comparison : ( 1 ) old , metal - poor gcs from @xcite ( _ filled circles _ ) ; ( 2 ) old , metal - rich es from @xcite ( _ filled triangles _ ) ; ( 3 ) swb type iv@xmath5vi large magellanic cloud ( lmc ) star clusters ( _ crosses _ ) .
the reddening - corrected ( @xmath7 ) and ( @xmath3 ) colours come from @xcite and @xcite .
the last sample is selected as `` intermediate age '' ssps , roughly 1 to 8 gyr according to @xcite . generally speaking ,
observables of these three samples locate at expected locations on the model grids .
+ ) vs. ( @xmath3 ) plot of b94 models .
the models are given at age@xmath18 gyr with @xmath19 .
solar metallicity ssps are marked as _
filled squares _ to guide the eye .
ssps of the same age are connected by _ solid lines _ , while ssps of the same metallicity are connected by _
dotted lines_. a vector for a@xmath20 mag is sketched to estimate the extinction effect .
observables from three samples are labelled by different symbols : _ filled triangles _ : es from @xcite . _
filled circles _ : gcs from @xcite .
_ crosses _ : swb type iv@xmath5vi large magellanic cloud star clusters from @xcite and @xcite . ]
+ ) vs. ( @xmath3 ) plot of b09 models .
the models are given at age@xmath18 gyr with @xmath21 .
other symbols are the same as in figure [ fig : pad94 ] . ] * b09 models + with improved understanding of tp - agb phase , padova group updated the stellar evolution models . compared with previous models of this group
( b94 ; @xcite ) , b09 models adopted a sophisticated tp - agb model which includes third dredge - up , hot bottom burning , and variable molecular opacities @xcite .
also , basic observables of agb stars in the magellanic clouds ( mcs ) were reproduced by the new models @xcite .
the ( @xmath7 ) vs. ( @xmath3 ) plot of the b09 models are shown in figure [ fig : pad11 ] .
observables are also plotted as in figure [ fig : pad94 ] .
isochrones of age larger than 5 gyr are consistent with that of the b94 models .
es and gcs locate on the metal - rich and metal - poor sides , respectively .
however , isochrones of 1 and 2 gyr change substantially : the ( @xmath3 ) colour of @xmath22 becomes redder than that of solar metallicity .
an unreported `` s '' shape is found if @xmath22 ssps are connected .
another unexpected change is about the isochrone of the youngest age , 100 myr : the ( @xmath3 ) colour of the lowest metallicity ( @xmath23 ) is even redder that of the highest metallicity ( @xmath24 ) .
prior to this work , @xmath25 or @xmath26 were paired with ( @xmath27 ) , ( @xmath28 ) , or other colours to estimate galaxy age @xcite . in this paper ,
the extinction independent q - indices @xmath0 $ ] and @xmath1 $ ] are assembled , but do a mediocre job breaking the age - metallicity degeneracy .
as seen in figure [ fig : mc11 ] , several observables lie outside the b09 model grids .
three reasons are hypothesized for these outliers : first , the conversion relations between _ uz , vz , bz , yz _ and _ uvby _ system presented in @xcite are empirical , which inevitably increases the uncertainty ; second , the dynamic range of @xmath0 $ ] index is not large compared to the observational error ; third , isochrones of age between 5 and 17 gyr are partially degenerate , which reduces the span of model grids .
the model isochrone of 100 myr lies low in the @xmath0 $ ] vs. @xmath1 $ ] plot .
this is similar to the statement in @xcite that most of the starburst galaxies have _ mz _ values ] less than @xmath29 , and they are well separated from other types of galaxies . to test the potential of @xmath1 $ ] index as an star formation rate indicator , s ( `` star formation rates equivalent to normal disk galaxy '' ) and s+(``starburst objects '' ) from @xcite are selected and shown in figure [ fig : mcs ] .
most of the _ mz _ values of this sample are less than @xmath30 , supporting the idea that starburst galaxies are bluer in @xmath1 $ ] .
however , we detect no clear @xmath1 $ ] value that separates es and star forming galaxies .
strmgren system photometry is far less developed than broad band photometry for galaxies , however . that , plus the large apparent observational scatter in figures [ fig : mceg ] , [ fig : mc11 ] and [ fig : mcs ] lead us to conjecture that strmgren photometry may still have great potential for breaking the age - metallicity degeneracy , but higher quality data are needed to be certain .
$ ] vs. @xmath1 $ ] plot of b09 models .
the models are discribed in figure [ fig : mc11 ] .
solar metallicity ssps are marked as _ filled squares _ to guide the eye .
_ filled circles _ : s and s+ of @xcite . ]
three sps models are assembled in this work : w94 , b94 , and b09 ( i.e. , primitive , competent , and sophisticated , respectively . ) .
figure [ fig : mceg ] , [ fig : mc11 ] , [ fig : pad94 ] , and [ fig : pad11 ] clearly illustrate the model differences in @xmath0 $ ] vs. @xmath1 $ ] and ( @xmath7 ) vs. ( @xmath3 ) plots . during the whole work
, we calculate the magnitudes of each bands in three sps models with the same codes , thus w94 , b94 , b09 models share the same spectral library , imf @xcite with lower mass limit @xmath31 , upper mass limit @xmath32 , and etc .
most of the model grid discrepancies rely on different stellar evolution descriptions : * w94 models are a merging of isochrones and tracks from many sources . compared to b94 ,
w94 models have a factor of two more stars in the upper rgb .
the core - helium burning phase was approximated by a single red clump .
the luminosities and lifetimes of agb stars were drawn from theoretical tracks .
* b94 models included all phases of stellar evolution until the remnant stage in all mass ranges .
convective core overshooting was also considered in the models . nevertheless , the analytic prescription of tp - agb phase was approximate . @xcite
* b09 models adopted the latest tp - agb model of @xcite . compared with previous ones
, this tp - agb model has included several crucial effects and it was calibrated by agb stars in the mcs .
bertelli et al .
reckoned b09 models as satisfactory achievement . in figure
[ fig : pad11 ] , ssps with @xmath22 have the reddest ( @xmath3 ) colours along isochrones of 1 and 2 gyr .
this feature is not seen in b94 model grids .
since b94 and b09 models mainly differ in tp - agb model treatments , which affect the near - ir colour seriously between 0.3 and 2 gyr @xcite , it is suggested different treatments of tp - agb stars are responsible for the changes of model grids .
considering the metallicities of star clusters in the lmc ( mean @xmath33 for clusters with @xmath34 gyr and @xmath35 for clusters with @xmath36 gyr , @xcite ) , swb iv@xmath5vi star clusters are better fitted in b09 models b94 models give much higher metallicities . nevertheless , noting that b09 models are calibrated by agb stars from the mcs , and that the metallicities of star clusters in the lmc are around @xmath22 , it is suspected that ssps of other metallicities , especially the extremely low metallicity , may not be well calibrated .
the scarcity of very low metallicity agb stars around solar neighbourhood impedes empirical calibration .
it is reasonable to suggest closer scrutiny for ssps with extremely low metallicities around 1 and 2 gyr .
another unexpected feature in figure [ fig : pad11 ] is the twist of the 100 myr isochrone .
this youngest isochrone is unique : the ( @xmath3 ) colour at the lowest metallicity is redder than that of the highest metallicity .
this is inconsistent with the usual sense that metal - rich ssps are redder than metal - poor ssps ( see the 100 myr isochrone of figure [ fig : pad94 ] ) .
the change of the 100 myr isochrones between b94 and b09 models challenge the method of estimating ages and metallicities of young ssps ( @xmath37 gyr ) from optical near - ir colour@xmath5colour plots @xcite . finally , we check b94 and b09 model differences in the popular ( @xmath38 ) vs. ( @xmath39 ) plot ( * ? ? ?
* ; * ? ? ?
* figure [ fig : brk ] ) .
_ solid lines _ are isochrones of b09 models ; _ dashed lines _ are isochrones of b94 models .
differences between two model grids are similar to ( @xmath7 ) vs. ( @xmath3 ) plots : ssps of @xmath22 have the reddest ( @xmath39 ) colours along the 1 and 2 gyr isochrones in b09 models ; the ( @xmath39 ) colour at the lowest metallicity is redder than that at the highest metallicity along the 100 myr isochrone of b09 .
thus , these features are dependent upon near - ir band ( @xmath40 ) , instead of visible bands ( @xmath41 or @xmath42 ) .
considering the tp - agb treament differences of b94 and b09 models and the tp - agb effects on @xmath40 band , it is again suggested tp - agb stars are responsible for the changes of young ssp isochrones ( @xmath43 gyr ) , which is consistent with @xcite and @xcite ) vs. ( @xmath39 ) plot of b94 and b09 models .
the b94 models are given at age@xmath18 gyr with @xmath19 , while the b09 models are given at age@xmath18 gyr with @xmath21 .
note that the two sets of tracks are not identical .
isochrones of b09 models are labelled by _ solid lines
_ , and isochrones of b94 models are labelled by _
dashed lines_. solar metallicity ssps are marked as _
filled squares _ to guide the eye . a vector for a@xmath20 mag
is sketched to estimate the extinction effect . ]
motivated by the need of estimating ages and metallicities from photometric systems , this work explored the ddo , batc , strmgren systems in the hope of finding age sensitive or metallicity sensitive indices .
three index@xmath5index plots were examined but only strmgren @xmath0 $ ] vs. @xmath1 $ ] plot showed moderate age - metallicity separation .
four indices of the ddo and strmgren systems changed substantially while increasing c and n abundances .
o traces an opposite trend compared to c and n , since o tends to form co with c. then , we turned to the literature - posited age - metallicity disentangling space the optical plus near - ir colour@xmath5colour plot .
updated stellar evolution gives no support for clear disentanglement of age .
even mean age , therefore , is a subtle effect in colors , and would require , seemingly , both finer - tuned models and observations of greater accuracy . 99 baade , w. , 1944 , apj , 100 , 137 bell , e. f. , de jong , r. s. , 2000 , mnras , 312 , 497 bertelli , g. , bressan , a. , chiosi , c. , fagotto , f. , nasi , e. , 1994 , a&as , 106 , 275 bertelli , g. , girardi , l. , marigo , p. , nasi , e. , 2008 , a&a , 484 , 815 bertelli , g. , nasi , e. , girardi , l. , marigo , p. , 2009
, a&a , 508 , 355 brocato , e. , castellani , v. , raimondo , g. , romaniello , m. , 1999 , a&as , 136 , 65 bruzual a. , g. , charlot , s. , 1993 , apj , 405 , 538 bruzual a. , g. , charlot , s. , 2003 , mnras , 344 , 1000 burstein , d. , faber , s. m. , gaskell , c. m. , krumm , n. , 1984 , apj , 287 , 586 cardiel , n. , gorgas , j. , sanchez - blazquez , p. , cenarro , a. j. , pedraz , s. , bruzual , g. , klement , j. , 2003 , a&a , 409 , 511 carter , d. et al . , 2009 , mnras , 397 , 695 charlot , s. , worthey , g. , bressan , a. , 1996 , apj , 457 , 625 conroy , c. , gunn , j. e. , white , m. , 2009 , apj , 699 , 486 cordier , d. , pietrinferni , a. , cassisi , s. , salaris , m. , 2007 , aj , 133 , 468 de jong , r. s. , 1996 , a&a , 313 , 377 dotter , a. , chaboyer , b. , jevremovic , d. , kostov , v. , baron , e. , ferguson , j. w. , 2008 , apjs , 178 , 89 faber , s. m. , friel , e. d. , burstein , d. , gaskell , c. m. , 1985 , apjs , 57 , 711 fan , x. et al . , 1996 , aj , 112 , 628 fioc m. , rocca - volmerange b. , 1997 , a&a , 326 , 950 fioc m. , rocca - volmerange b. , 1999 , preprint ( arxiv : astro - ph/9912179 ) girardi , l. , bressan , a. , bertelli , g. , chiosi , c. , 2000 , a&as , 141 , 371 girardi , l. , marigo , p. , 2007 ,
a&a , 462 , 237 gorgas , j. , faber , s. m. , burstein , d. , gonzalez , j. j. , courteau , s. , prosser , c. , 1993 , apjs , 86 , 153 hunt , l. k. , thuan , t. x. , izotov , y. i. , 2003 , apj , 588 , 281 james , p. a. , salaris , m. , davies , j. i. , phillipps , s. , cassisi , s. , 2006 , mnras , 367 , 339 lee , h. , worthey , g. , trager , s. c. , faber , s. m. , 2007 , apj , 664 , 215 maraston , c. , 2005 , mnras , 362 , 799 maraston , c. , daddi , e. , renzini , a. , cimatti , a. , dickinson , m. , papovich , c. , pasquali , a. , pirzkal , n. , 2006 , apj , 652 , 85 marigo , p. , girardi , l. , bressan , a. , groenewegen , m. a. t. , silva , l. , granato , g. l. , 2008 , a&a , 905 , 883 mcclure , r. d. , 1976 , aj , 81 , 182 mcclure , r. d. , van den bergh , s. , 1968 , aj , 73 , 313 odell , a. p. , schombert , j. , rakos , k. , 2002 , aj , 124 , 3061 peletier , r. f. , 1989 , ph.d .
thesis , univ .
groningen peletier , r. f. , valentijn , e. a. , jameson , r. f. , 1990 , a&a , 233 , 62 persson , s. , aaronson , m. , cohen , j. , frogel , j. , matthews , k. , 1983 , apj , 266 , 105 pessev , p. m. , goudfrooij , p. , puzia , t. h. , chandar , r. , 2008 , mnras , 385 , 1535 poole , v. , worthey , g. , lee , h .- c .
, serven , j. , 2010 , aj , 139 , 809 rakos , k. d. , kreidl , t. j. , schombert , j. m. , 1990 , astrophysics and space science , 170 , 239 rakos , k. d. , maindl , t. i. , schombert , j. m. , 1996 , apj , 466 , 122 rakos , k. d. , schombert , j. m. , kreidl , t. j. , 1991 , apj , 377 , 382 russell , s. c. , bessell , m. s. , 1989 , apjs , 70 , 865 salpeter , e. e. , 1955 , apj , 121 , 161 schombert , j. m. , hanlan , p. c. , barsony , m. , rakos , k. d. , 1993 , aj , 106 , 923 searle , l. , wilkinson , a. , bagnuolo , w. g. , 1980 , apj , 239 , 803 serven , j. , worthey , g. , briley , m. m. , 2005 , apj , 627 , 754 serven , j. , worthey , g. , toloba , e. , sanchez - blazquez , p. , 2011
, aj , 141 , 184 shang , z. et al . , 1998 ,
apj , 504 , l23 strmgren , b. , 1966 , ara&a , 4 , 433 thomas , d. , maraston , c. , bender , r. , 2003 , mnras , 339 , 897 | in the study of galaxy integrated light , if photometric indicators could extract age and metallicity information of high enough quality , photometry might be vastly more efficient than spectroscopy for the same astrophysical goals . toward this end
, we search three photometric systems : david dunlap observatory ( ddo ) , beijing - arizona - taiwan - connecticut ( batc ) , and strmgren systems for their ability to disentangle age and abundance effects .
only the strmgren @xmath0 $ ] vs. @xmath1 $ ] plot shows moderate age - metallicity disentanglement .
we also add to the discussion of optical to near - infrared johnson - cousins broad band colours , finding a great decrease in age sensitivity when updated isochrones are used .
[ firstpage ] galaxies : abundances galaxies : evolution galaxies : general galaxies : photometry galaxies : stellar content |
let @xmath0 and @xmath1 be integers and @xmath2 or @xmath3 .
a jacobi matrix with matrix entries is an operator of the form @xmath4 where @xmath5 are selfadjoint complex @xmath6 matrices and @xmath7 are invertible complex @xmath6 matrices . if @xmath3 , then the jacobi matrix is said to have dirichlet boundary conditions ( on the left and right edge ) and we denote it by @xmath8 . if @xmath9 for @xmath10 , the jacobi matrix is called periodic and will be denoted by @xmath11 .
both @xmath11 and @xmath8 are selfadjoint operators on the hilbert space @xmath12 .
matrices of type appear in a number of applications . in an example coming from solid state physics
, @xmath13 is the finite volume approximation of a @xmath14-dimensional tight - binding hopping model with periodic boundary conditions on a discretized cube @xmath15 .
the fibers are then of dimension @xmath16 and the @xmath17 describe the model within the fibers while the @xmath18 model the couplings between them .
the purpose of this note is to present an algorithm to calculate the eigenvalues of @xmath13 which extends the well - known sturm - liouville oscillation theory for one - dimensional second - order differential equations .
the main results are resembled in the following theorem : as a function of the energy for the hamiltonian @xmath19 with a simple fiber ( @xmath20 ) and for @xmath21 with @xmath22 , @xmath23 and @xmath24 ( left figure ) and @xmath25 ( right figure ) .
the eigenvalues are at the intersections with the axis , namely at @xmath26 , @xmath27 , @xmath28 , @xmath29 and @xmath30 for @xmath24 , and at @xmath31 , @xmath32 and @xmath33 for @xmath25 .
one also sees that one argument approaches @xmath34 from above as @xmath35 , while another approaches @xmath36 in that limit . for @xmath37
there is a similar behavior .
_ , title="fig:",scaledwidth=47.0% ] as a function of the energy for the hamiltonian @xmath19 with a simple fiber ( @xmath20 ) and for @xmath21 with @xmath22 , @xmath23 and @xmath24 ( left figure ) and @xmath25 ( right figure ) .
the eigenvalues are at the intersections with the axis , namely at @xmath26 , @xmath27 , @xmath28 , @xmath29 and @xmath30 for @xmath24 , and at @xmath31 , @xmath32 and @xmath33 for @xmath25 .
one also sees that one argument approaches @xmath34 from above as @xmath35 , while another approaches @xmath36 in that limit . for @xmath37
there is a similar behavior .
_ , title="fig:",scaledwidth=47.0% ] [ theo - osci ] associated to @xmath13 there is a real analytic path @xmath38 of @xmath39 unitary matrices such that the following hold : \(i ) as a function of energy @xmath40 , the eigenvalues of @xmath41 rotate around the unit circle in the positive sense and with non - vanishing speed .
\(ii ) as @xmath42 , half of the eigenvalues of @xmath41 converge to @xmath43 and the other half to @xmath44 .
\(iii ) the multiplicity of @xmath40 as eigenvalue of @xmath19 equals the multiplicity of @xmath45 as eigenvalue of @xmath41 .
the theorem is illustrated by two numerical examples in figure 1 and 2 .
of course , an important point explained below is how to calculate the unitaries @xmath41 .
equation below shows that @xmath41 can be calculated iteratively using the matrix mbius transformation . in the example of a tight - binding hamiltonian cited above , theorem [ theo - osci ]
thus reduces the linear algebra problem by one dimension from @xmath14 to @xmath46 , because the mbius transformation involves inverting @xmath6 matrices .
however , this has to be done many times in order to deduce the eigenvalues of @xmath19 using theorem [ theo - osci ] .
hence , whether this procedure is of any numerical interest when it comes to calculating the eigenvalues of matrices of type is not clear to the author . on a theoretical level ,
the arguments below leading to theorem [ theo - osci ] show that @xmath41 naturally describes the intersection theory of two lagrangian planes , one given by the formal solutions of the schrdinger equation associated to @xmath19 and the other by a boundary condition modeling the periodicity . as a function of the energy for the hamiltonian @xmath19 with a double fiber ( @xmath47 ) and for @xmath48 with @xmath49 , @xmath50 and @xmath51 ( left figure ) and @xmath52 ( right figure ) .
the @xmath53 eigenvalues are at the intersections with the axis and the asymptotics for large @xmath40 are also seen .
unfortunately , the graphics become less trackable for larger @xmath54 and @xmath55 . _ , title="fig:",scaledwidth=47.0% ] as a function of the energy for the hamiltonian @xmath19 with a double fiber ( @xmath47 ) and for @xmath48 with @xmath49 , @xmath50 and @xmath51 ( left figure ) and @xmath52 ( right figure ) .
the @xmath53 eigenvalues are at the intersections with the axis and the asymptotics for large @xmath40 are also seen .
unfortunately , the graphics become less trackable for larger @xmath54 and @xmath55 . _ ,
title="fig:",scaledwidth=47.0% ] before the more technical part of the paper , let us review the history of sturm - liouville oscillation theory for matrix - valued operators .
the first result is due to lidskii @xcite who considered hamiltonian systems and implicitly studied the corresponding operators with dirichlet boundary conditions .
independently and motivated by the closed geodesic problem ,
bott @xcite studied matrix - valued periodic sturm - liouville operators and proved a version of the above theorem . to our best knowledge
, the case of jacobi matrices with matrix - valued entries was first considered in our prior work @xcite .
it contained only separated left and right boundary conditions and , in particular , the case of @xmath8 .
this result is recalled in section [ sec - transfer ] . to round things up ,
section [ sec - hamiltoniansys ] briefly shows how the very same techniques can be applied to hamiltonian systems and thus provides an alternative proof of the result of bott @xcite .
the main reason why we believe that the present proof considerably simplifies prior ones is the following : the boundary conditions are ( hermitian ) lagrangian planes and these lagrangian planes can be identified with the unitary matrices ( proposition [ prop - diffeo ] ) .
for the spectral problem , one is then led to study the intersection of the formal solutions of the schrdinger equation with the boundary conditions .
the intersection of two lagrangian planes can easily be read off the spectrum of the associated unitaries ( proposition [ prop - wronski ] ) .
the bott - maslov index of a path of lagrangian planes counts the number of these intersections and can thus simply be defined by looking at the spectrum of the associated unitaries ( section [ sec - index ] ) .
the path relevant for the eigenvalue problem of @xmath8 is given by the energy dependence of lagrangian planes naturally associated to the solutions of the schrdinger equation @xmath56 .
for the study of the periodic operators @xmath19 , one also needs the bott or conley - zehnder index of a path of hermitian symplectic matrices . but
using a variant of the graph of each such matrix ( which is also lagrangian ) this index is just a special case of the bott - maslov index in doubled dimension ( section [ sec - index ] ) . many of the facts about intersection theory of lagrangian planes and symplectic paths are well - known ( even though in general stated for real symplectic and not hermitian symplectic structure ) , but the consequent use of the identification of lagrangian planes with unitaries leads to a particularly intelligible and compact presentation below .
the first such intersection theory seems to be due to bott @xcite and was developed precisely for the eigenvalue problem of matrix - valued sturm - liouville operators , see section [ sec - hamiltoniansys ] .
nevertheless , the theory is nowadays most often associated with maslov , due to his contributions @xcite .
conley and zehnder considered paths of symplectic matrices and defined their index @xcite , but this could also be considered a further development of bott s ideas .
a spectral flow definition of the index is due to robbin and salomon @xcite .
extensions , further geometric applications as well as a nice review on the indices as well as eigenvalue problems are given by long @xcite .
another recent review on various approaches to the above indices and links between them is @xcite .
sections [ sec - symplectic ] and [ sec - index ] resembles those results relevant for the proof of theorem [ theo - osci ] without further references , but also no claim of novelty .
* acknowledgement : * the author thanks a. knauf for discussions and a referee for a number of useful suggestions which allowed to improve the paper .
this work was supported by the dfg .
let the symplectic structure be given in its standard form : @xmath57 we recall that ( hermitian symplectic ) lagrangian planes in @xmath58 are @xmath55-dimensional planes on which the symplectic form @xmath59 vanishes .
such planes can be described by @xmath55 linearly dependent vectors @xmath60 , which we regroup to an @xmath61 matrix @xmath62 of rank @xmath55 . for a lagrangian plane
this matrix then satisfies @xmath63 and we then call @xmath64 also a lagrangian frame
. actually , several frames @xmath65 describe the same plane which is hence specified by an equivalence class @xmath66_\sim$ ] with respect to the relation @xmath67 for some @xmath68 . the lagrangian grassmannian @xmath69 is the set of all equivalence classes of lagrangian frames .
if a plane has a real representative @xmath64 , then it belongs to the real lagrangian grassmannian @xmath70 usually considered in symplectic geometry @xcite .
even though well - known @xcite , we recall the short argument leading to the following result for the convenience of the reader because it is central to the approach below .
[ prop - diffeo ] as real analytic manifold , the lagrangian grassmannian @xmath69 is diffeomorphic to the unitary group @xmath71 via the stereographic projection @xmath72 defined by @xmath73_\sim\bigr ) \;=\ ; \left[\binom{{{\bf 1}}_{l}}{\imath\,{{\bf 1}}_{l}}^*{\phi}\right]\ , \left[\binom{{{\bf 1}}_{l}}{-\imath\,{{\bf 1}}_{l}}^*{\phi}\right]^{-1}\;.\ ] ] this map also establishes a diffeomorphism from @xmath70 to the symmetric unitaries in @xmath71 .
* let @xmath74 be lagrangian where @xmath75 and @xmath76 are @xmath6 matrices satisfying @xmath77
. then @xmath78 it follows that the inverse in @xmath79_\sim\bigr)=(a-\imath \,b)(a+\imath \,b)^{-1}$ ] exists and thus that @xmath80 is well - defined .
setting @xmath81 and @xmath82 , one has @xmath83 and @xmath79_\sim\bigr)=\alpha\beta^{-1}$ ] .
therefore @xmath84_\sim)^*\pi([\phi]_\sim)=(\beta^*)^{-1}\alpha^*\alpha\,\beta^{-1}={{\bf 1}}_l$ ] so that @xmath84_\sim)$ ] is unitary .
further note that the definition of @xmath84_\sim)$ ] is indeed independent of the choice of the representative for @xmath85_\sim$ ] .
moreover , one can directly check that the inverse of @xmath80 is given by @xmath86_\sim \;.\ ] ] this is clearly real analytic and thus completes the proof of the first statement .
it is straightforward to implement the real symmetry , see @xcite where also a symplectic symmetry of the plane is dealt with . @xmath87
the dimension of the intersection of two lagrangian planes can be conveniently read off from the spectral theory of the associated unitaries , as shows the next proposition .
[ prop - wronski ] let @xmath65 and @xmath88 be lagrangian frames w.r.t .
to @xmath89 , and let @xmath90_\sim)$ ] and @xmath91_\sim)$ ] .
then @xmath92 for the first equality to hold , one only needs @xmath64 to be lagrangian .
* let us begin with the inequality @xmath93 of the first equality .
suppose there are two @xmath94 matrices @xmath95 of rank @xmath96 such that @xmath97 .
then @xmath98 so that the kernel of @xmath99 is at least of dimension @xmath96 .
inversely , given a @xmath94 matrix @xmath100 of rank @xmath96 such that @xmath101 , we deduce that @xmath102 .
as the column vectors of @xmath64 and @xmath103 are orthogonal and span @xmath58 , it follows that the column vectors of @xmath104 lie in the span of @xmath64 , that is , there exists an @xmath94 matrix @xmath105 of rank @xmath96 such that @xmath106 .
this shows the other inequality and hence proves the first equality of the lemma . for the second
, we first note that the dimension of the kernel of @xmath99 does not depend on the choice of the representative .
we use the representative of @xmath85_\sim$ ] given in and a similar one for @xmath107_\sim$ ] in terms of @xmath108 . but
a short calculation then shows that @xmath109 which implies the second equality .
lagrangian planes are mapped to lagrangian planes by matrices in the hermitian symplectic group defined by @xmath110 in the literature on krein spaces , the hermitian symplectic matrices are also called @xmath59-unitaries @xcite . if all entries of @xmath111 are real , then @xmath112 is in the symplectic group @xmath113 which acts on real lagrangian planes . by conjugation with the cayley
transform @xmath114 the hermitian symplectic group is isomorphic with the generalized lorentz group @xmath115 of signature @xmath116 conserving the quadratic form @xmath117 , namely @xmath118 given explicitly by @xmath119 as already stated , the hermitian symplectic group acts on the lagrangian grassmannian through the map @xmath120_\sim])\in\mbox{\rm hs}(2l,{{\mathbb c}})\times{{\mathbb l}}_l\mapsto [ { { \cal t}}\phi]_\sim\in{{\mathbb l}}_l$ ] .
we also denote it simply by @xmath121_\sim=[{{\cal t}}\phi]_\sim$ ] . under the stereographic projection @xmath80
this action becomes the action of the lorentz group @xmath115 via matrix mbius transformation on the unitary group : @xmath122_\sim\bigr ) \;=\ ; { { \cal c}}{{\cal t}}{{\cal c}}^*\cdot\pi\bigl([\phi]_\sim\bigr ) \;,\ ] ] where the dot denotes @xmath123 this formula also defines an action of @xmath115 on the siegel disc of matrices satisfying @xmath124 ( _ e.g. _ @xcite ) , but here we use the action on the maximal boundary of the siegel disc seen as a stratified space ( for further details , see @xcite ) .
next let us recall the well - known fact that the graph of a hermitian symplectic matrix @xmath112 is lagrangian w.r.t . the form @xmath125 .
this can be readily checked by representing the graph by a @xmath126-dimensional frame @xmath127 in @xmath128 . another way to state that is that @xmath129 is built from a fixed lagrangian frame @xmath130 w.r.t .
@xmath125 distorted by the matrix @xmath131 conserving the form @xmath125 .
this matrix @xmath125 also conserves the form @xmath132 and thus , given any lagrangian plane @xmath133 w.r.t .
this form @xmath132 , one obtains a lagrangian frame @xmath134 w.r.t .
this is the idea behind the following construction , except that we work with a different representation and a particular @xmath133 adapted to our purposes below .
hence let us associate to @xmath135 the @xmath136 matrix @xmath137 where the @xmath138 denotes the symplectic checker board sum given by @xmath139 this matrix is in the group @xmath140 of matrices conserving the symplectic form @xmath141 .
we denote the stereographic projection from @xmath142 to @xmath143 by @xmath144 and the @xmath136 cayley transform by @xmath145 .
then @xmath146 is the generalized lorentz group of signature @xmath147 .
furthermore , one can also define the symplectic sum of frames by @xmath148 if @xmath64 and @xmath133 are @xmath59-lagrangian , then @xmath149 is @xmath150-lagrangian .
we will use a particular @xmath150-lagrangian which is not of this form : @xmath151 one has @xmath152 so that @xmath153 defines a class in the lagrangian grassmannian @xmath142 . note that in our notations the hat always designates objects with @xmath55 replaced by @xmath126 , but , moreover
, @xmath154 is a particular matrix in @xmath140 associated to a given @xmath111 .
now as @xmath153 is lagrangian , so is @xmath155 and we can therefore associated a unitary @xmath156_\sim)$ ] to it . a modification of this
unitary will turn out to be particularly useful for the spectral analysis of @xmath111 ( see proposition [ prop - ttou ] below ) .
[ prop - ttou0 ] to a given @xmath111 let us associate a unitary @xmath157_\sim ) \;\in\ ; \mbox{\rm u}(2l ) \;.\ ] ] then @xmath158 where the matrices @xmath159 and @xmath160 are given by @xmath161 the map @xmath162 is a dense embedding with image @xmath163 thus @xmath143 is a compactification of @xmath164 . if @xmath165 , then @xmath166 is symmetric .
the group @xmath167 is embedded in the compact space of symmetric unitaries in @xmath143 .
* proof . *
first of all , @xmath168 is in the generalized lorentz group @xmath169 and thus by one has @xmath170 so that @xmath160 is invertible ( similarly @xmath171 is invertible ) .
now by , @xmath172_\sim ) \;=\ ; \left ( \begin{array}{cccc } { { \bf 1}}_l & 0 & 0 & 0 \\ 0 & a & 0 & b \\ 0 & 0 & { { \bf 1}}_l & 0 \\ 0 & c & 0 & d \end{array } \right ) \cdot \left ( \begin{array}{cc } 0 & -\imath\,{{\bf 1}}_l \\
-\imath\,{{\bf 1}}_l & 0 \end{array } \right ) \;,\ ] ] so that writing out the mbius transformation , one gets @xmath173_\sim ) & = & \left ( \begin{array}{cc } 0 & -\imath\,{{\bf 1}}_l \\ -\imath\,a & b \end{array } \right ) \left ( \begin{array}{cc } { { \bf 1}}_l & 0 \\ -\imath\,c & d \end{array } \right)^{-1 } \\ & = & \left ( \begin{array}{cc } 0 & -\imath\,{{\bf 1}}_l \\ -\imath\,a & b \end{array } \right ) \left ( \begin{array}{cc } { { \bf 1}}_l & 0 \\ \imath\,d^{-1}c & d^{-1 } \end{array } \right)\;.\end{aligned}\ ] ] from this the formula for @xmath166 follows and clearly its lower left entry , the matrix denoted by @xmath174 in , is invertible .
moreover , by unitarity , @xmath175 so that @xmath176 , which when combined with @xmath177 implies that also @xmath178 is invertible . finally let us show that the map @xmath162 is surjective onto the set .
indeed , given an element of this set , it is natural to set @xmath179 , @xmath180 , @xmath181 and @xmath182 . with some care one
then checks that the equations in indeed hold .
@xmath87 of course , the entries of @xmath183 can be read off from those of @xmath112 : @xmath184 hence @xmath166 can also be written out in terms of the entries of @xmath112 .
the following result justifies the above construction and , in particular , the choice of @xmath153 .
[ prop - ttou ] let @xmath112 and @xmath166 be as in proposition [ prop - ttou0 ] .
for @xmath10 introduce the unitary @xmath185 then @xmath186 * proof .
* let us introduce @xmath187 this is a lagrangian frame w.r.t .
suppose that this frame and the lagrangian frame @xmath155 have a non - trivial intersection .
this means that there are vectors @xmath188 such that such @xmath189 .
the first and third line of this vector equality imply @xmath190 and @xmath191 , the other two that @xmath192 .
this shows @xmath193 but now proposition [ prop - wronski ] can be applied to calculate the r.h.s .. as @xmath194_\sim\bigr ) \;=\ ; \left ( \begin{array}{cc } 0 & -\imath\ , e^{-\imath k}\,{{\bf 1}}_{l } \\
-\imath\ , e^{\imath k}\,{{\bf 1}}_{l } & 0 \end{array } \right ) \;,\ ] ] proposition [ prop - ttou0 ] completes the proof .
let us fix a lagrangian plane @xmath107_\sim\in{{\mathbb l}}_l$ ] and define the associated singular cycle @xmath195 as the stratified space @xmath196_\sim\in{{\mathbb l}}_l\;\left| \ ; \dim\bigl(\phi\,{{\mathbb c}}^l\cap\psi\,{{\mathbb c}}^l\bigr)=l\ , \right.\right\ } \;.\ ] ] under the stereographic projection one gets according to proposition [ prop - wronski ] @xmath197_\sim)^*u-{{\bf 1}}_l \bigr)=l\ , \right.\right\ } \;.\ ] ] let now @xmath198 be a ( continuous ) path in @xmath69 for which the number of intersections @xmath199 is finite and does not contain the initial point @xmath200 . we explain below that these transversality and boundary conditions can be considerably relaxed in a straightforward manner .
the index of interest here counts the number of intersections of the path with the singular cycle @xmath201 weighted by the orientation of the intersections . according to
, these intersection can be conveniently analyzed by the spectral flow of the unitaries @xmath202_\sim)^*\ , \pi(\gamma^e ) \;.\ ] ] at an intersection @xmath203 , let @xmath204 be those eigenvalues of the unitary @xmath205 which are all equal to @xmath45 at @xmath206 .
we call the @xmath207 also the eigenphases of @xmath205
. choose @xmath208 such that @xmath209 $ ] for @xmath210 and @xmath211 $ ] and that there are no other eigenphases in @xmath212 $ ] for @xmath213 and finally @xmath214 for those parameters .
let @xmath215 and @xmath216 be the number of those of the @xmath217 eigenphases less than @xmath218 respectively before and after the intersection , and similarly let @xmath219 and @xmath220 be the number of eigenphases larger than @xmath218 before and after the intersection .
then the signature of @xmath221 is defined by @xmath222 note that @xmath223 and that @xmath224 is the effective number of eigenvalues that have crossed @xmath45 in the counter - clock sense .
furthermore the signature is stable under perturbations of the path in the following sense : if an intersection by @xmath225 is resolved by a perturbation into a series of intersections by lower strata , then the sum of their signatures is equal to @xmath224
. finally let us remark that , if the phases are differentiable and @xmath226 for @xmath210 , then @xmath224 is equal to the sum of the @xmath217 signs @xmath227 , @xmath210 .
yet another equivalent way to calculate @xmath224 is as the signature of @xmath228 seen as quadratic from on the eigenspace of @xmath229 to the eigenvalue @xmath45 ( again under the hypothesis that the form is non - degenerate ) . now the intersection number or index of the path @xmath230 w.r.t .
the singular cycle @xmath201 is defined by @xmath231 if the initial point @xmath200 is on the singular cycle , but say the speeds of the eigenvalues passing through @xmath45 are non - vanishing , the index can still be defined . in order to conserve a concatenation property , we include only the initial point @xmath200 in and not the final point @xmath232 . furthermore , if a path @xmath230 is such that it stays on the the singular cycle for an interval of parameters @xmath40 , but it is clearly distinguishable how many eigenvalues pass through @xmath45 in the process , then the index can be defined as well .
we do not write out formal , but obvious definitions in these cases and restrict ourself for sake of simplicity to the transversal case with non - singular end points .
it is obvious from the definition that @xmath233 is a homotopy invariant under homotopies keeping the end points of @xmath230 fixed .
if the path @xmath230 is closed , namely @xmath234 , then there is an alternative way to calculate the index .
let @xmath235\mapsto e^{\imath\theta_l^e}$ ] be continuous paths of the eigenvalues of @xmath229 with arbitrary choice of enumeration at level crossings .
each of these paths has a winding number and clearly @xmath236 in particular , the r.h.s .
is independent of the choice of enumeration of the @xmath237 s at level crossings .
this leads to @xmath238 as @xmath239_\sim))$ ] it follows that the index @xmath233 can also be calculated as the winding number of @xmath240 and is hence independent of @xmath133 for a closed path . in this case
we therefore simply write @xmath241 .
otherwise stated , @xmath241 is the pairing of @xmath230 with a cocycle in the integer cohomology called the arnold cocycle @xcite .
the winding number can be calculated using a lift @xmath242 obtained with the multi - branched function @xmath243 .
let us now define also for a non - closed path its winding integral @xmath244 by @xmath245 this is independent of the choice of the lift . in case that the path is differentiable one has @xmath246 for a closed path , @xmath247 .
let us point out that all these definitions also work in the case of paths in the real lagrangian grassmannian @xmath70 which is more frequently studied in the literature .
then the unitaries @xmath229 are symmetric , but this does not change the above spectral flow picture . similarly , other symmetries than the complex conjugation can be implemented without alternating the definition of the index .
another example is the quaternion symmetry @xcite .
let us collect a few basic properties of the index .
[ prop - index ] for @xmath111 and @xmath230 as above , set @xmath248 .
\(i ) let the path @xmath249 denote the concatenation with a second path @xmath250 .
then @xmath251 \(ii ) given a second path @xmath252 and lagrangian frame @xmath253 , one has ( with notation ) @xmath254
\(iii ) one has @xmath255 .
\(iv ) for any lagrangian @xmath253 , @xmath256 \(v ) for a closed path @xmath230 , the index @xmath233 is independent of @xmath133 and @xmath257
. * proof . * items ( i ) through ( iii ) follow immediately from the definition . using proposition [ prop - diffeo ] , one can check that the compact subgroup @xmath258 acts transitively on @xmath69 .
hence there is an @xmath259 such that @xmath260
. then there exists @xmath261 such that @xmath262 is a rotation matrix by an angle @xmath263 , namely @xmath264 . now using ( iii ) , @xmath265 but @xmath266 so that all @xmath55 eigenvalues of @xmath267 are shifted by @xmath268 . by this
implies that the index difference can be at most @xmath55 . because @xmath269 the second estimate of ( iv ) is equivalent to the first one .
the third one follows either from first two or directly from the definitions . in ( v )
the independence of @xmath133 was already proved above and it implies the second claim . @xmath87
next we consider a path @xmath270 in the group @xmath164 . given a second such path @xmath271 , one has @xmath272 .
similarly the inverse path is @xmath273 .
combined with a lagrangian @xmath64 as base point or a path @xmath198 in @xmath69 , each such path @xmath274 induces paths in the lagrangian grassmannian @xmath69 : @xmath275_\sim)_{e\in[e_0,e_1 ) } \ ; , \qquad \gamma\,\gamma \;=\ ; ( { { \cal t}}^e\,\gamma^e)_{e\in[e_0,e_1 ) } \;.\ ] ] inversely , given a path @xmath198 in @xmath69 , there are many different ways to associate paths in @xmath164 .
one family of such paths is somewhat singled out as paths in @xmath276 : choose a representative @xmath277_\sim$ ] satisfying @xmath278 and then set @xmath279 .
[ prop - index2 ] for lagrangian @xmath64 and @xmath230 , @xmath274 , @xmath280 as above one has : \(i ) @xmath281 .
\(ii ) for any lagrangian @xmath282 , @xmath283 \(iii ) for any lagrangian @xmath253 and @xmath284 , @xmath285 \(iv ) for closed paths @xmath230 and @xmath274 and any @xmath64 , @xmath286 \(v ) for a closed path @xmath274 , @xmath287 is independent of @xmath64 . * proof . *
( i ) the index @xmath288 is defined in as a sum over @xmath40 with non - trivial intersections @xmath289 .
precisely for those @xmath40 the intersection @xmath290 is non - trivial .
moreover , the dimensions of these intersections coincide and a bit of thought shows that the signatures of the corresponding paths ( as @xmath40 varies ) have reversed signs .
now by ( i ) , @xmath291 , so proposition [ prop - index](iv ) implies ( ii ) . in order to prove ( iii ) we will check that @xmath292 then ( iii ) follows when combining this with proposition [ prop - index](iv ) . for the proof of , denote @xmath293_\sim)$ ] , @xmath294 and the entries of @xmath295 as in ,
that is , we drop the energy dependence throughout .
then , using that @xmath171 and @xmath160 are invertible ( _ cf .
_ proof of proposition [ prop - ttou0 ] ) , @xmath296 now let us take the winding integral .
the l.h.s .
gives @xmath297 , while @xmath298 and the second factor on the r.h.s .
give @xmath299 and @xmath300 .
further @xmath301 because @xmath302 , so that @xmath303 only needs one branch of the logarithm and hence the winding integral is bounded by @xmath304 .
similarly @xmath305 and thus the other three factors give also at most a contribution @xmath304 .
this implies . considering closed paths
, this argument also shows ( iv ) .
item ( v ) is clear from ( iv ) .
@xmath87 we are now interested in counting the number of passages of the eigenvalues of @xmath306 through @xmath307 weighted by orientation . in view of the discussions in section [ sec - symplectic ] , there is very natural way to do this : the graph of @xmath306 is again a path in the lagrangian grassmannian @xmath142 and the singular cycle associated to @xmath308 defined in locates the matrices @xmath306 with eigenvalue @xmath307 .
an index of @xmath274 can now be defined by just applying the previous definition to the graph .
hence , associated to @xmath274 let us define @xmath309 where as above @xmath310 and then @xmath311_\sim\ , , \widehat{\psi}_k\bigr ) \;.\ ] ] for a closed path @xmath274 , the index @xmath312 is again independent of @xmath307 and denoted by @xmath313 .
several of the other properties of proposition [ prop - index ] transpose immediately to @xmath314 , others are recollected in the next proposition .
[ prop - czindex ] let @xmath64 , @xmath133 , @xmath230 , @xmath274 and @xmath280 be as above .
\(i ) for a closed path @xmath274 , one has @xmath315 independently of @xmath64 .
\(ii ) for closed paths @xmath230 and
@xmath274 , one has @xmath316 .
\(iii ) for closed paths @xmath274 and @xmath280 , one has @xmath317 .
\(iv ) one has @xmath318
\(v ) one has @xmath319 * proof . * because the path on the r.h.s . of
is closed , proposition [ prop - index](v ) one can replace @xmath320 in by @xmath321 . then apply proposition [ prop - index2](i ) to shift the matrices @xmath322 on the other argument , and replace also @xmath153 by a direct sum @xmath323 .
but now all objects split into symplectic sums and thus one can apply proposition [ prop - index](ii ) . in the first argument
there is no @xmath40-dependence and hence a vanishing contribution to the index , and the second gives @xmath287 . both ( ii ) and ( iii )
follow directly from ( i ) and proposition [ prop - index2](iv ) . combining proposition [ prop - index](iv ) and proposition [ prop - index2](ii )
, we get @xmath324_\sim , \psi'\widehat{\oplus}\psi\bigr ) \,\right| \;\leq\ ; 2\,l\;.\ ] ] thus ( iv ) follows from proposition [ prop - index](ii ) .
item ( v ) results from ( iv ) and proposition [ prop - index](iii ) .
equivalent to , one can consider @xmath325 as a tridiagonal operator @xmath326 acting on sequences @xmath327 of vectors in @xmath328 as @xmath329 where @xmath330 , together with the boundary conditions @xmath331 these boundary conditions are dirichlet for @xmath3 and periodic for @xmath9 .
it is useful to take a more abstract point of view on these boundary conditions .
they are @xmath126 linear relations between the four @xmath55-dimensional vectors @xmath332 ,
@xmath333 , @xmath334 and @xmath335 , forcing them to lie in a particular @xmath126-dimensional plane of @xmath128 . in principle , any other @xmath126-dimensional plane of @xmath128 leads when combined with to an operator on a @xmath336-dimensional hilbert space .
such a plane can be described by a @xmath337 matrix @xmath338 built out of @xmath126 linearly independent vectors in @xmath128 spanning the plane . now a convenient way to formulate the generalized boundary condition is @xmath339 for some vector @xmath340 .
this defines a linear operator @xmath341 on @xmath12 .
the matrices @xmath330 are introduced in for later convenience .
the special case with @xmath9 corresponds to the choice @xmath320 defined in .
now not all boundary conditions @xmath338 lead to a self - adjoint operator @xmath341 . indeed , given
@xmath342 and @xmath343 corresponding to vectors @xmath105 and @xmath344 in , one has @xmath345 now selfadjointness requires the last term to vanish for all @xmath105 and @xmath344 .
hence a selfadjoint boundary condition is given if and only if @xmath346 .
this means that the plane @xmath347_\sim$ ] is lagrangian w.r.t .
@xmath150 . by proposition [ prop - diffeo ]
the set of self - adjoint boundary conditions can hence be identified with @xmath143 . as already pointed out ,
the boundary condition for @xmath19 is @xmath320 .
the dirichlet boundary conditions defining @xmath8 are given by @xmath348 where @xmath349 .
other boundary conditions separating left and right boundary can readily be dealt with and are simply given by modifications of @xmath350 and @xmath351 @xcite .
as for a one - dimensional jacobi matrix , it is useful to rewrite the schrdinger equation @xmath352 for a real energy @xmath353 in terms of the @xmath39 transfer matrices @xmath354 defined by @xmath355 for a real energy @xmath353 , each transfer matrix is in the hermitian symplectic group @xmath164 .
the schrdinger equation ( [ eq - schroedinger ] ) is satisfied if and only if @xmath356 and the boundary conditions ( [ eq - boundary ] ) hold .
let us first focus on the case of periodic boundary conditions where @xmath9 as this leads to the proof of theorem [ theo - osci ] stated in the introduction .
these boundary conditions can be written in a compact manner : @xmath357 next write the l.h.s . using . setting @xmath358 , this means that @xmath40 is an eigenvalue of @xmath19 if and only if @xmath307 is an eigenvalue of @xmath359 .
more precisely , @xmath360 as @xmath359 is in the group @xmath164 , one way to calculate the r.h.s
. is to appeal to proposition [ prop - ttou ] .
hence let the expanded transfer matrices @xmath361 be defined as in .
then define lagrangian frames by @xmath362 with @xmath153 as in .
finally let us denote the associated unitaries by @xmath363_\sim\bigr ) \ ; , \qquad \widehat{u}^{e , k}_n \;=\ ; \left ( \begin{array}{cc } 0 & \imath\,e^{-\imath
k}\,{{\bf 1}}_{l } \\ \imath\,e^{\imath k}\,{{\bf 1}}_{l } & 0 \end{array } \right ) \;\widehat{w}^e_n \;.\ ] ] due to , these unitaries can be calculated iteratively using the mbius transformation : @xmath364 now proposition [ prop - ttou ] applied to @xmath365 directly gives item ( iii ) of theorem [ theo - osci ] .
the other two statements are proved in sections [ sec - mono ] and [ sec - asymp ] .
now let us consider the case of dirichlet boundary conditions .
this could be done in exactly the same manner using the lagrangian frames @xmath366 defined in , but with the initial condition @xmath367 , and then analysing its intersection of @xmath368 with @xmath369 .
as all objects are symplectic direct sums , it is easier to consider directly the @xmath55-dimensional lagrangian frames @xmath370 where @xmath371 . by the intersection theory of section [ sec - symplectic ] ,
the multiplicity of @xmath40 as eigenvalue of @xmath372 is then equal to the dimension of the intersection of the lagrangian frames @xmath373 and @xmath374 , and hence equal to @xmath45 as eigenvalue of the unitary @xmath375_\sim)^*\ , \pi([\phi^e_n]_\sim ) \;=\ ; -\;\pi([\phi^e_n]_\sim ) \;.\ ] ] this provides the proof of the first statement of the following theorem .
the proof of the second is similar to that of proposition [ prop - derivbound ] below .
[ theo - osci2 ] @xcite the multiplicity of @xmath40 as eigenvalues @xmath372 is equal to the multiplicity of @xmath45 as eigenvalue of @xmath376 . as a function of energy @xmath40 ,
the eigenvalues of @xmath377 rotate around the unit circle in the positive sense and with non - vanishing speed .
let us point out that in the case @xmath20 the unitaries @xmath376 are called prfer phases .
hence it is natural to call them matrix prfer phases for general @xmath55 .
according to the above , the spectrum of @xmath19 can be read off the spectrum of the path of unitaries @xmath378 .
we next show that the eigenvalues of this unitary all rotate in the same direction , namely item ( i ) of theorem [ theo - osci ] .
the latter follows from the positivity of the operator appearing in the following proposition , because the rotation speeds of the eigenvalues of @xmath41 are given by the diagonal matrix elements of this operator w.r.t .
to the basis of eigenvectors of @xmath41 .
[ prop - derivbound ] for @xmath353 and @xmath379 , the matrix @xmath380 is positive semi - definite .
* proof . *
( in part , similar to @xcite ) let us introduce @xmath381 .
these are invertible @xmath39 matrices by the argument in proposition [ prop - diffeo ] and one has @xmath382 .
now @xmath383 ( \phi_+^e)^{-1 } \;.\ ] ] thus it is sufficient to verify positive definiteness of @xmath384 \;=\ ; 2\ ; ( \widehat{\phi}^e_n)^*\,\widehat{{{\cal j}}}\,\partial_e \widehat{\phi}^e_n \;.\ ] ] from the product rule follows that @xmath385 this implies that @xmath386 as one checks that @xmath387 and the matrices @xmath388 do not mix first and third columns and lines with the second and forth ones , it follows that @xmath389 clearly each of the summands is positive semi - definite . in order to prove the strict inequality , it is sufficient that the first two terms @xmath390 give a strictly positive contribution . hence let us verify that @xmath391 for this purpose let us show that the kernel of the matrix on the l.h.s .
is empty . as @xmath392
, we thus have to show that a vector @xmath393 satisfying @xmath394 actually vanishes . carrying out the matrix multiplications ,
one readily checks that this is the case .
@xmath87 the matrix @xmath395 can be calculated using the iterative mbius transformation , see . in parallel , the positive matrices @xmath396 can be calculated iteratively .
the corresponding equations can be readily written out , but as they are a bit lengthy and not used here we refrain from doing so .
the following result proves theorem [ theo - osci](ii ) .
[ prop - asymptotics ] the asymptotics of the unitaries are @xmath397 * proof . * it is sufficient to check @xmath398 .
we shall use @xmath399_\sim)$ ] and the formula given in proposition [ prop - ttou0 ] . hence let @xmath400 be the entries of @xmath401 and @xmath402 those of @xmath403 .
their link can be read off from .
one has @xmath404 while @xmath405 are all of order @xmath406
. hence @xmath402 are all equal to @xmath407 up to terms of order @xmath406 .
hence by proposition [ prop - ttou0 ] @xmath408 even though not much can be deduced directly about the lower left entry @xmath409 , unitarity of @xmath410 imposes @xmath411 .
@xmath87 proposition [ prop - asymptotics ] implies that the path @xmath378 can be closed using the one - point compactification of @xmath412 . as such
it can also be seen as a closed path @xmath413 in the lagrangian grassmannian @xmath142 if we set @xmath414_\sim$ ] . by propositions [ prop - index2](i ) and [ prop - index](ii )
its index is equal to the index of the path @xmath415 in @xmath69 with @xmath416 for any choice of @xmath417 .
the index can readily be calculated by a homotopy argument and is equal to @xmath336 @xcite .
furthermore , by the results above , this index of @xmath230 is equal to the total number of eigenvalues of @xmath19 and thus equal to @xmath336 .
it is reassuring that this can also be calculated independently .
in this section , we briefly illustrate how the techniques of sections [ sec - symplectic ] and [ sec - index ] can be used to study the solutions of linear hamiltonian systems of the form : @xmath418 where @xmath419 and @xmath420 are continuous functions on the finite interval @xmath421 $ ] into the hermitian and respectively non - negative matrices of size @xmath39 .
again @xmath353 is a spectral parameter .
let us focus on two examples .
if @xmath422 , then is the schrdinger equation for the quasi - one - dimensional dirac operator @xmath423 .
if @xmath424 , then is a rewriting of the schrdinger equation @xmath425 with @xmath426 for a matrix - valued sturm - liouville operator : @xmath427 where @xmath96 , @xmath428 and @xmath429 are continuous functions on @xmath421 $ ] into the @xmath6 matrices and @xmath96 is a continuously differentiable and positive with a uniform lower bound @xmath430 , @xmath431 .
indeed @xcite , the equivalence of @xmath425 with follows by setting @xmath432 now a solution @xmath64 of is continuous and has left and right limits @xmath433 and @xmath434 which intervene in @xmath435 with scalar product on the l.h.s . in @xmath436 , as well as in @xmath437 where the scalar product on the l.h.s
. is now in @xmath438 and @xmath64 , @xmath282 on the r.h.s .
are given by .
note that the r.h.s . in both of these equations
are the same and similar to those of the discrete case , so that also the following is completely analogous to the discrete case .
hence self - adjoint boundary conditions for @xmath439 and @xmath440 are given by @xmath441 where @xmath338 is lagrangian w.r.t .
@xmath150 . note that the permutation matrix @xmath442 is also underlying . again
the dirichlet boundary condition is given by @xmath443 and the periodic boundary condtion by @xmath320 .
further the transfer matrix of the jacobi matrices is replaced by the fundamental solution @xmath444 of , namely given by @xmath445 finally set @xmath446_\sim\bigr ) \ ; , \qquad \widehat{u}^{e , k}\;=\ ; \widehat{\pi}\bigl([\widehat{\psi}_k]_\sim\bigr)^*\ ; \widehat{\pi}\bigl([\widehat{{{\cal t}}}^e(1)\widehat{\psi}_0]_\sim\bigr ) \;.\ ] ] [ theo - osci3 ] the number of linear independent solutions of at energy @xmath40 with dirichlet boundary condition or respectively periodic boundary conditions is equal to the multiplicity of @xmath45 as eigenvalue of @xmath229 or respectively @xmath447 . as a function of energy @xmath40 , the eigenvalues of @xmath448 and @xmath447 rotate around the unit circle in the positive sense .
* proof . *
the proof of the first statement parallels that of the discrete case . for
the second , let us focus on @xmath229 an prove that @xmath449 . by an argument similar to the one in the proof of proposition [ prop - derivbound ] , it is sufficient to check that @xmath450 .
for that purpose , let @xmath451 . by
, @xmath452 as @xmath453 , one thus has @xmath454 because @xmath455 is positive , this implies the second claim .
@xmath87 one of the differences with the discrete case is that @xmath229 and @xmath447 do in general not have asymptotics as @xmath37 or @xmath35 .
both limits do nt exist in the case of @xmath439 , while for @xmath440 the limit for @xmath456 does exist , but not for @xmath37 .
the reason is that @xmath439 is neither bounded from above or below , while @xmath440 is bounded from below . as a final comment
let us exhibit another type of positivity that is intrinsic to , which is actually the one discovered by lidskii @xcite . set @xmath457_\sim\bigr ) \;.\ ] ] similar as in theorem [ theo - osci3 ] , this unitary allows to calculate the solutions of on the interval @xmath458 with dirichlet boundary conditions at @xmath218 and @xmath459 . by a calculation similar as in the proof of proposition [ prop - derivbound ] ,
one checks that @xmath460 where @xmath461 . hence as long as @xmath462 ,
the eigenvalues of @xmath463 rotate in the positive sense as a function of @xmath459 . in the case
where the hamiltonian system results from a sturm - liouville operator , this can be understood by the following quantum mechanical analogy : as @xmath459 grows , the particle described by @xmath440 becomes wider and hence the eigenvalue distances decrease so that the eigenvalues of @xmath463 have already made more rotations up to @xmath40 .
m. g. krein , _ principles of the theory of _
@xmath464_-zones of stability of a canonical system of linear differential equations with periodic coefficients _ , memory of a.a .
andronov , pp .
413 - 498 , izdat .
nauk sssr , moscow , 1955 ; english transl . in : m. g. krein ,
_ topics in differential and integral equations and operator theory _ , ( birkhuser , boston , 1983 ) | sturm - liouville oscillation theory for periodic jacobi operators with matrix entries is discussed and illustrated .
the proof simplifies and clarifies the use of intersection theory of bott , maslov and conley - zehnder .
it is shown that the eigenvalue problem for linear hamiltonian systems can be dealt with by the same approach . |
matter which is exposed to strong external magnetic fields changes its basic properties and structure and leads to a variety of new phenomena . as a result strong fields are of importance in different branches of physics like atomic , molecular or solid state physics . for atomic and molecular systems
there are two prominent possibilities to encounter the strong field regime : highly excited rydberg states in the laboratory and atoms and molecules in the atmospheres of magnetized white dwarfs ( see refs.@xcite for a compilation of the subject ) . from a theorists point of view particle systems in strong fields
pose a hard problem due to several competing interactions ( coulomb attraction and repulsion , para- and diamagnetic interactions ) . of particular interest
, but most complicated to investigate , is hereby the so - called intermediate regime which is characterized by comparable magnetic and coulomb binding forces .
focusing on the low - lying states of atoms and molecules we envisage this regime for those magnetic white dwarfs which possess field strengths in the regime @xmath2 .
each magnetic white dwarf possesses a characteristic regime of field strengths which varies , in case of a dipole , by a factor of two from the pole to the equator . to perform a first identification of observed spectra from the atmospheres of these objects
one uses the so - called stationary line spectroscopy : characteristic absorption features can appear only for those wavelengths which correspond to an extremum of the transition wavelength with respect to the field strength . in a second step one
then performs simulations of the radiation transport in the atmosphere in order to obtain synthetic spectra . in the eighties
the above approaches have been used to identify hydrogen in a number of magnetized white dwarfs @xcite .
recently the stationary line argument has been successfully @xcite used to obtain strong evidence for helium in the spectrum of the magnetic white dwarf gd229 whose absorption features have been mysterious ever since its discovery @xmath3 years ago .
this was only possible due to the enormous progress achieved with respect to our knowledge of the spectrum and transitions of the helium atom in strong magnetic fields @xcite .
however , this should not obscur the fact that there are a number of magnetic white dwarfs whose spectra remain unexplained and furthermore new magnetic objects are discovered ( see , for example , reimers et al @xcite in the course of the hamburg survey of the european southern observatory ( eso ) ) .
very recently strong candidates for quasimolecular absorption features have been discovered in magnetized white dwarfs @xcite .
this raises the demand for a theoretical investigation of molecular properties , in particular of the hydrogen molecule , in such strong fields .
several theoretical investigations were performed for molecular systems in strong magnetic fields .
most of them deal with the electronic structure of the @xmath4 ion ( see refs . @xcite and references therein ) .
very interesting phenomena can be observed already for this simple diatomic system . for the ground state of the @xmath4 molecule
the dissociation energy increases and the equilibrium internuclear distance simultaneously decreases with increasing field strength . furthermore it was shown @xcite that a certain class of excited electronic states , which possess a purely repulsive potential energy surface in the absence of a magnetic field , acquire a well - pronounced potential well in a sufficiently strong magnetic field .
moreover the electronic potential energies depend not only on the internuclear distance but also on the angle between the magnetic field and the molecular axis which leads to a very complex topological behavior of the corresponding potential energy surfaces @xcite .
in contrast to the @xmath4 ion there exist only a few investigations dealing with the electronic structure of the hydrogen molecule in the presence of a strong magnetic field .
highly excited states of @xmath5 were studied for a field strength of @xmath6 in ref .
for intermediate field strengths two studies of almost qualitative character investigate the potential energy curve ( pec ) of the lowest @xmath7 state @xcite .
a few investigations were performed in the high field limit @xcite , where the magnetic forces dominate over the coulomb forces and therefore several approximations can be used . very recently
a first step has been done in order to elucidate the electronic structure of the @xmath5 molecule for the parallel configuration , i.e. for parallel internuclear and magnetic field axes @xcite .
hereby refs.@xcite apply an exact , i.e. fully correlated approach , whereas refs.@xcite use hartree - fock calculations and focus exclusively on the identification of the global ground state of the molecule . in refs.@xcite
the lowest states of the @xmath1 and @xmath8 manifolds were studied for gerade and ungerade parity as well as singlet and triplet spin symmetry .
hereby accurate adiabatic electronic energies were obtained for a broad range of field strengths from field free space up to strong magnetic fields of @xmath9 a variety of interesting effects were revealed . as in the case of the @xmath4 ion ,
the lowest strongly bound states of @xmath1 symmetry , i.e. the lowest @xmath7 , @xmath10 and @xmath11 state , show a decrease of the bond length and an increase of the dissociation energy for sufficiently strong fields .
furthermore a change in the dissociation channel occurs for the lowest @xmath11 state between @xmath12 and @xmath13 due to the existence of strongly bound @xmath14 states in the presence of a magnetic field .
the @xmath10 state was shown to exhibit an additional outer minimum for intermediate field strengths which could provide vibrationally bound states .
an important result of refs .
@xcite is the change of the ground state from the lowest @xmath7 state to the lowest @xmath15 state between @xmath16 and @xmath17 this crossing is of particular relevance for the binding properties of the global ground state of the molecule : the @xmath15 state is an unbound state and possesses only a very shallow van der waals minimum which does not support any vibrational level .
therefore , the global ground state of the hydrogen molecule for the parallel configuration is an unbound state for @xmath18 .
furthermore it has been shown in ref .
@xcite that for very strong fields ( @xmath19 ) the strongly bound @xmath20 state is the global ground state of the hydrogen molecule oriented parallel to the magnetic field .
finally the complete scenario for the crossovers of the global ground state of the parallel configuration has been clarified in ref.@xcite which contains the transition field strengths for the crossings among the lowest states of @xmath21 and @xmath20 symmetry .
the above considerations show that detailed studies of the electronic properties of the hydrogen molecule in a magnetic field are very desirable .
the present investigation deals with the excited @xmath1 states of the hydrogen molecule in the parallel configuration which is distinct by its higher symmetry @xcite .
we employ a full configuration interaction ( ci ) approach which is most suitable for obtaining detailed information on the electronic structure .
our investigation is divided into two separate studies : the first and present work focuses on singlet states of both gerade and ungerade symmetry whereas a later investigation will focus on the corresponding triplet states again for both parities . due to the spin zeeman splitting the spin singlet and triplet manifolds
are increasingly separated with increasing field strength .
the spin character provides therefore a natural dividing line of our extensive work which contains a large amount of information and data on the behaviour of the excited states of the molecule .
the results of our calculations include accurate adiabatic pecs for the complete range of field strengths @xmath22 . up to seven excited states
have been investigated for each symmetry .
we present detailed data for the total and dissociation energies at the equilibrium internuclear distances as well as the equilibrium positions themselves for the lowest three excitations for each symmetry . due to the large amount of data the evolution of the higher excited states with increasing field strength
is presented only graphically
. however , further informations like , for example , the positions of the maxima and the accurate heights of the barriers or the complete data of the pec s as well as numerical data on the higher excited states , more precisely , on the fourth up to the seventh excited state , can be obtained from the authors upon request . in detail
the paper is organized as follows . in section
ii we describe the theoretical aspects of the present investigation , including a discussion of the hamiltonian , a description of the atomic orbital basis set and some remarks on the ci approach .
section iii contains the results and an elaborate discussion of the evolution of the electronic structure in the presence of the magnetic field with increasing strength .
our starting point is the total nonrelativistic molecular hamiltonian in cartesian coordinates .
the total pseudomomentum is a constant of motion and therefore commutes with the hamiltonian @xcite . for that reason
the hamiltonian can be simplified by performing a so - called pseudoseparation of the center of mass motion @xcite which introduces the center of mass coordinate and the conserved pseudomomentum as a pair of canonical conjugated variables .
further simplifications can be achieved by a consecutive series of unitary transformations @xcite . in order to separate the electronic and nuclear motion we perform the born - oppenheimer approximation in the presence of a magnetic field @xcite . as a first order approximation
we assume infinitely heavy masses for the nuclei .
the origin of our coordinate system coincides with the midpoint of the internuclear axis of the hydrogen molecule and the protons are located on the @xmath23 axis .
the magnetic field is chosen parallel to the @xmath23 axis of our coordinate system and the symmetric gauge is adopted for the vector potential .
the gyromagnetic factor of the electron is chosen to be equal to two .
the hamiltonian , therefore , takes on the following appearance : @xmath24 the symbols @xmath25 , @xmath26 and @xmath27 denote the position vectors , their canonical conjugated momenta and the angular momenta of the two electrons , respectively . @xmath28 and @xmath29 are the vectors of the magnetic field and internuclear distance , respectively and @xmath30 denotes the magnitude of @xmath29 . with @xmath31
we denote the vector of the total electronic spin . throughout the paper we will use atomic units .
the hamiltonian ( [ form1 ] ) commutes with the following independent operators : the parity operator @xmath32 , the projection @xmath33 of the electronic angular momentum on the internuclear axis , the square @xmath34 of the total electronic spin and the projection @xmath35 of the total electronic spin on the internuclear axis . in field
free space we encounter an additional independent symmetry namely the reflections of the electronic coordinates at the @xmath36 ( @xmath37 ) plane .
the eigenfunctions possess the corresponding eigenvalues @xmath38 .
this symmetry does not hold in the presence of a magnetic field ! therefore , the resulting symmetry groups for the hydrogen molecule are @xmath39 in field free space and @xmath40 in the presence of a magnetic field @xcite . in order to solve the fixed - nuclei electronic schrdinger equation belonging to the hamiltonian ( [ form1 ] ) we expand the electronic eigenfunctions in terms of molecular configurations . in a first step
the total electronic eigenfunction @xmath41 of the hamiltonian ( [ form1 ] ) is written as a product of its spatial part @xmath42 and its spin part @xmath43 , i.e. we have @xmath44 .
for the spatial part @xmath42 of the wave function we use the lcao - mo - ansatz , i.e. we decompose @xmath42 with respect to molecular orbital configurations @xmath45 of @xmath5 , which respect the corresponding symmetries ( see above ) and the pauli principle : @xmath46 \nonumber \\ & = & \sum\limits_{i , j } c_{ij } \left[\phi_i\left(\bbox{r_1}\right)\phi_j\left(\bbox{r_2}\right ) \pm \phi_i\left(\bbox{r_2}\right)\phi_j\left(\bbox{r_1}\right)\right ] \nonumber\end{aligned}\ ] ] the molecular orbital configurations @xmath47 of @xmath5 are products of the corresponding one - electron @xmath48 molecular orbitals @xmath49 and @xmath50 . the @xmath48 molecular orbitals are built from atomic orbitals centered at each nucleus .
a key ingredient of this procedure is a basis set of nonorthogonal optimized nonspherical gaussian atomic orbitals which has been established previously @xcite .
for the case of a @xmath51molecule parallel to the magnetic field these basis functions read as follows : @xmath52 the symbols @xmath53 and @xmath23 denote the electronic coordinates .
@xmath54 , @xmath55 and @xmath56 are parameters depending on the subspace of the h - atom for which the basis functions have been optimized and @xmath57 and @xmath58 are variational parameters .
we remark that the nonlinear optimization of the variational parameters @xmath57 and @xmath58 has to be accomplished for typically of the order of @xmath59 atomic orbitals and is done by reproducing many excited states of the hydrogen atom for each field strength separately .
it represents therefore a tedious and time consuming work which has , however , to be done with great care in order to obtain precise results for the following molecular structure calculations . for a more detailed description of the construction of the molecular electronic wave function
we refer the reader to ref .
@xcite . in order to determine the molecular electronic wave function of @xmath5 we use the variational principle which means that we minimize the variational integral @xmath60 by varying the coefficients @xmath61 .
the resulting generalized eigenvalue problem reads as follows : @xmath62 where the hamiltonian matrix @xmath63 is real and symmetric and the overlap matrix is real , symmetric and positive definite .
the vector @xmath64 contains the expansion coefficients .
the matrix elements of the hamiltonian matrix and the overlap matrix are certain combinations of matrix elements with respect to the optimized nonspherical gaussian atomic orbitals .
a description of the techniques necessary for the evaluation of these matrix elements is given in ref .
we mention here only that the electron - electron integrals needed a combination of numerical and analytical techniques in order to make its rapid evaluation possible .
the latter represents the cpu time dominating factor for the construction of the hamiltonian matrix . for the numerical solution of the eigenvalue problem ( [ form3 ] ) we used the standard nag library
the typical dimension of the hamiltonian matrix for each subspace varies between approximately 2000 and 5000 depending on the magnetic field strength .
depending on the dimension of the hamiltonian matrix , it takes between 70 and 250 minutes for simultaneously calculating one point of a pec of each subspace on a ibm rs6000 computer .
the overall accuracy of our results with respect to the total energy is estimated to be typically of the order of magnitude of @xmath65 and for some cases of the order of magnitude of @xmath66 .
it should be noted that this estimate is rather conservative ; in some ranges of the magnetic field strength and internuclear distance , e.g. , close to the separated atom limit , the accuracy is @xmath67 or even better .
the positions , i.e. internuclear distances , of the maxima and the minima in the pecs were determined with an accuracy of @xmath68 herefore about 350 points were calculated on an average for each pec .
it was not necessary to further improve this accuracy since a change in the internuclear distance about @xmath69 results in a change in the energy which is typically of the order of magnitude of @xmath65 or smaller .
the total cpu time needed to complete the present work amounts to several years on the above powerful computer .
to understand the influence of the external magnetic field on the electronic structure of the hydrogen molecule we first have to remind ourselves of the properties in the absence of the field .
accurate data for hydrogen are of great importance both in astrophysics as well as laboratory physics .
it is a paradigm for many molecular phenomena like charge transfer , excitation , ionization or scattering processes .
indeed our ci calculations on the basis of an anisotropic gaussian basis set provided also significant progress with respect to the knowledge of the field - free excitations of the molecule : several highly excited states have been calculated for the first time and some of the pecs for the lower lying states have been improved .
the corresponding results have been presented to some detail in ref.@xcite and contain elaborate information on the first eight excited singlet and triplet states for both gerade and ungerade parity . in the following we will first summarize the main properties of the excited singlet states in the absence of the field and then investigate the electronic structure in the presence of the magnetic field with increasing field strength .
we hereby first deal with the gerade and subsequently with the ungerade states . the investigation of the electronic states and pecs has been done for all internuclear distances considered ( @xmath70 ) with the same atomic orbital basis set .
the latter has been optimized to yield precise energies ( accuracy @xmath71 ) of the hydrogen atom for the six lowest states for both parities for vanishing atomic magnetic quantum number .
additionally , in order to describe correlation effects , we have included basis functions with atomic magnetic quantum numbers @xmath72
. the approximate number of two - particle configurations resulting from the above basis set is @xmath73 .
the accuracy of the electronic energies for the higher excited states @xmath74 ( @xmath75 indicates the degree of excitation ) is , due to the above choice of the optimized basis set , lower than that for less excited states .
figure 1 shows the pecs for the states @xmath76 where the dotted lines represent the curves for the higher excited states @xmath77 . there is a large energetical gap ( @xmath78 ) between the ground and the excited states of @xmath79 symmetry .
the first five excited states are well - known from the literature @xcite .
our calculations @xcite show in most cases an agreement within @xmath80 compared to the literature and in several cases also a variationally lower energy . as already mentioned the results on the higher excited states ( @xmath77 )
have for the first time been reported very recently in ref.@xcite . as can be seen from figure 1 all the pecs of the states @xmath76
possess a deep potential well around a minimum located approximately at @xmath81 .
a particular feature occuring for most of the considered @xmath82 states is the existence of a second outer minimum and therefore the corresponding pecs exhibit a double well .
vibrational states in these outer wells @xcite attracted recently significant experimental interest @xcite since they allow the experimental observation of long - lived and highly excited valence states of the hydrogen molecule .
the two minima of the @xmath83 state arise due to the fact that two different configurations of the same symmetry , namely the @xmath84 and the @xmath85 configurations , are energetically minimized at two significantly different internuclear distances
. the deep outer wells of the @xmath86 states arise due to a series of avoided crossings between the heitler - london configurations @xmath87 and the ionic configurations @xmath88 .
particularly the @xmath89 state possesses a very broad and deep ( @xmath90 ! ) outer potential well which is separated by a broad barrier from the inner well located at @xmath91 . the outer minimum is located at @xmath92 .
a series of avoided crossings at very large internuclear distances @xmath93 leads to the energetically equal dissociation limits @xmath94 of the @xmath95 states .
the dissociation channel of the @xmath96 state is the ionic configuration @xmath97 .
tables 1 to 3 contain ( among the data in the presence of a magnetic field ) the total and dissociation energies at the equilibrium internuclear distances , the equilibrium internuclear distances and the total energies in the dissociation limit for the first to third excited @xmath98 states in the absence of the magnetic field .
the subspace of @xmath7 symmetry contains the electronic ground state of the hydrogen molecule in field - free space and in the presence of a magnetic field in the regime @xmath99 for a detailed discussion of the appearance of this state and the global ground state with increasing field strength in general we refer the reader to refs.@xcite ( see also introduction of the present work ) . in the following
we investigate the evolution of the excited @xmath100 states with increasing magnetic field strength for the regime @xmath101 .
we will first study the changes of the pecs of individual states with increasing field strength and thereafter we present a global view of the evolution of the spectrum . in order to compare the pecs for the same state for different field strengths we subtract from the total energies the corresponding energies in the dissociation limit ( which is different for different field strengths ) ,
i.e. we show the quantity @xmath102 . in general the dissociation limit of a certain state of @xmath1 symmetry
changes with increasing field strength which is due to the reordering of the energy levels of the atoms ( hydrogen , hydrogen negative ion ) in the external field . for the atomic states we will use in the following the notation @xmath103 where @xmath75 specifies the degree of excitation and @xmath104 the atomic magnetic quantum number and z - parity , respectively .
let us begin our investigation of the evolution of individual states with increasing field strength with the @xmath105 state whose pecs are shown in figure 2a .
the positions of the two minima and the corresponding maximum decrease with increasing field strength .
the depth of the inner potential well decreases for @xmath106 and increases rapidly for @xmath107 . the depth of the outer well is monotonically increasing for the complete regime @xmath101 . for @xmath108 and @xmath109 the inner
well is therefore deeper than the outer well and vice versa for @xmath110 ( see figure 2a ) .
the dissociative behaviour of the pecs changes significantly with increasing field strength .
the origin of these changes is the fact that for @xmath111 the dissociation channel is @xmath112 whereas for @xmath113 we have the asymptotic behaviour @xmath114 ( the index @xmath115 stands for spin singlet ) .
the appearance of the ionic configuration as the dissociation channel for the low - lying electronic @xmath105 state can be explained as follows .
it is well - known that the hydrogen negative ion possesses infinitely many bound states in the presence of a magnetic field of arbitrary strength assuming an infinite nuclear mass @xcite .
certain of these bound states show a monotonically increasing binding energy with increasing field strength .
the latter surpass then more and more of the energy levels belonging to two hydrogen atoms one being in the global ground state and the other one in the corresponding excited state .
for a sufficiently strong magnetic field we therefore expect the configuration @xmath116 to become the dissociation channel particularly for the first excited state of @xmath7 symmetry .
due to the long range forces the onset of the asymptotic ( @xmath117 ) behaviour of the corresponding pecs with the ionic channel ( @xmath118 ) is qualitatively different from the pecs with a neutral dissociation limit ( @xmath119 ) .
this explains the different asymptotic behaviour of the pecs shown in figure 2a with increasing field strength .
finalizing the discussion of the @xmath105 state we remark that its pecs possesses a second maximum for @xmath120 which however occurs at large internuclear distances ( @xmath121 ) and is only of the order of @xmath122 above the dissociation limit .
table 1 contains relevant data of the pecs of the @xmath105 state with increasing field strength .
next we turn to the second excited i.e. the @xmath123 state whose pecs are shown in figure 2b .
the positions of the two minima and the corresponding maximum already present in field - free space decrease monotonically with increasing field strength .
starting from @xmath124 the depth of the inner well decreases with increasing field strength whereas it increases for @xmath125 .
besides a very small interval of field strengths the depth of the outer well increases with increasing field strength .
for @xmath126 the outer well is deeper than the inner one whereas for @xmath127 the deep inner well dominates the shape of the pec .
we remark that the curvature at the ( first ) maximum and the outer minimum increases significantly with increasing field strength .
the evolution of these increasingly sharper turns can only be fully understood if one looks at the complete spectrum ( see figure 3 and in particular 3(e ) ) with increasing field strength : they develop due to a number of narrow avoided crossing of the first to third excited states in strong fields .
an interesting property of the pec of the @xmath123 state is the existence of an additional outer ( third ) minimum for the interval @xmath128 which is shown in figure 2c .
this minimum arises due to the interaction with the ionic configuration @xmath118 . in field - free space
the lowest and only bound ionic channel @xmath116 is the dissociation channel of the @xmath129 state . with increasing field strength
the hydrogen negative ion becomes increasingly stronger bound ( see discussion above ) and therefore it occurs as the dissociation channel for the sequence of excited states @xmath130 finally becoming the dissocation channel of the @xmath105 state for @xmath113 .
the existence of the additional outer minimum becomes now understandable : due to the energetical lowering of the ionic dissociation channel with increasing field strength the higher excited states of @xmath7 symmetry evolve outer minima and corresponding wells for certain regimes of the field strength .
for the @xmath123 state this outer well is extremely shallow for @xmath131 and therefore almost invisible in figure 2c .
for @xmath132 it becomes however well - pronounced . between @xmath133 and @xmath134 there occurs a change with respect to the dissociation channel of the @xmath123 . for @xmath135
the dissociation channel is @xmath136 and for @xmath137 it is @xmath138 .
the similar asymptotic behaviour of the pecs belonging to different field strengths ( see figure 2c for @xmath139 and @xmath140 ) arises due to the fact that they possess all the ionic dissociation channel . in the latter regime
the position of the ( third ) outer minimum increases with increasing field strength ( for @xmath141 the outer minimum is located at @xmath142 ) .
finally there is a second change of the dissociation channel of the @xmath123 state to @xmath143 and therefore the outer minimum disappears for @xmath113 .
table 2 contains the total and dissociation energies at the equilibrium internuclear distances , the equilibrium internuclear distances and the total energies in the dissociation limit for the second excited @xmath144 state in the regime @xmath101 next we focus on the third excited @xmath145 whose pecs with increasing field strength are shown in figure 2d . in field - free space it possesses two minima and associated potential wells located at @xmath146 and @xmath147 , respectively .
the position of the inner minimum increases with increasing field strength whereas the corresponding dissociation energy decreases . finally for @xmath148 the associated well disappears but reappears for @xmath149 . with further increasing field strength the position of this inner minimum decreases and the depth of the corresponding well increases monotonically for @xmath107 . independently of this first inner minimum and the outer minimum
there appears for @xmath150 an additional third minimum and corresponding well ( see table 3 and figure 2d ) for small internuclear distances @xmath151 .
although this new minimum and well are energetically well below the dissociation limit for @xmath152 they are separated from the other inner minimum only by a tiny barrier .
these facts will become better understandable in the context of our discussion of the evolution of the whole spectrum with increasing field strength ( see below ) .
the properties of the pec of the @xmath145 state at large internuclear distances are somewhat analogous to that of the @xmath123 state .
the outer minimum has its origin in the interaction of the neutral @xmath119 and ionic @xmath153 configurations . starting with @xmath154 and increasing the field strength the depth of the outer well increases .
the first change of the dissociation channel from @xmath155 to @xmath156 occurs in the regime @xmath157 . in the regime
@xmath158 the position of the outer minimum increases with increasing field strength ( for @xmath133 it is already @xmath159 ) and the depth of the outer well decreases . due to the further increasing binding energy of the hydrogen negative ion @xmath160 state with increasing field strength
we encounter a second change of the dissociation channel at @xmath161 to @xmath162 which causes the disappearance of the outer minimum and well .
table 3 provides the corresponding data for the @xmath145 state .
the pecs of the @xmath163 and @xmath164 states are shown in figures 2e and 2f , respectively .
for both states the positions of the maxima and minima as well as the corresponding total energies show an irregular behaviour as a function of the field strength for @xmath165 .
we therefore focus on the main features of these states . for certain regimes of the field strength
we observe double well structures for the pecs .
analogously to the @xmath166 states there exist additional outer minima and wells due to the interaction with the ionic configuration for certain field strength regimes . for @xmath167 the position of the first inner minimum decreases rapidly with increasing field strength whereas the corresponding dissociation energy increases .
also we observe the existence of minima whose energies lie above the dissociation energy , i.e. the corresponding wells contain if at all metastable states .
we remark that some of the above - discussed features , in particular those associated with small energy scales , might not be visible in the corresponding figures 2 but only in a zoom of the relevant regimes of internuclear distances of the considered pecs .
we again emphasize that due to the large amount of data we do not present full pecs or data on the higher excited states @xmath168 which can be obtained from the authors upon request . in the present subsection we focus on the evolution of the complete spectrum of the excited @xmath169 states with increasing field strength .
this will give us the complementary information to the evolution of individual states presented above .
figure 3a - f shows the corresponding pecs for the field strengths @xmath170 , respectively .
the pecs of the five energetical lowest excited states @xmath171 are hereby illustrated with full lines indicating their higher accuracy whereas the pecs of the electronic states @xmath172 are less accurate and illustrated with dotted lines .
before we discuss the evolution with increasing field strength some general remarks are in order .
the energy gap between the ground state @xmath173 and the first excited state @xmath105 is of the order of @xmath174 in field - free space and increases montonically with increasing field strength . at the same time
the total energies of all states @xmath175 are shifted in lowest order proportional to @xmath176 with increasing field strength which is due to the raise of the kinetic energy in the presence of a magnetic field . in
field - free space many of the dissocation channels of the pecs of excited @xmath7 states are degenerate due to the degeneracies of the field - free hydrogen atom ( see figure 1 and 4 ) .
the major difference of the pecs in field - free space compared to those for weak fields is the removal of these degeneracies ( see , for example , figure 3a for @xmath177 ) . with increasing field strength figures 3a - d ( @xmath178 )
demonstrate the systematic lowering of the diabatic energy curve belonging to the ionic configuration @xmath97 .
this diabatic curve passes through the spectrum with increasing field strength thereby causing an intriguing evolution of avoided crossings and corresponding potential wells for the individual states . at @xmath179
the fourth excited @xmath163 state acquires the ionic dissociation channel .
the @xmath180 state thereby looses its outer potential well which was very well - pronounced in the absence of the external field .
in the same course the @xmath123 state shows a number of avoided crossings with the @xmath105 state : it develops an additional outer minimum and well which is rather deep at @xmath181 accompanied by the flattening of the first inner well and the deepening of the second inner well .
furthermore we observe for @xmath181 the appearance of a large number of avoided crossing among the higher excited states @xmath168 at @xmath182 . at @xmath183
the third excited @xmath145 state acquires the ionic dissociation channel .
subsequently , i.e. with further increasing field strength , the second excited @xmath123 state ( see figure 3d ) and finally the first excited @xmath105 state acquire ionic character for sufficiently large internuclear distances . in the high field situation
( see figure 3e for @xmath184 ) only the energetically lowest excited state possess a well - pronounced double well structure and the overall picture is dominated by the fact that the pecs of the considered states possess a very similar shape and are energetically very close to each other in particular around the inner minimum at small internuclear distances .
figure 3f shows for @xmath184 a zoom of the series of avoided crossings occuring for the higher excited states @xmath185 in the regime @xmath186 .
the four energetically lowest states of @xmath187 symmetry at @xmath154 have been investigated in detail and with high accuracy in the literature @xcite .
our results @xcite show a relative accuracy of @xmath65 for the energies of the @xmath188 state and of @xmath67 for the first two excited states i.e. the @xmath189 states .
the energies of the @xmath190 state are significantly lower than the data presented in @xcite .
the pecs for the @xmath191 presented in ref.@xcite for the first time are estimated to possess an accuracy of @xmath67 for the @xmath192 states and @xmath65 for the @xmath193 states .
figure 4 shows the pecs of the ground as well as eight excited states of @xmath187 symmetry in the range @xmath194 on a logarithmic scale .
the pec of the ground state @xmath188 of ungerade symmetry possesses a minimum at @xmath195 and a corresponding deep well
. a closer look at the wave function reveals its ionic character for @xmath196 . with further increasing internuclear
distance the ionic character of the wave function decreases and the corresponding dissociation channel is @xmath197 .
the pec of the first excited @xmath198 state is similar to that of the ground state @xmath188 : its equilibrium internuclear distance is @xmath199 the dissociation channel is identical to that of the @xmath188 state .
the depth of its single well is however only one third of the depth of the well of the @xmath188 state . for the higher excited states we observe a similar behaviour as in the case of the excited electronic states of @xmath200 symmetry .
the pecs of the @xmath201 states possess a deep well around a minimum located approximately at @xmath202 .
furthermore the @xmath203 states exhibit additional deep outer potential wells at large internuclear distances which arise due to the avoided crossings of the heitler - london configurations with the corresponding ionic configuration .
the outer minimum of the @xmath204 state is located at @xmath205 and the corresponding well possesses a remarkable depth of @xmath206 : it is expected to contain a large number of long - lived vibrational states .
tables 4 to 6 contain ( among the data in the presence of the magnetic field ) the total and dissociation energies at the equilibrium internuclear distances , the equilibrium internuclear distances and the total energies in the dissociation limit for the first to third excited @xmath207 states in the absence of the magnetic field .
first of all we remark that the dissociation channels of the @xmath208 states coincide with those of the @xmath209 states for @xmath210 in the complete regime @xmath211 . the qualitative behaviour of the pecs of the @xmath209 states at large internuclear distances is therefore similar to that of the @xmath208 states discussed in the previous section .
in particular many of the explanations and remarks provided there hold also for the present case of the @xmath209 states . before discussing the behaviour of the pecs of the individual excited @xmath209 states with increasing field strength some remarks concerning the lowest , i.e. ground state of @xmath11 symmetry
are in order ( for its pec with increasing field strength see figure 6 ) .
its dissociation energy increases monotonically with increasing field strength whereas its equilibrium internuclear distance increases slightly for weak fields and decreases significantly for increasingly stronger fields . as indicated above the asymptotic @xmath117 behaviour of the pecs of the @xmath212 and @xmath105 states
is very similar . for @xmath184
the pec of the @xmath212 state possesses a peculiar shape which is largely determined by the ionic dissociation channel @xmath213 ( see figure 6f ) . for more details on this state we refer the reader to ref.@xcite .
the first excited @xmath214 state possesses in field - free space an equilibrium internuclear distance @xmath215 .
figure 5a shows the corresponding pec with increasing field strength for @xmath216 whereas figure 5b illustrates particularly the behaviour at large internuclear distances . in the regime
@xmath217 the dissociation energy decreases slightly and the bond length increases . with further increasing field strength the dissociation energy increases drastically and
the bond length decreases . for @xmath218
there exists a maximum and a corresponding additional outer minimum at large internuclear distances ( see figure 5b ) whose origin is again the emergence of the ionic configuration for the wave function of the @xmath214 state .
figure 5b also demonstrates the similarity of the asymptotic @xmath219 behaviour of the pecs of the @xmath214 state in the regime @xmath220 .
the corresponding data for the pecs of the first excited @xmath214 state are given in table 4 .
turning to the second excited @xmath221 state we observe that the depth of the potential well located for @xmath124 at @xmath222 decreases for weak fields whereas it increases significantly for strong fields @xmath223 ( see figure 5c ) .
the existence of an additional outer minimum for this state can be seen in figure 5d . in many respects a similar behaviour to that of the @xmath214 state
is observed although , of course , the regimes of field strength for which the individual phenomena take place are different . table 5 contains the corresponding data of the pecs of the @xmath221 state .
finally figures 5e and 5f show the pecs of the @xmath224 states with increasing field strength , respectively .
they exhibit a number of maxima and minima most of which can however hardly be seen in figures 5e , f or occur at large internuclear distances .
the origin of their existence are again the different ( ionic and neutral ) dissociation channels .
these maxima and minima are present only for certain individually different regimes of the field strength . some of them are located above and some of them below the dissociative threshold .
as can be seen the bond length ( belonging to the inner minimum ) decreases monotonically and the dissociation energy increases significantly above some critical value @xmath225 .
the inner minimum and associated well possesses a remarkably large dissociation energy for strong fields .
table 6 provides data on the pecs of the @xmath226 state . to finalize our discussion on the @xmath11 subspace we show in figure 6 the evolution of the spectrum with increasing field strength .
figures 6a - f show the pecs for the @xmath227 states for the field strengths @xmath228 , respectively .
analogously to the case of the @xmath7 subspace we observe for weak fields the removal of the degeneracies due to the field - free hydrogen atom in the dissociation limit .
with increasing field strength we see the lowering of the diabatic energy line belonging to the ionic configuration which causes the appearance and disappearance of outer maxima , minima and corresponding outer potential wells until finally ( @xmath184 ) the @xmath212 state possess the ionic dissociation channel @xmath229 which is the origin of the peculiar shape of its pec .
a number of further observations made for the manifold of the @xmath230 states above can also be seen for the @xmath227 states in figure 6 like , for example , the similar shape of the potential wells of the excited states in the high field limit .
the hydrogen molecule is the most fundamental molecular system and of immediate importance in a variety of different physical circumstances . in spite of the fact that it has been investigated over the past decades in great detail and that our knowledge on this system has grown enormously there are plenty of questions and problems to be addressed even for the molecule in field - free space . as an example we mention certain highly excited rydberg states ( @xmath231 ) which , due to the ionic character of the binding for certain regimes of the internuclear distance , possesses a deep outer well at large distances which contains a considerable number of vibrational states . on the other hand
the detailed knowledge of hydrogen ( even of highly excited states ) is of utmost importance for our understanding and interpretation of the astrophysically observed interstellar radiation .
much less is known about the behaviour of the hydrogen molecule in strong magnetic fields .
with increasing field strength the ground state of the molecule undergoes two transitions which are due to a change of the spin and orbital character , respectively .
very recently the global ground state configurations have been identified for the parallel configuration ( there are good reasons which lead to the conjecture that the derived results hold for arbitrary angle of the internuclear and magnetic field axis ) both on the hartree - fock level @xcite and via a fully correlated approach @xcite . for low fields
the ground state is of spin singlet @xmath7 symmetry , for intermediate fields the spin triplet @xmath15 state represent the ground state whereas in the high field regime the @xmath20 state is the energetically lowest state .
the present work goes for the first time beyond the ground state properties and investigates excited states of the hydrogen molecule in the broad regime @xmath101 .
we hereby focus on singlet states of both gerade as well as ungerade symmetry : up to seven excited states have been studied for the parallel configuration with a high accuracy of the obtained pecs . a variety of different phenomena
have been observed out of which we mention here only the most important ones .
double well structures observed in particular for the field - free @xmath232 states are severly modified in the presence of the field thereby showing a coming and going of new maxima and minima as well as corresponding wells .
the overall tendency in the strong field limit is the development of deep inner wells containing a large number of vibrational states . in the course of the increasing field strength
a fundamental phenomenon occurs which has a strong impact on the overall shape of the pecs . due to the fact that the hydrogen negative ion becomes increasingly bound with increasing field strength we encounter changes in the dissociation channels of individual states from neutral @xmath233 to ionic @xmath234 character . for a certain regime of field strength
@xmath235 a certain excited state possesses therefore the ionic dissociation channel thereby modifying the asymptotic behaviour of its pec to an attractive coulombic tail . for weaker fields @xmath236 higher excited states possess this ionic dissociation channel whereas for stronger fields @xmath237 it belongs to increasingly lower excitations .
these facts influence the overall appearance of the spectrum thereby creating features like outer potential wells and/or largely changing avoided crossings .
the data on the pecs of the excited singlet states obtained here should serve as part of the material to be accumulated for the investigation of quasimolecular absorption features in magnetic white dwarfs .
the investigation of excited triplet states of @xmath1 symmetry or of @xmath8 states , which are of equal importance , are left to future investigations .
fruitful discussions with w.becken are gratefully acknowledged .
the deutsche forschungsgemeinschaft is gratefully acknowledged for financial support .
* figure 2 : * the evolution of the potential energy curves for some excited @xmath7 electronic states of the hydrogen molecule in the presence of a magnetic field @xmath101 . in detail
are shown the evolution of the pecs for the ( a ) @xmath105 , ( b ) @xmath123 , ( c ) zoom of @xmath123 , ( d ) @xmath145 , ( e ) @xmath163 and ( f ) @xmath164 states , respectively .
shown is the quantity @xmath102 where @xmath239 is the total energy .
* figure 3 : * the spectrum of potential energy curves for the excited @xmath100 electronic states of the hydrogen molecule in the presence of a magnetic field @xmath101 with increasing field strength . in detail
are shown the pecs for ( a ) @xmath240 ( b ) @xmath241 ( c ) @xmath242 ( d ) @xmath243 ( e ) @xmath244 and ( f ) zoom of @xmath245 , respectively .
* figure 5 : * the evolution of the potential energy curves for some excited @xmath11 electronic states of the hydrogen molecule in the presence of a magnetic field @xmath101 . in detail
are shown the evolution of the pecs for the ( a ) @xmath105 , ( b ) zoom of @xmath105 , ( c ) @xmath123 , ( d ) zoom of @xmath123 ,
( e ) @xmath145 and ( f ) @xmath163 states , respectively .
shown is the quantity @xmath102 where @xmath239 is the total energy .
* figure 6 : * the spectrum of potential energy curves for the excited @xmath227 electronic states of the hydrogen molecule in the presence of a magnetic field @xmath101 with increasing field strength . in detail are shown the pecs for ( a ) @xmath247 ( b ) @xmath241 ( c ) @xmath242 ( d ) @xmath248 ( e ) @xmath249 and ( f ) @xmath245 , respectively .
dddddddd & & & & & & & + 0.0 & 1.91 & 0.093122 & -0.718121 & 4.39 & 0.089241 & -0.714240 & -0.624999 + 0.001 & 1.91 & 0.093120 & -0.718117 & 4.39 & 0.089242 & -0.714239 & -0.624997 + 0.005 & 1.91 & 0.093061 & -0.718017 & 4.39 & 0.089252 & -0.714208 & -0.624956 + 0.01 & 1.91 & 0.092878 & -0.717703 & 4.39 & 0.089289 & -0.714114 & -0.624825 + 0.05 & 1.90 & 0.087831 & -0.708672 & 4.39 & 0.090459 & -0.711300 & -0.620841 + 0.1 & 1.88 & 0.077017 & -0.686953 & 4.38 & 0.093325 & -0.703261 & -0.609936 + 0.2 & 1.88 & 0.057630 & -0.633195 & 4.33 & 0.100709 & -0.676274 & -0.575565 + 0.5 & 1.89 & 0.040502 & -0.462472 & 4.08 & 0.122427 & -0.544397 & -0.421970 + 1.0 & 1.78 & 0.044819 & -0.135993 & 3.76 & 0.150618 & -0.241792 & -0.091174 + 2.0 & 1.54 & 0.067740 & 0.612336 & 3.37 & 0.191218 & 0.488858 & 0.680076 + 5.0 & 1.18 & 0.140991 & 3.130994 & 2.88 & 0.269224 & 3.002761 & 3.271985 + 10.0 & 0.95 & 0.245637 & 7.623918 & 2.55 & 0.351756 & 7.517799 & 7.869555 + 20.0 & 0.76 & 0.359056 & 16.960943 & 2.27 & 0.408351 & 16.911648 & 17.319999 + 50.0 & 0.56 & 0.575262 & 45.784366 & 1.96 & 0.471572 & 45.888056 &
46.359628 + 100.0 & 0.44 & 0.821768 & 94.614199 & 1.76 & 0.521822 & 94.914145 & 95.435967 + ddddddddddd & & & & & & & & & & + 0.0 & 2.03 & 0.035466 & -0.660465 & 3.27 & 0.038014 & -0.663013 & & & & -0.624999 + 0.001 & 2.03 & 0.035462 & -0.660458 & 3.27 & 0.038021 & -0.663017 & & & & -0.624996 + 0.005 & 2.03 & 0.035399 & -0.660305 & 3.27 & 0.038045 & -0.662951 & & & & -0.624906 + 0.01 & 2.03 & 0.035225 & -0.659851 & 3.27 & 0.038101 & -0.662727 & 18.97 & 0.000007 & -0.624633 & -0.624626 + 0.05 & 2.04 & 0.033199 & -0.649592 & 3.20 & 0.040478 & -0.656871 & 10.56 & 0.001510 & -0.617903 & -0.616393 + 0.1 & 2.03 & 0.030981 & -0.626596 & 3.06 & 0.048089 & -0.643704 & 10.02 & 0.006992 & -0.602607 & -0.595615 + 0.2 & 1.94 & 0.023894 & -0.563261 & 2.86 & 0.064882 & -0.604249 & 10.43 & 0.026352 & -0.565719 & -0.539367 + 0.5 & 1.90 & 0.030305 & -0.378302 & 2.73 & 0.079773 & -0.427788 & 10.69 & 0.066892 & -0.414907 & -0.348015 + 1.0 & 1.78 & 0.039842 & -0.041327 & 2.50 & 0.091331 & -0.092816 & 10.96 & 0.085978 & -0.087463 & -0.001485 + 2.0 & 1.54 & 0.042657 & 0.717581 & 2.22 & 0.089173 & 0.671065 & 12.14 & 0.079131 & 0.681107 & 0.760238 + 5.0 & 1.19 & 0.075290 & 3.248431 & 1.86 & 0.091565 & 3.232156 & 19.38 & 0.051644 & 3.272077 & 3.323721 + 10.0 & 0.96 & 0.131036 & 7.749942 & 1.63 & 0.093390 & 7.787588 & 85.58 & 0.011423 & 7.869555 & 7.880978 + 20.0 & 0.76 & 0.275591 & 17.095639 & 1.44 & 0.142076 & 17.229154 & & & & 17.371230 + 50.0 & 0.56 & 0.605249 & 45.931209 & 1.22 & 0.250618 & 46.285840 & & & & 46.536458 + 100.0 & 0.45 & 0.976059 & 94.770524 & 1.09 & 0.357751 & 95.388832 & & & & 95.746583 + ddddddddddd & & & & & & & & & & + 0.0 & 1.97 & 0.099408 & -0.654963 & & & & 11.21 & 0.049597 & -0.605152 & -0.555555 + 0.001 & 1.97 & 0.099397 & -0.654944 & & & & 11.21 & 0.049604 & -0.605151 & -0.555547 + 0.005 & 1.97 & 0.099051 & -0.654469 & & & & 11.21 & 0.049686 & -0.605104 & -0.555418 + 0.01 & 1.97 & 0.098011 & -0.653029 & & & & 11.21 & 0.049930 & -0.604948 & -0.555018 + 0.05 & 1.99 & 0.081038 & -0.625896 & 2.93 & 0.068063 & -0.612921 & 11.39 & 0.055502 & -0.600360 & -0.544858 + 0.1 & 2.05 & 0.075239 & -0.597699 & & & & 12.04 & 0.06419 & -0.586650 & -0.522460 + 0.2 & & & & 2.46 & 0.067543 & -0.546109 & 15.62 & 0.059554 & -0.538120 & -0.478566 + 0.5 & 1.91 & 0.022150 & -0.350029 & 2.35 & 0.035929 & -0.363808 & 50.06 & 0.020138 & -0.348017 & -0.327879 + 1.0 & 1.77 & 0.019181 & -0.010815 & 2.17 & 0.030750 & -0.022384 & & & & 0.008366 + 2.0 & 1.53 & 0.053756 & 0.750088 & 1.95 & 0.058173 & 0.745671 & & & & 0.803844 + 5.0 & 1.19 & 0.142822 & 3.283035 & 1.67 & 0.116673 & 3.309184 & & & & 3.425857 + 10.0 & 0.96 & 0.257211 & 7.786043 & 1.50 & 0.176832 & 7.866422 & & & & 8.043254 + 20.0 & 0.76 & 0.427463 & 17.133303 & 1.34 & 0.249318 & 17.311448 & & & & 17.560766 + 50.0 & 0.56 & 0.768452 & 45.971004 & 1.16 & 0.362134 & 46.377322 & & & & 46.739456 + 100.0 & 0.45 & 1.142090 & 94.811929 & 1.05 & 0.464011 & 95.490008 & & & & 95.954019 + dddddddddd & & & & & & & & & + 0.0 & 2.09 & 0.040772 & -0.665771 & & & & & & -0.624999 + 0.001 & 2.08 & 0.040770 & -0.665766 & & & & & & -0.624996 + 0.005 & 2.09 & 0.040697 & -0.665603 & & & & & & -0.624906 + 0.01 & 2.09 & 0.040477 & -0.665103 & & & & & & -0.624626 + 0.05 & 2.12 & 0.037068 & -0.653461 & & & & & & -0.616393 + 0.1 & 2.13 & 0.036237 & -0.631852 & 9.37 & 0.008169 & -0.603784 & 5.16 & -0.602345 & -0.595615 + 0.2 & 2.12 & 0.043756 & -0.583123 & 10.09 & 0.027637 & -0.567004 & 4.60 & -0.555876 & -0.539367 + 0.5 & 2.03 & 0.065201 & -0.413216 & 10.34 & 0.068386 & -0.416401 & 4.01 & -0.382377 & -0.348015 + 1.0 & 1.85 & 0.079251 & -0.080736 & 10.55 & 0.087513 & -0.088998 & 3.66 & -0.038398 & -0.001485 + 2.0 & 1.59 & 0.085790 & 0.674448 & 11.46 & 0.080259 & 0.679979 & 3.34 & 0.744375 & 0.760238 + 5.0 & 1.22 & 0.121736 & 3.201985 & 10.49 & 0.052138 & 3.271583 & 2.95 & 3.349758 & 3.323721 + 10.0 & 0.98 & 0.178133 & 7.702800 & 9.70 & 0.011729 & 7.869204 & 2.69 & 7.955928 & 7.880933 + 20.0 & 0.78 & 0.321481 & 17.049749 & 9.27 & 0.000271 & 17.370959 & 2.46 & 17.464313 & 17.371230 + 50.0 & 0.57 & 0.646721 & 45.889737 & 9.04 & 0.000168 & 46.536290 & 2.21 & 46.635075 & 46.536458 + 100.0 & 0.45 & 1.012902 & 94.733681 & 9.04 & & 95.746735 & 2.05 & 95.846945 & 95.746583 + dddddddddd & & & & & & & & & + 0.0 & 2.03 & 0.081441 & -0.636996 & 11.12 & 0.049964 & -0.605519 & 5.65 & -0.567698 & -0.555555 + 0.001 & 2.03 & 0.081426 & -0.636973 & 11.12 & 0.049971 & -0.605518 & 5.64 & -0.567692 & -0.555547 + 0.005 & 2.03 & 0.081009 & -0.636427 & 11.12 & 0.050053 & -0.605471 & 5.64 & -0.567528 & -0.555418 + 0.01 & 2.03 & 0.080107 & -0.635125 & 11.18 & 0.050291 & -0.605309 & 5.67 & -0.567116 & -0.555018 + 0.05 & 2.04 & 0.074463 & -0.619321 & 11.32 & 0.055802 & -0.600660 & 5.45 & -0.555308 & -0.544858 + 0.1 & 2.05 & 0.071675 & -0.594135 & 12.01 & 0.064313 & -0.586773 & 5.40 & -0.530941 & -0.522460 + 0.2 & 2.03 & 0.062081 & -0.540641 & 15.62 & 0.059632 & -0.538198 & 5.42 & -0.477396 & -0.478566 + 0.5 & 1.94 & 0.035148 & -0.363027 & 50.06 & 0.020138 & -0.348017 & 5.20 & -0.297160 & -0.327879 + 1.0 & 1.78 & 0.033120 & -0.024754 & & & & 4.92 & 0.058491 & 0.008366 + 2.0 & 1.54 & 0.068597 & 0.735247 & & & & 4.56 & 0.858143 & 0.803844 + 5.0 & 1.20 & 0.158473 & 3.267384 & & & & 4.06 & 3.490593 & 3.425857 + 10.0 & 0.96 & 0.272912 & 7.770342 & & & & 3.70 & 8.118636 & 8.043254 + 20.0 & 0.77 & 0.442635 & 17.118131 & & & & 3.38 & 17.648590 & 17.560766 + 50.0 & 0.56 & 0.782055 & 45.957401 & & & & 3.02 & 46.845692 & 46.739456 + 100.0 & 0.45 & 2.154264 & 94.799755 & & & & 2.81 & 96.074966 & 96.954019 + ddddddddddd & & & & & & & & & & + 0.0 & 2.00 & 0.078543 & -0.634098 & 5.67 & 0.010196 & -0.565751 & & & & -0.555555 + 0.001 & 2.00 & 0.078531 & -0.634077 & 5.67 & 0.010194 & -0.565740 & & & & -0.555546 + 0.005 & 2.00 & 0.078211 & -0.633537 & 5.67 & 0.010161 & -0.565487 & 33.89 & 0.000058 & -0.555384 & -0.555326 + 0.01 & 2.00 & 0.077103 & -0.631765 & 5.67 & 0.010059 & -0.564721 & 34.18 & 0.000338 & -0.555000 & -0.554662 + 0.05 & 2.01 & 0.063148 & -0.603904 & 5.56 & 0.009249 & -0.550005 & 42.09 & 0.004106 & -0.544862 & -0.540756 + 0.1 & 2.02 & 0.060038 & -0.577452 & 5.50 & 0.009923 & -0.527337 & 36.87 & 0.005048 & -0.522462 & -0.517414 + 0.2 & 2.00 & 0.057601 & -0.522906 & 5.79 & 0.002238 & -0.467543 & 33.92 & 0.000006 & -0.465309 & -0.465305 + 0.5 & 1.92 & 0.062864 & -0.343463 & & & & 34.57 & 0.000001 & -0.280600 & -0.280599 + 1.0 & 1.76 & 0.082384 & -0.003776 & & & & & & & 0.078608 + 2.0 & 1.53 & 0.123535 & 0.757397 & & & & & & & 0.880932 + 5.0 & 1.19 & 0.224179 & 3.290662 & & & & & & & 3.514841 + 10.0 & 0.96 & 0.348221 & 7.794138 & & & & & & & 8.142359 + 20.0 & 0.76 & 0.528386 & 17.142170 & 3.52 & & 17.707128 & & & & 17.670556 + 50.0 & 0.56 & 0.882660 & 45.981219 & 3.09 & & 46.897984 & & & & 46.863879 + 100.0 & 0.45 & 1.266401 & 94.823274 & 2.82 & & 96.127449 & & & & 96.089675 + | excited states of the hydrogen molecule subject to a homogeneous magnetic field are investigated for the parallel configuration in the complete regime of field strengths @xmath0 . up
to seven excitations are studied for gerade as well as ungerade spin singlet states of @xmath1 symmetry with a high accuracy .
the evolution of the potential energy curves for the individual states with increasing field strength as well as the overall behaviour of the spectrum are discussed in detail .
a variety of phenomena like for example the sequence of changes for the dissociation channels of excited states and the resulting formation of outer wells are encountered .
possible applications of the obtained data to the analysis of magnetic white dwarfs are outlined . |
a wide range of detectors is difficult to calibrate using gamma or x - rays , because of the absence of full absorption peaks in the spectrum .
in particular , it includes the detectors with spatial dimensions that are relatively small compared to the mean path of calibration source gamma rays , and also the ones which consist of a material with a low atomic number . as a result , the probability of the photoelectric effect is small compared to the probability of the compton scattering , and a small detector dimensions make it impossible to obtain a full absorption peak which could appear due to the multiple scattering .
the calibration by using the edges of the compton scattering has a significantly lower accuracy .
common examples include widespread plastic and liquid hydrocarbon scintillators @xmath0 or semiconductor detectors of a large volume based on silicon .
for example , the compton scattering cross - section of gamma rays at 0.1 ( 1.0 ) mev on carbon is @xmath1 = 2.9 ( 1.3 ) barn / atom , and the photoelectric effect cross - section is lower by two orders of magnitude @xmath2 = 0.021 @xmath3 barns / atom .
the corresponding gamma ray compton scattering mean paths are @xmath4 ( 7 cm at a density of @xmath5 ) and @xmath6 ( 16 cm ) . as a result
, the scintillation volume is mostly populated by the compton - produced electrons , and therefore it would be preferable to use those electrons for calibration . a classical approach @xcite utilizes a narrowly collimated beam of gamma - rays and mutually spaced scintillation detectors wired into the coincidence unit .
this design allows one to identify clearly the scattering angle and determine the amount of energy deposited in the scintillator .
the disadvantages of this method are that it requires an intense source of gamma - rays , and the data acquisition is quite time - consuming requiring a separate measurement for each particular value of scattered electron energy .
the key feature of our method is that the source of gamma - rays , and both scintillation and semiconductor detectors are located close to each other and this fact makes it possible to record simultaneously coincidence events in a large range of scattering angles , and thus to measure the scintillator light yield for the compton s electrons within a wide energy range .
the experimental setup with a liquid scintillation ( ls ) and semiconductor ( hpge ) detectors is shown on the inset of fig .
this layout allows one to tell the coincidence events of compton scattering in the scintillator from the photoelectric absorption events in ge - detector .
the sources with an activity of about @xmath7 bq ( included in the standard reference source kits ) are sufficient to obtain the required statistics , while the measurements are carried out at the same time for a range of electron energies that is limited by the maximum scattering angle .
we used two reference sources made of @xmath8co ( energies 1173 kev and 1333 kev ) and @xmath9bi ( gamma lines at 570 kev and 1063 kev ) .
for example , provided that the @xmath8co - source , ls- and ge - detectors are all aligned with the maximum scattering angle of @xmath10 , one should be able to study the recoil electrons energy within the 0 - 950 kev range ( fig .
[ fig1 ] ) . in order to register the scattered gamma - rays we used a cylindrical hpge - detector with the diameter of 60 mm and length of 60 mm , which had an energy resolution fwhm ( full width at half maximum ) of 1.0 kev at 122 kev ( gamma line of a @xmath11 source ) .
the detector was connected to the standard spectrometer channel , consisting of a charge - sensitive preamplifier with the resistive feedback and cooled field effect transistor ( fet ) , spectrometric amplifier with @xmath12 shaping time and 12-bit analog - to - digital converter ( adc ) .
the gain value was chosen for the adc energy scale to be equal to 0.4 kev / channel .
the ls - detector is designed as a hollow teflon ( ptfe ) cylinder with the internal diameter of 60 mm and length of 100 mm , which is filled with a liquid scintillator .
two photomultiplier tubes ( pmts ) r1307 are attached to the both ends of the cylinder .
the pmts are separated from the scintillator volume by 10-mm disc plates made of acrylic , which prevents the generation of events near the surface of photocathode .
linear alkylbenzene ( lab ) supplemented with ppo concentration 0.5 g / l @xcite was used as a scintillator .
each pmt had a dedicated spectrometer channel consisting of a preamplifier , an amplifier and a 10-bit adc .
the amplitude ratio between signals from pmt1 and pmt2 allows one to select a fiducial volume in the central region of the ls - detector .
the photo of hpge- and ls - detector is shown in fig .
[ fig2 ] .
the coincidence circuit was implemented using the @xmath13 standard . as a result ,
the computer memory accumulated the sequentially recorded amplitude values for three signals , corresponding to hpge detector coincidences with pmt1 and pmt2 of the ls - detector .
the stored data file was then processed off - line , once the sufficient statistics was obtained .
a three dimensional spectrum is formed during the data collection : the number of events n , the amplitude of the ge - detector and the sum of the pmt1 and pmt2 amplitudes . for monitoring purposes a three - dimensional plot is formed on - line during the data acquisition , picturing the number of events n versus the amplitude of ge - detector and sum of the amplitudes from pmt1 and pmt2 .
a 3d plot for a series of measurements with the @xmath9bi source is shown in fig .
the @xmath14 and @xmath15 axes correspond to the amplitudes of the ls - detector and hpge - detector accordingly .
the figure clearly shows two linear ridges , crossing the ( @xmath16 ) plane , which start at the energies of @xmath17 and @xmath18 kev and are governed by relation @xmath19 .
those ridges correspond to the compton scattering in the scintillator and the subsequent total absorption of the scattered gamma quantum in the ge - detector .
these features therefore allow us to select the compton electrons of the given energy .
for each energy value @xmath20 which was deposited within the ge - detector and stored in the channel i , there is a peak in the spectrum of ls - detector , that corresponds to the energy of the compton electron @xmath21 .
[ fig4 ] shows the spectrum of ls - detector , measured with a @xmath8co source , corresponding to the energy deposit of 1022 kev inside the ge - detector .
there are two peaks formed by the scattering of @xmath8co lines at 1173 kev and 1332 kev .
accordingly , the peaks in the ls - spectrum are located at 151 kev and 311 kev .
the position of the ls - detector peak @xmath22 is determined by fitting the spectrum using the sum of the gaussian and the baseline function selected to describe the continuous background ( fig .
[ fig4 ] ) .
in fact , it is sufficient to process a few peaks in order to determine the energy scale of ls - detector and the energy dependence of the resolution .
modern computational capabilities make it possible to process all of the obtained spectra .
the amplitude detector ls - electron energy dependence in a certain range ( 20 - 800 ) kev for 120 points , is shown in fig .
it should be noted that when using a point source at the fixed position , this method requires the uniformity of light collection as different scattering angles correspond to different regions of the scintillator .
however , this circumstance is not essential due to the poor energy resolution of scintillation detectors .
since the amplitude of the scintillation detector@xmath14 depends on the light output ( y ) , which is a fundamental property of the scintillator , this method allows us to study the dependence of the light yield on the electron energy y(e ) .
in particular , the deviation of y(e ) from a linear dependence at small electron energies ( @xmath23 kev),is called ionization quenching effect . a precise measurement of y(e ) is important for a number of physical problems in which the obtained scintillation spectrum is fitted by a theoretical electron energy spectrum .
the deviation from the linear dependence between the number of photons dy per length unit dx and energy losses de / dx is usually described by the phenomenological birks relation @xcite : @xmath24 where kb is the birks coefficient ( a quenching factor ) and @xmath25 is the light yield for kb = 0 .
these two values are inherent characteristics of the particular scintillator .
the typical values are kb = 0.01 cm / mev and @xmath25 = @xmath26 photons / mev @xcite .
the y(e , kb ) dependence is obtained by integrating the expression ( 1 ) over the energy , taking into account the energy losses de / dx for a particular substance . in case of lab ( @xmath27 ) the obtained y(kb)/y(kb=0 ) dependence on the electron energy for different values of kb are shown in the inset of fig .
the ratios are normalized to unity at e = 1 mev .
it can be seen that at the energy of 100 kev the deviation of y(kb)/y ( kb = 0 ) from 1 is ( 4 - 9)% for kb = ( 0.006 , 0.01 , 0.016 ) .
the method described above allows us to measure the entire y(e ) dependence , which can be used instead of the relation ( 1 ) , which is determined only by two parameters kb and @xmath25 . in our case , to achieve the position accuracy of less than 1% for the 100 kev peak , the statistics has to be greater than @xmath28 events in the peak ( fig .
[ fig4 ] ) .
there are two basic problems in this approach ( which are also present in the classical approach ) which are associated with the multiple compton scattering ( mostly double scattering ) in the ls - detector and the non - uniformity of light collection within the liquid scintillator volume .
the mean path of gamma - ray at 1.33 mev ( @xmath8co ) is 20 cm for lab . for the described ls - detector
the probability of double scattering at small angles , which represent the main interest in measurements of y(e ) dependence , is 15% .
the impact of both effects can be accounted for by performing the monte - carlo simulations .
the method of calibration of the scintillation detector energy scale with a gamma - ray based on the compton scattering in the scintillation detector and subsequent total absorption of the scattered photon in ge - detector is proposed . the gamma - ray source , scintillation detector and ge - detector are located almost next to each other .
the calibration method is effective for the detectors made of materials with a low atomic number or when the dimensions of the detector are comparable with the length of a gamma - ray mean path .
the technique can be used to study the dependence of the scintillator light yield on the energy of electrons . | in this paper we present a method of scintillation detector energy calibration using the gamma - rays .
the technique is based on the compton scattering of gamma - rays in a scintillation detector and subsequent photoelectric absorption of the scattered photon in the ge - detector .
the novelty of this method is that the source of gamma rays , the germanium and scintillation detectors are immediately arranged adjacent to each other .
the method presents an effective solution for the detectors consisting of a low atomic number materials , when the ratio between compton effect and photoelectric absorption is large and the mean path of gamma - rays is comparable to the size of the detector .
the technique can be used for the precision measurements of the scintillator light yield dependence on the electron energy .
scintillator and semiconductor detectors , energy scale calibration |
recently methods of noncommutative geometry ( @xcite ) were used to introduce a covariant regularization procedure for quantum fields on the 2 sphere ( @xcite ) .
there is hope that with these methods there will be a nonperturbative understanding of quantum effects . for a treatment of the chiral anomaly in the schwinger model
see @xcite .
the idea is to approximate the algebra of functions on the space by matrix algebras and to encode the geometrical information of the space in these algebras . in this way it becomes possible to construct models of qft on the virtual spaces ,
the matrix algebras are thought to be functions on . since the theory has only finite degrees of freedom the problems one usually deals with in qft are absent . for examples of these matrix geometries
see @xcite .
we will show that the @xmath0 , as a 4 dimensional riemannian manifold on which the group @xmath4 acts isometrically , can by treated in this way .
we construct the laplace operator on the fuzzy @xmath0 and show that the spectrum is the same as in the classical case , except for a truncation at higher eigenvalues , which can be interpreted as a uv cutoff . since @xmath4 acts on the matrix algebras by automorphisms we did not lose the @xmath4 covariance by this procedure .
we discuss the quantization of several vector bundles , among them a family of @xmath5-bundles , and compare the structures with the classical case . as well as for the functions we find that in the quantized sections higher order representations of @xmath6 are missing .
a straightforward generalization of the classical @xmath5-dirac operators is introduced and the spectra are calculated .
it turns out , that zero modes and index are different from the commutative case .
a possible solution to this problem is presented .
we start with a group theoretic construction of the classical structures on the @xmath0 .
let @xmath7 be the lie group @xmath8 and @xmath9 its lie algebra .
we choose a cartan subalgebra and a corresponding system of roots @xmath10 , positive roots @xmath11 , fundamental roots @xmath12 , and use the standard cartan - weyl basis @xmath13 .
@xmath14 is the root vector and @xmath15 the root element corresponding to the root @xmath16 .
, width=264 ] @xmath12 consists of two elements @xmath17 and @xmath18 ( see figure [ roots ] ) .
hence we have @xmath19 the compact real form @xmath20 is @xmath21 and the corresponding connected subgroup @xmath22 is the group @xmath4 .
the two fundamental weights are @xmath23 we denote the verma module with highest weight @xmath24 by @xmath25 the fundamental roots chosen such that @xmath26 corresponds to the fundamental representation of the @xmath4 .
the dimensions of these representations are given by @xmath27 and the image of the quadratic casimir element @xmath28 is the multiplication operator by the number @xmath29 to each verma module there corresponds an ( up to equivalence unique ) unitary irreducible representation of @xmath6 and a holomorphic irreducible representation of @xmath7 , which we denote by the same symbol .
let @xmath30 now be the parabolic subalgebra @xmath31 and @xmath32 the analytic subgroup of @xmath7 with lie algebra @xmath33 .
the quotient space @xmath34 is a complex flag manifold ( see e.g. @xcite ) and has a compact realization @xmath35 where @xmath36 is the subgroup of @xmath6 with lie algebra @xmath37 . in our case
this is the group @xmath38 .
this is just the subgroup that leaves the lowest weight subspace of @xmath26 invariant , and our space @xmath39 is just the complex projective space @xmath0 , the space of comlex lines in @xmath40 .
@xmath39 is in a canonical way a khler manifold .
the action of @xmath6 on @xmath39 is holomorphic and leaves the khler structure invariant .
there exists a unique normalized @xmath6-left invariant measure on @xmath39 which we denote by @xmath41 . in this section
we give the decomposition of the unitary representation of @xmath6 on the space @xmath42 , given by the action @xmath43 for the decomposition we use the fact that @xmath39 is a riemannian symmetric space of compact type ( @xcite ) .
this means that there is a real form @xmath44 with iwasawa decomposition @xmath45 , where @xmath46 is the abelian and @xmath47 the nilpotent part .
we denote by @xmath48 the corresponding real lie algebras .
@xmath49 can be extended to a maximal abelian lie subalgebra @xmath50 of @xmath51 such that its complexification @xmath52 is a cartan subalgebra of @xmath9 .
the weights are elements of the dual of the vectorspace @xmath53 .
we choose again fundamental roots and denote by @xmath54 the positive roots that do not vanish on @xmath49 and interpret them as linear forms on @xmath49 .
the killing form induces a scalar product @xmath55 on the dual @xmath56 .
we have a representation with highest weight @xmath57 contains a @xmath36-invariant vector if and only if 1 .
@xmath58 2 .
@xmath59 ( see @xcite ) in our case a short calculation shows that one can choose another cartan subalgebra such that @xmath60 and with @xmath61 @xmath62 is a real form of @xmath9 .
one obtains that the representations of @xmath6 that contain a @xmath36 invariant vector are just the representations with highest weight @xmath63 the space of @xmath36-invariant vectors in such a representation is furthermore one dimensional and by frobenius theorem exactly those representations occur with multiplicity one in the quasi - left regular representation .
is just the representation of @xmath6 induced by the trivial representation of @xmath36 ] hence we get for our case the following result [ decomp ] let @xmath65 with the canonical @xmath4 action . as @xmath4-modules we have @xmath66 and
the subspace of @xmath42 which is , as an su(3)-module , isomorphic to @xmath67 is just the space of functions of the form @xmath68 where @xmath69 is an ( up to a scalar unique ) k - invariant vector . identifying the lie algebra @xmath9 with the complex left invariant vector fields on the group @xmath6 and those with differential operators on @xmath70
, we get a natural action of the universal enveloping algebra @xmath71 on @xmath70 .
the quadratic casimir element @xmath72 becomes is a biinvariant differential operator @xmath73 of second order and is equal to the laplace - beltrami operator of the ( up to a scalar unique ) biinvariant metric on @xmath6 .
thus it can be restricted to the @xmath36-right invariant functions on @xmath6 and we get a differential operator @xmath74 .
this operator coincides with the metric laplacian on @xmath39 .
we get the laplace operator @xmath75 on the @xmath0 is essentially self - adjoint on @xmath76 .
the spectrum of its closure is purely discrete and the eigenvalues @xmath57 and multiplicities @xmath77 are @xmath78 proof : we restrict the domain to the finite linear span of vectors in the irreducible subspaces of the group action , on each of which @xmath75 acts by multiplication with @xmath79 .
hence the domain contains a total set of analytic vectors and we found the operator in its spectral decomposition .
@xmath80 the homogeneous vector bundles over @xmath39 are in one to one correspondence with the finite dimensional representations of the group @xmath36 .
if @xmath81 is a homogeneous vector bundle over @xmath39 and @xmath82 the representation of @xmath36 at the fibre @xmath83 at the identity class @xmath84 , then @xmath81 is isomorphic as a g - vector bundle to the associated bundle @xmath85 .
the sections of this bundle can be identified with the equivariant @xmath83-valued functions on @xmath6 , that is the functions @xmath86 that satisfy the equivariance condition @xmath87 .
the homogeneous vector bundles over @xmath39 have a canonical holomorphic structure since @xmath82 can be extended to a holomorphic representation @xmath88 of @xmath32 and the vector bundle can be constructed as @xmath89 .
we define several useful bundles on @xmath39 .
let @xmath90 for @xmath91 be the irreducible ( one dimensional ) representation of @xmath36 with highest weight @xmath92 .
we denote the homogeneous line bundle @xmath93 by @xmath94 . for example
@xmath95 is the hyperplane bundle and @xmath96 the tautological line bundle over @xmath39 .
let @xmath97 be the representation of @xmath36 with highest weight @xmath98 and @xmath99 the dual representation , we define @xmath100 and @xmath101 to be the bundles @xmath102 and @xmath103 respectively .
since the irreducible representations of @xmath36 are unitary these vector bundles have natural hermitian structures , and we denote by @xmath104 the hilbert space of square integrable sections and by @xmath105 the finite dimensional hilbert space of holomorphic sections of the bundle @xmath81 .
the representation of @xmath6 on @xmath106 is just the representation induced by @xmath82 . for the bundles @xmath94 we have the following isomorphisms of @xmath6-modules 1 . for @xmath107
: @xmath108 2 . for @xmath109 : @xmath110 sketch of proof : we specialize to @xmath107 since the proof is analogous for the other case .
the sections of @xmath94 will be identified with the equivariant functions on @xmath6 . for each section
@xmath86 there is a family of @xmath36-right invariant functions @xmath111 on @xmath6 , such that @xmath112 where @xmath113 is a nonzero highest weight vector and @xmath114 an orthonormal basis in @xmath115 . to see this simply choose @xmath116 . hence using theorem [ decomp ]
the subspace @xmath117 is dense in @xmath118 .
the subspaces @xmath119 define a filtration of @xmath120 .
a short calculation using @xmath121 gives @xmath122 where @xmath123 is a nonzero lowest weight vector in @xmath124 .
since @xmath125 and the vector @xmath126 is cyclic , we can choose a family of nonzero vectors @xmath127 such that @xmath128 and therefore @xmath129 this establishes the desired isomorphism of modules .
@xmath80 the same method allows us to decompose the representation on the space of sections of another class of bundles for the bundles @xmath130 we have the following isomorphisms of @xmath6-modules 1 . for @xmath107 :
@xmath131 2 . for @xmath132 : @xmath133 since @xmath6 acts on @xmath39 by holomorphic automorphisms the bundles of antiholomorphic @xmath134-forms @xmath135 are homogeneous vector bundles .
the vector space of antiholomorphic tangent vectors at @xmath84 can be identified with the nilpotent radical @xmath136 of the parabolic subalgebra @xmath30 , which is just @xmath137 .
the isotropy representation is the adjoint action of @xmath36 on @xmath136 . taking the dual representations this leads to the following isomorphisms of @xmath6-bundles .
@xmath138 it is well known that the @xmath0 does not admit a spin - structure .
it does however admit a @xmath5-structure and we give here a family of @xmath5-bundles . since @xmath39 is a khler manifold the bundle of antiholomorphic forms @xmath139 with the natural @xmath140-grading is a canonical @xmath5-bundle and the dolbeault - dirac operator @xmath141 is a @xmath5-dirac operator ( see e.g. @xcite ) .
tensoring this bundle with some line bundle we obtain other @xmath5-bundles .
tensoring the canonical @xmath5-bundle @xmath142 with the line bundle @xmath143 and using the above described isomorphisms of @xmath6-bundles we get the @xmath5-bundles @xmath144 : @xmath145 the sections of @xmath144 are identified with the @xmath146-valued functions @xmath147 on @xmath6 which satisfy the equivariance condition @xmath148 for all @xmath149 . here
the @xmath150 are the matrix elements of the representation @xmath97 .
the basis in the representation space of @xmath97 can be chosen in such a way that the dolbeault - dirac operator @xmath151 takes the form @xmath152 where @xmath153 is the differential operator corresponding to the left invariant vector field @xmath15 on @xmath6 .
this is an example of an abstract supersymmetric dirac operator .
we have the @xmath5-dirac operator @xmath151 is essentially self - adjoint on the space of smooth sections of @xmath144 .
the spectrum of its closure is purely discrete .
the eigenvalues @xmath57 and multiplicities @xmath77 are : 1 .
@xmath154 + @xmath155 2 .
@xmath156 + @xmath157 the index is @xmath158 sketch of proof : the essential self - adjointness follows from nelsons trick . using the transformation properties on the group an easy calculation shows @xmath159 the abstract foldy - wouthuysen transformation and the decompositions of the @xmath6-modules @xmath160 , @xmath161 , and @xmath162 gives the desired result.@xmath80
since @xmath39 is a compact khler manifold and @xmath95 a quantum line bundle we can use the berezin - toeplitz quantization procedure ( see e.g. @xcite ) .
we fix an @xmath163 and let @xmath164 be the orthogonal projection onto the subspace @xmath165 . for each @xmath166
we get an operator @xmath167 , defined by @xmath168 where @xmath169 is the multiplication - operator associated with @xmath86 .
the corresponding surjective map @xmath170 is called toeplitz quantization map .
the algebra @xmath171 is the algebra of quantized functions , which for high values of @xmath47 approximates the algebra @xmath172 in a certain sense . by the bott - borel - weyl theorem
( see @xcite ) we have the isomorphism of @xmath6-modules @xmath173 .
identifying these two spaces we obtain for the quantized algebra of functions the matrix algebra @xmath174 on which @xmath6 acts by inner automorphisms @xmath175 where @xmath176 is the representation which corresponds to the @xmath6-module @xmath177 . if @xmath178 is the orthogonal projection onto the highest weight subspace of @xmath179 we define the coherent state function on @xmath39 by @xmath180 and we get for the toeplitz quantization map @xmath181 which is equivariant , real , positive , and norm decreasing
. moreover the following results hold 1 . the restriction @xmath182 ist bijective .
2 . @xmath183 in the uniform topology on @xmath172 3 .
for @xmath184 we have with the @xmath185-norms on the @xmath186 : @xmath187 4 .
if @xmath188 is the normalized trace on @xmath186 , we have : @xmath189 5 . for @xmath184
holds : @xmath190 6 .
@xmath191 here @xmath192 is the space of truncated functions , that is the subspace that contains only the @xmath6-representations @xmath67 for @xmath193 .
it follows that as @xmath6-modules @xmath186 and @xmath194 are isomorphic .
this is a kind of cut - off automated by the quantization .
one introduces now the linear inclusions @xmath195 and the sequence @xmath196 the vector space limit of this sequence is in a canonical way a normed space and its closure has the structure of a @xmath197-algebra , the product being the limit @xmath198 of the products in the single algebras .
this algebra can be shown to be isomorphic to the algebra of continuous functions on @xmath39 .
moreover the isomorphism can be chosen to be equivariant . in this sense
the sequence approximates the @xmath0 .
we therefore understand the algebras @xmath186 as the algebra of functions on the virtual
fuzzy @xmath0 of order @xmath47 .
this requires however to have as a correspondence principle the toeplitz quantization maps or equivalently the above sequence in mind .
since @xmath6 acts on @xmath186 by inner automorphisms we get by differentiation an action of the lie algebra @xmath199 by inner derivations .
if @xmath176 is the representation of @xmath9 which corresponds to the @xmath9-module @xmath177 this action reads as follows @xmath200 \quad x \in g^{{\mathbb{c } } } , a \in \mathcal{a}_n\ ] ] this gives a representation of the universal enveloping algebra @xmath71 and we define in analogy to the commutative case the laplace operator @xmath201 to be the image of the quadratic casimir element , or more explicitly : @xmath202 since the quantization map is equivariant we have @xmath203 using the decomposition of the @xmath9-module @xmath186 one gets easily the eigenvalues @xmath57 of the laplacian @xmath204 and their multiplicities @xmath77 are : @xmath205 the spectrum shows the expected uv cutoff . the above action of @xmath9 on @xmath186 can be thought of as a lie algbera homomorphism from @xmath9 to the lie algebra of derivations @xmath206 of the algebra .
the image , as a subspace of @xmath206 , gives rise to a derivation based differential culculus on @xmath207 ( see @xcite ) . with the natural scalar product on the one - forms the laplace operator has the natural form @xmath208 , as in the commutative case
vector bundles in the noncommutative case should be finitly generated projective modules over the algebra . in @xcite
the concept of toeplitz quantization was generalized to the quantization of bundles .
let @xmath209 be a holomorphic hermitian vector bundle with base @xmath39 and @xmath210 be the dual bundle .
let furthermore @xmath211 be the hilbert space of holomorphic sections of the bundle @xmath212 , that is @xmath213 . using the orthogonal projection @xmath214 onto @xmath211
, we can define the surjective map @xmath215 according to @xcite this gives the quantization of the bundle @xmath209 in the following sense @xmath216 clearly @xmath217 is a finitely generated projective left @xmath186-module .
let @xmath176 be an irreducible representation of @xmath36 and @xmath209 the associated homogeneous vector bundle over @xmath39 .
if @xmath176 has highest weight @xmath218 , the isotropy representation of @xmath219 has highest weight @xmath220 then by the bott - borel - weyl theorem @xmath221 as g - modules , if we set by definition @xmath222 if @xmath223 or @xmath224 .
the isotropy representation of @xmath94 was @xmath90 , which had highest weight @xmath226 , and one obtains @xmath227 the quantization of the bundles @xmath94 is therefore given by the @xmath186-modules @xmath228 @xmath229 an easy calculation shows that as @xmath6-modules we have @xmath230 and @xmath231 the isotropy representation of @xmath130 is @xmath233 with highest weight @xmath234 .
this yields @xmath235 this gives as a quantization of @xmath130 the modules @xmath236 again as @xmath6-modules @xmath237 and @xmath238
in the classical case we had a family of @xmath5-bundles @xmath239 over @xmath39 , where @xmath240 furthermore we had the twisted dolbeault complex @xmath241 which we used to define the dirac operator @xmath242 where @xmath243 was a g - equivariant map fulfilling @xmath244 . + let @xmath245 be the quantization maps defined earlier .
we define the @xmath5-bundle in the quantum case to be the module @xmath246 in the following we consider the case @xmath247 .
there is a unique linear map @xmath248 such that the following diagram commutes @xmath249 @xmath250 is equivariant and satisfies @xmath251 . proof : uniqueness is guaranteed by the surjectivity of the maps @xmath245 . for existence
it is enough to show that @xmath252 this is easily derived from the decomposition of the modules and the fact that @xmath253 and the quantization maps are equivariant .
explicitely the map is given by @xmath254 , where @xmath255 is an arbitrary element of the set @xmath256 .
the @xmath5-dirac operator @xmath257 is the self - adjoint operator , given by @xmath258 as in the classical case we have @xmath259 and the map @xmath260 is equivariant .
we clearly have @xmath261 which gives , using the decomposition of the g - modules @xmath262 and @xmath263 : the eigenvalues @xmath57 of @xmath151 and their multiplicities @xmath77 are : 1 . @xmath264 + @xmath155 + @xmath265 + @xmath266 2 .
@xmath267 + @xmath157 + @xmath268 + @xmath269 3 .
@xmath270 + @xmath271 @xmath272 . the index and the number of zero modes are different from the commutative case .
that is because the maps @xmath250 vanish also on higher representations .
apart from this the spectrum is truncated , as we expected .
the dirac operator defines for each @xmath273 and @xmath274 an even spectral triple @xmath275 with @xmath276 which gives a differential calculus over the algebra @xmath186 .
taking for the space of spinor fields instead of ( [ spin1 ] ) the space @xmath277 we can use the same procedure to define a dirac operator , which except for a spectral cutoff has the same spectrum as the classical dirac operator .
in particular the index and zero modes are classical . in this case
we lose the @xmath186-module property , the space is however a module over the algebra @xmath278 .
this will be investigated elsewhere .
the tools provided in this paper make it possible to investigate scalar qft on the fuzzy @xmath0 , which is free of any divergencies , since the algebra of functions is finite dimensional and allows to construct a well defined functional integral .
the attempt to construct spinors led to similar difficulties as for the known case of the fuzzy sphere . there the use of supersymmetry provided a solution ( @xcite ) .
it seems promising to use a similar extension for the @xmath0 .
noncommutative generalizations of the dolbeault operators might as well lead to structures proposed in @xcite .
the authors would like to thank w. bulla , h. miglbauer , p. presnajder , j. pngel , g. reiter for helpful discussion . | we give a noncommutative version of the complex projective space @xmath0 and show that scalar qft on this space is free of uv divergencies .
the tools necessary to investigate quantum fields on this fuzzy @xmath0 are developed and several possibilities to introduce spinors and dirac operators are discussed .
h. grosse@xmath1 , a. strohmaier@xmath2 + @xmath3institute for theoretical physics , university of vienna , boltzmanngasse 5 , a-1090 vienna , austria + e - mail : [email protected] + @xmath2institute for theoretical physics , tu- graz , petersgasse 16 , a-8010 graz , austria + e - mail : [email protected] * keywords : * regularization , noncommutative geometry , geometric quantization , @xmath0 |
the enormous amounts of energy emitted by the centers of active galaxies are believed to be from accretion of matter onto supermassive black holes ( smbhs ) lying there ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
feedback from active galactic nuclei ( agn ) strongly influences the formation and cosmological evolution of structures , whereby the overall properties of a galaxy can be regulated by its central smbh , which also impacts the environment from pc to mpc scales ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the energy output is often observed as agn outflows in a wide variety of forms ( see * ? ? ?
* ; * ? ? ?
* for reviews ) , e.g. : collimated relativistic jets and/or overpressured cocoons in radio @xcite , blue - shifted broad absorption lines in the ultraviolet ( uv ) and optical @xcite , warm absorbers @xcite and ultra - fast outflows in x - rays @xcite , molecular gas in far - ir @xcite .
a major baryonic component pervading galaxy clusters is the diffuse hot ( temperature @xmath7 k ) gas present in between galaxies , called the intracluster medium ( icm ) .
it emits copiously in the x - ray band , as it undergoes radiative cooling by losing energy through the bremsstrahlung process , and emission lines of @xmath8kev energy associated to atomic transitions from metals .
observations of x - ray radial profiles of the so - called classical _ cool - core clusters _ show a sharp peak in the x - ray surface brightness at the core , which imply a low entropy , high density , hence a shorter cooling time of the central icm . as the cluster core cools faster , the central gas pressure also decreases .
the overpressured icm from the surrounding cluster region should inflow toward the core , as a slow , subsonic _ cooling flow _
@xcite in the standard scenario ; however observations find a lack of emission from cool gas in the cluster center ( e.g. , * ? ? ?
the core cooling times in @xmath9 of the clusters is much shorter than the cluster lifetimes , few gyrs @xcite .
therefore the dense icm at cluster cores should cool and fragment to form self - gravitating objects like molecular clouds and stars .
uv observations reveal recent star formation in the central brightest cluster galaxies of many cool - core clusters @xcite .
however , enough cold condensed material is not observed , which is called the _ cooling flow problem_. the observed mass deposition rates of cooling and star - forming gas in the cores of galaxy clusters with strong cooling flows is much lower than that predicted from x - ray observed gas density profiles and simple radiative cooling models @xcite .
the above observations imply that the cooling flow icm at cluster cores is being heated .
the most favored mechanism is heating by energy feedback of outflows driven by agn lying in brightest cluster galaxies @xcite .
other possible heating mechanisms are : star - formation and supernovae - driven outflows , thermal conduction from the hot outer icm to the center , cosmic rays , and magnetic fields . while x - ray observations reveal a standard pure cooling rate suppression ( by less than @xmath10 ) ; at the same time residual cooling of at least a few percent is observed ( e.g. * ? ? ? * and references therein ) .
hence the heating mechanisms at operation are avoiding overheating of cluster cores .
the majority ( @xmath11 ) of cool - core clusters are observed to host radio sources @xcite .
radio - loud agn drive relativistic jets on two opposite sides , surrounded by overpressured lobes , which expand supersonically and end in bow shocks @xcite .
radio jets and lobes have complex interactions rising through the cluster atmosphere : heat the cooling flow , displace the x - ray emitting gas along the jet direction , and/or locally compress the icm . structures like cavities and bubbles as depressions ( or holes ) in x - ray surface brightness images are observed in many clusters , coincident with the radio jet / lobe emission @xcite .
numerical hydrodynamical simulations are performed to follow the propagation of radio jets and lobes injected from the central agn into the icm , and their subsequent evolution imparting energy and entropy to the gas .
several 2d and 3d simulations using eulerian adaptive - mesh refinement codes have shown that mechanical , kinetic or thermal energy coupled from the agn to surrounding icm is able to offset its radiative cooling .
most of these use isolated cluster models , an initial isothermal spherically - symmetric analytical profile for the icm @xcite .
few studies have included full jet dynamics and a gas feedback model @xcite , used cluster extracted from a cosmological volume @xcite , implemented momentum - driven jets in zoom - in cluster resimulation @xcite .
mesh - code idealized simulations demonstrate that strong agn heating can transform a cool - core cluster to a non cool - core state @xcite .
lagrangian sph code simulations are also done to study agn feedback effects on icm evolution , which usually does not consider the radio jet s propagation .
these sph studies include additional gas physics ; star - formation , supernova feedback , and agn feedback as thermal heating : hot buoyant bubbles injected into the icm during active agn phases @xcite ; transition from quasar - mode to radio - mode feedback when the smbh accretion rate goes below a given limit , and the feedback efficiency is increased by a factor of @xmath12 @xcite .
recently , @xcite compared between adaptive mesh refinement and sph methods , performing cosmological zoom simulations of a galaxy cluster with agn feedback .
agn outbursts are fed by gas accretion from the icm , so self - regulation may arise naturally . when an agn is active for a period of time , its outflow heating stops the cooling flow .
this also limits the gas accreting onto the smbh , turning the agn dead .
afterwards , the icm cools again , cooling flow restarts , providing the central smbh with fuel and might trigger another phase of agn outburst .
the result is a self - regulating cyclic process with cooling flow alternating with bursts of agn activity .
idealized simulations have studied how agn feedback can successfully self - regulate the cooling and heating cycles of the icm at cluster cores @xcite .
explicitly simulations including agn jets have revealed the trigger of a circulation of gas , that is crucial in stabilizing cool cores and creating self - regulation ( e.g. , * ? ? ?
* ; * ? ? ?
the physical mechanisms of gas accreting onto smbhs and resulting energy feedback is complex , with the surrounding gas likely accelerated by thermal pressure and/or radiation pressure .
the relevant physical scales are orders of magnitude below the scales resolved in current galaxy formation simulations .
hence agn accretion and feedback are incorporated in the simulations using _ sub - resolution _ numerical prescriptions ( e.g. , * ? ? ?
the physics of the gas , on scales unresolved in cosmological simulations , is modelled using spatially averaged properties describing the medium on scales that are resolved .
galaxy formation simulations have investigated coupling of agn feedback using both mechanisms : thermal ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and kinetic ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
kinetic agn feedback in sph simulations has been implemented in a few previous work .
@xcite studied radiative- and momentum - based mechanical agn feedback in isolated and merging galaxies using the gadget code , and @xcite performed zoom - in cosmological simulations .
they implemented output of mass and mechanical momentum ( in addition to energy output ) from the bh to the ambient gas , by kicking each wind particle along the direction parallel or anti- parallel to its angular momentum .
more recently , a similar work was presented by @xcite .
our current study was developed completely independent of this ; and their aims / setup are quite different from ours .
they studied feedback from fixed - luminosity agn affecting turbulent gas spheres , using an initial condition of gas distributed from @xmath3 pc to @xmath13 kpc .
in addition to thermal feedback , there is momentum feedback , where each neighbour particle receives a fraction of the agn wind momentum in a direction
radially away from the smbh .
their bi - conical model , where the thermal and kinetic feedback energy are injected in a bi - cone , produces a far more complex gas structure around the smbh .
we investigate , in this paper , different sub - resolution models and implementations of agn feedback in cluster simulations using the sph technique .
we implement novel numerical methods to couple the energy from a central smbh to the surrounding gas , _ kinetic feedback - _ where the velocity ( or momentum ) of the gas is boosted . here
we present simulations of isolated clusters .
our goal is to explore the evolution of cluster cool cores , the properties of bh accretion , growth and feedback , especially its impact on the cooling icm .
we have not done any parameter tuning in this paper to match observations ; but our aim here is to explore the results of the new implementation , especially the differences from our thermal feedback . our previous studies ( * ? ? ?
* ; * ? ? ?
* and references therein ) used only thermal feedback , where the feedback efficiency needed to be increased by a factor of @xmath12 in the radio - mode .
@xcite ( following @xcite ) classified the power from agn into two parts : radiative and mechanical ; simulating both of them with thermal feedback .
in this paper , we progressively build a model where the effect of _ radiative power of the agn is simulated with thermal feedback _ , and the effect of _ mechanical power of the agn is simulated with kinetic feedback_. thus we add to and improve the _ unified model _ of agn feedback ( [ sec - num - bh - accr - feed ] ) , by integrating our new kinetic prescription with the mechanical outflow model of @xcite .
this renders an sph code ( including the full gas physics required for cosmological simulations ) able to simulate both quasar - mode and radio - mode agn , and generate bubble - like outflows in the radio - mode ( c.f .
[ sec - results - const - enrg ] ) in a self - consistent way .
this paper is organised as follows : we describe our numerical code and simulation setup in
[ sec - numerical ] , present , analyze and discuss our results in [ sec - results - const - enrg ] and [ sec - results - bh - accr - feedback ] ( two sections for two categories of simulations ) , discuss the caveats of our study in
[ sec - discussion ] , and in [ sec - conclusion ] we give a summary of our main findings and discuss future work .
we use a modified version of the treepm ( particle mesh ) - sph ( smoothed particle hydrodynamics ) code gadget-3 @xcite , which includes sub - resolution physics as described next .
the initial cluster models are presented in [ sec - num - ic ] , the bh modules including our new kinetic feedback prescription are detailed in [ sec - num - bh - kin]-[sec - num - implement ] , and our simulations are outlined in [ sec - num - sim ] .
besides agn feedback , the physical processes implemented in our simulations include : radiative physics , star - formation ( sf ) , chemical evolution , and are the same as in @xcite . as an exception in the simulations with no sf , radiative cooling and heating processes
are followed for a primordial mix of hydrogen and helium with no metals @xcite , by adopting the cooling rates from @xcite , including the effect of a uv background @xcite .
sf is implemented following the multiphase effective sub - resolution model by @xcite .
gas particles with density above a limiting threshold , @xmath14 @xmath15 ( units of number density of hydrogen atoms ) , contain cold and hot phases , and are star - forming .
collisionless star particles are spawned from gas particles undergoing sf , based on the stochastic scheme by @xcite .
we allow a gas particle to spawn up to four generations of stars .
stellar evolution and chemical enrichment are followed for 11 different elements : h , he , c , ca , o , n , ne , mg , s , si , fe , using the chemical evolution model of @xcite . we include a fixed stellar initial mass function according to @xcite .
there is no kinetic feedback from supernovae - driven galactic outflows in our simulations .
radiative cooling and heating is computed by adding metal - line cooling from @xcite , accounting for the 11 species , and photoionizing radiation from the cosmic microwave background and the @xcite model for the uv / x - ray background are considered .
[ cols="^,^,^,^,^,^,^,^,^,^,^,^ " , ] the initial condition ( ic ) is an isolated cluster in hydrostatic equilibrium ( e.g. , * ? ? ?
our fiducial cluster has a total mass ( including dark matter ) of @xmath16 , and virial radius of @xmath17 mpc .
a hubble parameter of @xmath18 km s@xmath19 mpc@xmath19 or @xmath20 , and baryon density @xmath21 are adopted ( e.g. * ? ? ?
table [ table - galaxies ] lists the relevant gas particle masses and force resolutions .
the baryon fraction of the cluster is @xmath22 .
we sample a sphere of more than @xmath12 times larger than the virial radius , in order to avoid void boundary conditions in the gas just at the virial radius , which makes the total gas mass to be @xmath23 .
there are no stars in the ic . in the runs including sf ,
new star particles form during the simulation via star formation in the gas .
all the particles ( gas , stars , bh ) follow collisionless gravitational dynamics , while in addition the gas particles undergo hydrodynamical interactions . the plummer - equivalent softening length for gravitational forces
is set to @xmath24 kpc for all the particles , in our default resolution simulations .
the sph computations are carried out using a @xmath25-spline kernel @xcite , which is the standard one used in the gadget code ( eq . 4 of @xcite ) .
the number of neighbors for each gas particle is taken as @xmath26 .
we set a minimum value of the sph smoothing length to be @xmath27 times the gravitational softening .
note that , using isolated ics , we could have reached much larger force and mass resolution . instead , our choice here is to mimic the typical resolutions of cosmological , zoomed - in simulations of galaxy clusters .
the procedure of generating our ic consists of the following steps , which are described in detail in the next paragraphs .
( a ) density profiles for dark matter and hydrostatic gas are computed analytically .
( b ) the gas density profile is sampled to produce the initial cluster .
( c ) the cluster is evolved with gravity and hydrodynamics , for several dynamical times , to reach true equilibrium and relaxation .
( d ) the cluster is then evolved with gravity , hydrodynamics and cooling , to develop a cool core .
the cluster dark matter halo is represented by a background fixed static potential , following the @xcite ( nfw ) density profile , @xmath28 where @xmath29 is a scale radius , @xmath30 is a characteristic ( dimensionless ) density , and @xmath31 is the critical cosmic density . some studies ( e.g. * ? ? ?
* ) find an _
adiabatic de - contraction _ effect on the dark matter halo due to massive outflows generated by agn .
nevertheless , we ignore any impact of agn on the dark matter halo ( by assuming it as a static potential ) ; because the goal of our idealized simulations is to study the different forms of feedback effects on the cluster - central gas .
a spherical distribution of gas denotes the icm , initially in hydrostatic equilibrium within the nfw potential .
the equilibrium solution for the gas can be found @xcite , assuming that the baryonic fraction of the halo is negligible and the gas ( of density @xmath32 , and pressure @xmath33 ) follows a polytropic equation of state with index @xmath34 : @xmath35 .
the equation of hydrostatic equilibrium is : @xmath36 here @xmath37 is the halo mass within radius @xmath38 , which can be derived from the nfw profile . with these assumptions , eq .
( [ eq - hydrostatic ] ) can be solved analytically @xcite , as done by @xcite , which we follow .
the analytic profiles for dark matter and hydrostatic gas are thus computed , and sampled to produce the initial gas distribution .
this initially generated isolated cluster is processed to form a cool - core , as displayed in fig .
[ fig - ic_profiles ] showing the radial profiles of gas density ( left panel ) , temperature ( middle ) , and entropy ( right ) .
the first generated ic is the red curve with square plotting symbols ; the cyan shaded area reveals the scatter ( most prominent in the density profile ) due to numerical noise .
the ic is first allowed to evolve using hydrodynamical interactions only up to @xmath39 gyr ( corresponding to a few dynamical times for such a cluster ) , the result of which is shown by the black curve ( triangles ) with yellow shaded area .
this process relaxes the ic while getting rid of the initial noise , and reaches an equilibrium .
the initial noise decays rather quickly during the hydrodynamics - only evolution , within a time of @xmath40 gyr , which is about @xmath41 of the dynamical time of the cluster .
the hydro - only result is subsequently evolved using cooling ( primordial composition gas with no metals ) and hydrodynamics , up to further @xmath39 gyr , which produces the blue curve ( asterisks ) with grey shaded area .
the cooling evolution forms a cold , dense core at the cluster center .
the cluster at this stage ( blue asterisks ) is taken as the default ic of our simulation runs .
the mass cooling rate ( to temperature @xmath42 k , within @xmath3 kpc of the cluster ) in our ic is @xmath43/yr .
after this , rescaled as @xmath44 , one bh is created in our simulations residing at the center of cluster , at rest .
it is a collisionless particle in the gadget-3 code , having a _ dynamical _ mass ( @xmath45 ) equal to a gas particle mass , given by simulation resolution .
the bh outputs feedback energy : a fixed power in the first cases ( [ sec - num - bh - const - enrg ] ) , and next undergoing mass growth and resulting feedback ( [ sec - num - bh - accr - feed ] ) .
feedback from bh corresponds to an output power @xmath46 injected and coupled to the surrounding gas .
it drives gas outward at a velocity @xmath47 and mass outflow rate @xmath48 .
the energy - conservation equation , with the kinetic energy of the outflowing gas equated to the feedback energy from the bh , is , @xmath49 this is similar to the _ energy - driven wind _
formalism of @xcite .
typical agn wind velocity values seen in observations is between @xmath50 km / s ( e.g. , * ? ? ?
we consider @xmath47 as a free model parameter .
this is a simplified assumption of our model , which is intended to be applied to cosmological simulations , although more physically the agn wind velocity should be self - regulated ( e.g. , * ? ? ? * ) .
in the first set of runs , the central bh is considered to output a fixed energy ( bh accretion is not taken into account here ) , i.e. @xmath46 = constant , at a fixed duty - cycle .
these runs have only radiative cooling , and no sf , in order to easily isolate the effect of agn energy injection on the gas .
the default bh power value that we explore is @xmath51 erg / s ; with two runs having higher and lower powers .
the bh activity is considered to be on for @xmath52 myr when the @xmath46 energy is ejected , and then off for @xmath3 myr when there is no energy output .
such a on / off fixed - power duty - cycle occurs periodically up to the end of the simulation , for a few gyrs .
the second set of runs includes star - formation , and a bh undergoing gas accretion , mass growth and resulting feedback .
the accretion of surrounding gas onto a smbh of mass @xmath53 is parametrized by the bondi - hoyle - lyttleton rate @xcite : @xmath54 here , @xmath55 is the gas density , @xmath56 is the sound speed , and @xmath57 is the velocity of the bh relative to the gas .
the bondi radius ( @xmath58 ) and sonic point ( where the gas passes through mach number @xmath59 , going from subsonic velocities at larger radii to supersonic at smaller radii ) are of the order of @xmath60 s of pc , hence are unresolved in current galaxy formation simulations .
the gas properties ( @xmath61 ) are estimated by smoothing on the resolution scale ( kpc to few @xmath3 s of pc ) at the bh location .
this results in artificially low densities compared to spatially resolving the bondi radius scale .
furthermore , smaller - scale simulations @xcite show that the cooling gas is multiphase , with a variable accretion rate
. this cold phase of the ism is not resolved in galaxy simulations . as a numerical correction ,
the accretion rate is enhanced in the simulations by setting the multiplicative factor to @xmath62 ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which we adopt .
most of our simulations are done using a single mode of accretion considering all the gas ( of any temperature ) around bh . in one run
we employ the cold accretion model by @xcite , using @xmath63 for computing the accretion rate of the cold gas ( temperature @xmath64 k ) , and @xmath65 for the hot gas ( @xmath66 k ) .
accretion onto the bh is limited to the eddington rate , making the resultant accretion rate , @xmath67 here , @xmath68 is the eddington mass accretion rate , expressed in terms of the eddington luminosity , @xmath69 where , @xmath70 is the gravitational constant , @xmath71 is the mass of a proton , @xmath72 is the speed of light , and @xmath73 is the thomson scattering cross - section for an electron .
a fraction of the accreted rest - mass energy is radiated away by the bh .
the radiation luminosity is , @xmath74 with @xmath75 being the radiative efficiency .
we adopt the mean value for radiatively efficient accretion onto a schwarzschild bh @xcite : @xmath76 , which is held at a fixed value .
a fraction @xmath77 of this radiated energy is coupled to the surrounding gas as feedback energy from the bh : @xmath78 we consider the feedback efficiency @xmath77 as a free parameter in our models .
we examine different ways by which the bh feedback energy can be coupled to the gas : @xmath79 * thermal * : we adopt the default scheme from @xcite .
the energy @xmath46 is distributed thermally to heat up the gas isotropically around the bh .
the temperature of the neighboring gas particles ( those contributing to eq .
[ eq - bh - smooth ] in [ sec - num - implement ] , particles lying within the smoothing length of the bh ) is incremented by an amount scaled by their sph kernel weights .
@xmath79 * kinetic * : the kinetic energy of the outflowing gas is derived from the bh radiation luminosity which is fed back .
the energy - conservation equation becomes : @xmath80 this gives the outflow rate in terms of the bh accretion rate , @xmath81 there are two free parameters in our sub - resolution model of kinetic feedback : @xmath77 and @xmath47 .
these parameters can be varied to obtain a closest match of the simulation versus observational @xmath82 $ ] relation @xcite .
the best - fit @xmath77 and @xmath47 however depends strongly on the simulation characteristics ( isolated galaxy / cluster , galaxy merger , cosmological box , etc ) .
we decide not to attempt for obtaining any best - fit parameters in this isolated cluster work , which is a study of how kinetic bh feedback affects cluster cool cores .
@xmath79 * unified * : the unified scheme is a combination of bh thermal and kinetic feedback , at every timestep .
following @xcite , this model considers two components of agn feedback ( based on * ? ? ?
* ) : _ outflow _ - mechanical feedback with outflow power @xmath83 , and _ radiation _ - radiative component with luminosity @xmath84 .
the total feedback energy per unit time is then the sum ( their eqs . 7 - 9 ) : @xmath85 @xcite implemented both the components as thermal feedback .
here in our unified model , the mechanical outflow power @xmath83 is coupled to the gas by kinetic feedback .
quantities in the local environment around the bh are computed in a kernel - weighted way , by smoothing over neighboring gas particles .
the kernel size , or the bh smoothing length @xmath86 , is determined at each timestep ( in analogy to finding gas particle smoothing length ) by implicit solution of the equation , @xmath87 here @xmath88 is the kernel estimate of the gas density at the position of the bh , and @xmath89 is the mass of @xmath90 neighboring gas particles .
the bh particle has @xmath12 times more neighbors ( e.g. , * ? ? ?
* ) than an sph gas particle , in order to have more particles for computing the physical properties of the bh sub - resolution model .
we distribute the bh kinetic feedback energy to the gas lying inside a bi - cone around the bh , with a motivation to create a bipolar narrow - angled outflows .
a bi - conical volume is defined with the bh at the apex , and two cones on opposite sides of the bh , having their axes along two diametrically opposed directions .
the slant height of each cone is @xmath86 , and the total opening angle of a single cone is taken as @xmath91 .
the cone - axes directions are considered as fixed in most of our runs , and we explore one case of varying this direction ( as written in table [ table - sims ] ) .
the fixed direction is taken along @xmath92-axis , for the bh located at the origin of our coordinate system .
the gas particles lying within this bi - cone volume around the bh are tracked , and their total mass @xmath93 is computed .
the feedback energy is distributed to gas within a distance @xmath86 from the bh . in our simulations
@xmath86 typically lies between @xmath94 kpc .
gas particles inside the bi - cone are stochastically selected and kicked into agn wind , by imparting a one - time @xmath47 velocity boost .
we use a probabilistic criterion ( similar to other sub - resolution prescriptions in gadget-3 ) for the implementation .
a probability for being kicked is calculated in a timestep @xmath95 for @xmath96th gas particle within bi - cone : @xmath97 here @xmath48 is the mass outflow rate obtained from eq .
( [ eq - energy - conservation ] ) . at a given timestep ,
all the gas particles within the bi - cone have the same probability to be ejected . a random number @xmath98 , uniformly distributed in the interval @xmath99 $ ] ,
is drawn and compared with @xmath100 .
for @xmath101 , the gas particle is given a wind velocity kick .
the inverse proportionality of @xmath100 with @xmath93 ensures that the number of particles kicked does not depend on the geometry of the volume , but depends on @xmath48 only .
the quantity @xmath102 is the mass of gas to be kicked .
the probability is constructed such that the available particles ( total mass @xmath93 within bi - cone ) are sampled to reproduce kicking at the rate given by @xmath48 , on average . during a simulation
, we always ensure that @xmath103 . after receiving agn wind kick ,
a gas particle s velocity becomes : @xmath104 the kick direction ( @xmath105 or @xmath106 ) is set along the axis of the cone inside which the gas particle lies , depending on the location of the particle w.r.t . the bh .
our choice represents a scenario where one bh injects kinetic energy along a fixed direction , chosen here as the @xmath107-axis for numerical simplicity .
such a prescription allows the energy injected to be channelized most efficiently , creating bi - directional outflows originating from the central bh . in future cosmological simulations , the fixed direction for agn kick would represent the spin of the bh . some studies ( e.g. , * ?
* ) implement hydrodynamic decoupling of the gas particles kicked into wind by supernovae , to enable the outflow to escape without affecting star - formation in the galaxy . here
we perform runs considering both decoupling and coupling of the agn wind particles from hydrodynamic interactions . in the presence of bh accretion / growth ( [ sec - num - bh - accr - feed ] )
, the collisionless bh particle ( [ sec - num - ic ] ) has a seed bh of initial mass @xmath108 . at each timestep
@xmath95 , it grows according to the bh accretion rate ( eq . [ eq - mdot - bh ] ) , its mass increases by an amount @xmath109 , with the dynamical mass @xmath45 ( [ sec - num - ic ] ) remaining the same .
after a bh has grown such that @xmath110 , it might accrete ( or _ swallow _ ) neighboring gas particles , using a stochastic methodology . when a gas particle is swallowed , it is removed from the simulation , and @xmath45 increases by the swallowed particle mass @xmath111
. this conserves dynamical mass within the computational volume .
the probability of swallowing gas is set to ensure that @xmath53 and @xmath45 track each other closely . once the bh is created residing at the cluster center at @xmath44 ( [ sec - num - ic ] ) , it is free to move at subsequent times .
we have checked that in our simulations , the bh remains within a displacement @xmath27 to @xmath112 kpc from the cluster center ; which means that the bh does not advect away .
@cccccccc run & agn accretion & agn feedback & @xmath46 & @xmath77 & @xmath47 & direction & hydrodynamic coupling + name & & & [ erg / s ] & & [ km / s ] & + + _ c5k _ & & constant energy & @xmath113 & & @xmath114 & fixed & coupled + _ c10k _ , _ c10khr _ & & constant energy & @xmath113 & & @xmath0 & fixed & coupled + _ c5kdec _ & & constant energy & @xmath113 & & @xmath114 & fixed & decoupled up to 50kpc + _ c10kdec _ & & constant energy & @xmath113 & & @xmath0 & fixed & decoupled up to 50kpc + _ c10kdechigh _ & & constant energy & @xmath115 & & @xmath0 & fixed & decoupled up to 50kpc + _ c10klow _ & & constant energy & @xmath116 & & @xmath0 & fixed & coupled + _
c10krand _ & & constant energy & @xmath113 & & @xmath0 & random & coupled + + _ sf _ & & no bh & & & & & + _ thr _ & all gas & thermal & & @xmath117 & & isotropic & + _ kin10k _ & all gas & kinetic & & @xmath117 & @xmath0 & fixed & coupled + _
unf10k _ & all gas & unified & & @xmath117 & @xmath0 & fixed & coupled + _
unf5k _ & all gas & unified & & @xmath117 & @xmath114 & fixed & coupled + _
unf10klo _ & all gas & unified & & @xmath118 & @xmath0 & fixed & coupled + _
unf10khi _ & all gas & unified & & @xmath112 & @xmath0 & fixed & coupled + _ unf10kloca _ & cold gas & unified & & @xmath118 & @xmath0 & fixed & coupled + table [ table - sims ] lists the series of simulations we perform .
they form two broad categories , as written below .
each category incorporate the same non - agn sub - resolution physics , and the different runs in it investigate various agn feedback models .
the evolution is followed for a time @xmath119 to @xmath120 gyr in the different runs . * _ cooling only , no sf , constant energy output from bh at fixed duty - cycle , coupled as kinetic feedback _ ( [ sec - num - bh - const - enrg ] ) : the gas is considered to be of primordial composition , i.e. , hydrogen and helium only , with no metals .
the run names correspond to the variation of the model parameters as written in the following . * * _ c5k _ : outflow velocity @xmath121 km / s . * * _ c10k _ : @xmath122 km / s . * * _ c10khr _ : high - resolution run with @xmath122 km / s . * * _ c5kdec _ , _ c10kdec _ : different-@xmath47 runs with the wind particles decoupled from hydrodynamic interactions up to a distance of @xmath52 kpc . *
* _ c10kdechigh _ : wind - decoupled run , with a higher output bh power @xmath123 erg / s . * * _ c10klow _ : lower value of output bh power @xmath124 erg / s . * * _ c10krand _ : direction of kicking wind particles ( bi - cone axis ) changes randomly between duty cycles , i.e. from one activity cycle of bh to another . * _ with cooling , sf and bh growth , energy output from bh accretion rate _ ( [ sec - num - bh - accr - feed ] ) : metals are generated according to the stellar evolution and chemical enrichment model described in [ sec - numerical ] .
six of our simulations have a single mode of accretion considering all the gas ( of any temperature ) around bh , using @xmath62 in the bondi accretion rate .
one run has the cold accretion model @xcite , using @xmath63 and @xmath65 . *
* _ sf _ : star - formation , stellar evolution , and chemical enrichment only ( no bh ) .
* * _ thr _ : thermal feedback from bh , with @xmath125 . * * _ kin10k _ : kinetic feedback from bh , with @xmath125 , @xmath122 km / s . * * _ unf@xmath126 _ : unified feedback model ( a combination of bh thermal and kinetic feedback ) , where the mechanical outflow power is coupled to the gas in the kinetic form . * * * _ unf10k _ : @xmath125 , @xmath122 km / s . * * * _ unf5k _ : @xmath125 , @xmath121 km / s . * * * _ unf10klo _ : @xmath127 , @xmath122 km / s . * * * _ unf10khi _ : @xmath128 , @xmath122 km / s . * * * _ unf10kloca _ : cold gas accretion model , unified feedback , with @xmath127 , and @xmath122 km / s .
note that the escape velocity of our cluster halo is @xmath129 km / s .
therefore both the values of @xmath47 explored in our models are above the cluster escape speed .
kinetic energy feedback imparted to the gas creates bipolar bubble - like outflows , originating from the central bh , and propagating radially outward up to a few @xmath3 kpc .
the morphology and structure of a representative outflow for the constant - energy case is plotted in fig .
[ fig - projected_map_run13_snap005 ] , showing the projected gas kinematics in the inner @xmath130 kpc@xmath131 region of run _
c5k _ at an evolution time @xmath132 gyr .
gas is kicked with @xmath121 km / s in bi - cone volume around the bh along a fixed direction ( revealed by the red outflowing velocity arrows in the bottom - right panel ) , and propagates outward .
it shocks with the surrounding slower - moving gas , and leads to the formation of outflowing bubble - like structures .
the developed outflowing bubbles are hot ( visible as red areas in the temperature plot ) , in the face - on plane ( middle - left panel ) shaped as spherical - shell , and in the edge - on plane ( middle - right panel ) as extended bipolar terminated with an outer elliptical - shaped shock .
the bubbles consist of low - density gas ( first row ) .
the time of @xmath132 gyr plotted in fig .
[ fig - projected_map_run13_snap005 ] corresponds to a part of the first outburst cycle of the bh .
the dense core of the cluster ( red areas in the density panels ) is still cool at this time , visible as the blue central region in the temperature panels .
however , the cool - core is destroyed within three or four activity cycles of the bh , and the center is heated up , as will be described later .
[ fig - gasproperties_vs_r_run14 ] presents the radial profiles of gas properties at three different time epochs in run _ c10k_. the curves denote the median value of the relevant property in radial bins , and the shaded areas enclose the @xmath133th percentiles above and below the median , showing the radial scatter .
the median quantities at the three different times are indistinguishable from each other at larger radii : @xmath134 kpc .
bh feedback imparted to the gas affects the median property at @xmath135 kpc , and the radial scatter up to further out @xmath136 kpc .
the central gas kicked into wind with @xmath122 km / s thermalizes its kinetic energy in the form of shocks , and generates bubble - like structures propagating outward .
the outflows are visible as rise in the radial scatter ( shaded areas ) in fig .
[ fig - gasproperties_vs_r_run14 ] : hot ( top - right panel ) reaching temperature as high as @xmath137 k , high - entropy ( bottom - right ) , low - density ( top - left ) , and high - pressure ( bottom - left ) .
the earliest time plotted : @xmath138 gyr ( black curve with square symbols ) is an epoch of the first active period of the bh .
the cyan shaded area shows the first outflow propagation , extending over @xmath139 kpc . during every agn - on active period ( @xmath52 myr )
, such a large outflow originates from the center and disperse out , moving up to a distance few @xmath3 kpc during the time the central agn is off ( @xmath3 myr ) ; before another one starts at the next active cycle .
the region @xmath140 kpc continues to accumulate these hot , high - entropy , low - density bubbles ; visible as the yellow and grey shaded areas fig .
[ fig - gasproperties_vs_r_run14 ] . the blue curve with triangles ( @xmath141 gyr ) is a later time agn - on active epoch at the @xmath142th cycle ; the yellow shaded area displays both an originating bubble at @xmath143s kpc , and the previous accumulated bubbles at @xmath3s kpc .
the red curve with asterisks ( @xmath144 gyr ) is an inactive epoch after the @xmath145th cycle when the agn is off ; the grey shaded area displays the previous accumulated bubbles at @xmath3s kpc .
the bubble - like outflows propagate radially outward up to @xmath146 kpc , signalled by a reduction in the radial scatter at larger radii .
the sharp drop of the radial scatter reveals the termination shock , where the outflowing gas motion slows down .
the initial cool core of the cluster is heated up in several simulations with constant - energy output from the bh , within a time - period depending on the parameter values of the kinetic feedback model .
[ fig - evol_gas_hot_cold ] presents the masses of cool and hot components of the gas at the cluster center , lying within distance @xmath147 kpc from the bh , versus evolution time , in the seven constant - energy runs .
the different curves are for various agn feedback models , distinguished by the colors and linestyles , as labelled and written in the caption . here
we define cool gas as that of temperature @xmath148 k which is shown in the left panel , and hot gas has @xmath149 k plotted in the right panel .
such a temperature was chosen for the cool / hot distinction based on the ic . as seen in fig .
[ fig - ic_profiles ] - blue curve with asterisks , the cool core has a median @xmath150 k in our initial cluster profiles , and the maximum median temperature is @xmath151 k at a larger radii . the hot - phase is the dominating component at the center over the cool - phase in all the runs , except _ c10kdec _ where the cool - phase dominates after @xmath152 gyr , reasons for which is elaborated below . according to the ic ( [ sec - num - ic ] ) , the cluster inner region consists of dense , cold , low - entropy gas ( blue curve in fig .
[ fig - ic_profiles ] ) , when the bh is created and allowed to eject feedback energy . at @xmath153 more gas cools , gets denser and sinks to the center , tending to enlarge the cool core in the absence of anything else . kinetic feedback from the bh acts as a heating source here .
a fraction of gas selected from the cool - core reservoir , in particular particles within a bi - cone volume inside the smoothing length of the bh , @xmath154 kpc , are chosen and kicked to a higher velocity .
this drives some of the innermost gas out , and ejects it to larger distances in the form of bubble - like outflows ( [ sec - res - outflow ] ) .
the mass kick out rate in our sub - resolution model , example for run _ c10k _ , is : @xmath155/yr .
when the agn is on for @xmath52 myr , a total mass of @xmath156 gas is kicked to a higher speed ( @xmath122 km / s ) .
the total mass outflown from the central @xmath143 kpc is mostly cool gas . the hot ( dominating phase ) , and consequently the total ( cool @xmath157 hot ) central gas mass remains almost constant with time .
the kicked fast wind undergoes shock with the slower - moving gas surrounding it , which are visible in fig .
[ fig - gasproperties_vs_r_run14 ] as the temperature jumps of the shaded areas , and were described in [ sec - res - outflow ] .
the interactions cause the imparted kinetic energy to be thermalized , heating up the cluster cool core .
it is revealed in the left panel of fig .
[ fig - evol_gas_hot_cold ] by the decrease of cool gas mass with time .
the default output bh power of @xmath51 erg / s is able to heat up the core in the cases where the kicked wind particles are always coupled to hydrodynamic interactions : _ c5k _ within @xmath158 gyr , _
c10k _ and _ c10krand _ by @xmath119 gyr . for a constant output bh power
, @xmath48 is proportional to @xmath159 ( eq . [ eq - energy - conservation ] ) .
so the mass outflow rate is @xmath12 times higher with @xmath121 km / s over @xmath122 km / s .
this causes @xmath12 times more central gas mass to be kicked in runs _
c5k _ and _ c5kdec _ , than in _ c10k _ and _ c10kdec_. this depletes the inner cool gas faster with @xmath121 km / s . as the extreme opposite case ,
when wind particles are kicked with @xmath122 km / s and decoupled from hydrodynamic interactions up to @xmath52 kpc ( run _ c10kdec _ , blue dotted curve ) , it produces increased cooling at the center , when the cool gas mass rises with time .
this is because the imparted kinetic energy is thermalized at @xmath134 kpc , and does not have any effect at the center .
the central dense gas continues to cool to @xmath160 k here .
the gas is kicked into wind at the same rate in runs _
c10k _ and _ c10kdec _ , however the cool core remains in case _
c10kdec_. this demonstrates that the hydrodynamical interactions of the kicked wind with the surrounding gas ( occurring in the coupled wind case _ c10k _ ) plays an important role in destroying the cool core .
a higher output bh power @xmath123 erg / s applied to the same wind - decoupled run ( _ c10kdechigh _ , red dash - dot - dot - dot curve ) heats up the core completely within @xmath161 gyr .
the wind - decoupled case with default output bh power and lower velocity @xmath121 km / s ( run _ c5kdec _ , black dashed curve ) depletes the cool central gas ( @xmath143 times reduction in cool mass in @xmath119 gyr ) , albeit rather slowly , since @xmath48 is @xmath12 times higher than in run _
c10kdec_. fig .
[ fig - evol_mdot_cold ] shows the time evolution of mass ( cooling - heating ) rate of the gas in the constant - energy runs . in computing this net cooling rate ,
we consider gas lying inside distance @xmath162 kpc and of temperature @xmath163 k as cold .
positive values indicate net cooling of the gas , and negative values indicate net heating .
the periodic negative spikes indicate the agn outburst heating at the rate @xmath164/yr . during agn - off periods
there is cooling at the rate @xmath165/yr , except run _
c10kdec _ where the cooling rate is higher .
the lower output bh power @xmath124 erg / s having always coupled wind ( run _ c10klow _ , cyan long - dashed curve ) maintains the cool core , with the inner cool gas mass ( fig .
[ fig - evol_gas_hot_cold ] ) remaining almost constant over time . in this case
, the smaller amount of energy injected allows cooling to occur at a rate few @xmath166/yr even after @xmath13 gyr , which keeps the cluster core in nearly thermal equilibrium .
we find that if the direction of imparting kinetic feedback changes randomly from one activity cycle of bh to another ( run _ c10krand _ , yellow dash - dot curve ) , it has a significant impact on the central hot gas mass .
it is revealed by the decrease of inner hot gas mass by @xmath143 times or more ( large spikes in the right panel of fig .
[ fig - evol_gas_hot_cold ] ) during certain activity cycles , while returning back to the original value after @xmath3 myr . here
the hot gas is expelled outside @xmath143 kpc from the central bh , by strong kinetic feedback , which is much more efficient with the change of the feedback direction between duty cycles of bh .
[ fig - gasproperties_vs_r_9runs ] displays the radial profiles of gas properties in the constant - energy runs at an evolution time @xmath141 gyr .
the radius is computed by the distance from the bh location , and we have checked that the bh remains within @xmath167 kpc from the cluster center , which is the origin of our spherical coordinates .
the profiles of the seven runs are indistinguishable from each other at larger radii : @xmath168 kpc for density ( top - left panel ) and pressure ( bottom - left ) ; @xmath169 kpc for temperature ( top - right ) and entropy ( bottom - right ) .
kinetic feedback from the bh affects the gas properties at only the regions inner to these radii .
it expels some central gas in the form of high - velocity outflows , which thermalize their energies and shock - heat the core .
the core heating is exhibited by the median temperature remaining high @xmath170 k in the inner regions .
this corresponds to a central entropy flattening , @xmath171 kev @xmath172 .
the cluster core is heated to @xmath170 k and having a flat entropy profile , by a time @xmath173 gyr in five runs : _
c5k _ , _ c10k _ , _ c5kdec _ , _
c10kdechigh _ , and _
c10krand_. kinetic feedback from the bh is less efficient in two runs , where the initial cool core ( @xmath148 k ) remains : _ c10kdec _ ( blue dotted curve in fig .
[ fig - gasproperties_vs_r_9runs ] ) where the kinetic energy is thermalized too far away from the center , and _ c10klow _ ( cyan long - dashed ) where the low output power keeps the cluster core in nearly thermal equilibrium .
these cool - core clusters have a @xmath174 times higher central density than the cases where the core is heated . the pressure variation between different models is smaller than the variations in other gas properties , because pressure mostly depends on the background potential which is static in our simulations .
we perform a higher resolution simulation _ c10khr _ with @xmath143 times more number of particles ( table [ table - galaxies ] ) than our default resolution cases .
we chose the constant - energy setup for our resolution study because it is the simplest with no star - formation ( i.e. free from resolution dependence of the star - formation model ) , thus here the resolution effect of the agn energy injection scheme can be better identified .
_ c10khr _ employs the same model parameters as run _
c10k_. fig .
[ fig - evol_gas_hot_cold_hires ] shows the masses of cool and hot components of the gas at the cluster center , lying within distance @xmath147 kpc from the bh , versus evolution time , in these two different resolution runs .
we find that the cluster cool core is heated up faster in the high - resolution case , as seen in the left panel of fig .
[ fig - evol_gas_hot_cold_hires ] . the inner cool gas mass decreases with time and is depleted within @xmath175 gyr in run _
( red dashed curve ) , which is almost half the time needed in the lower resolution case _
c10k _ ( green solid curve ) . fig .
[ fig - gasproperties_vs_r_hires ] presents the radial profiles of gas properties at two different epochs in runs _
c10k _ and _ c10khr_. the varying resolution brings a difference in the median profiles at @xmath176 kpc .
bubble - like outflows are visible as rise in the radial scatter ( shaded areas ) : hot ( middle panel ) reaching temperature as high as @xmath177 k , high - entropy ( right ) and low - density ( left ) .
the pressure profile remains the same in the different resolution runs , and we do not show it here .
the blue curves ( @xmath138 gyr ) depict an epoch of the first active period of the bh .
the red curves ( @xmath178 gyr ) are a later time agn - on active epoch . comparing the cyan and grey shaded areas
, we see that the inner bubble starts from a smaller radii at high - resolution , and extends up to a shorter distance .
the outer bubble signatures , extending over @xmath179 kpc , remain the same when resolution is varied . comparing between the solid and dashed curves at each epoch in fig .
[ fig - gasproperties_vs_r_hires ] , we see that the cluster core is heated more at high - resolution , entropy is higher , and the density is lower .
however the differences are not large and remain within a factor of @xmath119 .
moreover , the profiles at the two resolutions show an excellent degree of convergence at length scales larger than @xmath180 kpc .
we thus find that convergence is not reached strictly in our simulations .
our implementation of kinetic agn feedback is more effective at higher resolution .
full convergence can not be expected using the same parameter values within the framework of our sub - resolution models .
this is because feedback energy is distributed to the local gas , and higher gas densities are resolved in the high - resolution case . as discussed in @xcite ( 4 , paragraph 9 ) , the lack of convergence implies that different model parameters ( e.g. @xmath181 ) are required at different resolution in cosmological simulations to have the same fit to observational correlations ( e.g. bh mass versus host galaxy stellar velocity dispersion ) . in a contemporary work investigating numerical resolution ,
@xcite studied the behaviour of agn feedback when force and mass resolution are varied .
they found that an _ increase _ of resolution results in a _
less _ effective ejection of the gas by agn feedback .
the result we obtain here is at variance with theirs .
we however note that both the resolution and coupling of the agn energy to the gas is quite different in our work with respect to @xcite .
they simulated the interaction of an agn outflow with the clumpy turbulent gas in the inner part of the host galaxy on length scales from @xmath27 to @xmath13 kpc , which is smaller than ours ( @xmath182 kpc ) .
feedback - resistant high - density clumps are washed out at low effective resolutions in their simulations , whereas such clumps never form in our simulations .
our mass resolution is typical of that of a cosmological simulation of galaxy clusters , thus our gas particle mass is at least an order of magnitude higher than theirs .
moreover , in that paper the coupling of energy is done by injecting thermal energy .
hence in our opinion , the point of numerical convergence is still open , since it can easily depend on both the simulation force and length scales and on the precise scheme adopted for the energy deposition onto the gas .
a bh is seeded at the cluster center at @xmath44 according to our ic ( [ sec - num - ic ] ) , of initial mass @xmath183 ( [ sec - num - implement ] ) . in the next eight simulations with cooling ,
sf and bh accretion , the bh grows in mass by accreting gas from its surroundings , and consequently outputs feedback energy ( [ sec - num - bh - accr - feed ] ) .
we find that kinetic feedback - induced gas outflows are created , at the times when the bh undergoes a high accretion rate .
the cluster core is heated eventually when using higher values of the feedback efficiency parameter . the time evolution of bh mass accretion rate ( @xmath184 ) is presented in fig . [ fig - bhaccr_starform_rate ] , top panel .
the middle panel displays the eddington ratio @xmath185 .
the eight runs with different agn feedback models are distinguished by the colors and linestyles , as labelled in the bottom panel and caption . at @xmath44 , the @xmath186/yr is very low and rises slowly with time . from @xmath187 gyr
, the @xmath184 has a power - law increase , corresponding to the eddington - limited growth of the bh ( where eddington ratio @xmath59 ) , up to a time @xmath188 gyr , to reach a peak of @xmath184 . as an example , the peak @xmath184 of @xmath189/yr in run _
unf10k _ occurs at @xmath190 gyr . kinetic feedback ( run _ kin10k _ , red dash - dot curve ) produces a @xmath52 times larger peak @xmath184 than thermal feedback ( _ thr _ , blue solid ) using the same @xmath125 .
after its peak , the @xmath184 reduces in the runs with kinetic feedback ( _ kin10k _ and the _ unf _ cases ) , because significant amounts of gas are ejected from the center and/or heated up , limiting the gas feeding onto the bh .
while in the thermal feedback run _
thr _ , the @xmath184 is almost constant after the peak , because cold gas remains at the center here , retaining the cool core .
there are fluctuations in the @xmath184 , whereby it increases or decreases by a factor of up to @xmath143 in @xmath117 gyr .
a noteworthy feature is the huge variability of @xmath184 in run _ unf10kloca _ at @xmath191 gyr . here
the inclusion of cold gas accretion results in @xmath184 fluctuating by a factor of up to @xmath192 , with a periodicity of @xmath3 myr .
the bh mass versus evolution time is plotted in fig .
[ fig - evol_gas_star_bh ] , top - left panel .
all the feedback models have the bh growing in a qualitatively similar manner .
starting from the given seed mass of @xmath193 , each bh has an exponential growth . over the time range @xmath13 gyr to @xmath152 gyr , its mass increases by a factor @xmath194 to @xmath195 .
after @xmath196 gyr , @xmath53 comes to an almost steady state , having a very slow subsequent growth .
the final @xmath53 reached at @xmath119 gyr depends on the model : e.g. @xmath197 in run _
unf10k_. @xmath53 is inversely proportional to @xmath77 in the unified feedback model ( runs _ unf10klo _ : yellow dash - dot - dot - dot curve , _
: black dashed , _
unf10khi _ : purple dotted ; using @xmath198 ) .
this is due to self regulation of the bh growth ( see also e.g. * ? ? ?
* ) , and is analytically expected when the bh feedback energy is derived from @xmath184 ( eq . [ eq - edot - feed ] ) .
a higher @xmath77 imparts a stronger feedback ejecting out more central gas , and yields a less - massive bh than a lower @xmath77 .
the bh mass is also inversely proportional to the kick velocity @xmath47 for unified feedback ( runs _ unf5k _ : green long - dashed , and _
unf10k _ : black dashed ; using @xmath121 and @xmath0 km / s ) , which is contrary to naive expectations .
note that @xcite saw a direct proportionality of @xmath53 with @xmath47 , for kinetic feedback in isolated galaxies and mergers .
this is analytically expected because on increasing @xmath47 , @xmath48 decreases ( inverse proportionality in eq .
[ eq - mdotw - edw ] ) . the reduced mass outflow rate of kinetic feedback ejects out less gas . hence , more gas is available for accreting onto the bh , which grows more massive .
we posit that the discrepancy between the two studies arises from the simulation setup . here
the bh kicks out central gas within a cluster atmosphere , and at later times some of the ejected gas falls back in , after slowing down by interacting with the further - out intracluster gas .
whereas in @xcite , once the kicked gas escaped the galaxy , it never came back .
stars form in the isolated cluster from dense gas according to the sf prescription described in [ sec - numerical ] .
the sf occurs within a few kpc region at the cluster center where the gas inflows and cools , forming an initial cool core .
the presence of a bh decreases the central gas content , as some gas is accreted in , some gas is ejected by feedback , and some gas is heated ; which consequently quenches sf .
the total star formation rate versus time is displayed in the bottom panel of fig .
[ fig - bhaccr_starform_rate ] , for eight runs with different agn feedback models . at @xmath44
there is an initial burst of sf at a rate @xmath199/yr , because of our simulation ic : the cluster already has substantial amounts of cold , dense gas in the form of a cool core , which readily form stars when sf is allowed in the model .
it reduces by @xmath200 times soon afterward , because of reduced central gas , which has depleted by the initial sf burst .
the sfr increases almost linearly with time from @xmath112 gyr , as more gas cools , gets dense and is converted to stars .
there are periodic fluctuations in the sfr , when it would decrease by a factor of @xmath201 , occurring because of supernovae feedback in the stellar evolution model .
all the runs show this rising sfr initially , which continues to increase with time in the _ sf _ case without a bh .
the models start to deviate from @xmath202 gyr onwards , when the bh has grown to a high - mass and generates enough feedback suppressing sf .
thermal feedback ( run _ thr _ , blue solid curve ) quenches sf at @xmath203 gyr , the sfr is lower than the _ sf _ case ( magenta solid curve ) by up to a factor of @xmath143 .
kinetic feedback ( _ kin10k _ and the _ unf _ cases ) causes a greater suppression of sfr up to few @xmath3 times lower than the _ sf _ run , because significant amounts of gas are ejected from the center . as an example , in run _ unf10k _ ( black dashed curve ) the sfr is reduced from @xmath204/yr at @xmath205 gyr to @xmath206/yr at @xmath119 gyr . fig .
[ fig - evol_gas_star_bh ] , top - right panel displays the time evolution of mass of stars formed from gas via sf .
gas is continuously converted to stars by the rising sfr initially , for all the agn models .
the stellar mass hence increases with simulation time up to @xmath202 gyr , by when a total @xmath207 of stars has been formed , an amount which is comparable to the hot gas mass in the central @xmath143 kpc of the cluster .
after @xmath202 gyr , @xmath208 continues to rise with time in the _ sf _ case .
@xmath208 also increases with thermal feedback albeit with a smaller slope . while in the runs with kinetic feedback @xmath208 remains constant as no more new stars are formed by the hugely quenched sfr .
the initial cool core of the cluster is heated , by energy feedback from the growing bh , to different extents depending on the model parameter values .
the bottom row of fig .
[ fig - evol_gas_star_bh ] presents the masses of cool and hot components of gas within a distance @xmath147 kpc from the bh location , versus evolution time .
here cool gas is that of temperature @xmath148 k shown in the bottom - left , and hot gas has @xmath149 k plotted in the bottom - right panel .
the hot phase dominates at the cluster center over the cool phase in all the runs , except _ sf _
( magenta solid curve ) where the two phases become comparable after @xmath119 gyr .
the hot gas mass increases slowly with time , initially for all the models to reach @xmath209 at @xmath202 gyr .
this rising @xmath210 is caused by the inflow of hotter gas from the surroundings , because of gas depletion near the center by cooling - induced sf .
@xmath210 continues to rise for three runs : _ sf _ , _ thr _ , _ unf10klo _ , which has either no or low bh feedback . while @xmath210 reduces subsequently for five runs : _
kin10k _ , _
unf10k _ , _ unf5k _ , _ unf10khi _ , _ unf10kloca _ , decreasing by @xmath119 times from @xmath211 gyr to @xmath120 gyr , showing the influence of strong kinetic feedback in these cases where significant amount of central gas outflows .
a fraction of gas cools with time , gets denser , and sinks to the cluster inner region , forming a cool core .
initially the bh is small and does not exert enough feedback to influence the gas . therefore the cool gas mass within the inner @xmath212 kpc increases with time up to @xmath213 at @xmath214 gyr . by this time
the bh has grown to @xmath215 .
subsequently the bh exerts substantial kinetic feedback energy heating the cool core with some parameters , revealed in the bottom - left panel of fig .
[ fig - evol_gas_star_bh ] by the decrease of cool gas mass with time .
all the models with kinetic feedback ( _ kin10k _ and _ unf _ runs ) heat the core , reducing the cool gas to @xmath216 by @xmath217 gyr . in the case _
unf10klo _ ( yellow dash - dot - dot - dot curve ) , @xmath218 increases again forming a local peak between the time range @xmath219 gyr ,
because it uses the smallest @xmath127 and the low feedback allows some gas to cool at late times . with no bh feedback ,
run _ sf _ produces increased cooling at the center , when the cool gas mass monotonically rises with time .
thermal feedback ( _ thr _ , blue solid curve ) is able to deplete some of the cool central gas , which reduces by @xmath220 times at @xmath119 gyr , but increases again later , showing that fraction of the cool core remains .
run _ unf10kloca _ presents features similar to the variability of @xmath184 ( [ sec - res - bh - growth ] ) at @xmath191 gyr .
cold gas accretion results in the cool and hot gas mass fluctuating by a factor of up to @xmath143 , with a periodicity of @xmath3 myr .
@xmath221{gasproperties_vs_r_bhaccr.eps } \\
\includegraphics[width = 0.9 \linewidth]{z_tcool_vs_r_bhaccr.eps } \end{array } $ ] fig .
[ fig - gasproperties_vs_r_7runs ] displays the radial profiles of gas properties in the simulations with cooling , sf and bh growth , at an evolution time @xmath222 gyr .
the profiles of the eight runs are indistinguishable from each other at larger radii @xmath168 kpc . bh accretion and feedback influence the gas properties in the inner regions ( @xmath223 kpc ) only .
some central gas is expelled as high - velocity outflows by kinetic bh feedback , which thermalize their energies and shock heat the core .
the core heating is exhibited by the median temperature ( top - right panel ) remaining high @xmath170 k in the inner regions .
this corresponds to a central entropy flattening , @xmath224 kev @xmath172 ( middle - right panel ) .
the cluster core is heated to @xmath170 k with a flat entropy profile , by a time @xmath225 gyr in five runs : _
kin10k _ , _
unf10k _ , _ unf5k _ , _ unf10khi _ , _ unf10kloca_. the initial cool core ( central @xmath2 k ) remains in three runs : _ sf _ where there is no bh , _ thr _ since thermal feedback is less effective , and _
unf10klo _ where the kinetic feedback power is low because of the smallest @xmath127 value used .
these cool - core clusters have a @xmath174 times higher central density , and up to @xmath143 times higher central pressure , than the cases where the core is heated up .
we compute the total gas metallicity , @xmath226 , as the ratio of all metal mass to the total gas mass for each gas particle .
abundance ratios are expressed in terms of the solar metallicity , which is @xmath227 ( mass fraction of all metals in sun ) derived from the compilation by @xcite .
the gas metallicity radial profiles are plotted in the bottom - left panel of fig .
[ fig - gasproperties_vs_r_7runs ] .
the metallicity profiles are very steep , since metals are produced in the star - forming gas , concentrated inside the central @xmath143 kpc .
the median metal abundance at @xmath119 kpc reaches a few times solar in runs _ thr _ and _ unf10khi _ , and reduces by 4 orders of magnitude at @xmath143 kpc .
we checked that the run _ sf _
( where there is no bh , and hence the largest star formation rate ) generates the highest median metallicity among all the runs of @xmath228 at @xmath229 kpc .
the gas cooling time is : @xmath230 , where @xmath231 is the net cooling rate .
we adopt the cooling rate from the analytical formula ( equation 9 ) in @xcite , which is a function of temperature and metallicity .
it accounts for bremsstrahlung ( @xmath232 ) and metal line cooling ( @xmath233 ) terms , and approximates the cooling function derived by @xcite .
the @xmath234 radial profiles are plotted in the bottom - right panel of fig .
[ fig - gasproperties_vs_r_7runs ] . the cooling time becomes shorter than @xmath13 gyr at @xmath223 kpc . at the same time
we note that this @xmath234 estimate includes bremsstrahlung and metal line cooling only , and does not include radiative heating from photoionizing radiation of the uv / x - ray background .
the @xmath234 in run _
unf10kloca _ has a gently flattening core with no central abrupt fall , which are observationally favorable features .
this affirms that both cold accretion and a low efficiency @xmath235 are crucial to produce a proper self - regulated agn feedback loop .
in this study we explore a distinct mode of agn feedback : kinetic coupling of the energy output from the central bh .
it is implemented within our larger suite of sub - resolution models : metal - dependent radiative cooling and heating , star - formation , stellar evolution and chemical enrichment .
we perform simulations of isolated cool - core clusters , starting from an idealized set - up of a hydrostatic atmosphere with no stars .
our simulations produce a stellar mass of a few times @xmath236 after @xmath119 gyr ; since forming stars with our ic is a slow process , with timescale of gyrs , slower than timescale for feedback onset .
we do not want to perform any comparison with observational data , because our initial condition do not resemble a real poor cluster , and hence have not done any parameter tuning here to match observations .
our aim is to explore the results of the new kinetic agn feedback numerical implementation and assess its effectiveness , emphasising on the differences from our thermal feedback .
the current work is also a step toward building a model of unified agn feedback in cosmological simulations ; which we will run in the future , and where we will do the relevant parameter tuning . at our current resolution
, we do not resolve the physical blast of thermal feedback . also at our resolved scales
, we are not simulating the real agn jet propagation using our kinetic feedback numerical scheme . here
we only attempt to capture the effect of the energy deposition of the mechanical jets from agn at large ( @xmath237 kpc ) scales .
our finding that the cool core is destroyed within @xmath238 duty cycles using a fixed bh power of @xmath51 erg / s , is inconsistent with observational x - ray data . the lower bh power @xmath124 erg / s is a more observationally consistent parameter value to maintain the cool core .
previous simulations ( e.g. , * ? ? ?
* ) have found that bhs , which output a fixed power with fixed duty cycles , are not properly self - regulated , thus they tend to produce overheating or overcooling of cluster cool cores in the long run .
our results reveal the challenges of devising numerical schemes for sub - resolution models , which must capture the physical processes given the numerical resolution .
with regard to implementing agn feedback in galaxy - scale simulations , the agn output power is injected to the gas located near to the bh .
the gas in these central regions can be very dense , hence immediately and non - physically can radiate the injected energy away .
such problems are numerical in nature , which especially impacts our thermal feedback results .
we demonstrate that our kinetic agn feedback is more effective than thermal agn feedback , when implemented together with cooling , sf , stellar evolution and chemical enrichment .
this statement might sound unphysical .
but as @xcite discussed in detail , the implementation of bh thermal feedback is less efficient within the framework of the multiphase effective sub - resolution model of star - formation by @xcite , where sf is based on a density threshold only .
the thermal energy deposited to the gas particles which are multiphase ( and star - forming ) is radiated away quickly , since they are dense . and they attain the effective equation - of - state temperature dictated by their density .
the feedback efficiency @xmath77 ( eq . [ eq - edot - feed ] ) is an important parameter of agn feedback models , which determines what fraction of the energy from the bh couples to the surrounding gas .
cosmological simulations generally calibrate its value to reproduce the bh mass versus host galaxy stellar bulge mass or velocity dispersion relations at @xmath239 ( e.g. , * ? ? ?
various studies have used different values : @xmath240 together with @xmath76 @xcite ; increasing to @xmath128 during radio - mode when the bh accretion rate is smaller than @xmath241 times the eddington rate @xcite ; @xmath242 with @xmath128 in quasar - mode and @xmath243 in radio - mode @xcite ; @xmath244 @xcite ; @xmath244 in quasar - mode and @xmath245 in radio - mode @xcite .
using such high values of @xmath77 , on the other hand , leads to the overheating of the cool core ( fig .
[ fig - evol_gas_star_bh ] ) , which is in contrast to x - ray observations depicting a tight self - regulated balance between heating and cooling ( e.g. , preserving the observed positive x - ray temperature profile gradient for several gyr ) .
for instance , @xcite find , in high - resolution simulations of isolated clusters , that to quench cooling flows , at the same time avoiding both overheating and overcooling , the typical mechanical efficiency @xmath246 for poor clusters is @xmath247 .
@xcite saw that agn feedback is able to suppress cooling / sf using a feedback efficiency as low as @xmath248 , and the simulations show cold gas and jet cycles even after several gyr . factoring out the @xmath75 ,
these efficiencies translate to @xmath249 , i.e. , lower than that typically used in cosmological simulations .
such low @xmath77 can be analytically retrieved by equating the observed jet power to the core x - ray luminosity .
observationally , @xcite solved the continuity equation for the black hole mass function using the locally detected one as boundary condition , and the hard x - ray luminosity function of the agn growth rate distribution , and found that the kinetic efficiency is a few @xmath250 .
our choice of default @xmath125 in this work is motivated by cosmological simulations , which uses similar order values , also we have explored @xmath127 and @xmath112 to cover the parameter space . supernovae - driven kinetic feedback is not included in our simulations , so that the central smbh is the sole source of energy feedback , and to unambiguously infer any outflow as driven by agn . in most of our second - series runs with cooling , sf and bh growth , the cool core is heated up by strong agn feedback .
the influence of supernovae feedback is expected to be sub - dominant at the cluster center .
the more efficient kinetic feedback that we find here compared to thermal feedback ( used in all our cluster simulations before ) is promising in helping to solve problems pointed out by @xcite , in particular the still too large mass and too big effective radius of the brightest cluster galaxies .
we investigate different models of agn feedback in cluster simulations , using the sph code gadget-3 .
we implement novel methods to couple the feedback energy from a smbh in the _ kinetic _ form , where the velocity of the neighbouring gas is incremented .
gas particles lying inside a bi - conical volume around the bh ( cones of slant height @xmath86 and opening angle @xmath91 along two opposite directions ) are stochastically selected and imparted a one - time @xmath47 velocity boost .
this renders our code able to generate outflows driven by agn feedback in a self - consistent way .
we perform simulations of isolated cluster of total mass @xmath1 , containing @xmath251 gas particles , using a gravitational softening length of @xmath252 kpc .
the cluster is first evolved using hydrodynamical interactions only for @xmath39 gyr , subsequently using cooling and hydrodynamics for further @xmath39 gyr , which forms a dense cool core , which is taken as our initial condition .
a collisionless bh particle is created at this time ( renamed @xmath44 ) , residing at the cluster center , having a seed bh mass of @xmath193 .
our simulations form two broad series , each having the same non - agn sub - resolution physics , and the runs in it explore different agn feedback models .
the first series includes simulations with cooling - only , no - sf , constant energy feedback from bh at fixed duty - cycle ; the second series includes cooling , sf , bh undergoing gas accretion , mass growth and resulting feedback .
in particular , five of our runs in the latter series employ the `` unified '' agn model presented in @xcite . here , we use the mechanical outflow power as the source of energy for our kinetic feedback , and _ build an improved form of unified agn feedback model_. our main results are : * bipolar bubble - like outflows form , originating from the bh , and propagating radially outward to a distance of a few @xmath3 kpc .
the cluster cool core is shock heated in several runs , to a median temperature @xmath170 k , which corresponds to a central entropy flattening , @xmath171 kev @xmath172 .
a large outflow originates during each agn - on active period . *
the output bh power @xmath51 erg / s heats up the cool core in the runs where the kicked particles are always hydrodynamically coupled .
wind particles decoupled from hydrodynamics do not have any heating effect at the center , unless the power is increased by a factor of @xmath12 .
bh feedback is less efficient in two runs ( one with a lower bh power ) where the initial cool core remains . *
when allowing the bh to grow , the bh accretion rate has a power - law increase , to reach a peak of few times @xmath253/yr ( values depending on the feedback model ) .
after the peak , the @xmath184 reduces @xmath254 times in the runs with kinetic feedback and remains almost constant with thermal feedback . * the total star formation rate increases with time initially ; then the growing bh accretes in , heats up , and ejects out some central gas , quenching sf .
the sfr is reduced @xmath143 times with thermal , and @xmath3 times by kinetic feedback . _
thus , kinetic feedback is more efficient in quenching the sfr . _
the same is true for the inner cool gas mass . *
kinetic feedback is more effective and has a stronger impact on the cluster central gas than thermal feedback using the same feedback efficiency , within our implementation of star - formation model . *
the inclusion of _ cold gas accretion _ in the simulations with bh growth shows huge variability of the @xmath184 and inner cool / hot gas masses , thus _
produces naturally a duty cycle of the agn with a periodicity of @xmath3 myr_. * the unified feedback model , where we use our kinetic scheme to distribute the mechanical agn power , is basically as good as the pure kinetic one in quenching the sfr , heating the cool core and raising its central entropy level .
we emphasize again that the same result can not be obtained using purely thermal feedback .
in this paper we have presented an implementation of agn feedback in the kinetic form in an sph code .
the results presented here further demonstrate the relevance of explicitly describing the mechanical nature of the radio - mode agn feedback to create a self - regulated cooling - heating cycle in the core regions of isolated halos having the size of a galaxy cluster .
we consider this as quite encouraging in view of the forthcoming application of our implementation of mechanical agn feedback in cosmological simulations .
we are most grateful to volker springel for allowing us to use the gadget-3 code .
we thank lisa steinborn , klaus dolag , and alexander beck for useful discussions .
simulations were run at the cineca supercomputing center , with cpu time assigned through an iscra class - c proposal , and through an inaf - cineca grant .
data postprocessing and storage has been done on the cineca facility pico , granted to us thanks to our expression of interest .
this work is supported by the prin - miur 201278x4fl `` evolution of cosmic baryons '' funded by the italian ministry of research , by the prin - inaf 2012 grant `` the universe in a box : multi - scale simulations of cosmic structures '' , by the indark infn grant , and by consorzio per la fisica di trieste .
m.g . is supported by nasa through einstein postdoctoral fellowship award number pf-160137 issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory for and on behalf of nasa under contract nas8 - 03060 . | we implement novel numerical models of agn feedback in the sph code gadget-3 , where the energy from a supermassive black hole ( bh ) is coupled to the surrounding gas in the _ kinetic _ form .
gas particles lying inside a bi - conical volume around the bh are imparted a one - time velocity ( @xmath0 km / s ) increment .
we perform hydrodynamical simulations of isolated cluster ( total mass @xmath1 ) , which is initially evolved to form a dense cool core , having central @xmath2 k. a bh resides at the cluster center , and ejects energy .
the feedback - driven fast wind undergoes shock with the slower - moving gas , which causes the imparted kinetic energy to be thermalized .
bipolar bubble - like outflows form propagating radially outward to a distance of a few @xmath3 kpc .
the radial profiles of median gas properties are influenced by bh feedback in the inner regions ( @xmath4 kpc ) .
bh kinetic feedback , with a large value of the feedback efficiency , depletes the inner cool gas and reduces the hot gas content , such that the initial cool core of the cluster is heated up within a time @xmath5 gyr , whereby the core median temperature rises to above @xmath6 k , and the central entropy flattens .
our implementation of bh thermal feedback ( using the same efficiency as kinetic ) , within the star - formation model , can not do this heating , where the cool core remains .
the inclusion of cold gas accretion in the simulations produces naturally a duty cycle of the agn with a periodicity of @xmath3 myr .
[ firstpage ] cosmology : theory galaxies : clusters : general galaxies : jets ( galaxies : ) cooling flows black hole physics |
as is well - known unitarity triangle fits indicate the prominent role of the cabbibo - kobayashi - maskawa ( ckm ) mechanism@xcite in cp violation and flavour physics . the huge amount and the variety of cp collected data allows for a systematic search of physics beyond the standard model ( sm)@xcite@xcite .
a quite general and natural framework to go beyond the sm is : 1 .
allow for new physics in every place except in weak tree - level dominated processes .
2 . assume @xmath12 unitarity .
this framework is general enough to include practically all models of new physics ( np ) except those that explicitly violate @xmath12 unitarity .
going beyond @xmath12 unitarity makes the analysis almost impossible in a model independent way@xcite .
we will proceed in the first part with assumptions 1 and 2 . to go on with this analysis
it is very important to understand the key role played by a measurement of the unitarity triangle phase @xmath7 .
it is very convenient to write the ckm matrix in the following way@xcite@xcite:@xmath16where the rephasing invariant cp violating phases are@xcite@xmath17it has been shown from @xmath12 unitarity -@xmath18- that @xmath19 and @xmath20 even outside the sm@xcite .
so @xmath21 is too small to consider its measurement . @xmath22 and @xmath10 , accompanying @xmath23 and @xmath24 respectively , will enter only in loop processes : virtual transitions @xmath25 .
therefore processes that in the sm measure @xmath22 and @xmath10 could be contaminated by np .
_ processes that measure _ @xmath22 _ in the sm framework will be no longer evidence of cp violation in the ckm matrix in the presence of np_. the unique measurable phase that can be extracted from tree - level processes and therefore not contaminated by np is @xmath7 , because it can appear in a tree - level transition @xmath26 npbotella . we can conclude that just @xmath7 and the moduli of the first two rows are the unique parameters whose extraction from experimental data is not contaminated by np . or put in another way , these parameters -extracted from weak tree - level decays- are valid in all the models included in our assumptions 1 and 2 . in section
ii we will parametrize and clarify the np physics contributions to the contaminated flavour and cp observables . having seven experimental data of the ckm matrix in these general class of models , in section iii
we will address the question of the dominance of the ckm mechanism as the origin of cp violation in these models . at the same time , with the contaminated observables , we will set bounds on the np contributions . finally in section iv we will study potential large deviations of @xmath10 from its sm value in models that violates @xmath12 unitarity .
as we have explained the measurements of @xmath22 will be contaminated by np .
also it is known that @xmath22 measurements come always from the @xmath22 contribution to @xmath29 mixing .
the simple reason why @xmath22 contributions to @xmath14 decay amplitudes can not be measured - in a clean theoretical way- is because it always enter in the sm with a second weak phase .
therefore it is enough to parametrize the np contributions to the @xmath29 mixing .
we follow reference branco99 to write the off diagonal mixing matrix as@xmath30the sm corresponds to @xmath31 and @xmath32 . in this way we get@xmath33and therefore , by defining@xmath34we have for the experimental observables@xmath35 clearly an independent knowledge of the full ckm matrix together with the @xmath36 asymmetry @xmath37 and the mass difference @xmath2 will give us the opportunity to test for the np parameters @xmath38 and @xmath39 . as we will see in the next section ,
the actual knowledge of the moduli of the ckm matrix in the first two rows is not enough to know the entire ckm matrix , so the measurement of gamma is of paramount importance in order to complete the knowledge of the ckm matrix and to proceed with the np analysis .
gamma can be obtained from the phase of the rephasing invariant quartet @xmath40 @xcite .
a way of measuring it from pure tree - level decays is trough the interference of the two pure charged currents decay paths @xmath41 and @xmath42 @xcite@xcite .
babarbabar1 and belle@xcite have presented results using the dalitz plot analysis in @xmath43 with the subsequent decay @xmath44 . if any potential np in @xmath45 mixing is neglected@xcite , as seems reasonable , these analysis provide a measurement of gamma free from np contributions .
other relevant ways of measuring @xmath7 are the methods to measure @xmath46 @xcite .
by now it is well - known that using eq.([phases1 ] ) one has @xmath47 by definitionsilva04@xcite .
so a measurement of @xmath48 is nothing else than a measurement of @xmath49 .
the main channels are @xmath50 and @xmath51 . in this case
, the presence of penguin pollution could become np pollution , so one has to be much more careful in these processes .
the relevant observables are@xmath52to understand the effects of np let us first neglect the sm penguin pollution and treat the @xmath53 channel as if it were a pure tree level and therefore without np pollution in the decay amplitudes .
in these case we have in the @xmath54 channel@xmath55where @xmath56 .
the cp asymmetry measures @xmath57 .
once we include penguin pollution , the gronau and london isospin analysis can be partially summarized with the following formula@xcite@xcite@xmath58where @xmath59that tell us that @xmath60 can be extracted from the experimental data including branching ratios and @xmath61 . but
@xmath62 in the sm and therefore in our np scenario is a pure tree - level amplitude with weak phase gamma - is proportional to the @xmath63 piece- , so we will have that the observable @xmath60 will be@xmath64we conclude that the usual way of measuring @xmath48 in @xmath53 decays provides us with a measurement of @xmath65 even in the presence of np in the @xmath4 piece@xcite . obviously
the knowledge of @xmath66 from @xmath67 converts these @xmath48 methods in another way of extracting @xmath7 from tree - level pieces .
similar results are obtained for @xmath68 and @xmath69 .
within this class of models , in order to investigate whether the present experimental data already implies that ckm is complex , one has to check whether any of the unitarity triangles is constrained by data to be non - flat , i.e. to have a non - vanishing area .
if any one of the triangles does not collapse to a line , no other triangle will collapse , due to the remarkable property that all the unitarity triangles have the same area .
this property simply follows from unitarity of the 3@xmath83 ckm matrix .
the universal area of the unitarity triangles gives a measurement of the strength of cp violation mediated by a @xmath70-interaction and can be obtained from four independent moduli of @xmath71 .
the fact that one can infer about cp violation from the knowledge of cp - conserving quantities should not come as a surprise@xcite .
it just reflects the fact that the strength of cp violation is given by the imaginary part of a rephasing invariant quartet@xcite , @xmath72 , with @xmath73 , which in turn can be expressed in terms of moduli , thanks to 3@xmath83 unitarity . restricting ourselves to the first two rows of @xmath71 , to avoid any contamination from np , a possible choice of independent moduli would be @xmath74 , @xmath75 , @xmath76 and @xmath77 .
one can then use unitarity of the first two rows to evaluate @xmath78 , which is given , in terms of the input moduli@xcite , by @xmath79note that eq .
( [ j1 ] ) is exact , but the actual extraction of @xmath78 from the chosen input moduli , although possible in principle , it is not feasible in practice . to illustrate this point ,
let us consider the present experimental values of @xmath80 and @xmath77 , assuming gaussian probability density distributions around the central values .
we plot in fig.([jpdf ] ) the probability density distribution of @xmath81 , generated using a toy monte carlo calculation .
only @xmath82 of the generated points satisfy the trivial normalization constraints and , among those , only @xmath83 satisfy the condition that the unitarity triangles close @xmath84 . as we have mentioned , including the experimental data @xmath37 and @xmath85 still does not give evidence of a complex @xmath71
. we can not conclude if @xmath71 is the dominant contribution to cp violation , although we can set bounds on the np parameters @xmath38 and @xmath39 . in fig.([gamma0 ] ) we plot 68% ( black ) , 90% ( dark grey ) and 95% ( grey ) probability regions of the probability density function ( pdf ) of the apex @xmath86 of the @xmath87 unitarity triangle . in fig.([rfi0 ] ) we represent joint pdf regions in the plane @xmath88 , @xmath89 . because @xmath7 gives the apex of the triangle , it is clear from fig.([gamma0 ] ) that there is essentially no restriction on @xmath90 . on the contrary , because the moduli of the first two rows put an upper bound on @xmath91 and upper and lower bounds on @xmath92 , we can see in fig.([rfi0 ] ) significant constraints on @xmath93 and @xmath94 .
including the measurement of gamma in fig.([gamma0 ] ) will fix the unitarity triangle and the @xmath71 matrix .
the enormous effort developed at the b - factories belle and babar has resulted in the first measurements of @xmath7 in tree - level decays @xmath43 , @xmath95 , where the two paths to @xmath96 or @xmath97 interfere in the common decay channel @xmath98 . from a dalitz - plot analysis ,
belle @xcite has presented @xmath99 and babar @xcite @xmath100 , together with the solutions obtained by changing @xmath101 .
we average conservatively both measurements to the value @xmath102 ( @xmath103 ) , which we take as a quantitative measurement of a complex ckm matrix _ independent of the presence of np at the one - loop weak level_. babar has also presented a time - dependent analysis of the @xmath104 channel @xcite , that once supplemented with the @xmath105 and @xmath106 branching ratios @xcite and the measurement of the final polarization , can be translated @xcite into the measured value @xmath107 where the last error comes from the usual @xmath108 isospin bounds is the usual @xmath109 but where we have introduced @xmath66 instead of @xmath22 as the phase in the @xmath110@xmath111 mixing . ]
because the measurement is sensitive to @xmath113 , @xmath114 presents a fourfold ambiguity @xmath115 . in the @xmath116 channel the pentagon isospin analysis from quasi - two - body decays needs more statistics and/or additional assumptions .
a time - dependent dalitz - plot analysis in the channel @xmath117 has been presented by babar @xcite , with the result @xmath118.since this analysis is sensitive to both @xmath119 and @xmath120 , the resulting ambiguity is just a twofold one @xmath121 .
it is remarkable that these two solutions are in good agreement with two of the solutions coming from the @xmath51 channel .
this important property will be used to eliminate two of the four solutions coming from the @xmath51 channel .
the situation in the @xmath53 channel does not yet allow a full isospin analysis and the isospin bounds are quite poor . furthermore babar and
belle measurements are still in some conflict , so that we will not use these results . as before , we average the data from @xmath51 and @xmath68 but only keep the two solutions consistent with the @xmath68 channel data .
our averaged values are @xmath122 , ( @xmath123 ) . to analyze the implications for the dominance of the ckm mechanism for cp violation and the presence of np
, we add both measurements @xmath7 and @xmath124 to the previous analysis presented in figs.([gamma0 ] ) and ( [ rfi0 ] ) . in fig.([gamma1 ] ) we represent the analogue of fig.([gamma0 ] ) .
we conclude @xcite that a real ckm matrix is excluded at a 99.92% c.l .
and the probability of @xmath125\cup \lbrack -170^{\circ } ; -10^{\circ } ] $ ] is 99.7% . in fig([rfi1 ] ) we can see three solutions in the @xmath126 plane with @xmath127 and @xmath128 . in these plots
one has in general four solutions corresponding to the two values of @xmath7 ( @xmath57 ) and to the two signs of @xmath129 the last solution corresponds to @xmath130 .
it is the inclusion of both @xmath7 and @xmath57 constraints that almost eliminates the @xmath131 solutions .
the first solution is obviously the sm one , and the semileptonic asymmetry @xmath132 - not included here - is starting to play an important role@xcite in the exclusion of the solution @xmath133 .
it is important to stress that having an irrefutable piece of evidence for a complex ckm matrix , in a framework where the presence of np is allowed , has profound implications for models of cp violation . in the particular case of models with spontaneous cp violation
, a complex ckm matrix favours the class of models where , although yukawa couplings are real , the vacuum phase responsible for spontaneous cp violation also generates cp violation in charged - current weak interactions .
conversely , the evidence for a complex ckm matrix , even allowing for the presence of np , excludes the class of models with spontaneous cp violation and a real ckm matrix at a 99.92% c.l ..
within the sm and any extension where @xmath135 is unitary , like supersymmetric or multi higgs doublet models , we have the relationbotella03 @xmath136which shows that @xmath137 in any model where @xmath138 ckm unitarity holds . in particular , within
the sm one obtains at 90% cl @xmath139the only models in which @xmath10 can be significantly larger than @xmath140 are those in which @xmath135 is not unitary , what can only be achieved by enlarging the quark sector .
the most simple way of doing this is with the introduction of new quark singlets @xcite .quark singlets often arise in grand unified theories and models with extra dimensions at the electroweak scale @xcite .
they have both their left- and right - handed components transforming as singlets under @xmath141 , thus their addition to the sm particle content does not spoil the cancellation of triangle anomalies . in these models , the charged and neutral current terms of the lagrangian in the mass eigenstate basis are@xmath142where @xmath143 and @xmath144 , @xmath145 denotes the extended ckm matrix and @xmath146 , @xmath147 are hermitian matrices . @xmath148 and @xmath149 are not necessarily diagonal and thus flavour - changing neutral couplings ( fcnc ) exist at the tree level , although they are naturally suppressed by the ratio of the standard quark over the heavy singlet masses .
moreover , the diagonal @xmath150 couplings , which are given by the diagonal entries of @xmath148 and @xmath149 plus a charge - dependent term , are also modified . within the sm @xmath151 , @xmath152 for @xmath153 , @xmath154 and @xmath155 for @xmath153 .
the addition of up - type @xmath13 singlets modifies the first two of these equalities , while the addition of down - type @xmath156 ones modifies the last two . in models with a down quark singlet , from orthogonality of the second and third columns of @xmath145 ,
one obtains the generalization of eq.([sinchi1])@xmath157 , one obtains@xcite@xcite that at most @xmath158 , thus implying that in this class of models @xmath10 can not be significantly larger than @xmath159 . in model with an up quark singlet , from orthogonality of the second and third rows of @xmath145 , one gets@xcite@xmath160 in contrast with models containing down - type singlets , where the size of all fcnc is very restricted by experiment , present limits on @xmath161 are rather weak .
the most stringent one , @xmath162 with a 95% cl , is derived from the non - observation of single top production at lep , in the process @xmath163 and its charge conjugate .
this bound does not presently provide an additional restriction on the size of @xmath10 . in models with extra up singlets
@xmath164 can be of order @xmath165 , yielding @xmath166 . from eq.([sinchi3 ] ) , one derives some important phenomenological consequences .
first , we observe that a sizeable @xmath167 is associated to a fcnc @xmath168 , which _ leads to fcnc decays _
_ at rates observable at lhc__. in addition , the modulus of @xmath161 obeys the equality @xcite@xmath169this relation shows that conditions for achieving @xmath168 are to have a small deviation @xmath170 of @xmath171 from unity ( which is allowed by the measurement of @xmath172 and @xmath173 @xcite ) and a deviation of @xmath174 from unity of order @xmath159 .
this deviation is only possible if the mass of new top quark @xmath175 is bellow @xmath176 tev , again testable at lhc .
finally the @xmath177 mass difference sets bounds on @xmath178 that for large @xmath167 - @xmath179 - translate into bounds on @xmath180 in such a way that deviations of unitarity in the first row are not observable . by the same token , large values of @xmath10
will be correlated with an important contribution to @xmath181 mixing @xcite . the shaded area in fig.([rangechi ] )
represents the allowed interval of @xmath10 for a given @xmath182 .
note that lower values of @xmath183 are allowed for lighter @xmath175 quarks@xcite .
the fig.([rangechi ] ) has been obtained incorporating all the relevant constraints : the correction to the oblique parameter @xmath184 , @xmath172 and @xmath185 in the charm sector , @xmath181 mixing and several bounds from rare @xmath186 and @xmath14 decays . because @xmath10 is the phase of @xmath187
, important effects will appear in @xmath188 transition in the piece @xmath189 , nevertheless the presence of the new @xmath175 quark will introduce a contribution proportional to @xmath190 and dependent on the mass @xmath191 of this quark .
for the cp asymmetry of @xmath192 we getaguilar@xmath193where@xmath194gets contributions from @xmath10 and the new @xmath175 quark .
@xmath195 is in general a complex function fixed by wilson coefficients and matrix elements@xcite . by scanning all the allowed range of parameters we get
@xmath196 $ ] .
less dependent on hadronic matrix elements is the contribution to @xmath197 mixing:@xmath198@xmath199 ^{2}\right .
\notag \\ & & \left .
+ \left [ \frac{s(m_{t},m_{t})}{s(m_{t},m_{t})}-\left ( \frac{% s(m_{t},m_{t})}{s(m_{t},m_{t})}\right ) ^{2}\right ] \left ( \frac{v_{ts}^{\ast } v_{tb}}{v_{ts}^{\ast } v_{tb}}\right ) ^{2}\right\ } \label{chieffec1}\end{aligned}\]]with @xmath186 a constant factor and @xmath200 the usual inami - lim box functioninamlim . in any channel without a weak phase in the decay amplitude , for example in the @xmath201 and @xmath202 channels , the time dependent cp asymmetry is@xmath203again @xmath204 is equal to @xmath10 plus a @xmath175 dependent contribution .
the range of both contributions goes in opposite direction as @xmath191 changes . in fig.([sdd ] ) , we present the range of variation of @xmath205 in terms of @xmath206 . a potential spectacular departure from the sm
could be seen at lch .
the first measurements of @xmath7 clearly points towards the ckm mechanism as the dominant source of cp violation in the quark sector , even in the presence of np in all loops .
a real ckm matrix is excluded at the 99.92% c.l . including @xmath57 data , there are still alive three solutions : the most robust is the sm , one with @xmath207 is almost excluded and it remains another np solution where the semileptonic asymmetry can do a relevant job .
future improvements of the data will be crucial to left just very small deviations of the sm .
a large deviation of the sm value of the phase @xmath10 is only possible in models that violates @xmath12 unitarity .
more precisely , in models with an up singlet quark , @xmath10 can be order @xmath11 .
moderates effects appear in @xmath192 , more spectacular effects can be present in cp asymmetries in the @xmath208 system .
correlated effects of this scenario would be : rare top decays @xmath15 at a rate observable at lhc , production of a heavy top @xmath175 at lhc and important deviations from 1 of @xmath174 , measurable at ilc .
the present work has been done in different collaborations with j.a .
aguilar - saavedra , g.c .
branco , m. nebot , m.n .
rebelo , and j.p .
my acknowledgment to all of them and special greetings to gustavo branco in his 60th birthday .
this work has been partially founded by mec of spain under fpa2002 - 00612 .
utfit , m.bona _ et al .
_ , hep - ph/0501199 , ` http://www.utfit.org ` .
ckmfitter group , j. charles _ et al .
_ , hep - ph/0406184 , ` http://ckmfitter.in2p3.fr ` .
a. j. buras , f. parodi , and a. stocchi , jhep * 01 * , 029 ( 2003 ) , hep - ph/0207101 .
m. gronau and d. london , phys .
b253 * , 483 ( 1991 ) .
m. gronau and d. wyler , phys .
lett . * b265 * , 172 ( 1991 ) .
i. dunietz , phys . lett . * b270 * , 75 ( 1991 ) .
r. aleksan , i. dunietz , and b. kayser , z. phys . * c54 * , 653 ( 1992).d .
atwood , g. eilam , m. gronau , and a. soni , phys .
b341 * , 372 ( 1995 ) , hep - ph/9409229 .
d. atwood , i. dunietz , and a. soni , phys .
lett . * 78 * , 3257 ( 1997 ) , hep - ph/9612433 .
a. bondar and t. gershon , phys
* d70 * , 091503 ( 2004 ) , hep - ph/0409281 .
a. amorim , m. g. santos , and j. p. silva , phys .
* d59 * , 056001 ( 1999 ) , hep - ph/9807364 . c. c. meca and j. p. silva , phys .
lett . * 81 * , 1377 ( 1998 ) , hep - ph/9807320 . j. p. silva and a. soffer , phys . rev . *
d61 * , 112001 ( 2000 ) , hep - ph/9912242 .
m. gronau and d. london , phys .
lett . * 65 * , 3381 ( 1990 ) .
y. grossman and h. r. quinn , phys .
rev . * d58 * , 017504 ( 1998 ) , hep - ph/9712306 .
j. charles , phys . rev . * d59 * , 054007 ( 1999 ) , hep - ph/9806468 .
m. gronau , d. london , n. sinha , and r. sinha , phys .
b514 * , 315 ( 2001 ) , hep - ph/0105308 .
p. langacker and d. london , phys .
d * 38 * , 886 ( 1988 ) ; d. london , hep - ph/9303290 ; r. barbieri and l. j. hall , nucl .
b * 319 * , 1 ( 1989 ) ; y. nir and d. j. silverman , phys .
d * 42 * , 1477 ( 1990 ) ; g. c. branco , t. morozumi , p. a. parada and m. n. rebelo , phys .
d * 48 * ( 1993 ) 1167 ; e. nardi , e. roulet and d. tommasini , nucl .
b * 386 * , 239 ( 1992 ) ; v. d. barger , m. s. berger and r. j. phillips , phys .
d * 52 * , 1663 ; ( 1995 ) f. del aguila , j. a. aguilar - saavedra and g. c. branco , nucl .
b510 * , 39 ( 1998 ) ; g. barenboim and f. j. botella , phys . lett . * b433 * , 385 ( 1998 ) ; p. h. frampton , p. q. hung and m. sher , phys .
rept . * 330 * , 263 ( 2000 ) | we analyse the present experimental evidence for a complex ckm matrix , even allowing for new physics contributions to @xmath0 , @xmath1 , @xmath2 , @xmath3 , and the @xmath4 piece of @xmath5 and @xmath6 .
we emphasize the crucial rle played by the angle @xmath7 in both providing irrefutable evidence for a 3@xmath83 complex ckm matrix and placing constraints on the size of np contributions .
it is shown that even if one allows for new physics a real ckm matrix is excluded at a 99.92% c.l . ,
and the probability for the phase @xmath7 to be in the interval @xmath9\cup \lbrack 10^{\circ } ; 170^{\circ } ] $ ] is 99.7% .
large value of the phase @xmath10 , e.g. of order @xmath11 , is only possible in models where the unitarity of the @xmath12 cabibbo - kobayashi - maskawa matrix is violated through the introduction of extra @xmath13 quarks .
we study the allowed range for @xmath10 and the effect of a large @xmath10 on various low - energy observables , such as cp asymmetries in @xmath14 meson decays .
we also discuss the correlated effects which would be observable at high energy colliders , like decays @xmath15 , etc .. |
let @xmath0 denote @xmath1 letters of a process where each letter is drawn from a finite set @xmath2 with joint probability mass function ( pmf ) @xmath3 .
let @xmath4 be a realization and suppose that we wish to guess this realization by asking questions of the form `` is @xmath5 ? '' , stepping through the elements of @xmath6 until the answer is `` yes '' . we wish to do this using the minimum expected number of guesses .
there are several applications that motivate this problem .
consider cipher systems employed in digital television or dvds to block unauthorized access to special features .
the ciphers used are amenable to such exhaustive guessing attacks and it is of interest to quantify the effort needed by an attacker ( merhav & arikan @xcite ) .
massey @xcite observed that the expected number of guesses is minimized by guessing in the decreasing order of @xmath7-probabilities .
define the _ guessing function _
@xmath8 to be one such optimal guessing order .
@xmath9 implies that @xmath4 is the @xmath10th guess .
arikan @xcite considered the growth of @xmath11 $ ] as a function of @xmath1 for an independent and identically distributed ( iid ) source with marginal pmf @xmath12 and @xmath13 .
he showed that the growth is exponential in @xmath1 ; the limiting exponent @xmath14\ ] ] exists and equals @xmath15 with @xmath16 , where @xmath17 is the rnyi entropy of order @xmath18 for the pmf @xmath7 , given by @xmath19 malone & sullivan @xcite showed that the limiting exponent @xmath20 of an irreducible markov chain exists and equals the logarithm of the _ perron - frobenius eigenvalue _ of a matrix formed by raising each element of the transition probability matrix to the power @xmath18 . from their proof , one obtains the more general result that the limiting exponent exists for any source if the rnyi entropy _ rate _ of order @xmath18 , @xmath21 exists for @xmath16 .
pfister & sullivan @xcite showed the existence of ( [ eqn : limitingexponent ] ) for a class of stationary probability measures , beyond markov measures , that are supported on proper subshifts of @xmath22 @xcite .
a particular example is that of shifts generated by finite - state machines .
for such a class , they showed that the guessing exponent has a variational characterization ( see ( [ eqn : erho - iid ] ) later ) . for unifilar sources
sundaresan @xcite obtained a simplification of this variational characterization using a direct approach and the method of types .
merhav & arikan remark that their proof in @xcite for the limiting guessing exponent is equally applicable to finding the limiting exponent of the moment generating function of compression lengths . moreover , the two exponents are the same .
the latter is a problem studied by campbell @xcite .
our contribution is to give a large deviations perspective to these results , shed further light on the aforementioned connection between compression and guessing , and unify all prior results on existence of limiting guessing exponents .
specifically , we show that if the sequence of distributions of the _ information spectrum _
@xmath23 ( see han @xcite ) satisfies the _ large deviation property _
, then the limiting exponent exists .
this is useful because several existing large deviations results can be readily applied .
we then show that all but one previously considered cases in the literature satisfy this sufficient condition .
see examples [ example : iid]-[example : mixed - source ] in section [ sec : examples ] .
the large deviation theoretic ideas are already present in the works of pfister & sullivan @xcite and the method of types approach of arikan & merhav @xcite .
our work however brings out the essential ingredient ( the sufficient conditions on the information spectrum ) , and enables us to see the previously obtained specific results under one light . the quest for a general sufficient condition under which the information spectrum satisfies
a large deviation property is a natural line of inquiry , and one of independent interest , in view of the shannon - mcmillan - breiman theorem which asserts that the information spectrum of a stationary and ergodic source converges to the shannon entropy almost surely and in @xmath24 , for all @xmath25 ; see for example @xcite . in particular ,
the large deviation property implies exponentially fast convergence to entropy . in the several specific examples we consider ,
the information spectrum does satisfy the large deviation property .
one sufficient condition for the weaker property of exponentially fast convergence to entropy is the so - called _ blowing up property_. ( see marton & shields ( * ? ? ?
2 ) , or the survey article by shields @xcite ) .
one family of sources , that includes most of the sources we consider in this paper and goes beyond , is that of _ finitary encodings _ of memoryless processes , also called finitary processes .
these are known to have the blowing - up property , and therefore exponentially fast convergence to entropy ( see marton & shields ( * ? ? ?
it is an interesting open question to see if finitary processes , or what other sources with the blowing up property , satisfy the large deviation property .
the rest of the paper is organized as follows .
section ii studies the tight relationship between guessing and compression .
section iii states the relevant large deviations results and the main sufficiency results .
section iv re - derives prior results by showing that in each case the information spectrum satisfies the ldp .
section v contains proofs and section vi contains some concluding remarks .
in this section we relate the problem of guessing to one of source compression .
an interesting conclusion is that robust source compression strategies lead to robust guessing strategies . for ease of exposition ,
let us assume that the message space is simply @xmath2 .
the extension to strings of length @xmath1 is straightforward and will be returned to shortly .
a guessing function @xmath26 is a bijection that denotes the order in which the elements of @xmath2 are guessed . if @xmath27 , then the @xmath10th guess is @xmath28 .
let @xmath29 denote the set of natural numbers .
a length function @xmath30 is one that satisfies kraft s inequality @xmath31 where we have used the notation @xmath32 . to each guessing function @xmath33 , we associate a pmf @xmath34 on @xmath2 and a length function @xmath35 as follows .
given a guessing function @xmath33 , we say @xmath34 defined by @xmath36 is the pmf on @xmath2 associated with @xmath33 .
the quantity @xmath37 in ( [ eqn : q_g ] ) is the normalization constant .
we say @xmath35 defined by @xmath38 is the length function associated with @xmath33 .
observe that @xmath39 and therefore the pmf in ( [ eqn : q_g ] ) is well - defined .
we record the intimate relationship between these associated quantities in the following result .
( this is also available in the proof of ( * ? ? ?
* th . 1 , p.382 ) ) .
[ prop : guessingbounds ] given a guessing function @xmath33 , the associated quantities satisfy @xmath40 the first equality in ( [ eqn : guessing - pmfbounds ] ) follows from the definition in ( [ eqn : q_g ] ) , and the second inequality from the fact that @xmath41 .
the upper bound in ( [ eqn : guessing - lengthbounds ] ) follows from the upper bound in ( [ eqn : guessing - pmfbounds ] ) and from ( [ eqn : l_g ] ) .
the lower bound in ( [ eqn : guessing - lengthbounds ] ) follows from @xmath42 we now associate a guessing function @xmath43 to each length function @xmath44 .
[ defn : g_l ] given a length function @xmath44 , we define the associated guessing function @xmath43 to be the one that guesses in the increasing order of @xmath44-lengths .
messages with the same @xmath44-length are ordered using an arbitrary fixed rule , say the lexicographical order on @xmath2 . we also define the associated pmf @xmath45 on @xmath2 to be @xmath46 [ prop : lengthbounds ] for a length function @xmath44 , the associated pmf and the guessing function satisfy the following : 1 .
@xmath43 guesses messages in the decreasing order of @xmath45-probabilities ;
@xmath47 the first statement is clear from the definition of @xmath43 and from ( [ eqn : q_l ] ) .
letting @xmath48 denote the indicator function of an event @xmath49 , we have as a consequence of statement 1 ) that @xmath50 which proves the left inequality in ( [ eqn : g_lbounds ] ) .
this inequality was known to wyner @xcite .
the last inequality in ( [ eqn : g_lbounds ] ) follows from ( [ eqn : q_l ] ) and kraft s inequality ( [ eqn : kraftinequality ] ) as follows : @xmath51 let @xmath52 denote the set @xmath53 .
we then have the following easy to verify corollary to propositions [ prop : guessingbounds ] and [ prop : lengthbounds ] .
[ cor : inclusions ] for a given @xmath33 , its associated length function @xmath35 , and any @xmath54 , we have @xmath55 analogously , for a given @xmath44 , its associated guessing function @xmath43 , and any @xmath54 , we have @xmath56 the inequalities between the associates in ( [ eqn : guessing - lengthbounds ] ) and ( [ eqn : g_lbounds ] ) indicate the direct relationship between guessing moments and campbell s coding problem @xcite , and that the rnyi entropies are the optimal growth exponents for guessing moments , as highlighted in the following proposition .
let @xmath44 be any length function on @xmath2 , @xmath57 the guessing function associated with @xmath44 , @xmath58 a pmf on @xmath2 , @xmath59 , @xmath60 the length function that minimizes @xmath61 $ ] , where the expectation is with respect to @xmath58 , @xmath62 the guessing function that proceeds in the decreasing order of @xmath58-probabilities and therefore the one that minimizes @xmath63 $ ] , and @xmath37 as in ( [ eqn : c ] ) .
then @xmath64 } { \mathbb{e } \left [ g^*(x)^{\rho } \right ] } \leq \frac{\mathbb{e } \left [ \exp_2\{\rho l(x)\ } \right ] } { \mathbb{e } \left [ \exp_2\{\rho l^*(x)\ } \right ] } \cdot \exp_2\{\rho ( 1 + \log_2 c ) \}.\ ] ] analogously , let @xmath33 be any guessing function , and @xmath35 its associated length function . then @xmath65 } { \mathbb{e } \left [ g^*(x)^{\rho } \right ] } \geq \frac{\mathbb{e } \left [ \exp_2\{\rho l_g(x)\ } \right ] } { \mathbb{e } \left [ \exp_2\{\rho l^*(x)\ } \right ] } \cdot \exp_2\{- \rho ( 1 + \log_2 c ) \}.\ ] ] also , @xmath66 - \frac{1}{\rho } \log_2 \mathbb{e } \left [ \exp_2\{\rho l^*(x)\ } \right ] \right| \leq 1 + \log_2 c.\ ] ] observe that @xmath67 } \nonumber \\
\label{eqn:2a } & \geq & \mathbb{e } \left [ g_{l}(x)^{\rho } \right ] \\ & \geq & \mathbb{e } \left [ g^*(x)^{\rho } \right ] \nonumber \\
\label{eqn:2b } & \geq & \mathbb{e } \left [ \exp_2\{\rho l_{g^*}(x)\ } \right ] \exp_2\{-\rho ( 1 + \log_2 c ) \ } \\
\label{eqn:2c } & \geq & \mathbb{e } \left [ \exp_2\ { \rho l^*(x)\ } \right ] \exp_2\ { -\rho ( 1 + \log_2 c ) \},\end{aligned}\ ] ] where ( [ eqn:2a ] ) follows from ( [ eqn : g_lbounds ] ) , and ( [ eqn:2b ] ) from the left inequality in ( [ eqn : guessing - lengthbounds ] ) .
the result in ( [ eqn : ratioupperbound ] ) immediately follows .
a similar argument shows ( [ eqn : ratiolowerbound ] ) . finally , ( [ eqn : g*l*relation ] ) follows from the inequalities leading to ( [ eqn:2c ] ) by setting @xmath68 .
thus if we have a length function whose performance is close to optimal , then its associated guessing function is close to guessing optimal .
the converse is true as well .
moreover , the optimal guessing exponent is within @xmath69 of the optimal coding exponent for the length function .
let us now consider strings of length @xmath1 .
let @xmath6 denote the set of messages and consider @xmath70 .
let @xmath71 denote the set of pmfs on @xmath6 . by a source ,
we mean a sequence of pmfs @xmath72 , where @xmath73 .
we replace the normalization constant @xmath37 in ( 7 ) by @xmath74 and observe that @xmath75 if we normalize both sides of equation ( [ eqn : g*l*relation ] ) by @xmath1 , the difference between two quantities as a function of @xmath1 decays as @xmath76 , and vanishes as @xmath1 tends to infinity .
the following theorem follows immediately , with a change of base to natural logarithms .
[ thm : guesscodelimit ] given @xmath77 , the limit @xmath78\ ] ] exists if and only if the limit @xmath79\ ] ] exists .
furthermore , the two limits are equal .
it is therefore sufficient to restrict our attention to the campbell s coding problem @xcite and study if the limit @xmath80\ ] ] exists , where the infimum is taken over all length functions @xmath81 and exponentiation is with respect to the base of the natural logarithm .
before we proceed to studying the limit , we make a further remark on the connection between _ universal _ strategies for guessing and universal strategies for compression .
let @xmath82 denote a class of sources .
for each source in the class , let @xmath7 be its restriction to strings of length @xmath1 and let @xmath83 denote an optimal length function that attains the minimum value @xmath84 $ ] among all length functions , the expectation being with respect to @xmath7 .
on the other hand , let @xmath85 be a sequence of length functions for the class of sources that does not depend on the actual source within the class .
suppose further that the length sequence @xmath85 is asymptotically optimal , i.e. , @xmath86 } \\ & = & \lim_{n \rightarrow \infty } \frac{1}{n \rho } \ln \mathbb{e } \left [ \exp\{(\rho \ln 2 ) l_n^*(x^n)\ } \right],\end{aligned}\ ] ] for every source belonging to the class .
@xmath85 is thus `` univeral '' for ( i.e. , asymptotically optimal for all sources in ) the class .
an application of ( [ eqn : ratioupperbound ] ) with @xmath74 in place of @xmath37 followed by the observation @xmath87 shows that the sequence of guessing strategies @xmath88 is asymptotically optimal for the class , i.e. , @xmath89 } \nonumber \\
\label{eqn : growthexponent } & = & \lim_{n \rightarrow \infty } \frac{1}{n \rho } \ln \mathbb{e } \left [ g^*(x^n)^{\rho } \right ] \nonumber.\end{aligned}\ ] ] arikan and merhav @xcite provide a universal guessing strategy for the class of discrete memoryless sources ( dms ) . for the class of unifilar sources with a known number of states , the minimum description length encoding is asymptotically optimal for campbell s coding length problem ( see merhav @xcite ) .
it follows as a consequence of the above argument that guessing in the increasing order of description lengths is asymptotically optimal .
the left side of ( [ eqn : ratioupperbound ] ) is the extra factor in the expected number of guesses ( relative to the optimal value ) due to lack of knowledge of the specific source in class .
sundaresan @xcite characterized this loss as a function of the uncertainty class .
we begin with some words on notation . recall that @xmath71 denotes the set of pmfs on @xmath6 .
the shannon entropy for a @xmath73 is @xmath90 and the rnyi entropy of order @xmath91 is ( [ eqn : re ] ) .
the kullback - leibler divergence or relative entropy between two pmfs @xmath92 and @xmath7 is @xmath93 where @xmath94 means @xmath92 is absolutely continuous with respect to @xmath7 .
recall that a source is a sequence of pmfs @xmath72 where @xmath73 .
it is usually obtained via @xmath1-length marginals of some probability measure in @xmath95 .
also recall the definitions of limiting guessing exponent in ( [ eqn : limitingexponent ] ) and rnyi entropy rate in ( [ eqn : rer ] ) when the limits exist .
@xmath96 is an optimal guessing function for a pmf @xmath73 . from the results in section [ sec : guesscodeequivalence ] on the equivalence between guessing and compression , it is sufficient to focus on the campbell coding problem .
our first contribution is a proof of the following implicit result of malone & sullivan @xcite . the proof is given in section [ subsec : prop1 ] .
[ prop : renyi - guessing - equivalence ] let @xmath13 . for a source @xmath72 ,
@xmath20 exists if and only if the rnyi entropy rate ( [ eqn : rer ] ) exists .
furthermore , @xmath97 equals the rnyi entropy rate .
the question now boils down to the existence of the limit in the definition of rnyi entropy rate .
the theory of large deviations immediately yields a sufficient condition .
we begin with a definition .
ii.3.1 ) [ defn : ldp ] a sequence @xmath98 of probability measures on @xmath99 satisfies the _ large deviation property ( ldp ) _ with rate function @xmath100 $ ] if the following conditions hold : * @xmath101 is lower semicontinuous on @xmath99 ; * @xmath101 has compact level sets ; * @xmath102 for each closed subset @xmath103 of @xmath99 ; * @xmath104 for each open set @xmath33 of @xmath99 .
several commonly encountered sources satisfy the ldp with known and well - studied rate functions .
we describe some of these in the examples treated subsequently .
let @xmath105 denote the distribution of the information spectrum given by the real - valued random variable @xmath106 .
the following proposition gives a sufficient condition for the existence of the limiting rnyi entropy rate ( and therefore the limiting guessing exponent ) .
[ prop : ldpsufficientcondition ] let the sequence of distributions @xmath98 of the information spectrum satisfy the ldp with rate function @xmath101
. then the limiting rnyi entropy rate of
order @xmath107 exists for all @xmath13 and equals @xmath108 where @xmath109 .
consequently , the limiting guessing exponent exists and equals @xmath110 the function @xmath111 is the legendre - fenchel dual of the rate function @xmath101 .
proposition [ prop : ldpsufficientcondition ] says that , under the sufficient condition , the limiting guessing exponent equals @xmath112 , and is thus directly related to the large deviations rate function for information spectrum .
this is however different from merhav & arikan s ( * ? ? ?
2 ) for memoryless sources which states that the limiting guessing exponent is the legendre - fenchel dual of the source coding _
error exponent _ function .
we refer the reader to merhav and arikan ( * ? ? ?
iv ) for further interesting connections between source coding error exponent , guessing exponent , and two other exponents related to lossy source coding .
let us briefly discuss another approach to verify the existence of rnyi entropy rate ( see proposition [ prop : renyi - guessing - equivalence ] ) . with @xmath16 , we can rewrite @xmath113 times the rnyi entropy rate in ( [ eqn : rer ] ) as @xmath114 where @xmath115 and @xmath116 is the iid process on @xmath117 with uniform marginal on @xmath2 .
one can then view @xmath118 as the inverse temperature ( when @xmath13 ) of a statistical mechanical system , @xmath119 as the energy of the configuration @xmath4 , and the right side of ( [ eqn : normalizedpartition ] ) as a scaled version of ( i.e. , @xmath18 times ) the specific gibbs free energy of the corresponding statistical mechanical system , if the limit exists .
this view point is particularly useful because the iid process @xmath116 satisfies a sample path large deviation property . if the information spectrum sequence satisfies the continuity conditions in varadhan ( * ? ? ?
3.4 ) , then the limiting specific gibbs free energy exists , and so does the rnyi entropy rate .
our technical report @xcite treats an example via this more general approach . in order to study the examples in section [ sec : examples ] , we state some additional results on ldp of transformed variables .
( see ( * ? ? ?
* sec . 4.2 ) ) , ( * ? ? ?
* th . 6.12 and 6.14 ) ) .
[ prop : contractionprinciple ] let @xmath120 denote a sequence of @xmath121-valued random variables where @xmath121 is a complete separable metric space ( polish space ) .
let @xmath105 denote the distribution of @xmath122 for @xmath123 , and let the sequence of distributions @xmath98 on @xmath121 satisfy the ldp with rate function @xmath124 $ ] .
let @xmath125 be a continuous function .
the sequence of distributions of @xmath126 on @xmath99 also satisfies the ldp with rate function @xmath127 $ ] given by @xmath128 [ prop : exponential approximation ] suppose that the sequence of distributions of @xmath120 satisfies the ldp with rate function @xmath101 on @xmath99 .
assume also that the sequence of random variables @xmath129 is superexponentially close to @xmath120 in the following sense : for each @xmath130 @xmath131 then the sequence of distributions of @xmath129 also satisfies the ldp on @xmath99 with the same rate function @xmath101 .
the condition in ( [ eqn : superexponential ] ) is satisfied if @xmath132 where @xmath133 is the underlying sample space .
we are now ready to apply proposition 7 and related techniques to various examples . in first five examples that follow ,
our goal is to show that the sufficient condition for the existence of the limiting guessing exponent holds , i.e. , that the sequence of distributions of the information spectrum satisfies the ldp .
[ example : iid ] this example was first studied by arikan @xcite . recall that an iid source is one for which @xmath134 , where @xmath12 is the marginal of @xmath135 .
it is then clear that the information spectrum can be written as a sample mean of iid random variables @xmath136 it is well - known that the sequence @xmath98 of distributions of this sample mean satisfies the ldp with rate function given by the legendre - fenchel dual of the cumulant of the random variable @xmath137 ( see for example ( * ? ? ?
ii.4.1 ) or ( * ? ? ?
( 1.9.66 - 67 ) ) ) : @xmath138 & = & \ln \left ( \sum_{x \in \mathbb{x } } p_1(x)^{\alpha } \right ) \\ & = & ( 1 - \alpha ) h_{\alpha}(p_1).\end{aligned}\ ] ] the legendre - fenchel dual of the rate function is therefore the cumulant itself ( ( * ? ? ?
vi.4.1.e ) ) .
an application of proposition [ prop : ldpsufficientcondition ] yields that @xmath139 times this cumulant , given by @xmath15 , is the guessing exponent .
we thus recover arikan s result @xcite .
the rate function @xmath101 can also be obtained using the _ contraction principle _ ( proposition [ prop : contractionprinciple ] ) as follows .
this method will provide a recipe to obtain the limiting guessing exponent in subsequent examples .
consider a mapping that takes @xmath4 to its empirical pmf in @xmath140 .
empirical pmf is then a random variable .
the distribution of @xmath141 induces a pmf on @xmath140 .
it is well - known that the sequence of distributions of these empirical pmfs , indexed by @xmath1 , satisfies the _ level-2 _ ldp ) of sample means , level-2 refers to sample histograms , and level-3 to sample paths . ] with rate function @xmath142 .
see for example ( * ? ? ?
* th ii.4.3 ) . observe that the mapping from the empirical pmf to the information spectrum random variable is continuous .
we can therefore use the contraction principle to get a formula for @xmath101 in terms of @xmath143 as follows ( * ? ? ?
* th ii.5.1 ) . for any @xmath144 in @xmath99 ,
let @xmath145 i.e. , @xmath146 then @xmath147 using this , we can write @xmath148 thus yielding @xmath149 this formula extends to more general sources , as is seen in the next few examples .
[ example : markov ] this example was studied by malone & sullivan @xcite .
consider an irreducible markov chain taking values on @xmath2 with transition probability matrix @xmath150 .
our goal is to verify that the sufficient condition holds and to calculate @xmath20 defined by ( [ eqn : limitingexponent ] ) for this source .
let @xmath151 denote the set of _ stationary _ pmfs defined by @xmath152 denote the common marginal by @xmath153 and let @xmath154 we may then denote @xmath155 , where @xmath153 is the distribution of @xmath135 and @xmath156 the conditional distribution of @xmath157 given @xmath135 .
it is once again well known that the empirical pmf random variable satisfies the level-2 ldp with rate function @xmath158 , given by @xcite @xmath159 as in example [ example : iid ] , the contraction principle then yields that the sequence of distributions of information spectrum satisfies the ldp with rate function @xmath101 given by @xmath160 where for @xmath144 in @xmath99 , @xmath161 is defined by @xmath162 by proposition [ prop : renyi - guessing - equivalence ] , the limiting guessing exponent exists . perron - frobenius theory ( seneta ( * ? ? ?
1 ) , see also @xcite ) yields the cumulant directly as @xmath163 , where @xmath164 is unique largest eigenvalue ( perron - frobenius eigenvalue ) of a matrix formed by raising each element of @xmath150 to the power @xmath18 .
( recall that @xmath16 and @xmath165 ) .
thus @xmath166 , and we recover the result of malone & sullivan @xcite . it is useful to note that the steps that led to ( [ eqn : erho - iid ] ) hold in the markov case ( with appropriate changes to entropy and divergence terms ) and we may write @xmath167 where @xmath168 is the conditional entropy of @xmath157 given @xmath135 under the joint distribution @xmath169 , i.e. , @xmath170 [ example : unifilar ] this example was studied by sundaresan in @xcite .
a unifilar source is a generalization of the markov source in example [ example : markov ] .
let @xmath2 denote the alphabet set as before .
in addition , let @xmath171 denote a set of finite states .
fix an initial state @xmath172 and let the joint probability of observing @xmath173 be @xmath174 where @xmath175 is the joint probability of @xmath176 given the previous state @xmath177 .
the dependence of @xmath7 on @xmath172 is understood .
furthermore , assume that @xmath175 is such that @xmath178 , where @xmath179 is a deterministic function that is one - to - one for each fixed @xmath177 .
such a source is called a unifilar source .
@xmath180 and @xmath179 completely specify the process : the initial state @xmath181 is random with distribution that of marginal of @xmath182 in @xmath183 , the rest being specified by @xmath184 and @xmath179 .
example [ example : markov ] is a unifilar source with @xmath185 , @xmath186 , and @xmath187 where @xmath153 is the stationary distribution of the markov chain .
let @xmath188 denote the set of joint measures on the indicated space so that the resulting process @xmath189 is a stationary and irreducible markov chain .
let a @xmath190 be written as @xmath155 .
for any @xmath144 in @xmath99 , let @xmath191 then the sequence of distributions of information spectrum @xmath192 satisfies the ldp ( ( * ? ? ? * eqn .
( 1.9.30 ) ) ) with rate function given ( once again via contraction principle ) by @xmath193 the limiting exponent therefore exists .
following the same procedure that led to ( [ eqn : erho - iid ] ) in the iid case and ( [ eqn : erho - markov ] ) for a markov source , we get @xmath194 where @xmath168 and @xmath195 are analogously defined , and the result of sundaresan @xcite is recovered .
[ example : stationary ] pfister & sullivan @xcite considered a class of stationary sources with distribution @xmath196 that satisfies two hypotheses h1 and h2 of ( * ? ? ?
ii - b ) , which we will now describe . let @xmath197 denote the set of sources that satisfy @xmath94 for all @xmath123 , where @xmath7 and @xmath92 are restrictions of @xmath58 and @xmath169 to @xmath1 letters .
note that it may be possible that a @xmath198 is not absolutely continuous with respect to @xmath58 .
also , let @xmath199 denote the subset of stationary sources with respect to the shift operator @xmath200 defined by @xmath201 hypothesis h1 of pfister & sullivan @xcite assumes that for any neighborhood of a stationary source @xmath202 and any @xmath203 , there exists an ergodic @xmath204 in that neighborhood such that @xmath205 , where @xmath206 is the shannon entropy rate of source @xmath169 .
their hypothesis h2 is given by ( [ eqn : expodifference ] ) below .
under these hypotheses , pfister & sullivan @xcite proved that @xmath20 exists , and provided a variational characterization analogous to ( [ eqn : erho - unifilar ] ) , i.e. , @xmath207 where @xmath208 en route to this result , pfister & sullivan @xcite showed that the sequence of distributions of the _ empirical process _ satisfies the _ level-3 _ ldp for sample paths .
we first state this precisely , and then use this as the starting point to show the sufficient condition that the information spectrum satisfies the ldp . for an @xmath209 given by @xmath210 ,
we define @xmath211 as the first @xmath1 components of @xmath28 in the usual way .
consider a stationary source @xmath58 whose letters are @xmath212
. define the empirical process of measures @xmath213 this is a measure on @xmath214 that puts mass @xmath215 on the following strings : @xmath216 .
pfister & sullivan showed that the distributions of the @xmath95-valued process @xmath217 satisfies the level-3 ldp with rate function @xmath218 under hypotheses h1 and h2 of their paper ( ( * ? ? ? * prop .
2.2 - 2.3 ) ) . furthermore , @xmath219 so that we may restrict @xmath220 to @xmath221 .
we next use this to show that the information spectrum satisfies the ldp .
hypothesis h2 of pfister & sullivan assumes the existence of a continuous mapping @xmath222 satisfying @xmath223 where @xmath224 . by the compactness of @xmath214 , @xmath225 is uniformly continuous . under the weak topology on the complete separable metric space @xmath226 ,
the mapping @xmath227 defined by @xmath228 is a continuous mapping .
hence by the contraction principle , by setting @xmath229 we get that the sequence of distributions of @xmath230 satisfies the ldp with rate function @xmath101 given by @xmath231 where the restriction of the infimum to @xmath221 follows from ( [ eqn : restrictiontostationary ] ) .
furthermore , given hypothesis h2 and ( [ eqn : expodifference ] ) , an application of the exponential approximation principle ( proposition [ prop : exponential approximation ] ) indicates that the sequence of distributions of the information spectrum too satisfies the ldp with the same rate function @xmath101 , and we have verified that the sufficient condition holds .
what remains is to calculate this rate function . for this , we return to pfister & sullivan s work and use @xmath232 ( * ? ? ?
2.1 ) to write @xmath233 finally , the legendre - fenchel dual of the rate function is computed as in the steps leading to ( [ eqn : erho - iid])-([eqn : erho - unifilar ] ) , yielding ( [ eqn : erho - stationary ] ) .
[ example : mixed - source ] consider a mixture of two iid sources with letters from @xmath2 .
we may write @xmath234 where @xmath235 with @xmath236 the two marginal pmfs that define the iid components of the mixture .
it is easy to see that the guessing exponent is the maximum of the guessing exponents for the two component sources .
we next verify this using proposition [ prop : ldpsufficientcondition ] .
the sequence of distributions of the information spectrum satisfies the ldp with rate function given as follows ( see han ( * ? ? ?
( 1.9.41 ) ) ) .
define @xmath237 and for @xmath238 @xmath239 the rate function ( via the contraction principle ) is given by @xmath240 from proposition [ prop : ldpsufficientcondition ] we conclude that the limiting guessing exponent exists .
@xmath241 is then @xmath242 yielding @xmath243
we now prove propositions [ prop : renyi - guessing - equivalence ] and [ prop : ldpsufficientcondition ] . from theorem [ thm : guesscodelimit ]
it is sufficient to show that the limit in ( [ eqn : campbell ] ) for campbell s coding problem exists if and only if the rnyi entropy rate exists , with the former @xmath244 times the latter .
fix @xmath1 .
in the rest of the proof , we use the notation @xmath245 $ ] for expectation with respect to distribution @xmath7 .
the length function can be thought of as a bounded ( continuous ) function from @xmath6 to @xmath99 and therefore our interest is in the logarithm of its moment generating function of @xmath244 , the cumulant .
the cumulant associated with a bounded continuous function ( here @xmath85 ) has a variational characterization ( * ? ? ?
1.4.2 ) as the following legendre - fenchel dual of the kullback - leibler divergence , i.e. , @xmath246 } \nonumber \\ & = & \sup_{q_n \in \mathcal{m}(\mathbb{x}^n)}\big \ { ( \rho \ln 2 ) \mathbb{e}_{q_n}[l_n(x^n)]-d(q_n \parallel p_n)\big\}. \nonumber \\
\label{eqn : legendretransform}\end{aligned}\ ] ] taking infimum on both sides over all length functions , we arrive at the following chain of inequalities : @xmath247 } \\ & = & \displaystyle \inf_{l_n}\displaystyle \sup_{q_n \in \mathcal{m}(\mathbb{x}^n)}\big \{\mathbb{e}_{q_n}[(\rho \ln 2 ) l_n(x^n)]-d(q_n \parallel p_n)\big \ } \nonumber \\ & = & \lefteqn { \hspace*{-.1in}\sup_{q_n \in \mathcal{m}(\mathbb{x}^n ) } \inf_{l_n } \big \{\mathbb{e}_{q_n}[(\rho \ln 2 ) l_n(x^n)]-d(q_n \parallel p_n)\big \ } } \nonumber\\ \label{eqn : sup - inf } & & \hspace{2in}+ \theta(1)\\ \label{eqn : entropy } & = & \hspace*{-.1in}\sup_{q_n \in \mathcal{m}(\mathbb{x}^n ) } \big \ { \rho h_n(q_n ) - d(q_n \parallel p_n)\big \ } + \theta(1 ) \\
\label{eqn : variational } & = & \rho h_{\frac{1}{1 + \rho}}(p_n ) + \theta(1).\end{aligned}\ ] ] equation ( [ eqn : sup - inf ] ) follows because ( i ) the mapping
@xmath248-d(q_n \parallel p_n)\ ] ] is a concave function of @xmath92 ; ( ii ) for fixed @xmath92 and for any two length functions @xmath249 and @xmath250 , for any @xmath251 $ ] , the function @xmath252 is also a length function and @xmath253= \lambda \mathbb{e}_{q_n}[l_n^{(1)}]+(1-\lambda ) \mathbb{e}_{q_n}[l_n^{(2)}]+ \theta(1);\ ] ] ( iii ) @xmath71 is compact and convex , and therefore the infimum and supremum may be interchanged upon an application of a version of ky fan s minimax result @xcite .
this yields a compression problem , the infimum over @xmath85 of expected lengths with respect to a distribution @xmath92 .
the answer is the well - known shannon entropy @xmath254 to within @xmath255 nats , and ( [ eqn : entropy ] ) follows .
lastly , ( [ eqn : variational ] ) is a well - known identity which may also be obtained directly by writing the supremum term in ( [ eqn : entropy ] ) as @xmath256 \\ - ~ d(q_n \parallel p_n ) \big\}\end{aligned}\ ] ] and then applying ( [ eqn : legendretransform ] ) with @xmath257 in place of @xmath258 to get the scaled rnyi entropy .
normalize both ( [ eqn : unnormalizedcampbelleqn ] ) and ( [ eqn : variational ] ) by @xmath1 and let @xmath70 to deduce that ( [ eqn : campbell ] ) exists if and only if the limiting normalized rnyi entropy rate exists .
this concludes the proof .
this is a straightforward application of varadhan s theorem @xcite on asymptotics of integrals .
recall that @xmath105 is the distribution of the information spectrum @xmath259 .
define @xmath260 . since the @xmath98 sequence satisfies the ldp with rate function @xmath101 , varadhan s theorem ( see ellis ( * ? ? ?
ii.7.1.b ) ) states that if @xmath261 then the limit @xmath262 holds .
the integral on the left side in ( [ eqn : varadhan ] ) can be simplified by defining the finite cardinality set @xmath263 and by observing that @xmath264 take logarithms , normalize by @xmath1 , take limits , and apply ( [ eqn : varadhan ] ) to get the desired result .
it therefore remains to prove ( [ eqn : suffconditionvaradhan ] ) .
the event @xmath265 occurs if and only if @xmath266 the integral in ( [ eqn : suffconditionvaradhan ] ) can therefore be written as @xmath267 the sequence in @xmath1 on the left side of ( [ eqn : suffconditionvaradhan ] ) is then @xmath268 a constant sequence .
take the limit as @xmath269 to verify ( [ eqn : suffconditionvaradhan ] ) .
this concludes the proof .
we first showed that the problem of finding the limiting guessing exponent is equal to that of finding the limiting compression exponent under exponential costs ( campbell s coding problem ) .
we then saw that the latter limit exists if the sequence of distributions of the information spectrum satisfies the ldp ( sufficient condition ) .
the limiting exponent was the legendre - fenchel dual of the rate function , scaled by an appropriate constant .
it turned out to be the limit of the normalized cumulant of the information spectrum random variable .
while some of these facts can be gleaned from the works of pfister & sullivan @xcite and merhav & arikan @xcite , our work sheds light on the key role played by the information spectrum .
it will be of interest to find a rich class of sources beyond those listed in this paper for which the information spectrum satisfies the ldp . | the problem of guessing a random string is revisited . a close relation between guessing and compression
is first established .
then it is shown that if the sequence of distributions of the information spectrum satisfies the large deviation property with a certain rate function , then the limiting guessing exponent exists and is a scalar multiple of the legendre - fenchel dual of the rate function .
other sufficient conditions related to certain continuity properties of the information spectrum are briefly discussed .
this approach highlights the importance of the information spectrum in determining the limiting guessing exponent .
all known prior results are then re - derived as example applications of our unifying approach .
guessing , length function , source coding , information spectrum , large deviations . |
in this paper we consider the simplest of electrical networks , namely those that consist of only resistors .
the electrical properties of such a network @xmath1 are completely described by the _ response matrix _ @xmath2 , which computes the current that flows through the network when certain voltages are fixed at the boundary vertices of @xmath1 .
de verdire - gitler - vertigan @xcite and curtis - ingerman - morrow @xcite studied the _
inverse ( dirichlet - to - neumann ) problem _ for _ circular planar _ electrical networks . specifically , they considered networks embedded in a disk without crossings , with boundary vertices located on the boundary of the disk .
the following theorem summarizes their results . 1 .
any circular planar electrical network is electrically equivalent to some _ critical _ network , which is characterized by its _ medial graph _ being _ lenseless _ ( see ( * ? ? ?
* thorme 2 ) ) .
any two circular planar electrical networks having the same response matrix can be connected by simple local transformations : series - parallel , loop removal , pendant removal , and star - triangle transformations discussed in section [ ss : trans ] .
furthermore , if both networks are critical , then only star - triangle transformations are required ( see ( * ? ? ?
* thorme 4 ) or ( * ? ? ?
* theorem 1 ) ) .
the edge conductances of a critical circular planar electrical network can be recovered uniquely from the response matrix ( see ( * ? ? ?
* theorem 2 ) or ( * ? ? ?
* thorme 3 ) ) .
the response matrices realizable by circular planar networks are the ones having all _ circular minors _ nonnegative ( see ( * ? ? ? *
theorem 3 ) ) .
the space @xmath3 of response matrices of circular planar networks has a stratification by cells @xmath4 where each @xmath5 can be obtained as the set of response matrices for a fixed critical network with varying edge weights ( see ( * ? ? ?
* thorme 3 and 5 ) or ( * ? ? ?
* theorem 4 ) ) .
it is an open problem to extend these results to electrical networks embedded in surfaces with more complicated topology . in this paper
we make progress towards understanding the inverse problem for networks embedded in a cylinder .
our first main result is to construct a birational transformation we call the
_ electrical @xmath0-matrix_. this transformation acts on the edge weights of a local portion of an electrical network embedded into the cylinder , preserving all the electrical properties of the network ( corollary [ cor : resp ] ) .
furthermore , this electrical @xmath0-matrix satisfies the yang - baxter relation ( theorem [ thm : ryb ] ) , and is a close analogue of the `` geometric @xmath0-matrices '' of affine crystals , to be explained below . using the electrical @xmath0-matrix
, we formulate the following general conjecture .
a more precise version is given as conjecture [ conj : gen ] .
* any cylindrical electrical network is electrically equivalent to a critical cylindrical electrical network .
* any two cylindrical electrical networks @xmath6 and @xmath7 with the same _ universal response matrices _ are connected by local electrical equivalences .
furthermore , if @xmath6 and @xmath7 are both critical , then only star - triangle transformations , and electrical @xmath0-matrix transformations are needed . *
if a cylindrical electrical network is critical , then the edge conductances can be recovered up to the electrical @xmath0-matrix action . *
the space @xmath8 of universal response matrices of cylindrical electrical networks has an infinite stratification by @xmath9 where each @xmath10 is a semi - closed cell that can be obtained as the set of universal response matrices for a fixed critical network with varying edge weights .
the naive analogue of ( 4 ) does not hold see section [ ssec : tnn ] .
the universal response matrix in the conjecture is the response matrix of the universal cover of the cylindrical network @xmath6 . roughly speaking
, it allows us to not only measure the current flowing through the boundary vertices , but also how many times the current has winded around the cylinder . it may be possible to formulate this in a more electrically natural way by measuring magnetic fields .
thus the key difference between the planar and the cylindrical cases is that even for a critical network on a cylinder the edge conductances may not be uniquely determined from the response matrix .
this non - uniqueness comes from the existence of the electrical @xmath0-matrix , the action of which preserves both the underlying graph of the network and the response matrix , while changing the edge conductances
. the action of the electrical @xmath0-matrix can on the one hand be thought of as a galois group , and on the other hand as a monodromy group .
our second main result is to establish the above conjecture for a certain class of cylindrical networks we denote @xmath11 .
these critical networks can be thought of as the `` purely cylindrical '' networks .
there is no local configuration for which the star - triangle transformation can be applied in @xmath11 , but the electrical @xmath0-matrix generates an action of the symmetric group @xmath12 .
we show in theorem [ thm : sol ] that for the networks @xmath11 the edge conductances are recovered uniquely from the universal response matrix modulo this @xmath12 action , and in particular the inverse problem has generically @xmath13 solutions .
one way to formulate our main conjecture is that the networks @xmath11 exactly encapsulate the difference between the planar and cylindrical cases .
the proof of theorem [ thm : sol ] occupies the technical heart of the paper : we express the edge conductances as limits of certain rational functions of the universal response matrix ( theorem [ thm : main ] ) .
here we use crucially the work of kenyon and wilson @xcite .
kenyon and wilson study _
groves _ in circular planar electrical networks .
these are forests whose connected components contain specified boundary vertices .
kenyon and wilson connect ratios of grove generating functions with the response matrix of the corresponding network . by a careful choice of grove generating functions
, we can recover the desired edge conductances .
recall that a matrix is _ totally nonnegative _ if every minor of it is nonnegative .
there is a mysterious similarity between electrical networks and a different kind of networks arising in the theory of totally nonnegative matrices . in @xcite
, we presented an approach to understanding this similarity via lie theory . whereas the theory of total nonnegative is intimately related to the class of semisimple lie groups ( @xcite )
, we suggested in @xcite that a different class of `` electrical lie groups '' is related to electrical networks .
these electrical lie groups are certain deformed versions of the maximal unipotent subgroup of a semisimple lie group .
the main ideas of the present work are also motivated by this analogy , though our philosophy here is more combinatorial in nature .
the construction of the electrical @xmath0-matrix follows the techniques developed in @xcite .
there we constructed , using purely network - theoretic methods , the geometric ( or birational ) @xmath0-matrix of a tensor product of affine geometric crystals for the symmetric power representations of @xmath14 . in this paper , we use electrical networks instead of the `` totally nonnegative '' networks of @xcite , but nevertheless the underlying combinatorics is developed in parallel .
we plan to expand on this analogy in @xcite .
our formulation of conjecture [ conj : gen ] is also motivated by the analogy with total nonnegativity .
indeed , analogues of ( 1)-(5 ) ( and even the missing ( 4 ) ! ) for the _ totally nonnegative part of the rational loop group _ are established in @xcite . in particular , in @xcite we studied in detail from the totally nonnegative perspective , the networks @xmath11 , or more precisely , their medial graphs .
our solution here to the inverse problem for the networks @xmath11 follows the strategy in @xcite , where edge weights are recovered by taking limits of ratios of matrix entries ; this approach was originally applied by aissen - schoenberg - whitney @xcite to classify _ totally positive functions_. the situation we consider here is technically much more demanding , involving rather intricate kenyon - wilson grove combinatorics .
* acknowledgements . *
we cordially thank michael shapiro for stimulating our interest in the problem .
for more background on electrical networks , we refer the reader to @xcite . for our purposes ,
an electrical network is a finite weighted undirected graph @xmath1 , where the vertex set is divided into the _ boundary _ vertices and the _ interior _ vertices .
the weight @xmath15 of an edge is to be thought of as the conductance of the corresponding resistor , and is generally taken to be a positive real number
. a @xmath16-weighted edge would be the same as having no edge , and an edge with infinite weight would be the same as identifying the endpoint vertices .
we define the _ kirchhoff matrix _ @xmath17 to be the square matrix with rows and columns labeled by the vertices of @xmath1 as follows : @xmath18
let @xmath19 be a square @xmath20 matrix , and @xmath21 .
recall that the _ schur complement _
@xmath22 is the square @xmath23 matrix defined to be @xmath24-i,[n]-i } - m_{[n]-i , i}m_{i , i}^{-1 } m_{i,[n]-i}$ ] , where @xmath25 denotes the submatrix of @xmath19 consisting of the rows labeled by @xmath26 and the columns labeled by @xmath27 .
we define the _ response matrix _ @xmath2 to be the square matrix with rows and columns labeled by the boundary vertices of @xmath1 , given by the schur complement @xmath28 where @xmath29 denotes the submatrix of @xmath27 indexed by the interior vertices .
the response matrix encodes all the electrical properties of @xmath1 that can be measured from the boundary vertices .
note that our kirchhoff and response matrices are the negative of those commonly used in the literature .
we now discuss the local transformations of electrical networks which leave the response matrix invariant .
the following proposition is well - known and can be found for example in @xcite .
[ p : sp ] series - parallel transformations , removing loops , and removing interior degree 1 vertices , do not change the response matrix of a network .
see figure [ fig : elie1 ] .
the most interesting local transformation is attributed to kennelly @xcite .
assume parameters @xmath30,@xmath31,@xmath32,@xmath33,@xmath34 and @xmath35 are related by @xmath36 or equivalently by @xmath37 then switching a local part of an electrical network between the two options shown in figure [ fig : elec11 ] does not affect the response of the whole network .
[ prop : fm ] the sequence of @xmath38 transformations shown in figure [ fig : elec7 ] returns the conductances of edges to their original values . direct computation . in the terminology of kashaev , korepanov , and sergeev @xcite ,
the proposition states that the @xmath38 transformation is a solution of type ( @xmath39 ) to the _ functional tetrahedron equation_.
a _ planar partition _ of @xmath41 = \{1,2,\ldots , n\}$ ] is a partition @xmath42 such that there do not exist @xmath43 such that @xmath44 belong to the same part of @xmath45 , and @xmath46 belong to the same part .
a _ circular planar electrical network _
@xmath1 is an electrical network embedded into a disk , so that the intersection of @xmath1 with the boundary of the disk is exactly the boundary vertices of @xmath1 .
we suppose that the boundary vertices of @xmath1 are exactly @xmath41 $ ] , and that these vertices are arranged in order around the boundary of the disk .
a _ grove _ in @xmath1 is a spanning forest where each connected component intersects the boundary .
a grove @xmath47 has boundary planar partition @xmath42 if the connected components @xmath48 of @xmath47 are such that @xmath49 contains the boundary vertices labeled by @xmath50 .
the _ weight _ of the grove is the product of the weights of its edges .
kenyon and wilson @xcite study the probability @xmath51 that a random grove of @xmath1 is of type @xmath45 , where the probability of a grove is proportional to its weight .
let `` @xmath52 '' denote the partition of @xmath41 $ ] into singletons .
[ t : upr ] ( * ? ? ? * theorem 1.1 , lemma 4.1 ) let @xmath6 be a finite circular planar electrical network . 1 .
the ratio @xmath53 is an integer - coefficient polynomial in the @xmath54 , homogeneous of degree @xmath55 .
2 . suppose @xmath45 be a planar partition with parts of size at most two .
then the polynomial of ( 1 ) depends only on @xmath54 , for @xmath56 s which are not isolated parts of @xmath45 .
recall from @xcite that for a partition @xmath45 , we define @xmath57 where the sum is over spanning forests @xmath58 of the complete graph , for which the trees of @xmath58 span the parts of @xmath45 , and the product is over the edges of @xmath58 .
let s recall * rule 1 * from @xcite .
if a partition @xmath45 is non - planar then one can pick @xmath59 so that @xmath60 belong to one part of @xmath45 , and @xmath61 belong to another part of @xmath45 .
arbitrarily subdivide the part containing @xmath30 and @xmath32 into two sets @xmath33 and @xmath35 so that @xmath62 and @xmath63 , and similarly obtain @xmath34 and @xmath64 .
denote the remaining parts of the partition @xmath45 by `` rest '' .
then the rule is @xmath65 iterating this rule transforms each non - planar partition @xmath45 into a linear combination of planar ones , and kenyon and wilson show that the coefficients do not depend on how rule 1 is applied .
denote by @xmath66 the coefficient of a planar partition @xmath67 in the application of rule 1 to @xmath45 .
we have @xmath68 .
[ l : singles ] let @xmath69 be a cyclic interval .
suppose that a planar partition @xmath67 is such that no part @xmath70 of @xmath67 contains two elements of @xmath69 .
let @xmath45 be a possibly nonplanar partition such that @xmath67 occurs in the expansion of @xmath45 under repeated application of rule 1 ( until no more applications are possible ) .
then @xmath45 does not contain any part @xmath70 such that @xmath71 . by (
* theorem 1.2 ) , the result of repeatedly applying rule 1 does not depend on the choices made when applying rule 1 . let us suppose that @xmath45 contains a part @xmath70 such that @xmath71 . in applying rule 1 , if we ever encounter that @xmath72 , we will first try to choose @xmath33 and @xmath35 so that @xmath73 or @xmath74 .
this would guarantee that all the partitions occurring in rule 1 contain some part @xmath75 such that @xmath76 .
this choice is impossible only if @xmath77 , which would in turn imply that @xmath31 or @xmath78 lies in @xmath69 .
but in this case again all the partitions occurring in rule 1 contain some part @xmath75 such that @xmath76 .
thus @xmath67 can not occur in the expansion of @xmath45 .
[ l : combtype ] suppose @xmath67 is a partition such that the non - singleton parts of @xmath67 contain at most @xmath27 elements .
then there is a constant @xmath79 such that when @xmath80 is expanded as a polynomial in the @xmath81 s the coefficient of any monomial in the @xmath81 s is less than or equal to @xmath79 .
denote the set of boundary vertices by @xmath69 , and by @xmath82 the elements in the non - singleton parts of @xmath67 .
applying lemma [ l : singles ] , we see that every @xmath45 , such that @xmath67 occurs in the rule 1 expansion of @xmath45 , have parts that contain at most one element from each cyclic component of @xmath83 .
since the number of cyclic components is bounded by @xmath27 , we see that there is a bound , depending only on @xmath27 , on the number of combinatorial types of possible @xmath45 s . here
combinatorial type means the partition one obtains when singletons in @xmath84 are removed , and only the relative orders of the remaining elements are remembered .
it follows that one can find a constant @xmath79 so that the coefficient of @xmath67 in the rule 1 expansion of @xmath45 is bounded by @xmath79 , for any @xmath45 .
next we note that for the sum of , the edges of @xmath58 determine @xmath58 , which in turn determines @xmath45 .
so each monomial in the @xmath81 s occurs in at most one @xmath85 .
thus the coefficient of any monomial in the @xmath81 s is bounded by @xmath79 .
the ideas of this section follow closely the calculation of the `` whurl transformation '' in @xcite . in a special case the whurl transformation reduces to the _ birational _ , or
_ geometric @xmath0-matrix _ of certain affine geometric crystals .
this motivates the terminology of an _ electrical @xmath0-matrix_. fix @xmath86 and @xmath87 .
we define an electrical network denoted @xmath11 which is embedded in a cylinder .
it has @xmath88 boundary vertices all lying on the boundary of the cylinder , with @xmath89 on the left boundary component and @xmath90 on the right boundary component .
there are @xmath91 internal vertices , denoted @xmath92 where @xmath93 and @xmath94 . for convenience ,
the boundary vertices @xmath89 are denoted @xmath95 and the boundary vertices @xmath90 are denoted @xmath96 . for each @xmath97 and @xmath98
, we have edges from @xmath99 to @xmath100 , and from @xmath99 to @xmath101 . here
all lower indices are taken modulo @xmath102 .
now we focus on the network @xmath103 .
we label the edge weights of @xmath103 as follows . for each @xmath97
, we have edges with weights @xmath104 from @xmath105 to @xmath106 , weights @xmath107 from @xmath105 to @xmath108 , weights @xmath109 from @xmath110 to @xmath111 , and weights @xmath112 from @xmath108 to @xmath113 . here
all indices are taken modulo @xmath102 .
now define polynomials @xmath114 and @xmath115 as follows : @xmath116 and @xmath117 @xmath118 also define @xmath119 suppose @xmath120
. then @xmath121 @xmath122 now introduce additional parallel wires with parameters @xmath70 and @xmath123 from @xmath124 to @xmath125 .
this is a special case of the local electrical equivalence for parallel resistors .
( here the parameters @xmath70 and @xmath123 should be considered formally , instead of as nonnegative real numbers . )
we may perform @xmath38 operations to move the parameter @xmath70 through the resistor network , as shown in figure [ fig : elec5 ] .
[ l : unique ] there is a unique non - zero parameter @xmath70 , which is unchanged after moving through one revolution .
let us denote by @xmath126 the value of the `` extra edge '' after @xmath127 star - triangle transformations .
( the fourth network in figure [ fig : elec5 ] shows the location of @xmath128 . )
we claim that the parameter @xmath126 is a ratio of two linear functions in the original @xmath129 .
this is easily verified by induction using the star - triangle transformation .
thus after @xmath88 star - triangle transformations the equation @xmath130 we obtain is either a linear or a quadratic equation , and zero is clearly one of the roots .
therefore there is at most one non - zero solution , and as we shall see soon in the proof of theorem [ t : elecwhirl ] , a solution indeed exists .
suppose we perform the sequence of @xmath38 operations of figure [ fig : elec5 ] , using the parameter @xmath70 of lemma [ l : unique ] ( which we still have to prove exists ) .
then as illustrated in the final diagram of figure [ fig : elec5 ] , the `` extra edges '' with parameters @xmath70 and @xmath123 can be removed via the local electrical equivalence for parallel resistors .
we define the _ electrical @xmath0-matrix _ to be the transformation @xmath131 induced on the edge weights by this transformation .
[ t : elecwhirl ] the electrical @xmath0-matrix is given by @xmath132 we claim that the parameter @xmath133 is the parameter of lemma [ l : unique ] . define @xmath126 to be the parameter after @xmath105 pairs of @xmath134 and @xmath38 operations , and @xmath135 to be the parameter obtained from @xmath126 by one @xmath134 operation .
then @xmath136 indeed , calculating by induction @xmath137 and similarly for @xmath138 .
we may then calculate that the transformation is given by @xmath139 and similarly @xmath140 .
we also have @xmath141 and similarly @xmath142 .
[ prop : un ] the electrical @xmath0-matrix does not depend on where you attach the extra pair of edges , and is an involution .
if one applies the @xmath38 transformations to successively push through two edges , with weights @xmath143 and @xmath144 negative of each other , then there is no net effect on the weights of the other edges involved .
also , the weights on the two edges pushed through remain negatives of each other . thus if one pushes through @xmath70 to perform the electrical @xmath0-matrix , and then a @xmath123 , the latter will perform a transformation that will undo the first one .
furthermore , a working choice of parameter @xmath70 at one location this way yields a working choice of the parameter at any other location .
the uniqueness in lemma [ l : unique ] implies that all resulting electrical @xmath0-matrices are the same .
let us now consider the network @xmath145 .
the procedure of section [ ssec : r ] gives two different electrical @xmath0-matrices acting on @xmath145 : by acting on the part of the network involving vertices @xmath146 , or by acting on the part of the network involving the vertices @xmath147 .
we denote these @xmath0-matrices by @xmath148 and @xmath149 respectively .
[ thm : ryb ] the electrical @xmath0-matrix satisfies the yang - baxter equation @xmath150 first we note that to perform the electrical @xmath0-matrix , we can either add extra horizontal edges with conductances @xmath70 and @xmath123 between @xmath124 and @xmath125 , or we could split the vertex @xmath151 into three vertices and add vertical edges with conductances @xmath70 and @xmath123 between them ( see figure [ fig : elec16 ] ) . to perform the sequence of @xmath0-matrices @xmath152 , we will add horizontal edges for the first and third factor , but add vertical edges for the second factor . let the corresponding weights be @xmath70 , @xmath143 , @xmath153 as shown in figure [ fig : elec6 ] . according to lemma [ l : unique ] and theorem [ t : elecwhirl ] and proposition [ prop : un ] ,
these weights exist and are unique .
now , apply the @xmath134 transformation as shown in figure [ fig : elec6 ] to obtain weights @xmath75 , @xmath154 and @xmath155 , and their negatives on the other side .
we claim that if these new weights are pushed around the cylinder , they come out the same at the other end .
this follows from proposition [ prop : fm ] .
now if we push the weights @xmath70 , @xmath143 and @xmath153 through and apply the @xmath134 transformations , we obviously get again the weights @xmath75 , @xmath154 and @xmath155 . thus by lemma [ l : unique ] and theorem [ t : elecwhirl ] and proposition [ prop : un ] , while pushing @xmath75 , @xmath154 and @xmath155 through we are applying @xmath156 . since , by the electrical tetrahedron relation ( proposition [ prop : fm ] ) the two results are the same , the claim of the theorem follows .
[ cor : braid ] the electrical @xmath0-matrix gives an action of the symmetric group @xmath12 on @xmath11
. follows from theorem [ thm : ryb ] and proposition [ prop : un ] .
let @xmath6 be a cylindrical electrical network . thus @xmath6 is an electrical network embedded into a cylinder so that the intersection of @xmath6 with the boundary of the cylinder is exactly the boundary vertices of @xmath6 .
let @xmath157 denote the universal cover of @xmath6 .
it is an infinite periodic network embedded into an infinite strip .
given finite sets @xmath158 of vertices of @xmath157 which eventually cover all vertices of @xmath157 , we obtain a sequence of _ truncations _
@xmath159 , @xmath160 of @xmath157 as follows .
we let @xmath161 be the subgraph of @xmath157 consisting of all edges incident to a vertex in @xmath162 . furthermore
, we declare a vertex of @xmath161 internal if it lies in @xmath162 and is internal in @xmath157 .
each @xmath161 is a finite planar electrical network .
we suppose that the boundary vertices of @xmath157 are numbered @xmath163 on one side of the boundary , and by @xmath164 on the other side of the boundary .
( we will always picture the infinite strip as vertical , with vertex labels increasing as we go downwards . )
we define the _ universal response matrix _ of @xmath6 to be given by @xmath165 where @xmath166 where @xmath56 denote vertices of @xmath157 . for sufficiently large @xmath1 ,
any two fixed vertices of @xmath157 will lie in @xmath161 .
these limits exist and are finite , due to following lemma .
if @xmath167 is an electrical network , and @xmath168 a subset of its vertices , the _ response matrix of @xmath169 _ is the response matrix obtained by declaring all the vertices in @xmath168 to be interior .
[ lem : smon ] assume @xmath170 are two subsets of vertices of an electrical network @xmath167 , and assume @xmath105 and @xmath171 are two vertices not contained in @xmath172 .
then @xmath173 in the response matrix of @xmath174 is at least as large as @xmath81 in the response matrix of @xmath169 .
the schur complement with respect to some set @xmath168 can be taken as a sequence of schur complements with respect to single vertices in @xmath168 in some order ( see for example ( * ? ? ?
* ( 3.7 ) ) ) .
thus , it would suffice to prove the statement for @xmath175 consisting of a single vertex . in this case
the claim is obvious however , since off - diagonal entries of a response matrix are nonnegative , and diagonal entries are non - positive .
[ t : universalresponse ] the matrix @xmath165 is a well - defined infinite periodic matrix , which does not depend on which truncations @xmath161 are taken .
there are several parts to this statement .
the limits @xmath165 exist .
note that for sufficiently large @xmath1 , @xmath176 can be calculated by taking @xmath81 of the network @xmath177 , which is obtained from @xmath157 by declaring that only the internal vertices of @xmath161 are internal in @xmath177 .
the network @xmath177 is obtained from @xmath161 by adding extra boundary vertices attached only to boundary vertices of @xmath161 ( and by assuming @xmath1 is large enough , we may assume that these extra vertices are not incident to @xmath105 or @xmath171 ) .
but @xmath176 is by definition calculated by measuring the current flowing through @xmath171 when vertex @xmath105 is set to one volt and all other boundary vertices are set to zero volts . since current does not flow between zero volt vertices it follows that @xmath178 . by lemma [ lem : smon ]
the sequence @xmath179 is non - decreasing as @xmath180 , since each network is obtained from the previous one by declaring some extra vertices internal and taking the corresponding schur complement .
the limits @xmath165 do not depend on the sequence of truncations . assume we have two different sequences @xmath158 and @xmath181 . since we know that each eventually covers all vertices , we know that for each @xmath105 there is a @xmath171 such that @xmath182 and @xmath183 . then applying lemma [ lem : smon ]
we conclude that the two limits bound each other from above , and thus are equal .
the limits @xmath165 are periodic : @xmath184 . indeed , take two sequences @xmath158 and @xmath181 , one obtained from the other by a shift on the universal cover by the period @xmath102 .
we know that they give the same value of @xmath165 by the previous part . on the other hand ,
it is clear that the value one of them gives for @xmath165 is the value the other gives for @xmath185 .
the limits @xmath165 are finite .
any truncation of @xmath157 can also be viewed as a truncation of a finite cover @xmath186 $ ] ( obtained by lifting @xmath6 to a @xmath187-fold cover of the cylinder ) for a large enough @xmath187 . by lemma [ lem : smon ]
the conductance @xmath176 is bounded from above by the same conductance in @xmath186 $ ] , which in turn is bounded from above by the same conductance in the original network @xmath6 . indeed ,
if @xmath188 are vertices in @xmath186 $ ] that cover @xmath171 , then @xmath189)$ ] .
this can be seen as follows . using the linearity of the response and periodicity , the sum @xmath190)$ ]
measures the current through @xmath191 ( for any @xmath192 ) when all vertices @xmath193 that cover @xmath105 have potential @xmath124 , and all other vertices have potential @xmath16 . but projecting onto @xmath6 by identifying all covers of a vertex we see that this current flow is exactly @xmath54 . in particular , @xmath194 ) \leq l_{ij}(g)$ ] for any @xmath195 .
this shows that @xmath165 is bounded from above by @xmath54 , and thus if the latter is finite , so is the former .
thus for fixed @xmath56 , we can approximate @xmath165 arbitrarily well by calculating @xmath176 for some large @xmath1 .
the universal response matrix @xmath165 is invariant under the local electrical equivalences of @xmath1 . for any local electrical transformation in @xmath6 one can choose a sequence of truncations of @xmath157 containing completely several occurrences of this transformation . since the conductances in these truncations do not change , the limit is also invariant .
[ cor : resp ] the electrical @xmath0-matrix preserves the universal response matrix .
the only comment one needs to make is that the entries of the universal response matrix are limits of rational functions in the edge weights , and that these rational functions make sense formally even when negative conductances are used ( as in the derivation of the electrical @xmath0-matrix ) .
let @xmath196 $ ] be subsets of the same cardinality .
then @xmath197 is a _ circular pair _ if a cyclic permutation of @xmath198 is in order .
a @xmath20 matrix @xmath19 is _ circular totally - nonnegative _ if the minor @xmath199 is nonnegative for every circular pair @xmath197 .
curtis , ingerman , and morrow @xcite show that the response matrices of circular planar electrical networks are exactly the set of circular totally - nonnegative symmetric matrices for which every row sums to 0 .
( note that the response matrices in @xcite are the negative of ours . )
let us extend this to cylindrical electrical networks
. put the total order @xmath200 on @xmath201 .
let @xmath202 be two ordered subsets of the same finite cardinality .
then @xmath197 is a _ cylindrical pair _ if a cyclic permutation of @xmath198 is in order .
a matrix @xmath19 with rows and columns labeled ( and ordered ) with @xmath201 is _ cylindrically totally - nonnegative _ if the minor @xmath199 is nonnegative for every cylindrical pair @xmath197 .
[ p : tnn ] suppose @xmath203 is the universal response matrix of a finite cylindrical electrical network . then @xmath204 is cylindrically totally - nonnegative .
for each fixed cylindrical pair @xmath197 , and sufficiently large @xmath1 , the truncation @xmath161 is a finite circular electrical network including all the boundary vertices in @xmath205 and @xmath26 .
but then @xmath206 , using curtis - ingerman - morrow s result . the converse to proposition [ p : tnn ] , namely , which cylindrically totally nonnegative matrices are realizable as universal response matrices , is more subtle .
let @xmath207 be the lifts to the universal cover of a particular vertex in a finite cylindrical electrical network @xmath6 .
then for a fixed @xmath105 , the ( doubly - infinite ) sequence @xmath208 must satisfy certain recursions or convergence properties . in the different but closely related setting of total nonnegative points of loop groups ,
the correct property is to ask for the generating function of @xmath209 to be a rational function ( see ( * ? ? ?
* theorem 8.10 ) ) .
we now assume we are given a network @xmath210 .
the vertices have been labeled so that if we take the `` low '' edge ( from @xmath211 to @xmath212 ) at every step starting from @xmath105 we will end up at @xmath111 .
note that a shortest path from one side of the cylinder to the other consists of exactly @xmath187 edges .
from now on we consider groves in the universal cover @xmath157 of @xmath6 ( or in the truncations @xmath161 ) .
the boundary partition of such a grove would be a planar partition of @xmath201 arranged on the two edges of an infinite strip ( or in the truncations of this ) . [ l : short ] suppose @xmath105 and @xmath213 are can be connected by a path with @xmath187 edges .
there exists an integer @xmath19 such that there are no groves @xmath47 with the properties 1 .
there is a grove component @xmath214 with boundary vertices @xmath215 and which uses an edge below ( resp . above )
any of the shortest paths from @xmath105 to @xmath213 .
2 . there are grove components with boundary vertices @xmath216 ( resp .
@xmath217 ) . for a grove component @xmath214
let us call _ bad _ the edges it contains that are below the lowest path from @xmath105 to @xmath213 .
assume the grove component @xmath214 has bad edges , and furthermore without loss of generality assume that it has a bad high edge .
the case of a bad low edge is similar with the left and right sides of the network swapped .
assume @xmath195 is the first index such that bad high edge has one of the @xmath211 as its right endpoint .
we claim that the @xmath218 component has a bad high edge with right end having index @xmath219 or smaller .
indeed , consider the unique path from @xmath220 to @xmath221 inside @xmath218 . in order to avoid touching the bad edge of @xmath214 it has to turn , diverting from the lowest shortest path from @xmath220 to @xmath221 .
the first time it thus diverts gives a desired high edge .
now , the index of the first bad high edge can not decrease indefinitely , and in fact one sees that the statement of the lemma holds for @xmath222 .
the case of edges above the highest shortest path is similar .
define the radii @xmath223 for @xmath224 by @xmath225 where we have denoted the conductance of the edge joining two vertices @xmath226 by @xmath227 .
[ l : optimalpath ] let @xmath228 be an @xmath102-tuple of consecutive shortest paths , where @xmath229 connects @xmath230 to @xmath231 for each @xmath105 , and some fixed @xmath195 .
suppose that 1 .
all the @xmath229 have the same shape ; thus they high or low edges respectively at the same points along the path .
@xmath232 .
then the total weight @xmath233 is maximized exactly when the highest path is taken .
the weight of a subgraph is simply the product of its edge weights .
all the shortest paths are connected by switching from one side of a rhombus to the other , see figure [ fig : elec3 ] .
if this parallelogram involves vertices with upper index @xmath234 , @xmath195 and @xmath235 , then the higher path has greater weight exactly when @xmath236 .
let @xmath237 be the partition of @xmath201 with parts of size two @xmath238 for all @xmath239 , parts of size two @xmath240 for @xmath241 , and all other parts are singletons .
thus in particular , @xmath124 is a singleton .
we shall denote by @xmath242 the partition of @xmath201 obtained from @xmath237 by placing @xmath124 in the same part as @xmath243 to get a single part @xmath244 of size three .
we shall suppose that the truncation @xmath161 includes the boundary vertices @xmath245 and also the boundary vertices @xmath246 . for @xmath247 the partitions @xmath237 and
@xmath242 naturally gives rise to partitions of the boundary vertices of @xmath161 , where boundary vertices of @xmath161 which are not boundary vertices of @xmath157 are all considered singletons .
we will still denote these partitions of boundary vertices of @xmath161 by @xmath237 and @xmath242 .
[ thm : asw ] suppose that @xmath232 .
then @xmath248 where @xmath30 is the weight of the high edge connected to the vertex @xmath124 of @xmath6 .
let @xmath249 denote the vertex connected to both @xmath16 and @xmath124 .
let @xmath250 be the edge joining @xmath124 to @xmath249 , so that @xmath250 has weight @xmath30 .
let @xmath251 denote the other edge incident to @xmath124 . to approximate the lhs , we shall assume we have chosen @xmath252 .
let @xmath47 be a grove with boundary partition either @xmath242 or @xmath237 .
by lemma [ l : short ] , the grove component @xmath253 can not extend either above or below the set of edges contained in shortest paths from @xmath254 to @xmath255 .
in particular , there is a bound on the number of choices of @xmath253 , not depending on @xmath27 or @xmath1 . in the following
, we shall assume that we have fixed such a choice for @xmath253 .
let @xmath47 be a grove with boundary partition @xmath237 .
by lemma [ l : short ] , the grove component @xmath256 can not extend above the ( unique ) shortest path from @xmath16 to @xmath257 . in particular , the vertex @xmath249 must lie in @xmath256 . thus the edge @xmath250 is never used in @xmath47 , and @xmath258 is a grove with boundary partition @xmath242 . also by lemma [ l :
short ] , the grove component @xmath259 can not extend either above or below the ( unique ) shortest path from @xmath260 to @xmath261 , and therefore must be exactly this shortest path .
similar observations hold for a grove with boundary partition @xmath242 . in particular , the part of the grove above @xmath259 and the part below are essentially independent .
let us denote by @xmath262 ( resp .
@xmath263 ) the set of groves with boundary partition @xmath237 ( resp .
@xmath242 ) and by @xmath264 ( resp .
@xmath265 ) the total weight of that set of groves . let @xmath47 be a grove with boundary partition @xmath242 .
first we observe that all but a constant number of grove components @xmath266 are shortest paths .
furthermore , as @xmath105 goes from @xmath267 to @xmath27 , the shapes of the shortest paths are locally constant , and can only change when we encounter a grove component which is not a shortest path . furthermore ,
as we go down , the shape of the shortest path can only become lower .
we shall call this part of @xmath47 the lower half of the grove . note that there are @xmath187 different shapes @xmath268 of shortest paths , listed from highest to lowest , see figure [ fig : elec4 ] .
let @xmath269 denote the subset of groves where @xmath270 uses the edge @xmath250 , and let @xmath271 denote the subset of groves where @xmath270 does not use the edge @xmath251 .
given @xmath272 , we shall now show that for sufficiently large @xmath27 , one has @xmath273 .
let @xmath274 .
then the shortest paths which occur in the lower half of @xmath47 can only use the shapes @xmath275 . pick @xmath27 sufficiently large that we are guaranteed to have @xmath276 grove components in the lower half which are shortest paths
, where @xmath35 is a constant ( not depending on @xmath27 or @xmath1 ) we shall describe below .
then there is some @xmath277 ( say pick the least such @xmath105 ) which occurs at least @xmath278 times . define a set of new groves @xmath279 by : 1 . removing @xmath280 of the @xmath277 shaped paths , and shifting the lower half of @xmath47 downwards to fill in the removed area ; 2 . replacing the grove components @xmath281 with shortest paths ; 3 .
replacing @xmath282 with the shortest path from @xmath16 to @xmath257 union the edge @xmath250 ; 4 .
adding @xmath283 new shortest paths @xmath284 which are @xmath285-shaped ; 5 . adding one extra `` transition '' grove component @xmath286 which is @xmath285-shaped above , but which correctly fits with @xmath287 below .
figure [ fig : elec1 ] illustrates the procedure . in this case
@xmath288 , @xmath120 , @xmath289 , @xmath290 .
the @xmath270 component of the groves is shown in red , the part shifted down is shown in green .
the two removed groves of shape @xmath277 and the two new groves of shape @xmath285 that replaced them are shown in brown . note that @xmath291 belongs to @xmath292 .
since apart from the shortest paths the rest of the grove has only been modified at a bounded number of edges , by lemma [ l : optimalpath ] there is a constant @xmath293 not depending on @xmath294 , or @xmath39 such that @xmath295 .
furthermore , each grove in @xmath292 can occur in this way in at most @xmath296 different ways .
the @xmath297 counts the possible grove components @xmath281 and @xmath298 of which there is a universal bound for . the @xmath187 ways count the possible @xmath105 such that the shortest paths @xmath277 are being replaced by @xmath285 . setting @xmath299 , we conclude that @xmath300 . for every @xmath301 ,
the grove @xmath302 lies in @xmath263 , and this map is an injection from @xmath262 to @xmath263 which changes the weight of each grove by exactly @xmath30 . on the other hand
every @xmath303 is in the image of this map .
it follows that @xmath304 . letting @xmath305
, we obtain the statement of the theorem .
the partition @xmath237 is a planar partition of @xmath201 of the type described in theorem [ t : upr](2 ) , where all but finitely many vertices are isolated . define @xmath306 by taking the polynomial of theorem [ t : upr](2 ) and replacing @xmath54 by @xmath165 .
then we have @xmath307 and in particular , @xmath306 can be approximated arbitrarily well on some @xmath161 for very large @xmath1 .
unfortunately , this is not the case for the partition @xmath242 , which contains a part of size three , for which theorem [ t : upr](2 ) can not be applied .
instead one has @xmath308 for the sequence of polynomials @xmath309 of theorem [ t : upr](1 ) .
these polynomials depend on @xmath242 , which is suppressed from the notation .
[ p : polyest ] suppose @xmath310 .
then @xmath311 is a polynomial in the @xmath81 s such that every monomial involves some @xmath81 where @xmath312 and @xmath313 $ ] .
furthermore , there is some constant @xmath79 such that the coefficient of each monomial in @xmath309 is less than @xmath79 .
the first statement follows from lemma [ l : singles ] and our choice of @xmath242 .
the second statement is lemma [ l : combtype ] .
the assymmetry of the roles of @xmath105 and @xmath171 in proposition [ p : polyest ] is accounted for by the fact that @xmath81 is symmetric .
that is , we treat @xmath314 as the same variable .
note that we already know @xmath315 approaches a limit , and it follows from theorem [ thm : asw ] that @xmath316 approaches a limit as well .
[ l : ak ] for each @xmath27 and @xmath272 , one can find some @xmath33 such that @xmath317 fix @xmath27 .
we first show that there is @xmath33 such that for all @xmath318 we have @xmath319 it is known that the polynomials @xmath309 have degree @xmath320 , see section [ sec : kwpoly ] . by proposition [ p : polyest ]
, every monomial in @xmath321 has a factor @xmath81 where @xmath322 and @xmath323 , and has coefficient @xmath324 .
thus @xmath325 by the proof of theorem [ t : universalresponse ] , we know that for each @xmath105 we have @xmath326 . thus it is possible to find @xmath33 large enough that @xmath327 giving us @xmath328 since only finitely many @xmath81 s appear in @xmath329 , for sufficiently large @xmath1 we have @xmath330 furthermore , for sufficiently large @xmath1 we have @xmath331 finally , we combine the three estimates , , and use @xmath332 . [ thm : main ] fix @xmath6 , satisfying @xmath232 and fix one of the edges connected to one of the vertices @xmath333 . there is a sequence of polynomials ( depending on the universal response matrix of @xmath6 ) , @xmath334 , and @xmath335 such that @xmath336 where @xmath30 is the weight of the chosen edge . by symmetry , it is enough to establish the formula for the high edge connected to @xmath124 .
the polynomial @xmath337 is the one associated to @xmath237 from theorem [ t : upr](2 ) .
the polynomial @xmath338 is chosen via lemma [ l : ak ] so that @xmath339 , where @xmath340 is chosen so that @xmath341 as @xmath342 .
finally , we apply theorem [ thm : asw ] .
[ lem : wh ] assume a cylindrical network @xmath343 is obtained by concatenating two networks @xmath344 and @xmath345 .
then knowing @xmath8 and the universal response matrix of @xmath3 , one can recover the universal response matrix of @xmath346 .
assume the conductances in @xmath8 are @xmath104 for the high edges and @xmath107 for the low edges .
concatenate @xmath3 with a network @xmath347 with ( virtual ) conductances @xmath348 for low edges and @xmath349 for high edges .
we claim that the response of the resulting network is equal to that of @xmath346 . indeed ,
connect the opposite vertices of @xmath350 and @xmath8 by edges with conductances @xmath40 and @xmath351 , without changing the response .
changing resulting triangles into stars using the @xmath38 transformation , and letting @xmath352 , we see that there is an infinite conductance between opposite vertices and zero conductance between other pairs .
thus the two @xmath353 networks effectively cancel each other out , and we are left with a network with the same response as @xmath346 .
see figure [ fig : elec9 ] .
[ thm : sol ] there are generically @xmath13 sets of edge conductances which produce a given universal response matrix for the network @xmath11 .
all the solutions are connected by the @xmath12-action via the electrical @xmath0-matrix .
it is easy to see from theorem [ t : elecwhirl ] that the electrical @xmath0-matrix swaps the radii @xmath223 .
assume we have a solution for conductances in @xmath11 with given universal response matrix .
apply the electrical @xmath0-matrix to reorder the radii @xmath223 in non - increasing order .
then theorem [ thm : main ] allows us to recover the conductances in the leftmost @xmath353 part of the network . in particular , these conductances are the same for any solution . once we know that , we can use lemma [ lem : wh ] to recover the universal response matrix of the remaining @xmath354 part of the network .
then we repeat the procedure .
we see that once we require the radii to form a non - increasing sequence , the conductances are recovered uniquely by theorem [ thm : main ] .
therefore all other solutions can be obtained from that one by action of @xmath12 , which is what we want . in the generic case when all radii are distinct , the orbit has size @xmath13 .
[ rem : monodromy ] in the language of @xcite , theorem [ thm : sol ] says that the group @xmath12 generated by electrical @xmath0-matrices is exactly the _ monodromy group _ of the network @xmath11 .
it is the group acting on the edge weights of @xmath11 obtained by transforming the network via local electrical equivalences back to itself .
it is convenient to describe the general answer we expect using the language of _ medial graphs _ , see for example @xcite .
draw _ wires _ through the electrical network so that they pass through each edge and connect inside each face as shown in the first two pictures in figure [ fig : elec17 ] .
the third picture shows an example of an electrical network and its medial graph .
note that the medial graph always has four - valent vertices , and that the _ wires _ of the medial graph `` go straight through '' each vertex . in the terminology of @xcite a circular planar electrical network is _ critical _ if its medial graph avoids _ lenses _ , which is equivalent to saying that every pair of wires crosses as few times as possible , given their respective homotopy types .
if @xmath6 is a cylindrical electrical network , we say that @xmath6 is _ critical _ if the universal cover of @xmath6 satisfies this condition ; namely , the medial graph has wires which cross as few times as possible . let us call a cylindrical network _
canonical _ if the medial graph of its universal cover has the following form .
first , there are three kinds of wires : ( i ) some wires connect points on opposite boundaries , ( ii ) some wires connect points on the same boundary , and ( iii ) some wires do not intersect the boundary at all and correspond to ( simple ) cycles around the cylinder .
secondly , we require that the third kind of wires do not intersect the second kind , and furthermore , all points of intersection of wires of the first kind with themselves happen strictly before they intersect wires of the third kind .
an illustration is given in figure [ fig : elec19 ] .
* any cylindrical electrical network can be transformed using local electrical transformations ( those in section [ ss : trans ] and the electrical @xmath0-matrix ) into a critical cylindrical electrical network . *
any two cylindrical electrical networks @xmath6 and @xmath7 with the same universal response matrices are connected by local electrical equivalences .
furthermore , if @xmath6 and @xmath7 are both critical , then only star - triangle transformations , and electrical @xmath0-matrix transformations are needed .
* if a cylindrical electrical network is critical canonical , then the conductances corresponding to all crossings involving wires of types ( i ) and ( ii ) can be recovered uniquely .
the conductances corresponding to crossings of wires of type ( i ) and type ( iii ) can be recovered up to the electrical @xmath0-matrix action . *
the space @xmath8 of universal response matrices of cylindrical electrical networks has an infinite stratification by @xmath9 where each @xmath10 is a semi - closed cell that can be obtained as the set of universal response matrices for a fixed critical network with varying edge weights .
another way to phrase conjecture [ conj : gen](3 ) is that the _ monodromy group _ of a critical canonical cylindrical network is a symmetric group , generated by electrical @xmath0-matrices .
see remark [ rem : monodromy ] .
let us explain the semi - closed cells in conjecture [ conj : gen](5 ) .
let @xmath6 be a critical canonical cylindrical electrical network .
some edge weights can be recovered uniquely and these each give a @xmath355 in the parametrization .
the remaining part of the network is essentially one of the networks @xmath11 , whose edge weights can be recovered uniquely up to the electrical @xmath0-matrix action ( theorem [ thm : sol ] ) .
so the response matrices would be parametrized by the orbit space @xmath356 .
however , we can pick a distinguished element in each orbit : namely the one where the radii @xmath223 are non - increasing .
the corresponding response matrices would then be parametrized by @xmath357 , @xmath358 together with some collection of edge weights which can be freely chosen in @xmath355 .
thus the universal response matrices of critical canonical cylindrical electrical network is parametrized by a semi - closed cell @xmath359 . | in this paper we study the inverse dirichlet - to - neumann problem for certain cylindrical electrical networks . we define and study a birational transformation acting on cylindrical electrical networks called the electrical @xmath0-matrix .
we use this transformation to formulate a general conjectural solution to this inverse problem on the cylinder .
this conjecture extends work of curtis , ingerman , and morrow @xcite , and of de verdire , gitler , and vertigan @xcite for circular planar electrical networks .
we show that our conjectural solution holds for certain `` purely cylindrical '' networks . here
we apply the grove combinatorics introduced by kenyon and wilson @xcite . |
[ section : intro ] one of the great surprises in the search for worlds orbiting other stars was the discovery of a class of close orbiting extrasolar giant planets ( cegps ) .
the existence of such objects did not fit with the formation scenarios used to explain the arrangement of planets in our own solar system .
even before the first extrasolar gaseous giants were discovered , core accretion models predicted that they should only form @xcite at distances similar to those of the giant outer planets @xcite . here
temperatures must be cool enough to enable the formation of an ice ( water , methane and ammonia in solid or liquid form ) core from the proto - planetary disk before gas accretion onto the core can then take place . with the announcement of the discovery of the first cegp orbiting the f8 dwarf 51 peg @xcite at a distance of only 0.05 au came the realisation that existing theories of planetary formation needed to be revised .
@xcite argued , contrary to conventional belief @xcite , that gas giant planets could survive in such close proximity , although their formation beyond the ice point was still required .
mechanisms to enable an exchange of angular momentum to take place in a proto - planetary disk are believed to be responsible for the spiralling into and eventual halting of planets in a close orbit @xcite .
planetary atmospheric physics research in recent years has been motivated by the expectation that radiation processed by a large body close to its parent star should be detectable with modern instrumentation .
the first clear detection of sodium in the atmosphere of @xcite revealed a lower than expected concentration when compared with a cloudless planetary atmosphere .
several explanations for the lack of sodium were put forward , including the possible presence of a high cloud deck .
no observations had been able to differentiate between the possibility of no clouds ( combined with lower sodium abundance ) until recent spitzer observations of hd 209458b with the low resolution spectrometer irs @xcite indicated a spectral feature consistent with a high silicate cloud deck @xcite . the reflected light spectroscopic studies carried out by @xcite and more recent results from most photometry @xcite place albedo _
upper limits _ of 0.1 - 0.25 on the atmospheres of cegps .
@xcite has placed an upper limit of 0.12 on the geometric albedo of , suggesting that if clouds are present at all , they are highly non - reflective .
these are somewhat lower than the solar system gas giants , jupiter , saturn , uranus and neptune , which possess geometric albedos of 0.46 , 0.39 , 0.60 and 0.58 respectively @xcite .
these observed albedo limits for cegps rule out the high silicate cloud investigated in the models of @xcite , and contradict more recent spitzer observations @xcite .
a number of attempts to model the emergent spectrum of cegps have also been made in recent years .
@xcite have calculated spectra for planets with a range of orbital separations from the parent star , including the so called class of ` hot roasters ' .
these authors also presented spectra in the 0.4 - 5 region for specific systems .
for the well studied transiting planet , , they found a phase averaged planet - star flux ratio of log@xmath3 . a re - evaluation of the @xcite models by @xcite which include phase dependent effects have yielded 2.2 flux ratios of up to log@xmath4 for a cloudless atmosphere in the case of . here
, the incident radiation is completely absorbed and re - emitted on the day - side of the planet .
a more recent study by @xcite ( bha05 ) , which models day - night gradients , has yielded results consistent with previous studies . for no redistribution of heat ( i.e. re - emission of radiation on the day side ) , they find log@xmath5 , while uniform redistribution of the incident radiation yields log@xmath6 at all phases for .
several pioneering attempts were made to detect cegps in the near infrared ( nir ) from the ground in the combined star and planet light @xcite .
success in the ir was achieved with the spitzer space telescope , with detections of a reduction in thermal emission during secondary transits of @xcite and tres-1 ( k0v ) @xcite .
photometry indicates log@xmath7 at 24 while a 2.3 ground based measurement yields a value of log@xmath8 , albeit with a large degree of uncertainty @xcite .
the tres-1 photometry @xcite indicated log@xmath9 at 4.5 and log@xmath10 at 8.3 .
these observations are consistent with the @xcite base model .
@xcite found that the cegp yielded an even higher contrast ratio ( log@xmath11 at 24 ) due to the close orbit and later spectral type of the k2v star .
8 of @xcite plots a comparison of observed and predicted star - planet contrast ratios covering the spectral region . the and tres-1 observations are consistent with the models , with a 2.2 signal strength of 0.1% , whereas scaling the signal suggests a signal strength as high as 0.2% .
most recently , @xcite have used high s / n spitzer photometric observations to directly measure the planetary flux of the cegp @xmath12 and at 24 .
they find a phase dependent flux which is consistent with the bha05 models where heat is not significantly redistributed throughout the atmosphere but re - radiated on the dayside of the planet .
[ [ section ] ] [ section : hd75289 ] a companion to the main sequence g0 dwarf was first announced by @xcite following precision radial velocity monitoring with the coralie spectrometer at the 1.2-m euler swiss telescope . since its first discovery ,
@xcite have published a refinement of the system parameters in light of further observations . a comprehensive list of system parameters have been estimated and tabulated by @xcite .
subsequent to further refinement of parameters , as a result of a longer timebase of observations , @xcite give an ephemeris of .
this result , with , is marginally greater than the estimate of @xcite .
@xcite estimate an orbital radius semi - major axis of while @xcite estimate .
, representing three different models with an inclination of @xmath13@xmath14
. plotted are an empirical reflected light venus @xcite phase function , a lambert sphere phase function and the 2.2 phase function derived from the atmospheric models of bha05 . in the phase range
, the bha05 model closely mimics the venus phase function model.,height=330 ] [ fig : phasfunc ] [ tab : journal ] lccccc object & ut start & exp time per & number of abba & mean & comments + & & exposure [ s ] & sequences & s / n & + + hd 75289 & 01:46:55 & 120 & 6 & 401 & seeing = 0.68 + hd 75063 & 02:50:32 & 40 & 2 & 391 & g5v telluric standard + hd 75289 & 03:18:48 & 120 & 12 & 133 & slit mis - alignment ( counts decreasing ) + hd 75063 & 05:22:47 & 40 & 3 & 398 & + hd 75289 & 06:03:57 & 120 & 12 & 380 & low counts : slit re - alignment in middle of set . + hd 75063 & 08:46:42 & 40 & 3 & 515 & + + hd 75063 & 00:48:25 & 40 & 3 & 414 & + hd 75289 & 01:14:02 & 160 & 10 & 358 & + hd 75063 & 03:32:38 & 40 & 3 & 240 & + hd 75289 & 03:59:03 & 160 & 11 & 363 & software crash at end of sequence : frames rejected + hd 75063 & 06:25:46 & 40 & 3 & 385 & + hd 75289 & 03:48:51 & 160 & 11 & 337 & + hd 75063 & 09:06:56 & 40 & 3 & 405 & + we have carried out monte - carlo simulations to determine the most probable estimates of planetary mass , @xmath15 , and orbital inclination , @xmath16 .
an estimate of the orbital inclination provides an estimate for the most probable velocity amplitude of the planetary signal . as described in
[ section : results ] we can use this value as a guide when searching for a planetary signal .
the axial inclination of the system can simply be calculated from the equatorial rotation velocity , @xmath17 , determined from estimates of the stellar radius and rotation period , while can be measured directly .
we assume that the stellar equator and planetary orbit lie in the same plane in agreement with rossiter - mclaughlin effect measurements for the hd 209458 @xcite and hd 189733 @xcite systems .
monte - carlo simulations using updated values discussed above and values tabulated in table 1 of @xcite yield a most probable value of @xmath18@xmath14 .
a planetary mass of then follows from the known orbital amplitude of the stellar reflex motion of in light of observational and theoretical results , the adopted theoretical @xmath19 radius determined by @xcite is probably a significant over - estimate .
a recent plot @xcite of radii determined from all observations of transiting cegps indicates .
based on the recent theoretical evolution models from fig .
3 of @xcite for cegp radii , we estimate an approximate radius of @xmath20 . in this work
, we aim to detect the planetary signature of the close orbiting planet . in the nir , rather than searching for an attenuated copy of the stellar spectrum , we can search for the direct spectrum emerging from the heated atmosphere which is expected to possess temperatures in the range 1000 - 1200 k @xcite . the analysis presented below relies on the use of the predicted high density of absorption lines , due mainly to h@xmath21o , to search for the faint planetary signal . as such it has the potential to provide constraints on the reliability of model atmosphere opacities . in
[ section : detect ] we present the method used to detect a nir planetary signal while
[ section : obsdata ] outlines the data reduction and analysis of data .
we discuss our results in [ section : results ] and carry out further simulations in [ section : simplanet ] before discussing future prospects for this kind of survey in [ section : discussion ] .
[ section : detect ] the method of analysis uses a modification of the matched filter technique first presented in @xcite .
this technique has been refined and applied to several systems using optical data @xcite including @xcite where an upper limit of log@xmath22 has been determined . here
we seek to apply a similar method to nir spectra . in order to extract the planetary signal from a timeseries of spectra
, we model the planetary signature as a phase dependent spectrum superimposed on an unvarying stellar spectrum . the time dependent variations of the planet orbiting the star are : ( 1 ) the instantaneous doppler shift of the spectrum due to the relative position of the planet and ( 2 ) a phase dependent planet - star flux ratio , @xmath23 , which is dependent on atmospheric physics and heating due to the parent star . the ephemeris and velocity semi - amplitude , ( where @xmath16 is the orbital inclination ) , enable the instantaneous velocity shift of the planetary spectrum relative to the stellar spectrum to be determined at any observation phase .
since the rotation velocity of is only the planetary signature will be doppler shifted clear of the stellar lines at all phases except those close to @xmath24
. a clean spectral separation will thus be present at most phases .
because we observe the combined light from star and planet , the light is dominated by the former , meaning that we must express the planetary signature as a fraction of the stellar signature . the planet - star flux ratio
, @xmath25 can be expressed as a function of orbital phase ( @xmath26 ) and wavelength ( @xmath27 ) , such that @xmath28 the form of this function is similar to that used by in optical studies ( e.g. @xcite ) but with the geometric terms describing the albedo , planetary radius and orbital semi - major axis combined into a single function @xmath29 .
the value of @xmath30 can be derived from model spectra at the desired value of @xmath27 and represents the _ maximum _ planet - star flux ratio , observed at phase 0.5 , when @xmath31@xmath14 .
the phase angle @xmath32 combines the orbital inclination , @xmath16 , and orbital phase effects and is defined as @xmath33 the phase function @xmath34 may also either be empirically determined or modelled . for a cloud free model
, @xcite showed that planets reflect most efficiently shortward of , where photons undergo rayleigh scattering before being absorbed .
@xcite found that the form of the phase function is strongly dependent on the particle size at 5500 , with larger particles giving strong back scattering and albedos peaking at values similar to the jovian planets in out own solar system .
we have investigated the form of the phase function for based on the results of bha05 .
[ fig : phasfunc ] is a plot of @xmath34 for the case where @xmath31@xmath14 showing the close similarity of the venus phase function and bha05 phase functions for since we are concerned with the detection of the planetary signal and not the characterisation ( i.e. we are no attempting to distinguish between phase function forms ) in this paper , we have adopted the venus phase function in subsequent analyses .
this function has the advantage of being able to mimic inclination effects through use of the @xmath32 parameter , not possible with our empirical function , and is a valid approximation as our spectra were obtained with phases .
[ section : obsdata ] we present the observations made with the phoenix spectrograph @xcite at gemini south on 2006 january 15 & 16 .
densely sampled timeseries were recorded using the 256 @xmath35 1024 insb aladdin ii array .
in addition to the spectra , observations of the bright a1 iii star , , were made to enable careful monitoring of atmospheric absorption features .
because of vignetting and cosmetic issues , the useful region of spectrum was trimmed in the dispersion direction during processing to give a useful area of .
the spectral range of 81.74 covers the 21807.32 to 21889.06 ( 2.18 -2.19 ) region of the nir at a spectral resolution of . with the 1024 pixel array , this gives a mean pixel resolution of 0.096 per pixel which is equivalent to a 1.32 velocity increment at the central wavelength .
the journal of observations is recorded in table [ tab : journal ] .
[ section : detector ] the aladdin array suffers from a number of significant cosmetic defects .
these include a number of dust spots and elongated features .
hot pixels are also present and were flagged during the extraction process to ensure that they were not included .
the left side of the detector appears to exhibit sensitivity which alternates between adjacent rows .
this stripe pattern produces a ripple at the @xmath04% level but is not present on the right hand portion of the detector .
the characteristics of the aladdin detector also necessitated an observing strategy to monitor the faint residual charge persistence signature common in this kind of detector when working in the infrared @xcite .
observations were thus made by alternating the position of the star on the slit and thus the spectrum on the detector in an abba type pattern to allow any persistence signature to be monitored .
each of the 4 sub - exposures in an abba sequence comprised a number of further sub - exposures .
short exposure times help to minimise dark current and may help to reduce persistence levels . for 40s exposures ,
2x20s exposures were used .
similarly , 120s
= 2x60s and 160s = 4x40s .
@xcite has however found that continual flushing of the array can increase the dark current floor , and _ increase _ the persistence current .
+ + [ fig : ccd_flatfield ] we could not detect persistence of the spectra which typically peaked at 5000 e@xmath36s .
the spectra however yielded a maximum of 12500 e@xmath36s and did leave a faint persistence trace at a level of 1.6@xmath2 of the sky background or 0.13% of the peak counts when switching between a and b positions .
for , telluric lines may shift wavelength throughout the night by a few tens of @xcite , but this effect translates to a shift of 1/130th of a pixel .
hence the persistence effect when coadding these spectra is negligible .
the effect is more crucial , if significant variation in the position of features is expected , such as from an orbiting planet whose signature is doppler shifted during its orbit . if a signal of similar relative strength to that in the spectra is seen in the spectra , the blurring effect is still expected to be a second order phenomenon and we do not expect to detect such a signature .
[ section : data_reduction ] pixel to pixel variations were removed using flat - field exposures taken with an internal tungsten reference lamp . in order to create a reliable balance frame to remove the pixel sensitivity variations
, we divided a gaussian blurred ( using a fwhm of 7 pixels ) version of the master flat field image by the original master flat field image .
[ fig : ccd_flatfield ] ( top ) shows the flatfield image normalised to unity .
the large - scale variations at the 5 per cent level shown in fig .
[ fig : ccd_flatfield ] ( bottom ) are a consequence of broad sensitivity variations on the aladdin ii array .
we found that these variations were dependent on the illumination of the array , being both source and time dependent , resulting in an inability to remove the pattern using standard flatfielding techniques .
for example , we extracted the flat - field variation with a profile used for extraction of our object frames .
the morphology of the flatfield ripples was found to differ from the continuum ripple in the extracted object frame .
similarly , since the ripple pattern was source dependent , we were unable to flatfield the spectra by dividing by the standard star , .
while the pattern for a given object remained stationary to first order throughout each night of observations , second order time dependent variation was also seen . nodding the telescope ( see below ) between two positions on the slit resulted in a drift of the a and b positions of several pixels throughout the night . during extraction
we therefore traced each spectrum independently .
since the ripple pattern described above appeared different for hd 75289b than for the flat field , we removed it by fitting splines to the continuum of the extracted spectra .
the worst cosmic ray events were removed at the pre - extraction stage using the starlink figaro routine bclean @xcite . instead of making use of the abba sequences to reject sky lines by extracting from the a - b and b - a differences @xcite , we found that the s / n ratios of the extracted spectra were optimised when the sky background was modelled by fitting polynomials of degree 3 to the pixels either side of the spectral profile at each x - position in each frame .
an iterative fitting was used to reject deviant x - position sky fits , thereby rejecting any sky lines .
only a single sky line at @xmath37 was present in the data at a level of @xmath38 per cent in the worst spectra ( on january 15 ) and 1 per cent in a typical spectrum .. the spectra were extracted using echomop , the chelle reduction package developed by @xcite .
the spectra were extracted using echomop s implementation of the optimal extraction algorithm developed by @xcite .
echomop propagates error information based on photon statistics and readout noise throughout the extraction process .
+ [ fig : wavcalib ] at the time of observations , no arc lamp giving sufficient lines to perform a calibration in the 2.18 -2.19 wavelength range was available .
we were thus unable to carry out a conventional calibration using the positions of known emission lines .
instead , we used a spectrum generated from a hitran line list @xcite to identify corresponding features in an observed spectrum of the telluric standard star , .
we used the simple emission line fitting routine emlt in the starlink figaro package to fit gaussians to the inverted spectra in order to identify the positions and width of the lines .
the latter were used as uncertainties in the measurements . by fitting a cubic polynomial to the 12 identified telluric lines ( fig .
[ fig : wavcalib ] ) , we obtained a wavelength calibration which was subsequently used in our analysis .
the @xmath39 of the fit using a cubic polynomial provided improvements by a factor of 31 and 1.8 when compared with a straight line and a quadratic fit respectively .
the 0.023 rms residual scatter in the fit corresponds to at the centroidal wavelength of 21484 , or 0.06 of a resolution element .. [ fig : planet_spectrum ] [ section : extracting ] we first constructed a high s / n ratio master stellar spectrum template to accurately subtract the unshifted starlight from each observed spectrum .
this has the additional benefit of removing ( to first order ) the telluric water vapour lines prevalent in this part of the spectrum . by monitoring the strengths of the telluric lines alone
, we found , as expected , that they vary in strength throughout the night , generally being stronger when was observed at high airmass .
the telluric lines did not all behave in this way however , with some lines weaker than the mean at times while others were stronger . using a single master frame comprising both stellar and telluric lines left residuals of up to 5-@xmath2 times the expected noise level after subtraction of the starlight from each spectrum .
instead we made individual use of the bright star , , which contains only telluric lines in the observed spectral range of our data .
this star was observed at regular intervals on each night of observations .
a mean telluric template spectrum was created for each night of observations by aligning and coadding all the spectra observed throughout that night .
the telluric spectrum was fitted to each spectrum by taking derivatives of the spectra and using splines to calculate the scale factor at points across the spectra .
this process can account for lines which behave independently over the night ( i.e. all telluric lines do not necessarily vary in strength by the same factor at any given time ) and is described in @xcite ( appendix a ) .
the procedure provides a spectrum which closely matches the telluric lines for each observed spectrum and is used to divide out these features , leaving only stellar and planetary features .
a master stellar - plus - planet spectrum is then created for each telluric corrected observation , but excluding the current telluric corrected spectrum .
every stellar - plus - planet spectrum is then divided by an aligned and shifted version of the current master stellar - plus - planet spectrum in order to remove the stellar lines . since a planetary signal
is doppler shifted according to phase relative to the stellar lines only the planetary signature should remain .
the master stellar - plus - planet spectrum which is subtracted from each spectrum in turn contains a blurred copy of the planetary signal .
this will thus attenuate the planetary signal in the frame on which subtraction is being performed , to some degree .
the effect is worst at @xmath40 & @xmath41 where the planet moves very little in velocity space from spectrum to spectrum .
the attenuation correction applied to the matched travelling gaussian used to model the planetary motion when searching for a signal is described in detail in 6 and appendices d2 & d3 of @xcite .
we found however that further time varying systematic residuals remained in the spectra and believe that these remaining residuals are due to the time varying ripple which we were unable to remove from the spectra at the extraction stage .
since we remove this effect using splines , we expect time varying differences in the fits which do not exactly match the observed changes in the ripple pattern due to finite signal to noise in the data .
we therefore implemented a method which removes the residual signal which we treat as pattern noise .
this method using principal component analysis is described by @xcite ( appendix b ) and removes time varying patterns at fixed positions in the spectra .
when the first few principal components are removed , this has little or no effect on any planetary signal since it changes position from spectrum to spectrum .
we discuss the results using this algorithm further in section
[ section : results ] .
[ section : modelspectrum ] the residual spectrum will contain a doppler - shifted copy of the planetary spectrum , which at this stage is still buried in the noise . to reduce the effective noise
, we model the planet s spectrum as the convolution of a doppler - shifted mean line profile ( which we wish to determine ) and an absorption line pattern for an irradiated _ model _ atmosphere spectrum with parameters equal to those of .
our model spectra were generated using the cloud - free `` rainout '' method described in bha05 which improves the earlier models of @xcite ( based on the `` ames - cond '' models of @xcite ) by iteratively reducing the elemental abundances involved in grain formation at a given layer and recomputing the chemical equilibrium at each new set of stratified elemental abundances .
the resulting equilibrium chemistry and opacity sampling after rainout of species is fully self consistent , whereas earlier cond models simply excluded the grain opacities .
thus in the models used in this work , elements which are important in the atmospheres of cool stars such as ti and v are significantly depleted leading to negligible concentrations of tio and vo .
day - night temperature gradients were modelled under the assumption that concentric neighbouring rings of different temperature interact very little via radiative transfer processes . for a fuller description of the model opacities and setup
see @xcite , @xcite and bha05 . the temperatures on the dayside of the planet
lead to an atmospheric chemistry dominated by h@xmath21 , he , h@xmath21o and co ( bha05 ) . the dominant opacities in the 2.18 - 2.19 wavelength span of our data however are due to h@xmath21o .
we derived the necessary list of line positions and depths for least squares deconvolution from a model spectrum before instrumental and stellar line broadening were added to the spectra and for the case when the planet is in conjunction with the star ( i.e. at orbital phase 0.5 ) . in fig .
[ fig : planet_spectrum ] ( top ) , we plot the model spectrum generated for .
the spectrum has been convolved with a gaussian to mimic the resolution of the observations and rotationally broadened to under the assumption that the planet is tidally locked to the star .
[ fig : planet_spectrum ] also shows a typical spectrum ( middle ) and spectrum ( bottom ) .
the tick marks below then spectrum indicate the theoretical positions of 10 stellar lines ( fe and si opacities ) which are listed in the vienna atomic line database , vald @xcite .
additional opacities not included in vald can also be seen ( see fig . [
fig : planet_spectrum ] caption ) .
the spectra are thus clearly dominated by telluric lines .
the mean absorption profile is recovered through use of a least squares deconvolution method first demonstrated by @xcite .
a list of absorption lines is derived from the _ model _ atmosphere ( see above ) and the depths , before any kind of broadening , are used to optimally weight each line in the _ observed _ spectrum .
the deconvolution is effectively a sophisticated cross - correlation method which optimally coadds all the aligned absorption profiles while removing any side lobes due to line blending .
the resulting profile is a mean absorption line with s / n ratio boosted , thereby improving the chance of detecting weak signals which are dominated by noise . the current version of the code @xcite which propagates errors from the input spectra has been used extensively for reflected light searches in the optical by @xcite and @xcite . for , we expect @xmath098 planetary absorption features in the wavelength range of our data .
the weighted mean normalised depth @xcite of these lines relative to the planetary `` continuum '' is 0.096 .
the optimal nature of the deconvolution means that all 98 lines in the wavelength range of our observations contribute , but only 8 ( indicated by tick marks below the model planet spectrum in fig .
[ fig : planet_spectrum ] ) of the lines possess a normalised depth relative to the planetary continuum greater than the mean weighted value of 0.096 . of the 98 lines ,
50 possess depths greater than 0.01 of the normalised planetary continuum .
attenuation by a further factor in the combined star+planet spectrum is one of the main parameters which we wish to determine . from photon statistics ( @xmath42 per spectrum , where @xmath43 = number of electrons )
, we expect an _
mean _ s / n limit in our spectra of for both nights of data .
the mean s / n ratios , measured from flat regions in the normalised spectra , were 268@xmath44 for the nights of january 15 & 16 respectively .
the upper and lower limits represent the minimum and maximum recorded s / n ratio on each night .
slit alignment and tracking problems were the cause of low counts in a number of the january 15 spectra with a minimum s / n = 25 ( readout noise = 40 e@xmath45 ) .
the gain in s / n of 3.4 for the deconvolved line profile yielded s / n ratios of 797 @xmath46 over the two nights ( i.e. minimum s / n @xmath0112 and maximum s / n = 1149 ) .
[ section : results ] [ section : matchedfilt ] we model the time - dependent changes in doppler shift and brightness , in the manner described in [ section : detect ] .
this matched - filter enables us to search for features in the timeseries spectra and is described in @xcite ( appendix d ) .
we asses the relative probabilities of the @xmath39 fits to the data by varying @xmath47 and @xmath48 and plotting the improvement in @xmath39 , which is normalised to the best - fitting model . to calibrate any candidate detection
, we construct a simulated planet signal of known @xmath30 that is added to the extracted spectra prior to removal of stellar and telluric features and before least squares deconvolution is carried out . by ensuring the fake planet
is recovered correctly by our procedures , we can be confident of a calibrated detection in the presence of a genuine planet signal .
the significance of the result is assessed using bootstrap statistical procedures based on the random re - ordering of the data in a way that scrambles phases while preserving the effects of any correlated systematic errors @xcite .
the order of the observations is randomised in a set of 3000 trials which will scramble any true planetary signal while enabling the data to retain the ability to produce spurious detections through the chance alignment of systematic errors .
the least squares estimate of log@xmath49 and associated @xmath39 as a function of @xmath50 enable us to plot 68.4 , 95.4 , 99.0 and 99.9 per cent bootstrap limits on the strength of the planetary signal .
cc + & + + & + + [ fig : data_planet ] [ section : searching ] fig .
[ fig : data_planet ] ( left ) shows the phased timeseries deconvolved spectra and the resulting @xmath39 map of log@xmath49 vs @xmath50 . the effective wavelength after performing deconvolution is .
[ fig : data_planet ] ( right ) shows the relative probability @xmath39 map of the planet - star flux ratio @xmath51 vs @xmath50 .
the darkest regions in the map represent the greatest improvement in @xmath39 when fitting the model described in [ section : matchedfilt ] .
we can obtain an estimate of the upper limit of the velocity amplitude of the planet . with and , we find . with the orbital inclination of @xmath14 reported in
[ section : hd75289 ] , we estimate a most probable velocity amplitude of . in
we plot the corresponding motion and recovered position for _ _ k__@xmath52 ( dashed curves and vertical lines respectively ) as a visual guide to the reader .
it is possible to see the non - gaussian systematics in the phase plot in fig .
[ fig : data_planet ] ( top left ) .
we believe that these features are the result of our inability to fully remove the varying telluric signatures and uncharacterised changes in the observed ripple function described in [ section : data_reduction ] . to remove the most significant trends in the timeseries data at fixed positions in wavelength
, we used principal component analysis ( @xcite , appendix b ) .
the eigenvalues of the first two principal components indicated a strong contribution while a less strong contribution from the next five components was then followed by a linear decrease in significance .
we thus investigated removal of the first two , and the first five components , with the resulting input timeseries for the latter case being significantly cleaner of systematic noise features . when we compare fig .
[ fig : data_planet ] ( bottom right ) with fig .
[ fig : data_planet ] ( top right ) , we see how the feature is reduced in significance , leaving several features of comparable strength and low statistical significance .
the feature only marginally lies in the range of possible values since the uncertainty on our @xmath53 measurement yields an upper ( 1-@xmath2 ) limit of .
a planet with this velocity amplitude would imply @xmath54@xmath14 , indicating a planet likely to show transits , an observation not reported in the literature .
the feature is of no greater significance and implies an orbital inclination of @xmath14 respectively .
this is unlikely to give a detection given the relatively low maximum flux we would receive under the reasonable assumption @xcite that heat is not effectively redistributed and the planet is hottest on its dayside .
all candidate features however can confidently be rejected on the basis of the observed flux ratios for a number of systems ( [ section : intro ] ) since the mean implies a planet significantly hotter than even ( see fig .
8 of @xcite ) . under the assumption that all candidate features are spurious
, the overall level of noise in the data enables us to rule out the presence of a planet at _ _
k__@xmath55 with log@xmath56 at the 2-@xmath2 level ( fig .
[ fig : data_planet ] , top right ) . cc + & + + & + + [ fig : sim_planet ]
[ section : simplanet ] in order to assess the feasibility of detecting a planetary signature in the nir , we have carried out simulations using fake data based on the known physical parameters of . by considering the wavelength coverage possible with current nir multiorder cross - dispersed chelle technology ( i.e. ircs / subaru @xcite and nirspec / keck 2 @xcite ) , we estimate that approximately 50 spectra with r @xmath57 , a mean s / n ratio of 300 and encompassing a wavelength range of 1.90 to 2.45 could be obtained per of observations @xcite . with a typical spectral coverage of 70% of this
range our planetary model atmosphere for hd 75289 yields @xmath58 useful lines with depths between 0.05 and 1.0 of the normalised continuum .
we generated model spectra for a g0v star using models @xcite and a hitran model of the telluric lines @xcite .
these spectra were combined with our model planetary spectrum assuming an orbital inclination of , a relative maximum planet - star flux ratio of and a velocity amplitude of .
gaussian random noise was added to the spectra to simulate a s / n ratio of 300 .
we assumed observations were taken over two nights with a ten night phase gap .
this strategy optimises observations at phases close to @xmath59 , while minimising the timebase .
we attempted to recover of the planetary signal in the same way as described in previous sections .
the phased timeseries spectra are presented in fig . [
fig : sim_planet ] ( top left ) , and show a detection well clear of the 99 per cent confidence level ( fig .
[ fig : sim_planet ] , top right ) .
since the semi - major axis of the orbit and the period are both known , we can use our estimate of @xmath60 to determine the inclination .
our matched - filter analysis gives @xmath61 leading to an inclination determination of 72@xmath14 .
hence we overestimate the inclination by 5@xmath14 .
the detection limit at the simulated velocity of the planet is log@xmath62 .
equivalently , at the projected velocity of the planet , we expect to recover the signal at the limiting confidence for data with s / n @xmath63 200 .
we carried out a further test to estimate the feasibility of recovering the planetary signal in the case where the line strengths are incorrect .
we assumed that the positions of all lines in our input model planetary spectrum could be incorrect by some fraction .
the depth of each line , but not the position , was modified by a gaussian random fraction of the original strength .
we then attempted to recover the planetary signal with the modified line list ( used for deconvolution ) and found that for data with s / n = 300 , we still recover the planet with a 99.9 per cent confidence limit for strengths which have been modified by on average 50 per cent .
this simulation essentially mimics the effects of non - optimal extraction of the lines using our deconvolution code . our ability to recover
the planet is more sensitive to incorrect positioning of lines however .
we find that a clear 99.9 per cent confidence detection is achieved in the limiting case of randomly re - positioning 15 per cent of the lines before deconvolution .
the main reason these mismatches , which we discuss in the next section , is likely the uncertainties in opacities , opacity oscillator strengths and atmospheric physics of the planet .
we note that in reality , removal of the telluric lines at the shorter wavelengths of the above range to 2.08 ) does not affect our ability to make a clear detection .
we also simulated recovery of the planet for a spectral resolution of r @xmath64 .
the main advantage of high resolution data would be the increase in depth of the absorption lines relative to the continuum .
the results are shown in fig .
[ fig : sim_planet ] ( bottom ) and show the clear gain attained by doubling the resolution .
our limiting 99 per cent confidence threshold in this instance would enable us to detect a planetary signal at a level of log@xmath65 .
the planet is recovered with @xmath66 , leading to an inclination determination of 69.2@xmath14 . since a narrower slit
will be used to make observations at higher resolution , we may not observe this degree of improvement in reality .
our confidence indicates we would still detect the planet with or approximately 1/4 the number of photons required in the r @xmath57 limiting case .
we discuss the prospect of obtaining spectra of this nature in section
[ section : discussion ]
[ section : discussion ] the methods employed in this paper provide a very encouraging indication that a planetary signal can potentially be extracted from high resolution timeseries spectra .
the fact that we do not detect a signal can be attributed largely to systematic noise introduced from difficulties in consistently normalising each spectrum in our timeseries .
this arises largely because of the systematic and varying large scale flat fielding errors associated with the detector employed in the observations as discussed in
[ section : data_reduction ] .
we have shown that the feature detected in our spectra is unlikely to be a real planetary feature since its strength is much reduced when additional principal components , which describe ( systematic ) trends in the data , are removed .
the eigenvector describing the contribution of principal components shows a very strong contribution from the first 2 components with an exponential decay which levels off after the 7th principal component . at this level ,
the detected feature becomes consistent with the shot noise in the spectra .
the performance of nir detectors is clearly of great importance if we wish to detect small signals dominated by a nearby bright object .
more modern detectors such as the 1024@xmath67 array used in conjunction with nirspec at keck 2 @xcite do not show the characteristics @xcite of the older array used to secure the data in this work .
future detectors promise even greater stability with @xcite finding 0.05 per cent flatfield stability over a period of one day for a 2048@xmath67 hawaii-1rg engineering array .
current observational evidence for a planetary signal at therefore remains inconclusive .
initial attempts to search for extrasolar planetary spectra ( by the method of difference analysis where the spectrum of the host star seen during eclipse of the planet is subtracted from the spectrum of the combined star and planet , seen out of eclipse ) revealed null results .
the analysis by @xcite failed to detect the predicted continuum ` bump ' at 2.1 - 2.2 at a level of log@xmath68 .
@xcite did not detect the eclipse shape from 2.3 secondary eclipse observations of , but nevertheless the mean drop in flux ( based on two measurements ) of 0.115 @xmath69 0.139 per cent indicates a planet - star flux ratio of 0.0010 @xmath69 0.0009 . while this confirms model predictions for a planet - star flux ratio , @xcite note that the size of the uncertainty does not exclude the blackbody result . while advances have been made in detecting spectral features in the mid - infrared 7.5 - 13.2 interval @xcite using spitzer space telescope observations
, models would suggest that the higher contrast ratio in the nir will necessitate the kind of signal enhancement techniques used in this work .
it is clear that there is still much to learn about the atmospheres of cegps .
both @xcite and @xcite have found that there is little evidence for the h@xmath21o opacities present in model spectra .
@xcite however find evidence for silicate clouds in the spectrum of .
the simulations we have carried out in this work make the assumption that the opacities in current models are correct , even if their strengths are not .
if many species are omitted or are simply not present in reality , this could have significant impact on the detection method , either for better or worse .
as with the @xcite search for the 2.1 - 2.2 bump , our models contain unreliable oh opacities and oscillator strengths and missing ch@xmath70 opacities which are seen in t dwarf spectra @xcite .
this is clearly important when using methods which aim to detect the ` continuum ' shape , while we found ( [ section : simplanet ] ) that incorrect positioning of greater than 15 per cent of lines would affect our ability to make a clear detection .
conversely , competing models could be used to obtain an optimal extraction of the absorption signature of a planetary spectrum . switching off model opacities for species which are known to have unreliable positions
may be necessary to enable recovery of the planetary signal . in the case of a clear detection
, it should then also be possible to distinguish between differing phase function models by monitoring the improvement in @xmath39 in our matched filter analysis .
if the albedo spectrum is largely independent of wavelength in the regions where absorbing species do not play a significant role , including 2.2 ( see fig .
8 of @xcite ) , we would expect a very small reflected light component in the nir , given the upper limits found in the optical .
our data contain only 10 lines arising from the spectrum of itself , with a mean depth 2.7 times the mean depth of the 98 lines we expect to see in the planetary atmosphere .
nevertheless , if the 10 lines are seen as a component reflected from possible clouds in the atmosphere of , we expect little gain in the signal if the upper albedo limit of 0.12 @xcite is taken into account .
the maximum reflected light signal in this case is only log@xmath71 , over an order of magnitude less than the nir log@xmath72 estimate adopted throughout this work
. [ section : summary ] we have presented an analysis and feasibility study for the detection of the nir spectroscopic signature of the close orbiting extrasolar giant planet , . we do not detect the planet with the present gemini / phoenix data but through simulations based upon models and observational constraints we expect that current cross - dispersed spectrographs , such as ircs / subaru @xcite and nirspec / keck 2 @xcite , operating at resolutions in the region of r @xmath57 are sufficient to obtain a detection with 99.9 per cent confidence .
facilities which could offer higher resolution and broad wavelength coverage will be an enormous benefit . with a large spectral capability
, it will be possible to determine the nir spectral energy distribution in j , h & k regions , complementing observations currently being made at longer wavelengths .
this paper is based on observations obtained with the phoenix infrared spectrograph , developed and operated by the national optical astronomy observatory .
we would like to thank the anonymous referee for useful comments .
jrb was supported by a pparc funded research grant during the course of this work .
n. , ives d. , 2006 , in high energy , optical , and infrared detectors for astronomy ii .
edited by dorn , david a. ; holland , andrew d .. proceedings of the spie , volume 6276 , pp .
62760o ( 2006 ) .
high - speed multiple window readout of hawaii-1rg detector for a radial velocity experiment d. , richardson l. j. , seager s. , harrington j. , 2006 , in arnold l. , bouchy f. , moutou c. , eds , tenth anniversary of 51 peg - b : status of and prospects for hot jupiter studies infrared radiation from hot jupiters .
pp 218225 k. h. , blum r. d. , joyce r. r. , sharp n. , ridgway s. t. , bouchet p. , van der bliek n. s. , najita j. , winge c. , 2003 , in guhathakurta p. , ed . ,
discoveries and research prospects from 6- to 10-meter - class telescopes ii . edited by guhathakurta , puragra
. proceedings of the spie , volume 4834 , pp .
353 - 363 ( 2003 ) .
4834 of presented at the society of photo - optical instrumentation engineers ( spie ) conference , the phoenix spectrograph at gemini south n. , tokunaga a. t. , terada h. , goto m. , weber m. , potter r. , onaka p. m. , ching g. k. , young t. t. , fletcher k. , neil d. , robertson l. , cook d. , imanishi m. , warren d. w. , 2000 , in iye m. , moorwood a. f. , eds , proc .
4008 , p. 1056
- 1066 , optical and ir telescope instrumentation and detectors , masanori iye ; alan f. moorwood ; eds .
ircs : infrared camera and spectrograph for the subaru telescope .
pp 10561066 i. s. , becklin e. e. , bendiksen o. , brims g. , canfield j. , figer d. f. , graham j. r. , hare j. , lacayanga f. , larkin j. e. , larson s. b. , levenson n. , magnone n. , teplitz h. , wong w. , 1998 , in fowler a. m. , ed . , proc .
3354 , p. 566 - 578 , infrared astronomical instrumentation , albert m. fowler ; ed .
design and development of nirspec : a near - infrared echelle spectrograph for the keck ii telescope . | we present a search for the near infrared spectroscopic signature of the close orbiting extrasolar giant planet .
we obtained @xmath0230 spectra in the wavelength range using the phoenix spectrograph at gemini south . by considering the direct spectrum , derived from irradiated model atmospheres , we search for the absorption profile signature present in the combined star and planet light .
since the planetary spectrum is separated from the stellar spectrum at most phases , we apply a phase dependent orbital model and tomographic techniques to search for absorption signatures . because the absorption signature lies buried in the noise of a single exposure we apply a multiline deconvolution to the spectral lines available in order to boost the effective s / n ratio of the data .
the wavelength coverage of 80 is expected to contain @xmath1 planetary lines , enabling a mean line with s / n ratio of 800 to be achieved after deconvolution .
we are nevertheless unable to detect the presence of the planet in the data and carry out further simulations to show that broader wavelength coverage should enable a planet like to be detected with 99.9 per cent ( 4 @xmath2 ) confidence .
we investigate the sensitivity of our method and estimate detection tolerances for mismatches between observed and model planetary atmospheres .
[ firstpage ] line : profiles methods : data analysis techniques : spectroscopic stars : late - type stars : individual : stars : planetary systems |
because of the limited accuracy of available functionals , density functional theory ( dft ) fails for a large number of correlated systems .
there are numerous examples for which dft can not describe neither their ground state properties , nor their excitation properties .
thus , in order to describe many - body effects arising in the strongly correlated systems containing for instance transition elements or @xmath3 electrons , theories were designed to take into account the interaction among correlated orbitals explicitly .
the dft+@xmath0 method@xcite or the combination of dft with dynamical mean field theory method ( dft+dmft ) @xcite have been successfully applied to a large number of systems in the last twenty years . in particular , these methods have been particularly useful to study the ground state and the photoemission spectra of mott insulators such as bulk actinide@xcite and lanthanide oxide@xcite .
for instance , to date dft+dmft is the only method to give a good description of photoemission spectra of both @xmath1 and @xmath2 cerium @xcite .
however , in these frameworks and applications , the interaction among correlated orbitals , named @xmath0 , remained most often an input parameter .
as a consequence , there is a stringent need to calculate the magnitude of the interaction @xmath0 , in order to recover a truly _ ab initio _ scheme .
methods were then proposed to evaluate @xmath0 from first - principles .
the constrained local density approximation ( clda ) method@xcite deduces the value of @xmath0 from the variation of energy with respect to the number of correlated electrons on an atom .
later , cococcioni _ et al _ @xcite generalized this method to a non basis dependent scheme .
finally , the constrained rpa method@xcite uses the linear response theory to compute the value of the screened interaction .
screening processes corresponding to electron hole transitions among the correlated orbitals are however excluded from the calculation .
indeed an exact many - body scheme would already contain all the screening processes associated to the degrees of freedom involved in the calculation .
therefore , some transitions have to be disregarded to avoid double - counting .
the crpa scheme very clearly defines which screening processes have to be taken into account .
the method has been implemented in several electronic structure codes using the lmto@xcite , flapw@xcite , fplmto@xcite , plane wave@xcite and projector augmented - wave ( paw)@xcite methods and applied to different systems in the last few years@xcite .
an important point emphasized in several works is that the calculation of the crpa screened interaction based on a previous dft calculation depends crucially on the definition of a many - body model@xcite .
the many - body model is defined by a set of local orbitals together with the interactions among them .
a crpa calculation of a model would require first the definition of a set of local orbitals , and second a consistently calculated screened interaction . whereas the choice of the angular momenta of the selected orbitals unambiguously defines the angular part of the local orbitals , the definition of the radial part is more subjected to variation .
generally , it relies on the use of localized wannier functions , which are built as a unitary transform of kohn - sham orbitals in an energy window@xcite .
the key point is to construct the model specific crpa screened interaction .
it is especially important to study this dependence as a function of the localization of the wannier orbitals .
indeed , correlated orbitals used in dft+@xmath0 and dft+dmft can be formulated as wannier orbitals with different energy windows@xcite depending on implementation choices .
thus the coherence of the dft+@xmath0/dft+dmft calculations and of the crpa calculation can only be guaranteed if both methods use the same wannier functions .
implementations of dft+@xmath0 in popular codes very often use atomic orbitals@xcite , whereas implementations of dft+dmft use wannier functions@xcite .
it is thus expected that the value of @xmath0 used in dft+@xmath0 and dft+dmft should differ .
some works indeed discuss the calculation of @xmath0 for a given energy window@xcite but there are no calculation of a screened interaction in the same basis as the one used in dft+@xmath0 codes . even for dft+dmft calculation ,
the definition of an energy window is especially important for systems with entangled correlated bands@xcite .
moreover and especially for very localized systems such as mott insulators , self - consistent calculations over @xmath0@xcite are desirable because the erroneous lda or generalized gradient approximation ( gga ) band structures can not correctly describe the screening in these systems .
the coherence of basis between the dft+@xmath0 and the crpa calculation is then of the utmost importance .
this is in particular the case for cerium and uranium dioxide for which no self - consistent calculations of @xmath0 exist .
in this work , we report an implementation of the crpa method using the paw method in the abinit package@xcite .
the implementation is versatile enough to allow for the calculation of screened interaction in both the same wannier basis as our recent dft+dmft implementation@xcite and in atomic orbitals and as in our dft+@xmath0 implementation@xcite .
then , we show that the self - consistent dft+@xmath0 calculation of the screened interaction in the crpa method in strongly correlated systems is essential to describe the static and dynamical screened interaction .
we exemplify our study with two important applications : uo@xmath4 and cerium .
as the crpa method is described in details elsewhere @xcite , we only sketch the most important points here .
the screened interaction is in general a four index matrix that is defined as @xmath5 with @xmath6 the bare coulomb interaction and @xmath7 the indices of the correlated orbitals . in this work ,
the use projected wannier functions as defined in refs . and .
similar wannier functions have been used in crpa calculations@xcite .
we first introduce the auxiliary wavefunctions @xmath8 as @xmath9 for a given atomic site @xmath10 , we call @xmath11 the bloch transform of isolated atom kohn sham orbitals with projected angular momentum @xmath12 .
@xmath13 are kohn - sham orbitals for k - point @xmath14 , band index @xmath15 and spin @xmath16 .
@xmath8 is thus a weighted sum of kohn - sham orbitals .
this sum extends over a given number of kohn sham orbitals that can be defined by an index range or alternatively by an energy window @xmath17 .
the orthonormalization of @xmath8 leads to well defined wannier functions @xmath18 , unitarily related to @xmath13 . in the limit of a large number of kohn sham bands ,
the projection in eq .
( [ eq : tchi ] ) becomes complete and the wannier functions @xmath18 become equivalent to atomic orbitals @xmath11 .
the definition of @xmath0 is very much similar to the coulomb integrals used in quantum chemistry , but with the screening of the frequency dependent crpa dielectric matrix , @xmath19 .
this dielectric matrix can be expressed as a function of the crpa non - interacting polarizability @xmath20 and the bare interaction @xmath6 as ( in the matrix notation ) @xmath21
@xmath22 contains all electron - hole screening processes except the ones that are internal to the correlated orbitals of the model .
it can be conveniently written as @xmath23 in eq .
( [ eq : pola ] ) , @xmath24 are band indices , @xmath25 are k - points in the brillouin zone and @xmath26 is the occupation number for band @xmath27 , spin @xmath16 and k - point @xmath28 .
if the correlated bands in the model are completely isolated from the other ones , then we can assume that@xcite @xmath29 when ( @xmath30 ) and ( @xmath31 ) are both correlated bands and @xmath32 otherwise .
for example , a model could define correlated orbitals as wannier orbitals constructed only from the @xmath3 bands .
nevertheless , with this specific choice , the wannier orbitals have some weight on other orbitals : oxygen-@xmath33 for oxides@xcite or @xmath34 for pure metals .
the intensity of this weight depends on the hybridization of @xmath3 orbitals with the other orbitals .
this last definition of correlated orbitals is not the one used in most implementations of dft+@xmath0 in modern codes@xcite . in these implementations ,
correlated orbitals are most often atomic orbitals which thus corresponds to wannier functions @xmath18 for a large window of energy@xcite . if the bands are completely entangled or if one defines wannier functions @xmath18 from a larger energy window , then the preceding assumption of eq .
( [ eq : pola0 ] ) can not be made@xcite and some authors have proposed the more general assumption@xcite : @xmath35 \\ \times \left[\sum_{m_2 } |\bra\psi^\sigma_{n'\mathbf{k'}}| w^\sigma_{m_2\vk'}\ket|^2\right ] .
\label{eq : weightpola}\end{gathered}\ ] ] if the correlated bands are not entangled and if the wannier functions are defined from these correlated bands only , then eq . ( [ eq : weightpola ] ) simply reduces to eq .
( [ eq : pola0 ] ) .
fully screened coulomb interaction @xmath36 corresponds to @xmath32 . in this work
, we use an implementation of the calculation of the dielectric function in paw@xcite . from the screened interaction expressed in the kohn sham basis ,
we compute the screened interaction in eq .
( [ eq : ummmm ] ) , using the wannier functions as defined in refs . and . the weight of wannier functions necessary for eq .
( [ eq : weightpola ] ) is evaluated within paw following ref . .
then , the values of the famous hubbard @xmath0 and hund @xmath37 are simply extracted by taking the average among the considered localized orbitals : @xmath38 @xmath39 note that this definition of @xmath0 is also sometimes referred to as the @xmath40 slater integral .
we emphasize that this definition is different from the average of the diagonal elements of the coulomb interaction matrix @xmath41 ( see e.g ref . ) . in particular diagonal elements
are usually larger , and thus @xmath42 .
however our definition is coherent with the @xmath0 used in the dft+@xmath0 approach@xcite and physically describes the average interaction between electrons in all orbitals .
appendix [ sec : udetails ] gives the details of the implementation and the peculiarities of the paw formalism for the calculation of @xmath0 .
appendix [ sec : appendix ] gives a benchmark of our implementation with respect to recent calculations on srvo@xmath43 .
the calculation are performed within the projector augmented - wave ( paw ) method as implemented in abinit@xcite .
the valence states include @xmath44 for oxygen , @xmath45 for cerium and @xmath46 for uranium respectively .
two projectors per angular momentum are used , and completeness of the projector basis is checked by increasing their number .
the parameters of calculation are chosen such that the precision on the static values of @xmath0 and @xmath37 is better than 0.2 ev .
for uo@xmath4 , we thus use a 4x4x4 k - point grid , and energy cutoffs for the wavefunctions , the dielectric function and the bare coulomb interaction are respectively 15 ha , 5 ha , and 35 ha . 100 bands are sufficient for the calculation of the polarisability . for cerium , we use a 8x8x8 k - point grid and energy cutoffs for the wavefunction , the dielectric function and the bare coulomb interaction are respectively 15 ha , 10 ha and 35 ha ( for large values of the volume , a 4x4x4 k - point grid was sufficient ) . for the static screened exchange @xmath37 in cerium ,
a value of 120 ha was however necessary but a 4x4x4 k - point grid is sufficient as well as 100 bands for the calculation of the polarisability .
this high value of the cutoff originates from the calculation of oscillator matrix elements in the paw formalism ( see appendix [ sec : udetails ] ) .
a smearing of the kohn sham occupations of 0.1 ev is used . for all systems ,
experimental structural parameters are used : 5.47 for uo@xmath4 , 4.83 for @xmath1 cerium , and 5.16 for @xmath2 cerium . unless specified , all dft+@xmath0 calculations use the full localized limit ( fll ) double counting correction@xcite . a discussion on the role of the double counting correction is given in appendix [ app : dc ] .
for uo@xmath4 , dft+@xmath0 are performed for simplicity in the ferromagnetic configuration , which requires a symmetry breaking@xcite and we use the correlated @xmath3 density matrix found in ref . .
the goal of this section is to define models of correlation for uranium dioxide and cerium . for each model ,
one thus defines an energy window that encompasses at least the bands which have the same main character as the selected orbitals . from the definition of the window energy , wannier functions of the correlated orbitals
are built according to the scheme of ref . .
from the choice of the correlated orbitals , the crpa polarizability is built by excluding some screening channels corresponding to correlated orbitals .
we distinguish different ways to exclude the screening according to eq .
( [ eq : pola0 ] ) or to eq .
( [ eq : weightpola ] ) .
.different models for the description of correlation in uo@xmath4 . from the top to the bottom
, the @xmath3 wannier function are expected to be more and more localized .
for the first three model ( resp .
last four models ) , eq .
( [ eq : pola0 ] ) ( resp eq .
( [ eq : weightpola ] ) ) is used to compute @xmath47 .
the three models @xmath3-ext ( b@xmath48 ) ( b@xmath49 ) ( b@xmath50 ) use a fixed number of bands ( respectively 28 , 38 and 48 ) that corresponds to the energy windows given . [ cols="^,^,^,^ " , ]
whereas all results in tab . [
tab : uo2.static ] and [ tab : cerium.static ] are obtained with the fll double counting correction@xcite , we give in this appendix results obtained with the around mean field ( amf ) double counting correction@xcite .
this is _ a priori _ less justified than the fll double counting correction because in our dft+@xmath0 calculations electrons are not delocalized among all f - orbitals . for uranium dioxide ( resp . @xmath2
cerium ) using the @xmath3-ext ( b@xmath48 ) ( resp .
@xmath3-ext ( b@xmath51 ) ) model , we find @xmath52 = 4.5 ev ( resp .
4.2 ev ) instead of 5.0 ev ( resp .
5.9 ev ) for the fll double counting .
these differences can be understood from the analytical expression of the dft+@xmath0 kohn sham potential in amf ( see e.g. ref . ):
the hubbard bands and especially the minority spin bands are lowered in energy with respect to fll .
it can thus be expected that the contribution of upper hubbard bands to the screening processes is larger , hence the lower value of @xmath0 . for cerium ( with only one correlated electron ) , the shift of hubbard bands is -1.5 ev , and even larger for the minority bands .
the value of @xmath0 is thus greatly reduced .
moreover , the cerium dos hence obtained in amf would badly compare with experimental photoemission spectra@xcite . as the fll dos is better
, we can roughly expect that the value of @xmath0 obtained in fll is better .
however , and more generally , this comparison calls for a more general self - consistent scheme with a more justified double counting correction such as in the gw+dmft@xcite scheme .
we thank gabriel antonius , jordan bieder , silke biermann , matteo giantomassi , franois jollet , cyril martins , priyanka seth and marc torrent for discussions about this work .
we acknowledge prace for awarding us access to resource marenostrum iii based in spain at bsc .
this work was granted access to the hpc resources of ccrt and cines under the allocation 2012096681 made by genci ( grand equipement national de calcul intensif ) . | we report an implementation of the constrained random phase approximation ( crpa ) method within the projector augmented - wave framework .
it allows for the calculation of the screened interaction in the same wannier orbitals as our recent dft+@xmath0 and dft+dmft implementations .
we present calculations of the dynamical coulomb screened interaction in uranium dioxide and @xmath1 and @xmath2 cerium on wannier functions .
we show that a self - consistent calculation of the static screened interaction in dft+@xmath0 together with a consistent wannier basis is mandatory for @xmath2 cerium and uranium dioxide .
we emphasize that a static approximation for the screened interaction in @xmath1 cerium is too drastic . |
it is clear from the papers presented at this workshop that studies of ultrasoft narrow - line seyfert 1 galaxies ( nls1s ) have undergone exciting growth over the past few years .
progress has been made defining their phenomenological properties ; nls1s show extreme spectral shapes and variability at a variety of wavelengths ( e.g. , x - ray , optical and radio ) . while a solid physical understanding of the origin of their extreme properties has not yet emerged , they are plausibly objects with high values of the mass accretion rate relative to the eddington rate ( @xmath0 ) . in this paper
, we will briefly discuss potential connections between luminous nls1s and broad absorption line qsos ( balqsos ) , concentrating on the x - ray point of view .
we will informally advocate such connections to the greatest extent possible in an attempt to stimulate further work in this area . in the x - ray band ,
even the basic phenomenological properties of balqsos are poorly known ( see 2 ) .
because of their low x - ray fluxes , the present observations do not allow one to draw a characteristic x - ray spectral energy distribution for a balqso .
however , with the advent of the new generation of x - ray observatories , one hopes that x - ray studies of balqsos will undergo exciting growth in the coming decade comparable to that achieved for nls1s in the past few years .
most luminous qsos are thought to have bal outflows which comprise a major part of the nuclear environment ( e.g. , weymann 1997 , hereafter w97 ) .
furthermore , the bal phenomenon has interesting connections with the ` radio volume control ' of qsos ( e.g. , 4 of becker et al .
2000 and references therein ) , and bal outflows may clear gas from qso host galaxies and thereby affect star formation and qso fueling over cosmic timescales ( e.g. , fabian 1999 ) . with x - rays , one would like to study the bal wind in absorption to determine properties such as its column density , ionization state , abundances , and covering factor .
one would also like to probe the nuclear geometry ( e.g. , using the iron k@xmath1 line and x - ray variability ) and determine if balqsos have normal qso x - ray continua underlying their absorption .
studies with , , and have established that balqsos are very weak emitters in the 0.12.0 kev band , and that with a few notable exceptions they are also weak when studied with more penetrating 210 kev x - rays ( e.g. , green & mathur 1996 ; gallagher et al .
1999 ; mathur et al .
2000 ; see brandt et al .
2000 for a review ) .
the higher detection fraction obtained in hard x - rays ( @xmath2% ) is generally consistent with the presence of heavy internal x - ray absorption . if balqsos have normal underlying qso x - ray continua , then large column densities of @xmath3 @xmath4 are required to extinguish the x - ray emission in several cases ( see figure 1 ) .
these large x - ray column densities have important physical implications .
for example , if the x - ray absorption occurs in gas at a distance @xmath5 cm outflowing with a significant fraction of the terminal velocity measured from the uv bals , large mass outflow rates ( @xmath6 m@xmath7 yr@xmath8 ; @xmath9 ) and kinetic luminosities ( @xmath10 ) are derived .
while these can be reduced if the x - ray and uv absorbers in balqsos differ , the possibility of such powerful outflows demands further study . at present
, it has been possible to measure the intrinsic x - ray continuum shape well for only one ` bona - fide ' balqso , pg 2112 + 059 ( @xmath11 ; gallagher et al .
2000 , in preparation ; see jannuzi et al .
1998 for the uv spectrum ) .
this object is one of the most luminous pg qsos at @xmath12 with @xmath13 .
our recent asca spectra show strong absorption below @xmath14 kev , but the rest - frame 310 kev continuum appears relatively unaffected by absorption ( see figure 2 ) .
we measure a power - law photon index of @xmath15 ( 90% confidence ) .
this first detection of a typical power - law x - ray continuum for a balqso is reassuring and argues that balqsos indeed have normal x - ray power sources underneath their absorption .
several papers have qualitatively commented on observational similarities between nls1s end of boroson & green ( 1992 , hereafter bg92 ) eigenvector 1 ( i.e. , the ` i zwicky 1 objects ' ) .
we use the term ` nls1s ' loosely to refer to such objects . ] and balqsos ( e.g. , boroson & meyers 1992 , hereafter bm92 ; brandt et al .
1997 ; laor et al .
1997 ; lawrence et al .
1997 ; leighly et al .
two demographic facts must immediately be considered : ( 1 ) nls1-type objects are found over a wide range of luminosity ranging from low - luminosity seyferts such as ngc 4051 to luminous qsos such as phl 1092 , while bals are seen only in luminous qsos , and ( 2 ) most and perhaps all qsos have bal regions while most qsos do not show the characteristic ` nls1 ' properties . the first suggests that only the luminous nls1s may be unified with balqsos , and the rest of this paper shall focus on such objects .
fact 2 is double - edged : while it is implausible to try to unify _ all _ balqsos and luminous nls1s , it appears that there must be _ some _ connection between them .
the objective is then to identify the subset of balqsos that are the ` cousins ' of luminous nls1s .
based on a comparison of na i emission in i zwicky 1 and balqsos , bm92 suggested that `` i zwicky 1 is a promising candidate for a mg ii balqso in which our line of sight does not pass through a bal cloud . ''
low - ionization balqsos , showing absorption by species such as mg ii and al iii , comprise 1015% of the balqso population and often have reddened continua and extremely weak [ o iii ] emission ( e.g. , weymann et al .
1991 , hereafter w91 ; bm92 ; sprayberry & foltz 1992 ) .
they are thought to have large amounts of relatively cool gas and dust in their nuclei , and they are sometimes postulated to be young , recently activated qsos .
it appears that low - ionization balqsos have larger bal - region covering factors ( @xmath16 ) than typical balqsos ( e.g. , bm92 ; turnshek et al .
1997 ) .
luminous nls1s and low - ionization balqsos appear to share many common properties , lending plausibility to a connection between them .
these are listed in table 1 , where a few relevant references for the low - ionization balqso properties are also given .
in several cases the low - ionization balqso properties have only been measured in small and statistically incomplete samples ; better systematic studies are clearly needed so that one can reason armed with more than anecdotal evidence .
systematic h@xmath17 fwhm measurements are in particular required .
the four low - ionization balqsos in bm92 have h@xmath17 fwhm of @xmath183000 km s@xmath8 , fairly narrow for high luminosity objects .
however , there also appear to be at least several low - ionization balqsos that have substantially broader h@xmath17 fwhm of 40006000 km s@xmath8 ( e.g. , mcintosh et al .
1999 ; m. brotherton , private communication ) . while a few radio - loud nls1s and low - ionization balqsos exist ( e.g. , becker et al .
2000 ; grupe et al .
2000 ) , both classes are in general conspicuously radio quiet relative to broad - line seyfert 1s and high - ionization balqsos . [ cols="<,<,<",options="header " , ] @xmath19 the low - ionization balqsos also have stronger uv fe ii emission than non - balqsos ( w91 ) .
+ empirically , luminous nls1s and low - ionization balqsos appear quite different in x - rays .
luminous nls1s are strong emitters of soft x - rays while low - ionization balqsos are very weak in this band ( see 2 )
. however , this may just be due to absorption in low - ionization balqsos that prevents us from seeing the underlying x - ray continuum ; the intrinsic x - ray continua of low - ionization balqsos could be very similar to those of nls1s .
if better data eventually show , for example , that many low - ionization balqsos have unusually steep x - ray power laws with photon indices of @xmath202.5 , this would strongly suggest a connection with luminous nls1s ( compare with brandt , mathur & elvis 1997 ) .
it appears likely that absorbing gas capable of completely obscuring the intrinsic x - ray continuum is present in the nuclei of some nls1s . a range of absorption strengths is already known to be present and is being extended to nearly balqso - level column densities with new observations .
several nls1s are known to have x - ray warm absorbers ( e.g. , brandt et al . 1997 ) , and ` anomalous ' absorption around 1 kev may be present in some nls1s as well ( e.g. , leighly et al . 1997 ) .
the 1 kev absorption features have been speculated to be associated with highly ionized gas ( @xmath21 @xmath4 ) flowing out of the nucleus at 0.20.3@xmath22 .
however , other explanations are also plausible ( e.g. , nicastro , fiore & matt 1999 ) , and the evidence requires substantial improvement .
heavier absorption still is known to be present in mrk 507 ( iwasawa , brandt & fabian 1998 ) and the soft x - ray weak pg 1535 + 547 ( mrk 486 ; gallagher et al .
2000 , in preparation ; also see brandt , laor & wills 2000 ) .
the x - ray absorption in mrk 507 is complex but appears to be mainly due to mildly ionized gas with a column density of @xmath23 ( 23)@xmath24 @xmath4 .
better uv spectra are now needed to search for uv absorption lines from this object .
pg 1535 + 547 shows the heaviest x - ray absorption known in a nls1 ; modeling it with neutral gas requires a column density of @xmath25 @xmath4 and a covering fraction of @xmath26% ( see figure 3 ) .
unfortunately , the photon index of the underlying power law is poorly constrained at present ( @xmath27 ) due to limited photon statistics ; spectra from should reveal if the power law is unusually steep .
the characteristics of the x - ray absorption are likely related to the observed uv absorption and high polarization ( e.g. , smith et al . 1997 ) , and the strength of the x - ray absorption is remarkable in a type 1 object .
the derived column density is close to that seen for balqsos as well as the tori of some seyfert 2 galaxies . in another example ,
based on a steep x - ray power - law continuum and strong variability , iwasawa et al .
( 1996 ) have suggested that an obscured nls1 nucleus may reside in the seyfert 2 galaxy iras 183255926 .
x - ray studies of ultrasoft nls1s suggest that they are characterized by extreme values of a primary physical parameter , probably @xmath0 ( e.g. , brandt & boller 1999 ) . observations at optical and uv wavelengths have also revealed that nls1s have large amounts of dense ( @xmath28 @xmath29 ) , low - ionization , line - emitting gas in their broad line regions ( blrs ; e.g. , bg92 ; wills et al . 1999 ; kuraszkiewicz et al .
however , the physical relation between high @xmath0 and dense blrs remains unclear .
wills et al .
( 1999 ) suggest that gas injected into the nucleus by starburst activity may lead to both high @xmath0 and high blr gas densities .
the most straightforward way to connect luminous nls1s and low - ionization balqsos would be to postulate that both are characterized by high @xmath0 .
it would then just be the presence or absence of bal material along the line of sight that determines the classification of a given high @xmath0 object .
the amount of line - of - sight material could naturally be set by the nuclear orientation , although random fluctuations in the number of intervening bal ` clouds ' could perhaps also be relevant . the average value of @xmath16 for high @xmath0 objects would set the relative number densities of luminous nls1s and low - ionization balqsos .
while @xmath30 for high @xmath0 objects could plausibly be as large as @xmath31% given likely selection effects against low - ionization balqsos ( e.g. , goodrich 1997 ; krolik & voit 1998 ) , a value much larger than this would probably overpredict the number of low - ionization balqsos relative to luminous nls1s .
the scenario of the previous paragraph has a couple of attractive features .
first of all , the strong , high @xmath16 outflows of low - ionization balqsos might well result from high @xmath0 .
accreting systems with high @xmath0 have a greater ability to radiatively drive outflows due to their larger photon luminosities per unit gravitational mass .
furthermore , radiation - dominated analogues of advective inflow - outflow solutions may become important at high @xmath32 ( e.g. , 5 of blandford & begelman 1999 ; blandford & begelman , in preparation ) .
the scenario of the previous paragraph may also be able to explain why nls1s have unusually dense blrs without requiring a secondary agent , such as starburst activity .
some of the gas accelerated as a result of the high value of @xmath0 is likely to end up in the blr , where it can cool and increase the local density ( see 8.2 of brandt , laor & wills 2000 ) .
if high - velocity outflows with large covering factors are present in the nuclei of all luminous active galaxies with high @xmath0 , one might hope to occasionally see ` occultations ' when material moves through the line of sight .
such occultations have been seen in several high @xmath0 x - ray binaries where they can cause dramatic x - ray variability , sometimes without strong spectral changes ( e.g. , dower , bradt & morgan 1982 ; brandt et al .
an intriguing , speculative possibility is that _ some _ of the extreme x - ray variability seen in the most variable nls1s is due to occultations .
the large - amplitude variability seen in iras 132243809 , for example , is difficult to explain as true luminosity changes on energetic grounds .
an occultation model is able to explain the variability seen on a timescale of days provided the occulting material is quite thick and moves rapidly across the line of sight ( boller et al .
1997 ) . while bal winds are both thick and rapidly moving , they probably could not cause the variability because they originate too far from the nucleus . however , highly ionized material launched at smaller radii ( e.g. , the ` hitchhiking gas ' of murray et al .
1995 ) may be relevant .
observations with the new generation of x - ray observatories should be able to test some of the ideas discussed above .
for example , systematic measurements of the x - ray continuum shapes of both high - ionization and low - ionization balqsos can test if low - ionization balqsos have unusually steep power - law continua with photon indices of @xmath202.5 . if they do , this would immediately suggest a connection with luminous nls1s and high @xmath0 , and it is encouraging to see that some progress is now being made measuring balqso x - ray continua ( see 2 and 3 ) . unusually strong x - ray variability from low - ionization balqsos would also suggest a connection with luminous nls1s .
furthermore , high - quality x - ray and uv spectroscopy of moderately absorbed nls1s should help to bridge the gap between unabsorbed nls1s and the heavily absorbed balqsos . for at least one nls1 ,
pg 1535 + 547 , the x - ray column density is comparable to that inferred for balqsos , and uv absorption is also present ( see 3 ) .
the nls1s with the strongest known absorption ( e.g. , mrk 507 and pg 1535 + 547 ) unfortunately fall short of having qso - level luminosities ; hopefully more luminous examples will be found . finally ,
if low - ionization balqsos and luminous nls1s are indeed related , one might hope to see ` transitions ' between these two types of objects .
these could occur due to abrupt changes in the flow geometry of the absorbing material .
strong changes in the absorption - line properties of some low - ionization balqsos have already been seen ( e.g. , boroson et al .
1991 ; junkkarinen , cohen & hamann 1999 ) , although these appear to occur on fairly long timescales .
perhaps at the y3k workshop on nls1s there will be a celebration when mrk 231 undergoes such a transition and ends up looking something like phl 1092 .
this would be a sight to rival the y2k launch of !
we acknowledge the support of nasa ltsa grant nag5 - 8107 ( wnb ) and the pennsylvania space grant consortium ( scg ) .
we thank b.j .
wills for helpful comments .
brandt , w.n . ,
boller , th .
1999 , in structure and kinematics of quasar broad line regions , ed .
gaskell , c.m . ,
brandt , w.n .
, dietrich , m. , dultzin - hacyan , d. , eracleous , m. ( asp press : san francisco ) , p. 265 | luminous ultrasoft nls1s and low - ionization balqsos share many properties , and they both represent important extremes of the active galaxy phenomenon . we briefly discuss their observational similarities as well as potential physical connections between them , concentrating on the x - ray point of view .
we present several ways by which potential connections might be further tested .
galaxies : active ; qsos : general ; x - rays : galaxies |
disordered optical potentials used in current experiments are created by shining coherent laser beams through diffusive plates , as sketched in fig . [
fig : speckle ] .
the shift of atomic energy levels due to light - matter coupling is proportional to the intensity of radiation @xmath0 , yielding an effective optical potential @xmath1 for the center - of - mass motion of the atoms , where @xmath2 is the detuning between the laser and atomic transition frequencies @xcite .
we restrict our study to blue - detuned speckles , corresponding to @xmath3 , which are widely used in experiments .
the optical disorder is then always positive and its potential distribution obeys the rayleigh law . without loss of generality
, it is customary to shift @xmath4 by its average value @xmath5 , so that the distribution takes the form : @xmath6 @xmath7 being the unit step function . with this choice , @xmath8 and the variance is @xmath9 .
the fact that the local potential distribution @xmath10 is not gaussian and strongly asymmetric in contrast with model potentials often used in theoretical calculations has important consequences for the behaviour of the mobility edge @xcite .
since the kinetic energy is positive , we see from eq .
( [ eq : pv ] ) that no energy level exists below @xmath11 , implying that the mobility edge @xmath12 must satisfy the condition @xmath13 .
optical speckle potentials also exhibit non - local spatial correlations .
in particular , the two - point correlation function @xmath14 is characterized by finite correlation lengths which , as shown below , play a crucial role for the mobility edge . in the simplest model of a three - dimensional isotropic speckle potential ( created by a monochromatic laser of wavevector
@xmath15 coming from all directions of space ) , the correlation function is @xmath16 ^ 2 $ ] , where @xmath17 for this model potential not easily realized experimentally
the mobility edge has been numerically computed in ref . .
it was shown that the mobility edge lies _ way below _ the average potential , @xmath18 , a non - trivial behavior confirmed by various approximate theoretical approaches based on the self - consistent theory of localization@xcite .
however , all measurements reported in ref . and a subset of those appearing in ref . have positive mobility edges . in the simple case of isotropic correlations in the potential , the asymptotic behavior of the mobility edge can be deduced by considering the different energy scales of the system .
there is a natural energy scale associated with the correlation length @xmath19 , called the correlation energy@xcite : @xmath20 where @xmath21 is the atomic mass . for @xmath22
the de broglie wavelength of an atom with energy @xmath5 is much larger than the correlation length of the potential , so that the particle can tunnel through the disorder barriers .
this is the so - called `` quantum '' regime , where the mobility edge is expected to be very close to zero .
in contrast , for @xmath23 the matter wave resolves all the fine details of the disordered potential . in this `` classical '' regime
, the mobility edge is expected to be close to the percolation threshold of the potential , which is very close to @xmath11@xcite . among the three characteristic energy scales ,
@xmath24 only their ratios matter , so that one has a unique ( unknown ) scaling function such that @xmath25 with @xmath26 and @xmath27 .
the question most relevant to experiments is whether this scaling behavior applies to realistic speckles with _
correlation functions .
two different experimental setups have essentially been developed . in ref .
, a single diffusive plate was used to create a speckle pattern with a rather strong anisotropy , depending on the numerical aperture @xmath28 of the imaging system . for moderate values of @xmath29 the correlation lengths in the directions orthogonal and parallel to the speckle laser beam ( of wavelength @xmath30 )
are given by @xmath31 and @xmath32 , respectively @xcite .
for experimental setups in which two crossed coherent laser speckles are made to interfere @xcite , there is an additional characteristic length scale corresponding to the fringe spacing , @xmath33 , generated by the two perpendicular beams , as can be seen in fig .
[ fig : speckle ] . in both cases ,
different correlation lengths are present in the system , so that it is no longer clear how to define the correlation energy in eq . .
and @xmath34 directions ) , as used in refs . , with same numerical aperture @xmath35 .
* b * , two - point correlation function of the resulting disorder potential as function of position ( in units of the laser wavelength @xmath30 ) in the @xmath36 , @xmath34 and @xmath37 directions .
the three correlation lengths @xmath38 , @xmath39 and @xmath40 ( see text ) are markedly different in such anisotropic experimental configuration.,title="fig : " ] and @xmath34 directions ) , as used in refs . , with same numerical aperture @xmath35 .
* b * , two - point correlation function of the resulting disorder potential as function of position ( in units of the laser wavelength @xmath30 ) in the @xmath36 , @xmath34 and @xmath37 directions .
the three correlation lengths @xmath38 , @xmath39 and @xmath40 ( see text ) are markedly different in such anisotropic experimental configuration.,title="fig : " ]
in order to understand how the position of the mobility edge is affected by the anisotropy of realistic optical disorder , we have performed quasi - exact numerical calculations for two different speckle geometries and different values of the numerical aperture . in fig .
[ fig : sspme ] we plot the normalized mobility edge @xmath41 for the _ single _ laser speckle configuration as a function of the disorder amplitude @xmath5 , expressed in units of @xmath42 , for five different values of the numerical aperture @xmath28 controlling the anisotropy . in fig . [
fig : cspme ] , the same quantity is plotted for a disordered potential generated by _ two _ crossed identical laser speckles , for two different values of @xmath28 .
the first striking observation is that the mobility edge is always negative ( i.e. below the average potential @xmath5 ) for all anisotropic configurations , as for the isotropic speckle whose results are also plotted in the figures .
the second striking observation is that , although the mobility edge values are widely different for different degrees of anisotropy , the behavior in each case is strongly similar .
this suggests that the scaling law , eq .
( [ eq : scaling_law ] ) , can be extended to the anisotropic case by an appropriate redefinition of the correlation energy .
we thus _ define _ the correlation energy in the anisotropic case by ( see supplementary information ) : @xmath43 @xmath44 and @xmath40 being the correlation lengths of the disorder along the major axes .
there is however still some ambiguity on how to define the correlation length in a given direction .
depending on the detailed shape of the correlation function , different definitions may lead to different numerical values ( up to a constant numerical factor depending on the shape ) .
we use a pragmatic definition , namely that @xmath19 is the half - width at half - maximum of the central correlation peak ( see fig .
[ fig : speckle]b ) divided by the numerical factor @xmath45 such that @xmath46 this adhoc definition ensures the compatibility with the usual definitions of the correlation length and correlation energy in the isotropic case . in the inset of figs .
[ fig : sspme ] and [ fig : cspme ] , we replot the same numerical values of the mobility edge as function of rescaled disorder strength @xmath47 for the various anisotropic configurations , each configuration having a specific @xmath48 horizontal rescaling factor as defined in eq .
( [ eq : correlation_energy_anisotropic ] ) .
amazingly , all curves collapse on a _ single universal _ curve , independently of the anisotropy and whether the speckle pattern is created by a single laser beam or two crossed beams .
this constitutes a major result of this paper , as it allows to take into account all complications introduced by the inherent anisotropy of experimental speckle disorder through a simple rescaling of the correlation energy .
strictly speaking , this universality can not be exact , as details of the disorder correlation function must have an influence on the position of the mobility edge . however , this influence is actually rather moderate@xcite .
indeed , the system being strongly scattering at the mobility edge ( according to the ioffe - regel criterion , the mean free - path is shorter than the de broglie wavelength ) , long - range correlations in the potential are essentially irrelevant .
this universality is more accurate in the quantum regime @xmath49 where the matter wave averages out the disordered potential . in the regime of current cold - atom experiments ,
the universal function @xmath50 allows to predict the position of the mobility edge within _ a few percent _ , which is largely enough considering that the discrepancy between experimental results and theoretical predictions is of the order of 100% or higher .
recently , the position of the mobility edge in the single speckle configuration has been estimated in ref . , using a different method based on the statistical properties of the energy spectrum .
we have added the corresponding points in the inset of fig . [
fig : sspme ] .
although they lie slightly above our results , these data confirm that the mobility edge is very significantly below the average potential , even in the anisotropic case . of atoms in the single - speckle configuration , plotted as a function of the disorder amplitude @xmath5 in units of @xmath42 ( @xmath15 being the wavevector of the laser beam and
@xmath21 the atomic mass ) .
different curves correspond to different numerical apertures @xmath51 ( from lower to upper coloured lines ) which control the degree of anisotropy .
black symbols extending to @xmath52 correspond to the predictions of ref . for the isotropic case
( error bars are not visible as they are smaller than symbol size ) .
the inset shows that the normalized mobility edges @xmath53 versus normalized disorder amplitude @xmath54 , with average correlation energy @xmath55 , approximately collapse to a single curve on top of the isotropic speckle result ( black line ) for all anisotropic configurations ; @xmath56 and @xmath57 being the correlation lengths in the directions orthogonal and parallel to the laser beam , respectively .
brown squares show the results of an independent estimation of the mobility edge@xcite , in fair agreement with our results . ] of atoms in the crossed - speckles configuration depicted in fig .
[ fig : speckle ] , plotted as a function of the disorder amplitude @xmath5 in units of @xmath58 for two different values of the numerical aperture @xmath59 ( respectively lower and upper coloured lines ) .
two additional values @xmath60 ( respectively lower purple and upper blue point ) for a given normalized disorder strength @xmath61 , as well as the result for isotropic speckle ( black line ) are also plotted .
the inset shows the same data for the normalized mobility edge @xmath41 but versus normalized disorder amplitude @xmath54 , with average correlation energy @xmath62 , which collapse to a single universal curve . ]
to date , three different cold - atom experiments have reported measurements of the mobility edge in blue - detuned speckle disorder @xcite , by monitoring the temporal expansion of a localized matter wavepacket in the presence of disorder . although all three experiments do not use the same setup for the generation of the optical disorder potential and hence have different potential correlation functions , they can be quantitatively compared in the quantum regime using the above universal scaling of the mobility edge .
the experiment of ref . used a very anisotropic single speckle configuration . the use of spin - polarised fermionic atoms allowed to have a relatively well defined energy of the atoms , which could be an important advantage for pinpointing the precise location of the mobility edge .
however , the duration of the experiment may have been too short to detect slow diffusion above the mobility edge@xcite , thus leading to a possible overestimation of the lowest diffusive energy . as can be seen in the inset of fig .
[ fig : expme ] , their inferred mobility edge is indeed much higher than other experimental and numerical estimates .
moreover , it is always above the average potential , in contradiction with all numerical and theoretical predictions .
we thus conclude that these results are impaired by some severe imperfections .
the experiment of ref . used two crossed speckles to achieve a more isotropic configuration , but suffered from a broad initial energy distribution of the atoms extending on both sides of the mobility edge making a direct measurement of the mobility edge rather difficult .
an indirect estimate was given assuming that the mobility edge @xmath12 scales quadratically with the disorder amplitude , @xmath63 a `` naive '' implementation of the self - consistent theory of localization @xcite combined with the estimated energy distribution of the atoms gave a prediction for the localized fraction vs. disorder strength which was not in good agreement with the experiment .
however , a fair agreement was observed by correcting the prediction of the self - consistent approach by a `` heuristic shift '' quadratic in @xmath64 this yields the straight red line @xmath65 shown in fig .
[ fig : expme ] , in decent agreement with our numerical predictions and notably below the average potential .
unfortunately , these measurements were limited to a disorder strength much smaller than the correlation energy @xmath66 .
the more recent experiment of ref . used a similar two - speckles configuration , with a resulting large localized fraction of atoms thanks to a better control of atom - atom interactions via a feshbach resonance .
when the disorder amplitude is rescaled by the appropriate @xmath67 , the qualitative behavior of the mobility edge with @xmath5 is similar to our numerical results , but quantitatively too high .
the origin of this deviation is not entirely clear to us at this moment .
we note , however , that the energy distribution of atoms has not been directly measured in the experiment , but rather inferred from numerical simulations of fairly small systems .
an error in this non - trivial reconstruction process could have biased their estimation of the mobility edge . moreover , in these crossed - speckle configurations as can be seen in fig .
[ fig : speckle].b small numerical apertures result in multiple correlation peaks in the direction of fringes ( @xmath34 direction in the aforementioned figure ) ; this could lead to an effective correlation energy smaller than the one we define in eq .
( [ eq : correlation_energy_anisotropic ] ) , hence pushing the mobility edge below the `` universal '' curve . has been computed using the correlation lengths provided in each article . the universal curve for the mobility edge is also represented for reference ( black dashed line ) .
we computed numerically the mobility edge for non - interacting cold atoms in three - dimensional speckle potentials , taking into account the long - range anisotropic disorder correlations that characterize realistic experimental configurations . in the quantum regime of low disorder , which is relevant for most experiments
, it was shown that the mobility edge displays a robust scaling property when the anisotropy and/or speckle geometry is varied , with the correlation energy of the potential
defined from the geometric mean of the correlation lengths as the scaling parameter .
early experimental studies of 3d anderson localization@xcite as well as more accurate measurements@xcite show significant discrepancies from this universal behavior , calling for further studies .
this work opens the way for more quantitative comparisons between current cold - atom experiments on 3d anderson localization and theory / numerics .
* definition of the correlation energy for anisotropic speckles . * for isotropic disorder , the characteristic energy scale separating the low - energy `` quantum '' regime from the high - energy `` classical '' regime is the correlation energy in eq .
( [ eq : correlation_energy_isotropic ] ) . for anisotropic disorder , it is customary @xcite to define the correlation energy using the geometrical average of the correlation lengths along the three directions , eq .
( [ eq : correlation_energy_anisotropic ] ) .
our numerical results confirm that this is indeed the energy scale relevant for the mobility edge , and we give below two complementary arguments justifying this assumption .
first , at very low disorder strength , the mobility edge is also at very low energy , in fact only slightly below the average potential .
then , the typical de broglie wavelength @xmath68 of the atom is larger than the correlation lengths of the potential , so that the particle averages out all short - range fluctuations of the potential landscape .
the typical size of a potential spot is the product @xmath69 the quantum wave will thus average about @xmath70 uncorrelated potential spots . from the central limit theorem , it will result in an effective renormalized potential with gaussian fluctuations of typical strength @xmath71 and _ isotropic _ correlation length @xmath72 at this stage , the details of the correlation function of the initial potential are irrelevant , leaving only the geometric average @xmath73 as a relevant parameter .
calculations performed for specific models@xcite confirm that the dynamics at very low energy tends to be isotropic .
a second justification comes from the calculation of the weak localization correction to classical diffusive transport .
the starting point is the modification of the classical boltzmann diffusion picture by quantum interference due to closed loops . in the anisotropic case , where one must use a diffusion tensor ,
the correction to the boltzmann tensor @xmath74 is @xcite : @xmath75 where @xmath76 is the average density of states per unit volume .
the integral has a large @xmath77 divergence which can be regularized with a convenient cut - off .
the simplest choice for this momentum cut - off is to choose along each direction @xmath78 the inverse of the transport mean free path @xmath79 . as discussed in refs . , the anisotropy of the diffusion tensor essentially reflects the anisotropy of the transport mean free path , so that the transport mean free time @xmath80 is essentially identical along all directions .
the diagonal element of the diffusion tensor along direction @xmath81 is then @xmath82 with this choice of cut - off , the integral can be simply performed , and the correction scales like@xcite @xmath83^{-1/2}.$ ] in turn , this implies that , instead of the usual isotropic ioffe - regel criterion @xmath84 for the determination of the mobility edge , we have @xmath85^{1/3}\approx 1.$ ] as , in any case , the transport mean free path in a given direction is always larger than the correlation length of the disorder in the same direction , we end up again with the conclusion that the relevant length scale is @xmath86 note however that the weak localization correction is here predicted to have no effect on the anisotropy of the diffusion tensor@xcite , but only leads to a global renormalization .
more sophisticated approaches @xcite as well as numerical results @xcite show that it is not the case .
close to the mobility edge , transport is less anisotropic than in the classical boltzmann regime .
a quantitative understanding of this phenomenon is highly desirable , but beyond the scope of this paper .
preliminary results indicate that some anisotropy remains at the mobility edge . in any case
, it does not seem to invalidate that the anisotropic correlation energy eq .
( [ eq : correlation_energy_anisotropic ] ) is the relevant energy scale of the problem .
* numerical generation of the disordered potential .
* in order to reproduce the correct disorder correlation functions
i.e. the ones used in real experiments we generate numerically a `` random '' electromagnetic field by combining many plane waves with different wavevectors and uncorrelated random phases . because a monochromatic laser is used to create the speckle , the wavevectors ( of unit modulus ) are distributed on the surface of a sphere , typically with a gaussian distribution like in most experiments , inside a single cap with half - angle @xmath28 ( in the `` single speckle '' configuration ) or two perpendicular caps ( in the `` crossed speckles '' configuration ) .
all wavevectors are taken with the same polarization .
the disorder at a given position is then simply the modulus square of the electric field computed by coherently combining the various plane waves .
we carefully checked that we obtain in the small @xmath28 limit the correlation functions predicted in refs . .
* transfer - matrix method .
* the mobility edge is calculated using the same procedure outlined in ref . .
starting from a spatial discretization of the 3d schrdinger equation on a grid , whose spacing @xmath87 is much smaller than the correlation lengths of the potential and de broglie wavelength ( we used @xmath88 for all computations ) , we end up with an effective anderson model with hopping term @xmath89 and correlated on - site disorder .
we then proceed in the standard way @xcite by computing recursively the total transmission of a bar - shaped sample of width @xmath90 and length @xmath91 , such that @xmath92 at fixed energy @xmath93 and disorder amplitude @xmath5 . as quasi-1d systems
are always exponentially localized , the transmission decays exponentially with @xmath91 over the localization length @xmath94 using the finite - size scaling technique on the adimensional localization length @xmath95 allows one to probe the 3d anderson transition , see ref . for details .
compared to the isotropic case computed in ref .
, the difficulty lies in the anisotropic correlation function , requiring significantly larger system sizes especially along the directions of largest correlation lengths to reach convergence . typically , we used transverse sizes of up to @xmath96 sites for the finite - size scaling analysis and lengths around @xmath97 sites .
small values of @xmath29 as used in ref . , are thus very challenging numerically ; more than two million hours of cpu time on supercomputer resources have been used to obtain the results of figs . [ fig : sspme ] and [ fig : cspme ] . 36 anderson , p. w. absence of diffusion in certain random lattices . _ phys .
_ * 109 * , 1492 ( 1958 ) .
wiersma , d. s. , bartolini , p. , lagendijk , a. & righini , r. localization of light in a disordered medium .
_ nature _ * 390 * , 671 ( 1997 ) .
st " orzer , m. , gross , p. , aegerter , c. m. & maret , g. observation of the critical regime near anderson localization of light .
* 96 * , 063904 ( 2006 ) .
schwartz , t. , bartal , g. , fishman , s. & segev , b. transport and anderson localization in disordered two - dimensional photonic lattices .
_ nature _ * 446 * , 52 ( 2007 ) .
lahini , y. , avidan , a. , pozzi , f. , sorel , m. , morandotti , r. , christodoulides , d. n. & silberberg , y. anderson localization and nonlinearity in one - dimensional disordered photonic lattices .
* 100 * , 013906 ( 2008 ) .
hu , h. , strybulevych , a. , page , j. h. , skipetrov , s. e. & van tiggelen , b. a. localization of ultrasound in a three - dimensional elastic network .
phys . _ * 4 * , 945 ( 2008 ) .
chabanov , a. a. , stoytchev , m. & genack , a. z. statistical signatures of photon localization .
_ nature _ * 404 * , 850 ( 2000 ) .
billy , j. , josse , v. , zuo , z. , bernard , a. , hambrecht , b. , lugan , p. , cl ' ement , d. , sanchez - palencia , l. , bouyer , p. & aspect , a. direct observation of anderson localization of matter waves in a controlled disorder .
_ nature _ * 453 * , 891 ( 2008 ) .
roati , g. , derrico , c. , fallani , l. , fattori , m. , fort , c. , zaccanti , m. , modugno , g. , modugno , m. & inguscio , m. anderson localization of a non - interacting bose - einstein condensate .
_ nature _ * 453 * , 895 ( 2008 ) .
katsumoto , s. , komori , f. , sano , n. & kobayashi , s. fine tuning of metal - insulator transition in @xmath98 using persistent photoconductivity .
_ j. phys .
jpn . _ * 56 * , 2259 ( 1987 ) .
waffenschmidt , s. , pfleiderer , c. & lhneysen , h. v. critical behavior of the conductivity of si : p at the metal - insulator transition under uniaxial stress .
lett . _ * 83 * , 3005 ( 1999 ) .
kondov , s. s. , mcgehee , w. r. , zirbel , j. j. & demarco , b. three - dimensional anderson localization of ultracold matter .
_ science _ * 334 * , 66 ( 2011 ) .
jendrzejewski , f. , bernard , a. , muller , k. , cheinet , p. , josse , v. , piraud , m. , pezz , l. , sanchez - palencia , l. , aspect , a. & bouyer , p. three - dimensional localization of ultracold atoms in an optical disordered potential .
phys . _ * 8 * , 398 ( 2012 ) .
semeghini , g. , landini , m. , castilho , p. , roy , s. , spagnolli , g. , trenkwalder , a. , fattori , m. , inguscio , m. & modugno , g. measurement of the mobility edge for 3d anderson localization .
phys . _ * 11 * , 554 ( 2015 ) .
evers , f. & mirlin , a. d. anderson transitions .
phys . _ * 80 * , 1355 - 1417 ( 2008 ) .
delande , d. & orso , g. mobility edge for cold atoms in laser speckle potentials .
lett . _ * 113 * , 060601 ( 2014 ) .
pasek , m. , zhao , z. , delande , d. & orso , g. phase diagram of the three - dimensional anderson model for short - range speckle potentials .
a _ * 92 * , 053618 ( 2015 ) .
yedjour , a. & van tiggelen , b. a. diffusion and localization of cold atoms in 3d optical speckle . _ eur .
j. d _ * 59 * , 249 ( 2010 ) .
piraud , m. , sanchez - palencia , l. & van tiggelen , b. anderson localization of matter waves in three - dimensional anisotropic disordered potentials .
_ phys . rev .
a _ * 90 * , 063639 ( 2014 ) .
cohen - tannoudji , c. , dupont - roc , j. & grynberg , g. _ atom - photon interactions : basic processes and applications _
( wiley , 1992 ) .
r. c. , miniatura , c. , delande , d. , sigwarth , o. & mller , c. a. localization of matter waves in two - dimensional disordered optical potentials .
lett . _ * 95 * , 250403 ( 2005 ) .
r. c. , sigwarth , o. , miniatura , c. , delande , d. & mller , c. a. coherent matter wave transport in speckle potentials .
_ new j. phys . _ * 9 * , 161 ( 2007 ) .
pilati , s. , giorgini , s. , modugno , m. & prokofev , n. dilute bose gas with correlated disorder : a path integral monte carlo study .
_ new j. phys . _ * 12 * , 073003 ( 2010 ) .
goodman , j. _ speckle phenomena in optics : theory and applications _ ( dover , 2007 ) .
fratini , e. & pilati , s. anderson localization of matter waves in quantum - chaos theory .
a _ * 92 * , 063621 ( 2015 ) .
mller , c. a. & shapiro , b. comment on `` three - dimensional anderson localization in variable scale disorder '' .
* 113 * , 099601 ( 2014 ) .
piraud , m. , pezz , l. & sanchez - palencia , l. quantum transport of atomic matter waves in anisotropic two - dimensional and three - dimensional disorder .
_ new j. phys . _ * 15 * , 075007 ( 2013 ) .
vollhardt , d. & wlfle , p. in _ electronic phase transitions _
( eds hanke , w. & kopaev , y. ) 1 ( elsevier , 1992 ) .
wlfle , p. & bhatt , r. n. electron localization in anisotropic systems .
b _ * 30 * , 3542 ( 1984 ) .
piraud , m. , pezz , l. & sanchez - palencia , l. matter wave transport and anderson localization in anisotropic three - dimensional disorder . _ europhys .
lett . _ * 99 * , 50003 ( 2012 ) .
zambetaki , i. , li , q. , economou , e. n. & soukoulis , c. m. localization in highly anisotropic systems .
lett . _ * 76 * , 3614 ( 1996 ) .
zambetaki , i. , li , q. , economou , e. n. & soukoulis , c. m. localization in weakly coupled planes and weakly coupled wires .
b _ * 56 * , 12221 ( 1997 ) .
li , q. , katsoprinakis , s. , economou , e. n. & soukoulis , c. m. scaling properties in highly anisotropic systems .
b _ * 56 * , r4297 ( 1997 ) . panagiotides , n. a. , evangelou , s. n. & theodorou , g. localization - delocalization transition in anisotropic solids . _
b _ * 49 * , 14122 ( 1994 ) .
bhatt , r. n. , wlfle , p. & ramakrishnan , t. v. localization and interaction effects in anisotropic disordered electronic systems .
b _ * 32 * , 569 ( 1985 ) . lopez , m. , clment , j .- f .
, lemari , g. , delande , d. , szriftgiser , p. & garreau , j. c. phase diagram of the anisotropic anderson transition with the atomic kicked rotor : theory and experiment .
_ new j. phys . _ * 15 * , 065013 ( 2013 ) .
mckinnon , a. & kramer , b. the scaling theory of electrons in disordered solids : additional numerical results .
_ z. phys .
b _ * 53 * , 1 ( 1983 ) .
we acknowledge discussions with v. josse and thank s. pilati for sending us the raw data of ref . .
m. pasek was supported by erc ( advanced grant `` quantatop '' ) and the region ile - de - france in the framework of dim nano - k ( project qugasp ) .
the authors were granted access to the hpc resources of tgcc under the allocations 2014 - 057301 , 2015 - 057301 , 2015 - 057083 , 2016 - 057629 and 2016 - 057644 made by genci ( `` grand equipement national de calcul intensif '' ) . | recent experiments on non - interacting ultra - cold atoms in correlated disorder have yielded conflicting results regarding the so - called mobility edge , i.e. the energy threshold separating anderson localized from diffusive states . at the same time , there are theoretical indications that the experimental data overestimate the position of this critical energy , sometimes by a large amount . the non - trivial effect of anisotropy in the spatial correlations of experimental speckle potentials have been put forward as a possible cause for such discrepancy . using extensive numerical simulations
we show that the effect of anisotropy alone is not sufficient to explain the experimental data .
in particular , we find that , for not - too - strong anisotropy , realistic disorder configurations are essentially identical to the isotropic case , modulo a simple rescaling of the energies .
when a wave travels in a random medium , the interference between multiple scattering paths caused by the disorder barriers can completely stop its diffusion .
this phenomenon , known as anderson localization @xcite , is completely general and applies to any kind of waves including light waves in diffusive media @xcite or in photonic crystals @xcite , ultrasound @xcite , microwaves @xcite and atomic matter waves @xcite , the latter describing the behavior of non - interacting atoms in the low temperature quantum regime .
localization experiments using cold atomic gases have several advantages over their solid - state counterparts@xcite .
the inhibition of transport may be directly measured by probing the atomic wavefunction using the time - of - flight technique .
the effect of interactions , that unavoidably hinders localization measurements in solid - state systems , can here be reduced by going from a bose - einstein condensate to a dilute gas or by using feshbach resonances .
lastly , thanks to light - matter interaction , far - detuned laser speckles can be used as tunable disordered potentials for the atoms , allowing the exploration of the localization phase diagram .
several experiments@xcite on cold atomic gases have attempted to characterize metal - insulator anderson transitions in three dimensions , starting from a precise measurement of the mobility edge .
this turns out to be surprisingly difficult , both theoretically and experimentally .
since the anderson transition is a second - order quantum phase transition@xcite , the localization length diverges on the localized side and the diffusion coefficient slowly vanishes on the diffusive side .
distinguishing the two behaviors requires long observation times , which is not easily achievable in experiments .
moreover , an atomic wavepacket expanding in a random potential inevitably contains states at different energies , thus mixing localized and extended components . for these reasons ,
results on the position of the mobility edge are widely spread . on the theoretical side ,
the main difficulty comes from the fact that the metal - insulator transition takes place in the strongly scattering regime where perturbative / diagrammatic expansions in the disorder strength fail and one has to resort to the indignity of numerical calculations .
the first quasi - exact numerical estimations @xcite of the mobility edge turned out to lie significantly below most experimental measurements .
it has been proposed that the peculiar anisotropic correlation functions of realistic disordered potentials , which were not taken into account in ref . , could be at the origin of this discrepancy .
however , reproducing all minute details of experimental speckle configurations in numerical simulations is cumbersome , if desirable at all .
it would be more profitable to gain physical understanding of how the mobility edge changes with speckle geometry , and know its more universal features . in the present paper , we show that the effect of anisotropy on the mobility edge can be taken into account very simply to a surprizingly good accuracy by a proper rescaling of the characteristic energies .
as most reported experimental measurements of the mobility edge do not seem to follow such scaling property , we conclude that they are actually problematic . |
as a consequence of the inelasticity of collisions , granular fluids exhibit a series of behaviors that are in sharp contrast with those of molecular fluids .
one of them is the absence of energy equipartition in a granular mixture of mechanically different species .
the granular temperatures of the components of the mixture , defined from the average kinetic energies , are different .
although this possibility was already pointed out many years ago @xcite , it has not been until recently that a systematic study of the effect has been started . for the homogeneous state of a freely cooling binary mixture of inelastic spheres , an explicit expression for the ratio of temperatures of the two components has been obtained from an approximate solution of the kinetic enskog equations @xcite .
the accuracy of this prediction for weak dissipation and low density has been confirmed by molecular dynamics ( md ) simulations @xcite .
the above homogeneous cooling state ( hcs ) is not accesible experimentally . in order to maintain a granular system fluidized ,
an external energy supply is required .
this is often carried out by means of vibrating walls or external fields , that generate macroscopic gradients in the system because of inelasticity . at a theoretical level , _
homogeneously driven _
granular fluids have also been considered . in these models ,
stochastic forces injecting energy are added , then allowing a system otherwise isolated to reach a steady state .
the lack of energy equipartition in the steady state of homogeneously driven granular mixtures has been also analyzed @xcite , by extending the methods of @xcite .
nevertheless , the relationship between this kind of driving and actual experiments is uncertain .
the non - equipartition has also been confirmed by md simulations of simple shear flows @xcite and of vibrated granular gases ( in absence of gravity ) @xcite .
experimental evidence of the coexistence of different temperatures in strongly vibrated granular mixtures has been reported in both two - dimensional @xcite and three - dimensional @xcite systems .
moreover , the dependence of the temperature ratio on the several parameters characterizing the system was investigated .
quite surprisingly , it has been observed that the above ratio remains practically constat in the bulk of the system , in spite of the rather large gradients exhibited by each of the partial temperatures .
the lack of energy equipartition has consequences on the expressions of the transport coefficients of a granular mixture @xcite .
therefore , it must also affect the shape of the hydrodynamic profiles and , in particular , the density distribution of each of the components in inhomogeneous mixtures .
consequently , it seems clear that the temperature difference can play a role in the segregation phenomena occurring in vibrated granular mixtures @xcite . here ,
the particular case of a low density binary mixture in the tracer limit , i.e. when the mole fraction of one of the components is very low , will be considered .
the simplicity of the system allows a detailed and controlled discussion of the hydrodynamic profiles of the tracer component , that can be expressed analytically in terms of those of the other ( excess ) component .
the starting point are the kinetic boltzmann equations for the mixture , and the analysis is based on the chapman - enskog method . the theory is first formulated for a general kind of states and later on particularized for the steady state of an open vibrated system .
then , the theoretical predictions are compared with md simulation results , and a quite satisfactory agreement is found over a wide range of values of the parameters defining the mechanical properties of the system .
a short summary of some of the results presented here was given in ref .
@xcite .
the plan of the paper is as follows . in sec . [
s2 ] , some on the fundamental ideas on which the chapmann - enskog method , as applied to a one - component system , is based are shortly reviewed .
they are used in sec .
[ s3 ] , where an explicit expression for the ratio between the local temperatures of the two components is derived .
this ratio only depends on the mechanical properties of the particles , being independent of the gradients of the hydrodynamic fields .
an extensive comparison with md results for a vibrated granular system in presence of gravity is carried out .
the shape of the density profile for the tracer component is investigated in sec .
[ s4 ] , where an expression for the mass flux relative to the local flow is derived to first order in the gradients of the mixture fields ( navier - stokes order ) .
for the particular case of the steady state of an open vibrated system , an expression for the density profile of the tracer component in terms of the fields of the excess component is obtained . since analytical forms for these fields
are known , the expression can be integrated numerically .
some details of this are given in the appendix .
again , the theoretical predictions compare well with md simulation results , although some significant discrepancies show up as the difference in mass of the particles is increased .
special emphasis is put on the relationship between the breakdown of energy equipartition and the relative position of the centers of mass of the two species .
finally , sec .
[ s5 ] contains some further discussion of the results , as well as some comments on their applicability to arbitrary states .
as already said , the aim of this work is to analyze some properties of the hydrodynamic profiles of a dilute granular binary mixture in the tracer limit , i.e. when the mole fraction of one of the species is very small .
more precisely , the emphasis is put on the relationship between the profiles of the tracer component and those of the mixture in an arbitrary state .
it is assumed that the concentration of the tracer component is so small that its presence does not affect the state of the granular mixture that is , therefore , determined by the state of the other ( excess ) component . at a kinetic theory level , this implies that the evolution of the one - particle distribution function of the excess component obeys a closed nonlinear boltzmann equation .
moreover , in the evolution of the distribution function of the tracer component , the mutual interactions between their particles can be neglected , as compared with the interactions with the particles of the excess component .
consequently , its distribution obeys a linear boltzmann - lorentz equation .
this is formally equivalent to consider an impurity or intruder immersed in a dilute granular gas , and this will be the terminology used in the following .
let us start by shortly reviewing some basic aspects of the chapman - enskog method applied to a one - component granular gas that will be needed in the following .
consider a low density gas composed by smooth inelastic hard spheres ( @xmath0 ) or disks ( @xmath1 ) of mass @xmath2 and diameter @xmath3 .
the inelasticity of collisions is modelled by a velocity - independent coefficient of normal restitution @xmath4 , defined in the interval @xmath5 .
the particles are submitted to an external force of the gravitational type , so the force acting on each particle has the form @xmath6 , where @xmath7 is a constant and @xmath8 the unit vector in the positive direction of the @xmath9 axis .
it is assumed that in the low density limit , the time evolution of the one - particle distribution function of the gas , @xmath10 , is accurately described by the nonlinear inelastic boltzmann equation @xmath11,\ ] ] where @xmath12 is the boltzmann collision operator describing the scattering of pairs of particles , @xmath13 & = & \sigma^{d-1 } \int d{\bm v}_{1 } \int d\widehat{\bm \sigma}\ , \theta ( { \bm g } \cdot \widehat{\bm \sigma } ) \nonumber \\ & & \times { \bm g } \cdot \widehat{\bm \sigma } \left[\alpha^{-2 } f({\bm r},{\bm v}^{\prime},t)f({\bm r},{\bm v}^{\prime}_{1},t ) \right .
\nonumber \\ & & \left .
-f({\bm r},{\bm v},t)f({\bm r},{\bm v}_{1},t ) \right].\end{aligned}\ ] ] here @xmath14 is the heaviside step function , @xmath15 is the unit vector pointing from the center of particle @xmath16 to the center of the other colliding particle at contact , and @xmath17 the relative velocity .
moreover , @xmath18 and @xmath19 are the precollisional or restituting velocities , i.e. the initial values of the velocities leading to @xmath20 and @xmath21 following the binary collision defined by @xmath22 , @xmath23 macroscopic fields number density , @xmath24 , flow velocity @xmath25 , and granular temperature , @xmath26 , are defined in the usual way as velocity moments of the distribution function , @xmath27 @xmath28 @xmath29 where @xmath30 is the peculiar velocity .
balance equations for the above fields are directly obtained from the boltzmann equation , @xmath31
@xmath32 @xmath33 the pressure tensor @xmath34 and the heat flux @xmath35 are given by @xmath36 @xmath37 respectively . the cooling rate , @xmath38 , in eq .
( [ 2.7 ] ) is due to the energy dissipation in collisions and it is also a functional of the distribution function , @xmath39.\ ] ] the balance equations ( [ 2.5])-([2.7 ] ) only become a set closed hydrodynamic equations for the fields once @xmath34 , @xmath40 and @xmath38 are expressed as functionals of them .
this requires to find a solution to the boltzmann equation such that all the space and time dependence occurs through @xmath41 , @xmath42 , and @xmath43 . in the chapman - enskog procedure , this _ normal _ solution is generated by expressing it as a series expansion in a formal non - uniformity parameter @xmath44 , @xmath45 where each factor of @xmath44 means an implicit gradient of a macroscopic field .
use of the above expansion in the definitions of the fluxes and the cooling rate gives a corresponding expansion for them .
finally , the time derivatives of the fields are also expanded in the gradients , @xmath46 , by means of the balance equations .
then , it is @xmath47 where the zeroth order cooling rate @xmath48 is given by @xmath49.\ ] ] in this way , to zeroth order in @xmath44 it is obtained : @xmath50.\ ] ] the solution to this equation is chosen such that it verifies the conditions @xmath51 @xmath52 @xmath53 i.e. , it leads to the same hydrodynamic fields as the complete distribution function .
this choice is consistent with the solubility conditions for the set of equations generated by the chapman - enskog method @xcite .
thus @xmath54 is easily generated from the distribution function for the homogeneous cooling state ( hcs ) of a dilute gas @xcite by substituting the homogeneous density and temperature fields by the local values @xmath24 and @xmath26 , and replacing the velocity @xmath20 by the peculiar one , @xmath55 .
therefore , it has the form @xcite @xmath56 where @xmath57 is a local thermal velocity . as a consequence , eq .
( [ 2.14 ] ) can be rewritten as @xmath58.\ ] ] using eq .
( [ 2.16 ] ) it is easily obtained : @xmath59 @xmath60 @xmath61 in the last expression , use has been made of the fact that @xmath62 because of symmetry considerations @xcite .
moreover , @xmath63 is the pressure of the gas .
the procedure can be continued and navier - stokes hydrodynamic equations for a dilute granular gas , with explicit expressions for the transport coefficients in the first sonine approximation , have been derived @xcite .
let us suppose now that an impurity or intruder of mass @xmath64 and diameter @xmath65 is added to the system .
it is assumed that the presence of the intruder does not affect the state of the gas , so that the one - particle distribution function of the gas particles is still determined by the nonlinear boltzmann equation ( [ 2.1 ] ) .
moreover , the macroscopic flow velocity and temperature for the mixture formed by the gas plus the intruder are the same as those for the gas alone , i.e. they are given by eqs .
( [ 2.4 ] ) .
the distribution function of the intruder , @xmath66 , obeys the linear boltzmann - lorentz equation @xmath67,\ ] ] where the collision operator now is @xmath68 & = & \overline{\sigma}^{d-1 } \int d{\bm v}_{1}\ , \int d\widehat{\bm \sigma}\ , \theta ( { \bm g } \cdot \widehat{\bm \sigma } ) \nonumber \\ & & \times { \bm g } \cdot \widehat{\bm \sigma } \left[\alpha_{0}^{-2 } f_{0}({\bm r},{\bm v}^{\prime \prime},t)f({\bm r},{\bm v}^{\prime \prime } _ { 1},t ) \right .
\nonumber \\ & & \left . -f_{0}({\bm r},{\bm v},t)f({\bm r},{\bm v}_{1},t ) \right].\end{aligned}\ ] ] here , @xmath69 and @xmath70 is the coefficient of normal restitution for impurity - gas collisions .
the precollisional velocities in this case are given by @xmath71 with @xmath72 .
the number density for the intruder is @xmath73 the chapman - enskog procedure when applied to a mixture , assumes the existence of a normal solution of the boltzmann equations for the mixture in which all the space and time dependence of the one particle distribution function of each of the species occurs through a functional dependence on the hydrodynamic fields of the mixture @xcite .
these fields can be chosen in different , equivalent ways . here
they are taken to be the concentration of the impurity @xmath74 , the pressure @xmath75 , the local flow velocity @xmath42 , and the temperature @xmath43 , i.e. @xmath76.\ ] ] then , an expansion similar to the one given in eq .
( [ 2.11 ] ) is considered for the distribution function of the impurity , @xmath77 .
to lowest order in @xmath44 , eq .
( [ 2.21 ] ) becomes @xmath78.\ ] ] taking into account that dimensional analysis requires that @xmath79 is of the form @xmath80 eq . (
[ 2.28 ] ) is seen to be equivalent to @xmath81.\ ] ] the solution of this equation is chosen such that @xmath82 @xmath83 @xmath84 the last equality in eq . ( [ 2.33 ] ) defines the local temperature , @xmath85 , of the intruder .
the above requirements are consistent with the solubility conditions of the equations generated by the chapman - enskog method .
it is worth to stress that eq .
( [ 2.30 ] ) holds with independence of the specific form of the hydrodynamic fields of the mixture .
thus multiplication of that equation by @xmath86 and integration over @xmath87 yields @xmath88 where @xmath89\ ] ] is the lowest order in the gradients of the cooling rate for the temperature of the intruder .
the evaluation of the cooling rates @xmath48 and @xmath90 requires to solve eqs .
( [ 2.14 ] ) and ( [ 2.30 ] ) . this can be done in a systematic way by expanding @xmath54 and @xmath79 in terms of an ensemble of orthogonal polynomials @xcite . here
we will consider a leading order approximation that is expected to give quite accurate results , at least for not very strong inelasticity .
the zeroth order distributions are approximated by gaussians , @xmath91 @xmath92 note that these expressions are consistent with the conditions ( [ 2.14 ] ) and ( [ 2.31])-([2.33 ] ) . employing them , it is straightforward to calculate the cooling rates .
the technical details needed to evaluate the integrals have already been discussed several times @xcite , and they will not be reproduced here . the result is @xmath93 @xmath94 where @xmath95 , @xmath96 is a dimensionless collision rate , @xmath97 and @xmath98 is the ratio of the mean square velocities for the intruder and fluid particles , @xmath99 substitution of eqs .
( [ 2.38 ] ) and ( [ 2.39 ] ) into eq.([2.34 ] ) gives @xmath100 where the parameter @xmath101 is given by @xmath102 the solution of this equation provides the expression for @xmath103 .
there is a unique real solution for all allowed values of @xmath104 and @xmath101 . for elastic collisions ( @xmath105 ) ,
the solution is @xmath106 and energy equipartition follows . if only the intruder - gas collisions are inelastic ( @xmath107 ) , the result @xmath108 derived by martin and piasecki for an inelastic impurity in an equilibrium elastic gas is recovered @xcite .
the simplicity of eq .
( [ 2.42 ] ) may appear as a surprise , specially taking into account its generality .
no particular state of either the gas or the intruder has been considered and , nevertheless , the temperature ratio is given as a function of only the mechanical properties of the particles .
the existence of an expression for the temperature ratio is a consequence of the assumption that there is a hydrodynamic description of the mixture in terms of only @xmath109 , @xmath75 , @xmath42 , and @xmath43 .
therefore , although the partial temperatures of the intruder and the gas are different , they are not needed to specify the macroscopic state of the mixture . nevertheless , the theory provides a relationship to determine them . the fact that eq .
( [ 2.42 ] ) does not involve any of the other hydrodynamic fields follows from the tracer and dilute limits we are considering .
the above relation between the partial temperatures is not modified when applied to the hcs , consistently with previous results @xcite .
this does not imply by itself any kind of _ local hcs approximation _ , in the same way as energy equipartition in molecular systems does not imply local equilibrium , and it is valid independently of the gradients present in the system . on the other hand , the above results do not hold if the characterization of the macroscopic state of the mixture also requires to specify the partial temperatures of the gas and the intruder and , therefore , the set of independent hydrodynamic variables ( and also the number of equations ) must be expanded .
this seems to be the kind of approximation followed in @xcite . in order to check the accuracy of the theoretical prediction given by eq
( [ 2.42 ] ) , we have performed md simulations with an event driven algorithm .
the particular situation we have considered is a two - dimensional system confined in a rectangular box of width @xmath110 , open on the top , and submitted to an external field of the gravitational type , as defined above eq.([2.1 ] ) .
periodic boundary conditions are enforced in the direction perpendicular to the field . to maintain the system fluidized
, energy is continuously supplied through the wall located at @xmath111 , which is vibrating . for simplicity , and given that we are interested in the bulk properties of the system , the wall moves in a sawtooth way , with very small amplitude and high frequency , so that it can be considered that all the particles colliding with the wall find it at @xmath111 and with the same velocity @xmath112 @xcite .
the particle - wall collisions are elastic . under the above conditions
, it has been proven that a one - component granular gas reaches a stationary state with only gradients in the @xmath9-direction @xcite .
nevertheless , this state becomes unstable and develops transversal inhomogeneities when the size of the vibrating wall is larger than a critical value @xcite .
in fact , the same happens for a closed system in absence of gravity @xcite . in all the simulations to be reported in the following ,
the value of @xmath110 has been taken small enough so the system stays homogeneous in the transversal direction .
the above steady state is highly inhomogeneous .
an example is given in fig [ fig1 ] , where the density and temperature profiles for a monodisperse system of @xmath113 particles with @xmath114 are shown . in order to plot the temperature profile on the same scale as the density one ,
the former has been scaled with an arbitrary value @xmath115 . here and in the following , the units defined by @xmath116 , @xmath117 , and @xmath118 will be used . in these units , the data in the figure correspond to @xmath119 and @xmath120 .
these values lead to a fluidized state without transversal inhomogeneities , and with a density low enough as to expect the boltzmann equation to give an accurate description of the system .
it is seen that the density profile exhibits a maximum at which @xmath121 .
the existence of a density maximum is a general feature of open vibrated systems as long as the number of monolayers at rest is large enough @xcite . on the other hand ,
the temperature profile presents a minimum which is related with the existence of a term proportional to the density gradient in the expression of the heat flux @xcite .
figures [ fig2]-[fig4 ] show the profile of the temperature ratio @xmath122 for different values of the mechanical properties of the particles .
it is observed that , aside statistical fluctuations , the ratio remains practically constant in a wide region of the system , even if @xmath43 ( and @xmath85 ) varies significantly . in fig .
[ fig2 ] , the influence of the inelasticity of the gas - intruder collisions on the temperature ratio is analyzed .
as expected , the deviation from equipartition increases as @xmath70 decreases , becoming more different from @xmath4 .
the influence of the mass ratio @xmath123 is studied in fig .
[ fig3 ] , where it is seen that the temperature ratio significantly deviates from unity when @xmath124 but @xmath70 differs significantly from @xmath4 .
this is in contrast with previous findings in both experiments @xcite and md simulations @xcite of vibrated granular mixtures ( not in the tracer limit ) , probably because the several coefficients of normal restitution in those works were too close one another as to observe this deviation .
the figure also shows that the boundary layer near the vibrating wall , defined as the region in which the hydrodynamic prediction is not verified , increases as the ratio @xmath123 increases .
this can be explained as a consequence of the larger increase of the kinetic energy of the impurity than of the gas particles when colliding with the vibrating wall .
in addition , the collisions of the intruder with the gas become less efficient in `` thermalizing '' the former .
the same explanation applies to the strong fluctuations at large heights .
finally , the sensitivity of @xmath125 on the diameter ratio is illustrated in fig .
[ fig4 ] , from where it can be concluded that it is quite weak . a quantitative comparison between the theoretical prediction given by eq .
( [ 2.42 ] ) and the md results is presented in figs .
[ fig5 ] and [ fig6 ] . according with eq .
( [ 2.42 ] ) , for a given value of @xmath101 , the value of @xmath98 is a function of only the parameter @xmath104 .
then , the comparison will be presented by considering series of simulations in which the value of the coefficient of restitution of the gas , @xmath4 , and the diameter ratio , @xmath126 , and therefore @xmath101 are kept constant , while the mass ratio @xmath123 and the restitution coefficient @xmath70 and , therefore @xmath104 are changed . in fig.[fig5 ] , two series of data , one corresponding to @xmath127 , @xmath128 ( @xmath129 ) and @xmath127 , @xmath114 ( @xmath130 ) are presented . in each case , several values of the mass ratio have been considered , as indicated in the figure .
in addition , the value of the coefficient of restitution for gas - intruder collisions has also been varied in the range @xmath131 . to keep the figure readable ,
we have used the same symbols for the data corresponding to the same mass ratio although different values of @xmath70 .
a quite good agreement between theory and simulation is observed , specially for the largest value of @xmath4 .
the small but systematic discrepancies for @xmath128 and @xmath132 are probably due to the failure of the hydrodynamic navier - stokes equations to accurately describe the state of the gas , as discussed in @xcite . the validity of the theory to describe the dependence on the diameter ratio is checked in fig . [ fig6 ] .
again , the same symbol is used for the md results corresponding to situations differing only in the value of the coefficient of restitution @xmath70 . as before , a weak dependence of the temperature ratio on the diameter of the particles is observed .
it can be concluded that eq .
( [ 2.42 ] ) provides an accurate description of the temperature ratio over a wide range of values of the mechanical parameters characterizing both the gas and the intruder .
an equation for the number density of the impurity @xmath133 is directly obtained from eq .
( [ 2.21 ] ) , @xmath134 where @xmath135 is the mass flux for the impurity relative to the local flow @xmath25 , @xmath136 the isotropy of the function @xmath137 implies that the zeroth order of the flux mass for the intruder vanishes , i.e. @xmath138 .
then , we proceed to study the next order , that requires to determine @xmath139 . to first order in @xmath44 , eq .
( [ 2.21 ] ) reads @xmath140-j_{0}[f_{0}^{(1)},f^{(0 ) } ] \nonumber \\ & = & -\partial_{t}^{(1)}f_{0}^{(0)}-{\bm v}\cdot \nabla f_{0}^{(0)}+ g_{0 } \frac{\partial}{\partial v_{z}}f_{0}^{(0)}.\end{aligned}\ ] ] the action of the time derivatives @xmath141 and @xmath142 on the hydrodynamic fields is given by eqs.([2.12 ] ) and ( [ 2.18])-([2.20 ] ) , respectively .
then , evaluating the right hand side of eq .
( [ 3.1 ] ) gives @xmath143 - j_{0}[f_{0}^{(1)},f^{(0)}]\ ] ] @xmath144 where the coefficients of the gradients of the fields on the right hand side are given by @xmath145 @xmath146 @xmath147 { \bm v},\ ] ] @xmath148 here , @xmath149 is the unit tensor in @xmath150 dimensions and @xmath109 was defined above eq .
( [ 2.27 ] ) .
note that the external field does not occur in eq .
( [ 3.1 ] ) .
this is because of the particular form for it we are considering ( gravitational type ) . to proceed , we need the expression of @xmath151 .
it has been derived in ref .
@xcite and reads @xmath152 where the coefficients @xmath153 , @xmath154 , and @xmath155 are functions of the peculiar velocity @xmath87 and the hydrodynamic fields .
then , the solution of eq .
( [ 3.2 ] ) is of the form @xmath156 substitution of eqs .
( [ 3.7 ] ) and ( [ 3.8 ] ) into eq .
( [ 3.2 ] ) and identifying coefficients of independent gradients yields @xmath157 -j_{0}[\mathcal{a}_{0},f^{(0 ) } ] = { \bm a},\ ] ] @xmath158 \mathcal{b}_{0}-j_{0}[f_{0}^{(0)},\mathcal{b } ] -j_{0}[\mathcal{b}_{0},f^{(0)}]\ ] ] @xmath159 @xmath160 \mathcal{c}_{0}-j_{0}[f_{0}^{(1)},\mathcal{c}]-j_{0}[\mathcal{c}_{0},f^{(0)}]\ ] ] @xmath161 @xmath162-j_{0}[f_{0}^{(0)},\mathcal{d}]-j_{0}[\mathcal{d}_{0},f^{(0)}]\ ] ] @xmath163 use of eq .
( [ 3.8 ] ) into eq .
( [ 2.26 ] ) taking into account symmetry considerations , gives the expression for the mass flux of the intruder to first order in the gradients , @xmath164 where @xmath165 is the diffusion coefficient , @xmath166 is the pressure diffusion coefficient , and @xmath167 is the thermal diffusion coefficient . in the following
, the quantities @xmath168 , @xmath169 , and @xmath170 will be evaluated in the first sonine approximation , i.e. we consider @xmath171 with @xmath172 being the maxwellian given by eq.([2.37 ] ) .
consistently , @xmath153 and @xmath154 are approximated by @xmath173 and it happens that both @xmath174 and @xmath175 are zero ( see appendix c in ref .
substitution of the above expressions into eqs .
( [ 3.9])-([3.11 ] ) gives a set of closed equations for @xmath176 , @xmath177 , and @xmath178 .
multiplication of these equations by @xmath179 and integration over @xmath87 yields @xmath180 a_{0}n_{0}t_{0}=-\frac{n_{0}t_{0}}{x_{0}},\ ] ] @xmath181 b_{0}n_{0}t_{0}\ ] ] @xmath182 @xmath183 c_{0}n_{0}t_{0}\ ] ] @xmath184 where @xmath185\ ] ] is a collision frequency .
it can be easily evaluated by using standard integration techniques with the result @xmath186 where @xmath187is the elastic limit .
an expression for @xmath188 for the case @xmath0 , valid for arbitrary concentrations of the two components of the mixture and incorporating some non - gaussian contributions , has been obtained in @xcite .
eq.([3.23 ] ) is consistent with that result .
> from dimensional analysis , it follows that @xmath189 , @xmath190 , and @xmath191 .
then , the temperature and pressure derivatives in eqs .
( [ 3.19])-([3.21 ] ) can be evaluated and the equations become @xmath192 @xmath193 @xmath194 therefore , @xmath195^{-1},\ ] ] @xmath196^{-1},\ ] ] @xmath197 the expressions of the transport coefficients follow by using the above results into eqs .
( [ 3.14])-([3.16 ] ) , @xmath198 @xmath199 @xmath200 these expressions are consistent with those reported in ref .
@xcite for @xmath0 and without external field .
let us now particularize the above results for the steady state considered in the previous section , i. e. an open vibrated system with no macroscopic flow and gradients only in the @xmath9 direction . in this case , eq .
( [ 2.25 ] ) implies that @xmath201 , since the intruder mass flux must vanish for @xmath202 .
thus eqs .
( [ 3.13 ] ) and ( [ 3.31])-([3.33 ] ) yield @xmath203 using eqs .
( [ 2.38 ] ) and ( [ 3.23 ] ) , and also the equation determining @xmath98 , eq .
( [ 2.42 ] ) , it is found @xmath204.\ ] ] this shows that @xmath205 is a function of only @xmath101 and @xmath104 , i.e. of the mechanical properties of the particles ( @xmath98 is given in terms of them by eq .
( [ 2.42 ] ) ) , but it does not depend on the hydrodynamic fields . therefore , eq.([3.34 ] ) can be expressed in compact form as @xmath206 where @xmath207 and @xmath208 integration of eq .
( [ 3.36 ] ) directly provides an expression for the probability density @xmath209 of finding the impurity at height @xmath9 , @xmath210 with @xmath211 being the normalization constant .
the hydrodynamic profiles for the steady state of an open vibrated dilute gas ( without impurity ) where studied in @xcite and the results are shorty summarized in the appendix .
there , it is also indicated how the numerical evaluation of the integral appearing on the right hand side of eq .
( [ 3.39 ] ) is implemented . in figs .
[ fig7]-[fig9 ] the theoretical and simulation results for @xmath212 and @xmath209 are compared for different values of the parameters . in all cases , those of the gas are the same as in fig . [ fig1 ] , i.e. @xmath114 , @xmath113 , and @xmath119 , and the diameter ratio is @xmath213 . on the other hand , in fig .
[ fig7 ] it is @xmath214 , in fig . [ fig8 ] , @xmath215 , and in fig .
[ fig9 ] , @xmath132 .
finally , in each of the figures results for two different values of the coefficient of restitution @xmath70 are reported .
for the cases reported in fig .
[ fig7 ] and [ fig8 ] the agreement between theory and simulation can be considered as quite satisfactory , given that no fitting parameter is being used when constructing the density profile for the impurity from the hydrodynamic profiles of the gas . on the other hand , for @xmath216 ,
[ fig9 ] shows strong discrepancies between theory and simulation results .
there are several possible causes for it .
one is that the heavy intruder is close , on the average , to the vibrating wall and in this region the hydrodynamic description for it is not accurate , as indicated in connection with fig .
in fact , the theoretical prediction for @xmath209 is maximum at the wall .
another closely related possible origin of the discrepancy , is that we are assuming that the intruder does not affect the hydrodynamic profiles of the gas .
this may be a bad approximation when the intruder is too massive as compared with the gas particles , specially taking into account that the number of the latter is not very large .
we remind that this number can not be increased too much , in order to keep the system fluidized with low density and avoiding the transversal instability mentioned in sec .
[ s3 ] .
we have investigated also the density profile of the intruder for other values of the diameter ratio @xmath217 , namely @xmath218 and @xmath219 , and obtained similar results .
> from @xmath209 , the average height of the intruder @xmath220 can be computed through @xmath221 the ratio @xmath222 , where @xmath223 is the height of the center of mass of the gas , provides a measurement of the relative position of the intruder with respect to the gas .
this ratio is plotted in fig .
[ fig10 ] as a function of @xmath98 for several systems with @xmath114 and @xmath224 ( @xmath225 ) .
the solid line is the theoretical prediction and the symbols simulation results for different values of the mass ratio and @xmath70 . figures [ fig11 ] and [ fig12 ] show the same function but for @xmath213 ( @xmath226 ) and @xmath227 ( @xmath228 ) .
the agreement between theory and simulation is good in all cases , specially in the region close to @xmath229 .
in fact , for @xmath132 it is better than it could be expected from fig .
this good agreement is not surprising , since we are looking now to the simplest global property of the distribution @xmath209 . on the other hand ,
the simulation results for @xmath230 are systematically above the theoretical prediction . in general
, it seems that the slope of the simulation results is smaller than the theoretical prediction , although we have no explanation for it .
a relevant conclusion to be reached from the above results is the dominant role played by the value of @xmath98 in determining the relative position of the intruder with respect to the gas . in an elastic system , @xmath98 is equal to the mass ratio @xmath231 , but in a granular gas @xmath98 can be larger or smaller than the mass ratio due to the lack of energy equipartition .
consequently , the relative position of an intruder in a granular gas depends in a rather complicated way on the coefficients of restitution of the system , the mass ratio , and the size ratio .
in this work , we have studied the hydrodynamic profiles of an impurity immersed in a low density granular gas that is in an arbitrary state .
analytical predictions for the temperature and density profiles have been derived and compared with molecular dynamics simulations for the particular case of a vibrated system .
a satisfactory qualitative and quantitative agreement has been found , although the discrepancies increase as the difference between the masses of the gas particles and the intruder increases . when this is the case , the boundary layer next to the vibrating wall becomes rather wide and , moreover , the intruder influences the density profile of the gas , contrary to the tracer limit assumed in the theory .
the theoretical study is based on the validity of the usual chapman - enskog procedure to generate the hydrodynamic equations of the system when starting from the kinetic ( boltzmann ) equations . then , although the results are quite general and can be applied to a variety of situations as the specific one considered here , some caution is needed for other particular cases . for instance , in the so - called steady simple shear state , there is an intrinsic coupling between inelasticity and gradients @xcite .
this implies that the shear rate and the coefficient of restitution can not be considered as independent quantities , and the usual chapman - enskog procedure becomes inaccurate .
this is in contrast with the situation considered here , in which the gradients are controlled by the vibrating wall and the external gravitational field . in general , each physical state
should be checked to verify whether a direct expansion in the gradients is legitimate . for the ratio between the local temperatures of the intruder and the gas
, the theory predicts that it is constant in the bulk of the system .
this has been confirmed by the md simulations .
the same behavior has been found in experiments @xcite and md simulations @xcite of vibrated systems , outside the dilute and tracer limits .
nevertheless , an extension of the analysis presented in this paper @xcite , indicates that the ratio depends also on the local densities ratio .
what happens is that this dependence is rather weak and gives corrections inside the statistical uncertainties for the values of the parameters considered in the above mentioned works .
a qualitative discussion of the relative position of the intruder with respect to the gas was given in @xcite , where the crucial role played by the non - equipartition of energy was discussed .
more precisely , it was shown that the intruder was above or below the center of mass of the gas depending on the sign of the pressure diffusion coefficient @xmath232 , that in turn is mostly determined by the sign of @xmath233 , as seen in eq.([3.32 ] ) .
this is in agreement with the simulation results reported in figs .
[ fig10]-[fig12 ] .
it can be concluded that the difference in temperature of the components of a mixture must be taken into account when studying segregation phenomena , i.e. the demix of the components when shaken .
of course , at higher densities and finite volume fraction of the components , other mechanisms @xcite are also important and even dominant .
it is worth to mention that the work reported here provides another example of the accuracy of hydrodynamics to describe granular systems under rather extreme conditions of inelasticity .
although the inelasticity of the gas has been kept small in order to avoid the transversal instability , predicted by the hydrodynamic theory itself , the coefficient of inelasticity for the gas - impurity collisions has been varied over a wide range ( typically between @xmath234 and @xmath235 ) and a reasonable good agreement between theory and simulation has been found .
this research was supported by the ministerio de educacin y ciencia ( spain ) through grant no .
fis2005 - 01398 ( partially financed by feder funds ) .
in ref . @xcite , analytical expressions for the density and temperature profiles of an open vibrated granular gas in the steady state were derived .
they are given by @xmath236^{2}}\ , , \ ] ] @xmath237 \xi^{-\gamma},\ ] ] where @xmath238 for @xmath1 and @xmath239 for @xmath0 , @xmath240 and @xmath241 are constants to be determined from some boundary conditions , and @xmath242 and @xmath243 denote the modified bessel functions of first and second kind , respectively .
the dimensionless length scale @xmath244 is defined by @xmath245 with @xmath246 being the local mean free path .
finally , @xmath247 and @xmath248 are functions of the coefficient of normal restitution , both vanishing in the elastic limit @xmath249 .
their explicit expressions are given in @xcite .
note that the maximum value of @xmath244 is @xmath250 @xmath251 being the total number of particles and @xmath110 the length or area of the vibrating wall . using the hydrodynamic relation @xmath252 we can write @xmath253 , defined in eq .
( [ 3.38 ] ) , as @xmath254 the expression for @xmath255 is easily obtained by using eqs .
( [ a.1])-([a.3 ] ) and the properties of the modified bessel functions @xcite , @xmath256}{a i_{\gamma}(\xi)+b k_{\gamma}(\xi)}\ , .\ ] ] in this way , we have an explicit expression for the integral appearing in eq .
( [ 3.39 ] ) that , therefore , can be evaluated numerically for each set of values of the parameters .
of course , this requires to determine first the constants @xmath240 and @xmath241 . in the results to be reported here ,
it has been done from the simulation data themselves , namely by fitting the position and value of the temperature minimum @xcite .
another possibility that has been proven to give accurate results would be to require the hydrodynamic heat flux to vanish at infinite height @xcite .
then for a granular gas with @xmath114 , @xmath113 , @xmath120 , and @xmath119 , leading to @xmath257 , @xmath258 , and @xmath259 , the values @xmath260 and @xmath261 are found .
the solid lines in fig .
[ fig1 ] are the fit to the hydrodynamic profiles of the gas obtained in this way .
once @xmath240 and @xmath241 are determined , it is possible to evaluate numerically the right hand side of eq .
( [ 3.39 ] ) for each set of values of the parameters defining the mechanical properties of the intruder . | the hydrodynamic state of an impurity immersed in a low density granular gas is analyzed .
explicit expressions for the temperature and density fields of the impurity in terms of the hydrodynamic fields of the gas are derived .
it is shown that the ratio between the temperatures of the two components , measuring the departure from energy equipartition , only depends on the mechanical properties of the particles , being therefore constant in the bulk of the system .
this ratio plays an important role in determining the density profile of the intruder and its position with respect to the gas , since it determines the sign of the pressure diffusion coefficient .
the theoretical predictions are compared with molecular dynamics simulation results for the particular case of the steady state of an open vibrated granular system in absence of macroscopic fluxes , and a satisfactory agreement is found . |
the first major question in the study of high - mass star formation is whether or not the process is a scaled - up version of the formation of low - mass stars , which are known to form by accretion via a disk , and to disperse their natal envelopes through bipolar outflows . at the other extreme , high - mass stars may form by the coagulation of lower - mass stars or protostellar cores ( see reviews by churchwell and by evans in this volume ) .
the second major question of high - mass star formation is its relation to clustered star formation : does the stellar density influence the emergent mass spectrum ? to what degree
do the conditions in the initial molecular cloud determine the properties of the stellar population it produces ?
the study of high - mass star formation is hampered by the short time - scale of the process .
for example , a late o - type star lasts only @xmath01 myr , equal to the pre - main sequence lifetime of a 1 m@xmath1 star .
moreover , 15% of this time is spent embedded in natal gas and dust that hides the photosphere from optical view . as a consequence ,
known regions of high - mass star formation lie at distances of several kpc , so that details @xmath2 au are unresolved at typical @xmath3 resolution .
the large distances cause considerable confusion in these regions , and make it difficult to construct source samples with a common luminosity and distance , necessary for systematic studies . to answer these questions , it is necessary to build up a statistically significant sample of regions forming high - mass stars , and to study their physical and chemical properties .
this review describes recent studies of molecular material in which stellar groups including high - mass stars are forming .
the focus is on detailed studies of small samples ; larger source samples are described by bronfman et al .
( 1996 ) , brand et al .
( 2001 ) and sridharan et al .
the first sample consists of nine deeply embedded objects of @xmath4 l@xmath1 , @xmath5 kpc , which have strong mid - infrared and weak cm - wave emission .
the second set contains nine molecular clumps which have @xmath6 l@xmath1 , @xmath7 kpc and weak mid - infrared emission , and are close to ultracompact h ii regions .
both samples are characterized by strong submillimeter continuum and line emission , associated h@xmath8o and ch@xmath9oh masers , and molecular outflows . in many cases ,
the distances are only kinematic estimates .
better distance estimates may come soon from near - infrared spectroscopy ( hanson et al .
1997 ; kaper , this volume ) .
the determination of the physical structure of high - mass star - forming regions consists of two parts : first the measurement of the total mass and the temperature structure , and second the determination of the density structure .
the mass and temperature structure can be obtained by fitting dust models to the broad - band infrared spectrum .
masses derived from molecular line data are less reliable due to uncertain molecular abundances . as illustrated in fig .
[ fig : gl2591 ] ( solid line ) , a single model can not fit the entire spectrum , so that to measure the mass , the model must be fitted to the optically thin part , at @xmath10 m .
early models ( wolfire & cassinelli 1986 ; churchwell et al . 1990 ) mostly fitted the near - infrared part of the spectrum , which indicates @xmath05 times lower optical depths than the submillimeter part due to deviations from spherical geometry .
this result holds for both bright and weak mid - infrared sources ( harvey et al .
2000 ; hatchell et al .
2000 ) . the source gl 2591 offers a rare opportunity to measure the submillimeter dust opacity by comparing dust masses to those derived from co isotopic emission .
while the co mass is usually a lower limit since an unknown fraction of co is frozen out onto dust grains , mid - infrared observations of gl 2591 indicate a ratio of gas - phase to solid - state co column density of @xmath11 ( mitchell et al .
1989 ) , reducing the uncertainty in co abundance to better than a factor of two .
the dust masses of gl 2591 quoted in van der tak et al .
( 1999 ) for draine and lee ( 1984 ) and mathis et al .
( 1983 ) dust are a factor of 8 too high due to the use of a grain size of @xmath12 m rather than @xmath13 m . however , the conclusion that the dust model of ossenkopf & henning ( 1994 ) matches the mass derived from c@xmath14o best is still valid .
this model for dust in star - forming regions includes grain coagulation and the formation of ice mantles .
the derived temperature profile indicates @xmath15
@xmath16100 k at @xmath17 @xmath181000 au from the central star . within this radius ,
ice mantles on the grains should have evaporated .
the dotted line in fig . [ fig : gl2591 ] is a calculation using the coagulated grain model without ice mantles .
this model clearly fits the mid - infrared part of the spectrum better than the model with ice mantles .
the density structure can be derived either by fitting dust models to the spatial profile of dust continuum emission , or by fitting the line spectrum of a molecule with a high dipole moment ( such as cs ) that is sensitive to density .
the first approach assumes that the grain optical properties do not vary with position ; the second that the molecular abundance does not .
while either assumption may be invalid , the methods can serve as tests of each other .
van der tak et al .
( 2000a ) found the two methods to agree well for six embedded high - mass star - forming regions , while for a seventh , s 140 , the assumption of a centrally heated envelope appeared invalid .
the good agreement between the dust and cs results may have been fortuitous due to the use of a gaussian beam shape .
models of the 350 @xmath19 m dust emission profile of gl 2591 using the actually measured beam shape indicate an @xmath20 density profile , steeper than the @xmath21 profile found from cs ( mueller et al . , this volume ) .
if this result also holds for the scuba data and the other sources , it indicates a depletion of cs at small radii .
the alternative explanation of enhanced dust emissivity at small radii is not predicted by dust models , in which the main effect would be ice mantle evaporation , which drops the opacity .
the fractional ionization of molecular clouds controls both the influence of magnetic fields on their dynamics and their gas - phase chemistry . in regions shielded from ultraviolet starlight ,
the main source of ionization is by cosmic rays which produce h@xmath22 , followed by proton transfer reactions into hco@xmath23 and h@xmath9o@xmath23 .
current best estimates of the cosmic - ray ionization rate @xmath24 based on the abundances of oh and hd measured in visual absorption toward nearby diffuse clouds have a factor of ten uncertainty ( federman et al .
an alternative approach using the abundances of hco@xmath23 and dco@xmath23 measured through millimeter emission lines towards nearby dark clouds is no more accurate due to uncertain and potentially large depletion of co and o ( caselli et al .
1998 ; williams et al . 1998 ) . at the current sensitivity , both these methods
are limited to the solar neighbourhood and can not give information on variations of @xmath24 on galactic scales . the infrared h@xmath22 detections ( geballe & oka 1996 ; mccall et al .
1999 ) for well - characterized lines of sight make the embedded high - mass objects good targets for a more direct determination of @xmath24 .
van der tak & van dishoeck ( 2000 ) combine the temperature and density structure of seven embedded sources with a small chemical network to compute their ionization structure ( fig .
[ fig : zeta ] ) and fitted the observed h@xmath22 absorption and h@xmath25co@xmath23 emission lines .
the mean value of @xmath26 s@xmath27 is consistent with the diffuse and dark cloud estimates , and with recent voyager and pioneer spacecraft data at distances of up to 60 au from the sun ( webber 1998 ) .
the source - to - source spread in @xmath24 of a factor of 2 - 3 on scales of a few kpc agrees with results from @xmath28ray observations ( hunter et al .
this agreement may be a coincidence since the @xmath28rays are of higher energy ( @xmath0gev ) than the particles providing the bulk of the ionization ( a few 10 mev ) .
intervening clouds affect the h@xmath22 but not the h@xmath25co@xmath23 data .
alternatively , the spread may be due to shielding against cosmic rays .
the data suggest a trend of decreasing @xmath24 with increasing total column density , but not at a significant level .
however , hco@xmath23 probes the outer envelopes ( fig . [
fig : zeta ] ) , because in regions with @xmath29 k , hco@xmath23 reacts with evaporating h@xmath8o ice . using tracers of the innermost parts of a low - mass core , caselli et al .
( 2001 ) found @xmath30 s@xmath27 .
follow - up studies may reveal by how much cosmic rays are attenuated inside dense cores .
the models described in 2 have allowed us to study the chemical composition of the envelopes of embedded high - mass stars as a function of distance from the central star .
the first example is the ch@xmath9oh molecule studied by van der tak et al .
( 2000b ) .
modeling of spectra indicates three types of sources ( fig . [
fig : ch3oh ] ) : ` cold ' sources with ch@xmath9oh / h@xmath8 @xmath31 , ` warm ' sources with ch@xmath9oh / h@xmath8 @xmath32 , and sources where the abundance ` jumps ' from @xmath31 to @xmath33 .
this ` jump ' occurs at @xmath34 k , where solid h@xmath8o , the bulk of icy grain mantles , evaporates .
combined with the high observed abundances of solid ch@xmath9oh in these sources , these ` jumps ' suggest an evolutionary sequence where ch@xmath9oh forms on dust grains in ` cold ' sources , then evaporates into the gas phase in ` jump ' sources , and is broken down by reactions with ions in ` warm ' sources .
however , gas - phase ch@xmath9oh abundances measured in the submillimeter are factors of @xmath35100 lower than those derived from solid - state data .
since the nearby orion region is an exception , resolution could play a role .
observations of gas - phase ch@xmath9oh in infrared absorption with ground - based instruments with high spectral resolution may clarify this point .
a much earlier phase of evolution is probed by the co molecule .
abundances derived from c@xmath14o line emission increase systematically by a factor of @xmath363 as @xmath37=20@xmath3840 k , where @xmath37 is the average temperature of the envelope , weighted by mass . since these temperatures are the sublimation points of pure co and mixed co
h@xmath8o ices , freeze - out and evaporation appear to control the co abundance ( van der tak et al .
2000a ) . in low - mass cores ,
kramer et al . ( 1999 ) found a similar relation of co abundance with temperature .
a third example of depth - dependent chemistry is the enhancement of hcn at temperatures @xmath39 k ( boonman et al . 2001 ) . on the other hand ,
the abundance of h@xmath8co was found to be constant within the range @xmath40 k , suggesting that this species is formed either in the gas phase or in co - dominated ice .
the latter idea is consistent with evidence from d@xmath8co ( ceccarelli et al .
the co@xmath8 molecule may be tracing the thermal history of star - forming regions . while the co@xmath8 ice abundance is high and evaporation of ices
is observed , its gas - phase abundance is low , so that co@xmath8 must be rapidly destroyed after it evaporates ( van dishoeck 1998 ) . however , there is no obvious low - temperature destruction path .
the required @xmath41 k for co@xmath8 destruction is well above measured values , suggesting impulsive heating events in the past .
possibilities include outflow shocks ( charnley & kaufman 2000 ) , if desorption of the ices is also caused by these shocks .
other options are episodic accretion ( doty et al .
2001a ) and the infrared flares discussed by bally ( this volume ) .
the evolution of the temperature and density structure of embedded objects was studied by van der tak et al .
( 2000a ) .
conventional tracers of physical evolution , used for low - mass objects , do not change much over the narrow age range of the objects studied in detail .
in particular , the shape of the spectral energy distribution does not change with time , as most material remains at low temperatures .
the radio continuum flux does not seem indicative either , partly due to the unknown contribution of shocks , and partly because the surface temperatures of the ionizing stars are not well known .
however , the embedded objects can be ordered in a temperature sequence , based on the gas / solid ratios of h@xmath8o and co@xmath8 , the fraction of heated solid @xmath25co@xmath8 , the 45/100 @xmath19 m colour , and the excitation temperatures of co , c@xmath8h@xmath8 and ch@xmath9oh .
these indicators cover a wide range of temperatures and spatial scales , demonstrating that geometry is not an effect . since crystallization is an irreversible process , the heated @xmath25co@xmath8 ice measures the maximum temperature .
hence , the temperature change must be a systematic increase rather than random fluctuations , although gas - phase co@xmath8 argues the opposite .
still , the sequence may be evolutionary since it is correlated with the ratio of envelope mass to bolometric luminosity .
if confirmed through a larger source sample , this correlation indicates the progressive dispersal of the envelope . as for the time order of particular phenomena ,
the small body of existing data suggests that ch@xmath9oh masers occur first , followed by h@xmath8o masers , and later by oh masers .
the situation is complex as each kind of maser may trace more than one physical component ( minier et al .
very massive and energetic outflows occur throughout the embedded phase , like for low - mass objects ( beuther , this volume ) . the chemical composition of massive protostellar envelopes may serve as an independent probe of their structure and evolution .
a good test case is gl 2591 , for which a large body of data has been analyzed , both in submillimeter emission ( jcmt , swas ) and in infrared absorption ( iso - sws , irtf ) .
doty et al . ( 2001b ) coupled the temperature and density structure of gl 2591 by van der tak et al .
( 2000 ) to a complete gas - phase chemical network .
it appears that the measured molecular abundances can be reproduced to within factors of a few for a source age of @xmath42 yr .
the results discussed so far pertain to a limited range of ages and luminosities . to see
just how limited , we are investigating the set of line - rich molecular line sources near ultracompact h ii regions from hatchell et al .
. observations of ch@xmath9oh ( fig .
[ fig : g7578 ] ) indicate the presence of both cold ( 20 - 50 k ) and warm ( 110 - 500 k ) gas .
the column density ratio of these components ranges from @xmath180.1 to @xmath1660 , comparable to the values found in the embedded bright mid - infrared sources .
the large implied masses of cold gas argue against a more evolved status of these regions , as would seem plausible based on their higher abundances of complex organic molecules . to analyze the data in more detail
, we are developing models for the temperature and density structure , based on tracers of dust ( scuba ) and gas ( cs lines ) .
regions forming high - mass and low - mass stars have a different observational appearance , but their underlying structure and evolution seem similar .
regions forming massive stars are larger and hence more massive , but their absolute densities and their density distributions are similar to those found in low - mass star - forming regions .
the more rapid dispersal of high - mass envelopes may be mainly driven by the faster pace of evolution of the central stars .
there is mounting evidence for disks around high - mass stars , outflows are found to be ubiquitous , and compact dust and gas cores seem common as well .
the source samples with and without nearby ultracompact h ii regions appear to evolve in parallel , suggesting that the proximity of radio emission and the shape of the spectral energy distribution are not good clocks . however , the excitation and abundance of selected molecules appear excellent tracers of evolution . ,
a. m. s. , stark , r. , van der tak , f. f. s. , et al.2001 , , 553 , 63 , j. , cesaroni , r. , palla , f. , molinari , s. 2001 , , 370 , 230 bronfman , l. , nyman , l .- a . , may , j. 1996 , , 115 , 81 , p. , walmsley , c. m. , terzieva , r. , herbst , e. 1998 , , 499 , 234 , p. , walmsley , c. m. , zucconi , a. , et al.2001 , , in press , c. , loinard , l. , castets , a. , et al.2001 , , 372 , 998 , s. b. , kaufman , m. j. 2000 , , 529 , l111 churchwell , e. , wolfire , m. g. , wood , d.o.s . 1990 , , 354 , 247 doty , s.d . ,
van dishoeck , e.f . ,
van der tak , f.f.s . , et al.2001a , , in press doty , s. d. , van dishoeck , e. f. , van der tak , f. f. s. , boonman , a. m. s. 2001b , , submitted , b. t. , lee , h. m. 1984 , , 285 , 89 , s. r. , weber , j. , lambert , d. l. 1996 , , 463 , 181 , t. r. , oka , t. 1996 , nature , 384 , 334 hanson , m.m . , howarth , i.d . , conti , p.s .
1997 , , 489 , 698 , p. m. , butner , h. m. , colom ' e , c. , di francesco , j. , smith , b. j. 2000 , , 534 , 846 hatchell , j. , fuller , g.a . ,
millar , t.j . ,
thompson , m.a . ,
macdonald , g.h .
2000 , , 357 , 637 hatchell , j. , thompson , m.a . ,
millar , t.j . ,
macdonald , g.h .
1998 , , 133 , 29 , s. d. , bertsch , d. l. , catelli , et al .
1997 , , 481 , 205 , c. , alves , j. , lada , c. j. , et al.1999 , , 342 , 257 , j. s. , mezger , p. g. , panagia , n. 1983 , , 128 , 212 , b. j. , geballe , t. r. , hinkle , k. h. , oka , t. 1999 , , 522 , 338 , v. , conway , j. e. , booth , r. s. 2001 , , 369 , 278 , g. f. , curry , c. , maillard , j .-
p . , allen , m. 1989 , , 341 , 1020 , v. , henning , t. 1994 , , 291 , 943 sridharan , t.k . ,
beuther , h. , schilke , p. , et al .
2001 , apj , in press , f. f. s. , van dishoeck , e. f. 2000 , , 358 , l79 , f. f. s. , van dishoeck , e. f. , evans , n. j. , ii , bakker , e. j. , blake , g. a. 1999 , , 522 , 991 , f. f. s. , van dishoeck , e. f. , evans , n. j. , ii , blake , g. a. 2000a , , 537 , 283 , f. f. s. , van dishoeck , e. f. , caselli , p. 2000b
, , 361 , 327 , e. f. 1998 , faraday discussion 109 , 31 , w. r. 1998 , , 506 , 329 , j. p. , bergin , e. a. , caselli , p. , myers , p. c. , plume , r. 1998 , , 503 , 689 wolfire , m.g . , cassinelli , j.p .
1986 , , 310 , 207 | this paper is a review and an update on recent work on the physical and chemical structure of the envelopes of newly born massive stars , at the stages preceding ultracompact h ii regions .
it discusses methods and results to determine total mass , temperature and density structure , ionization rate , and depth - dependent chemical composition . |
the outer atmosphere of the sun , the solar corona , is most clearly discernible when seen in soft x - rays . since the early rocket experiments in the late 1960s
, it has become widely recognized that soft x - ray imagery of the sun provides a powerful means to investigate physical conditions of hot plasmas ( whose temperature often exceeds 1 mk ) that prevail in the corona .
coronal imaging with the _ soft x - ray telescope _ ( sxt )
onboard the _ yohkoh _ ( _ solar - a _ ) satellite @xcite ( operation period : 19912001 ) covered a full solar activity cycle and have revealed that magnetic reconnection plays an essential role in the energy release processes in the dynamic solar corona .
the _ solar - b _ satellite was launched at 21:36 ut on 22 september 2006 .
it was named _ hinode " _ , which means sunrise in japanese (;
figure [ fig : hinode ] ) . the _ x - ray telescope _ ( xrt )
onboard _ hinode _ is a successor of the _ yohkoh_/sxt .
it also employs grazing - incidence optics , but with improved spatial resolution ( consistent with 1@xmath0 ccd pixel size compared with @xmath1 2.5@xmath0 pixel size in the case of sxt ) while maintaining similar exposure cadence to that of sxt ( only order of few milli - seconds for flare and order of few seconds for active region ) .
the xrt has the capability of imaging emission formed at much longer wavelengths than sxt .
the combination of a backside thinned ccd and thin al - mesh filter allow xrt to image emission significantly longward of 60 .
this is a major difference from _
yohkoh_/sxt in that the xrt can observe not only high - temperature plasmas ( @xmath2 2 mk ) seen with sxt , but also low - temperature ( @xmath3 2 mk , reaching even below 1 mk ) plasmas which comprise a significant amount of the corona . with this enhanced temperature range for observing coronal plasmas , coupled with increased spatial resolution ,
the xrt is able to perform detailed imaging observations of a wide variety of coronal plasmas in a temperature range covering , continuously , from below 1 mk to well above 10 mk .
one of the most significant scientific features of the xrt is its coronal - temperature - diagnostic capability , namely capability to make temperature maps , for such plasmas . in order to have xrt perform coronal temperature diagnostics with its full capability
, we carefully calibrated the effective area of the xrt and its response to coronal temperatures using not only ground - based test data but also on - orbit data observed in x - rays and visible light .
the effect of the on - orbit contamination , which manifested itself as decreasing intensity of the sun s corona with time as imaged by xrt , was also calibrated as accurately as possible . on the basis of our calibration results ,
we review the coronal - temperature - diagnostic capability of xrt with the filter - ratio method . in section [ sec : optical elements of xrt ] , the optical elements of xrt are briefly mentioned . in section [ sec :
overview of calibration ] , we show the overview of the calibration performed in this article . in section [ sec : on - orbit calibration ] , we summarize how to identify the contaminant and how to measure its accumulating thickness on the focal - plane analysis filters ( fpafs ) and ccd as a function of time .
this then identifies the effective area and temperature response of xrt including the contamination at every period . in section [ sec :
temperature ] , we evaluate the coronal - temperature - diagnostic capability of xrt , explain the filter - ratio method to derive the coronal temperature , and present suitable filter pairs for each coronal temperature .
an example of coronal temperature distribution with the xrt data is also shown .
finally , we summarize the result of this article in section [ sec : summary ] .
additionally , in appendix [ sec : ground - based calibration ] , the calibration of the nine x - ray fpafs with ground - based end - to - end test data is explained . in 20022003
, the x - ray transmission measurement of the fpafs was performed at the x - ray astronomy calibration and testing ( xact ) facility of the istituto nazionale di astrofisica / osservatorio astronomico di palermo g.s . vaiana " @xcite .
however , some of the calibrated fpafs were unfortunately damaged . in this article
, we characterize the complete flight set of fpafs , namely both the un - damaged and re - manufactured fpafs , with another ground - based test performed in 2005 .
our calibrated thicknesses are consistent with the thicknesses available from xact data .
an overview of the xrt scientific objectives , design , and performance of the xrt telescope are summarized by , while the x - ray camera is described by .
in order to describe in detail the calibrations performed on the xrt , we first briefly describe the optical elements .
figure [ fig : optical elements ] shows the optical elements and optical paths of xrt .
the xrt can take not only x - ray images but also visible light ( g - band ) images for co - aligning images between xrt and other instruments mounted on _
the solid and dashed lines in figure [ fig : optical elements ] show the optical paths in x - rays and visible light , respectively . the xrt optical elements for x - rays consist of pre - filter ( pf ) , two grazing - incidence mirrors ( m1 and m2 ) , two focal - plane analysis filters ( fpaf1 and fpaf2 ) , and ccd .
* the pre - filter consists of eight fan - shaped annular filter segments each made of thin aluminum on a polyimide film ( figure [ fig : pf ] ) , which were manufactured by luxel corporation .
the aperture shape was designed to obscure the locations of mirror - bonding pads to avoid possible degradation in image quality caused by surface deformation of the mirror at the bonding pads . in figure
[ fig : pf ] , the aperture areas are shown in color with their respective opening angles . the average thickness of the entire pre - filter assembly is then given as @xmath4 , where @xmath5 and @xmath6 are the thickness and opening angle of each filter segment , respectively .
the average pre - filter consists of 1538 al and 2030 polyimide based on the measurement by manufacturer .
* the m1 and m2 mirrors are grazing - incidence annular mirror manufactured by goodrich , each made of zerodur . on the basis of measurements by the manufacturer , the annular entrance aperture of the xrt primary mirror is located between radii of @xmath7 = 17.042446 cm and @xmath8 = 17.074051 cm . considering the open angle of pre - filter ( 242.04@xmath9 , see figure [ fig : pf ] ) , the geometric aperture area of xrt
is calculated to be @xmath10 = 2.28 @xmath11 .
the grazing angle of x - rays at each xrt mirror is about 0.91@xmath9 on average .
* xrt has two filter wheels : fw1 and fw2 .
each filter wheel has six positions to mount focal - plane analysis filters ( fpafs ) .
as shown in figure [ fig : fw ] , five x - ray analysis filters are mounted on fw1 , and fw2 has four x - ray filters and one filter for visible - light observation ( g - band ) .
these nine x - ray filters are made of several kinds of metal and support , with different thicknesses as summarized in table [ tbl : fpaf ] .
the filters are designed to observe the corona in a temperature range from less than 1 mk to more than 10 mk .
focal - plane analysis filters on fw1 and fw2 ( hereafter fpaf1 and fpaf2 , respectively ) can be selected independently , even allowing combinations of filters both from fw1 and fw2 in series . *
x - rays are focused on a back - illuminated ccd @xcite . the effective area [ @xmath12 of xrt is defined by the product of the geometric aperture area [ @xmath13 and the efficiency of all of the optical elements : @xmath14 where @xmath15 is the transmission of the pre - filter , @xmath16 and @xmath17 reflectivities at the primary and secondary mirrors , @xmath18 and @xmath19 the transmissions of fpaf1 and fpaf2 , and @xmath20 the quantum efficiency of the ccd . because each efficiency is a function of wavelength , @xmath21 is also a function of wavelength . in this article , we calibrate @xmath15 , @xmath18 , and @xmath19 , and then derive the effective area .
when xrt takes visible - light images in g - band , the visible - light shutter is opened , and fw1 and fw2 are set to the open and g - band filter position , respectively .
this g - band filter is a bandpass filter with a central wavelength of 4305.6 and a bandwidth of 172.8 ( fwhm ) .
the visible light is focused onto the same ccd as x - rays ( figure [ fig : optical elements ] ) .
x - rays can not reach the ccd through the g - band ( dotted line ) or the x - ray ( solid line ) paths , because those x - rays are blocked by g - band filters employed at the front of the xrt and on fw2 , respectively .
.focal - plane analysis filters and pre - filter .
[ cols="^,^,^,^,^,^,^,^,^ " , ] @xmath22 : : the percentage in parenthesis shows the ratio of each component to the total photon flux . using the best - fit results in figure [ fig : fpc data ] , we obtained the incident spectrum into xrt ( see figure [ fig : xrcf spectrum ] ) .
the errors shown by gray areas in figure [ fig : xrcf spectrum ] are estimated from the photon noise and the energy range of fpc data points .
for example , the error bars in o - k and c - k line are larger than for the other lines , because the total photon number is small in the o - k line case , and because the energy range of the c - k line does not completely cover the continuum .
the photon - number fluxes of characteristic x - ray lines and continua are summarized in table [ tbl : xrcf comp ] .
the contribution of continuum is not negligible even for the al - k line .
the support of the al - mesh filter is a stainless steel mesh .
figure [ fig : mesh ] is a picture of the mesh taken by the filter manufacturer with a high - precision digital microscope .
according to the manufacturer , the width of the mesh wire and the open area are 38 @xmath23 m and 330 @xmath23 m , respectively . on the basis of this measurement ,
the open area occupies 77% of whole area .
we adopt this measured value of 77% for the geometrical open area . here
we define annulus transmission " as the fraction of the x - rays passed though the open area of the mesh to the incident x - rays into the mesh .
note that the x - rays pass through the open area without any loss of intensity , while at the wires , the x - rays are completely blocked .
hence , the average of annulus transmission should be consistent with the geometrical open area of the mesh .
the annulus transmission is not uniform and depends on the relative position of the shape of x - rays on the filter with respect to the mesh pattern .
however , we do not know this relative position for the flight xrt .
instead , we derive distribution of the annulus transmission for all possible relative positions , take the average of the distribution as the annulus transmission of the al - mesh filter , and employ the deviation of the distribution as that for the annulus transmission . the procedure is detailed in the following .
the shape of x - rays imaged on the fpafs is an annulus , because the shape of the xrt aperture is an annulus ( see figure [ fig : pf ] ) .
the radius and width of annular x - ray image are about 4.1 mm and 7.6 @xmath23 m on the fpafs , respectively , because the radius and width of xrt annular aperture , focal length , and distance between focus position and fpaf are about 170 mm , 0.32 mm , 2700 mm , and 65 mm , respectively .
figure [ fig : mesh trans image ] shows some schematic examples of the relationship between the annular x - ray image and the mesh .
the gray area indicates the mesh of al - mesh filter and the black ring shows the x - ray annulus on the filter .
the variation in annulus transmission is shown in figure [ fig : mesh trans ] .
the left panel shows a cell ( unit opening ) of the mesh .
the width and height of this cell are both 368 @xmath23 m on the basis of manufacturer s measurement .
the black and white regions indicate mesh wire and open area , respectively .
the right panel shows the annulus transmission in gray scale at the position where the center of the annular x - ray image is located at the corresponding left panel .
figure [ fig : mesh trans histogram ] shows distribution of the annulus transmission over the varying center position of the x - ray annulus with respect to the mesh pattern ( this is the histogram of calculated annulus transmission in the right panel of figure [ fig : mesh trans ] ) .
the average of annulus transmission is consistent with the geometrical open area of the mesh .
however , the annulus transmission has a scatter with the standard deviation @xmath24 of 2.3% .
this @xmath24 is also considered in the calibration of al - mesh filter thickness .
the uncertainty in the annulus transmission originated from the known relative position of the annulus center with respect to the mesh pattern can thus be evaluated in this way . in this evaluation
, we assumed that the shape of the x - ray image on the filter is a perfect annulus .
but , in reality , the shape corresponds to the colored area in figure [ fig : pf ] and is not symmetrical .
hence , we should consider not only the relative position but also the relative angle between the patterns along the image annulus and the direction of the mesh wires , which is also not known . in order to assess the amount of additional uncertainty caused by this
, we calculated the scatter of annulus transmission due to the uncertainty of the relative position and angle .
the standard deviation @xmath25 of this scatter ( imperfect annulus case ) turned out to be almost the same as @xmath26 ( perfect annulus case ) .
hence , we claim our treating the image shape as a perfect annulus is sufficient for deriving the average fraction and @xmath24 of the annulus transmission for the x - ray image on the al - mesh filter .
after the first bakeout , spot - like patterns appeared in both x - ray and g - band data as shown in figure [ fig : spot ] .
subsequent analyses indicated that the estimated thickness of contaminant on the ccd was more than 3000 before the first bakeout ( see figure [ fig : contam on ccd ] ) .
the patterns most likely originate from the condensation of contaminant and are triggered by a bakeout with the thickness of the contaminants accumulated on the ccd exceeding a certain threshold value ( the value itself is not known ) .
these spots were not removed at all even with the second bakeout that was performed after the first bakeout and lasted for a month .
the ratio of spot area to full ccd area was 2.6% at this point ( see table [ tbl : bakeout ] ) .
as it took five months before we agreed , as the instrument team , to proceed with performing the next ( third ) bakeout , the resultant thickness of the contaminants again exceeded 3000 , reaching even beyond 3500 ( see figure [ fig : contam on ccd ] ) .
this resulted in the creation of additional contamination spots across the ccd , with an increased spot area ratio of 5.2% .
note , however , this additional increase was rather an intentional one ; we took the option of keeping a low - temperature diagnostic capability by removing the accumulated contaminants while accepting an increase in the spot area .
the resultant spot distribution is indicated by the black areas in figure [ fig : spot map ] . after the third ccd bakeout
, we regularly perform the bakeout every three to four weeks . as we guessed , no more spots were created after the third bakeout as shown in table [ tbl : bakeout ] .
the ratio of spot area to full - ccd area has kept to be 5.2% .
we investigate the relation between the thickness of some spots measured with their newton - ring patterns in g - band and the absorbed x - ray intensities by such spots , and conclude that the material of spot contaminant is most likely the same material as the laminar contaminant .
this result will be presented as a separate paper with more detailed analysis .
in appendix [ subsubsec : contam monitor method ] , in order to measure the contaminant thickness on the ccd , we used the transmission of visible light into the ccd through contamination layers , equation ( [ eq : t_g ] ) . here , we derive this equation . the reflected and transmitted amplitudes [ @xmath27 and @xmath28 of rays at a boundary of two layers whose refractive indexes are @xmath29 and @xmath30
are written as @xmath31 @xmath32 where @xmath33 is the amplitude of incident rays from the @xmath29 layer to the boundary .
we consider transmission of three layers as shown in figure [ fig : contam ccd ] . here
, we define @xmath33 as the amplitude of incident rays in vacuum ( @xmath29 layer ) , and @xmath34 the complex amplitude of rays which are reflected @xmath35 times at the borders of layers ( @xmath29@xmath30 border and @xmath30@xmath36 border ) and transmitted into the ccd ( @xmath36 layer ) .
note the amplitudes of rays shown by white and black arrows in figure [ fig : contam ccd ] are @xmath37 and @xmath38 .
@xmath34 is expressed as @xmath39 where @xmath40 ( @xmath41 ) gives the phase difference between @xmath34 and @xmath42 .
hence , the total amplitude [ @xmath43 of incident rays into the ccd is derived as @xmath44^{-1 } .
\label{eq : a_t}\end{aligned}\ ] ] using the relation between amplitude @xmath45 and intensity @xmath46 in a layer of refractive index @xmath47 given as @xmath48 where @xmath49 denotes complex conjugate , the intensity @xmath50 in vacuum ( @xmath29 layer in figure [ fig : contam ccd ] ) is @xmath51 and the total intensity @xmath52 detected by the ccd is @xmath53 equation ( [ eq : t_g ] ) is derived from equations ( [ eq : a_t ] ) , ( [ eq : i_n1 ] ) , and ( [ eq : i_t ] ) .
when the molecular formula of pure metal and oxidized metal are x and x@xmath54o@xmath55 , and the molecular weight of x and o are @xmath56 and @xmath57 , respectively , the reduced thickness [ @xmath58 of pure metal by the creation of oxidized metal with a thickness of @xmath59 is derived to be @xmath60 based on the conservation of the number of metal atoms .
the information about materials used in our calibration is summarized in table [ tbl : material ] .
the photon noise can be derived with equation ( [ eq : sigma_i ] ) . for the condition ( b ) in section [ subsec : suitable filter pair ] , we investigated how much photon noise is expected in the observed data with the single temperature of the filter - ratio temperature .
however , the actual solar corona may have a multi - temperature structures .
hence , here we discuss the meaning of photon noise estimate with the filter - ratio temperature .
figure [ fig : filter ratio temperature ] shows the correlation between the fractional photon noise derived with dem [ @xmath61 and with the filter - ratio temperature [ @xmath62 for four dems ( regions ) summarized in table [ tbl : coronal structure ] .
@xmath63 is the photon noise calculated with the filter - ratio temperature [ @xmath64 and equation ( [ eq : sigma_i ] ) .
@xmath65 is the photon noise derived with the dem as following steps : _ i _ ) we calculate the photon noise [ @xmath66 for each temperature bin of dem with equation ( [ eq : sigma_i ] ) .
_ ii _ ) we derive the net photon noise of step _ i _ ) as @xmath67 .
the @xmath63 and @xmath65 is calculated for all filters which are included in the suitable filter pairs in figure [ fig : suitable filter pair 800 ] with the typical values of coronal structures in table [ tbl : coronal structure ] . on the basis of this figure
, we can say that @xmath63 well matches @xmath65 which is the actual photon noise .
this equality can be understood as following : since the amount of plasma at the dem peak is dominant , its photon noise mainly affects @xmath65 .
in the temperature range where the dem peak is located , the conversion factor [ @xmath68 from observed dn to @xmath69 ( see equation [ eq : sigma_i ] ) is almost constant for the filters which are included in the suitable filter pairs ( see figures [ fig : suitable filter pair 0 ] and [ fig : suitable filter pair 800 ] ) .
for example , in a temperature range of 34 mk where the active region has the dem peak ( see figure [ fig : filter pair2 ] ) , @xmath70 for al - poly , ti - poly , and med - be filters varies within factors of 1.07 , 1.11 , and 1.02 , respectively ( see the bottom panel in figure [ fig : k ] ) .
hence , @xmath65 well matches @xmath63 , since @xmath63 is derived from @xmath70 at @xmath71 , and @xmath71 is close to the dem peak temperature ( see section [ subsec : filter ratio temperature ] ) .
we note that , for temperature ranges other than the dem peak , @xmath70 varies with a factor of typically 24 .
however , since _ i _ )
the photon noise is a function of @xmath72 ( see equation [ eq : sigma_i ] ) and furthermore , _ ii _ ) in this temperature range , the contribution of photon noise to @xmath65 is smaller than at the dem peak , the variation of @xmath70 in these temperature ranges does not significantly affect @xmath65 .
in appendix [ subsubsec : x - ray analysis in phase 3 ] , we identified the contaminant material as a long - chain organic compound whose chemical composition , density , and refractive index are similar to those of dehp , although dehp is not used in the xrt .
this means that the actual contaminant material is narrowed down to the material whose refractive index and density are close to 1.5 and 1 g @xmath73 , respectively .
next let us discuss the chemical composition of the actual contaminant material .
since the atomic number of silicon is about two times larger than carbon and oxygen , _
i.e. _ the cross - section of silicon to x - rays is certainly larger , the existence of silicon affects on the x - ray transmission certainly .
in fact , when we assume that the contaminant contains some silicon atoms , the discrepancy between the observed and expected x - ray intensity ratios rises as shown in figure [ fig : x - ray si filter contam ] .
hence , we conclude that there are no silicon atoms in the contaminant material .
on the other hand , since the atomic numbers of carbon and oxygen ( also nitrogen ) are comparable and that of hydrogen is much smaller than carbon and oxygen , _
i.e. _ hydrogen is much more transparent to x - rays , some slight differences in numbers of such atoms among a long - chain organic compound do not make any significant difference in the transmission of x - rays .
hence , although we can not exactly identify the chemical composition of the contaminant material , it is acceptable for the calibration of the xrt that we tentatively employed dehp as the contaminant material .
we note that the identified property of the contaminant , where chemical composition is a long - chain organic compound without silicon , refractive index is @xmath1 1.5 , and density is @xmath1 1 g @xmath73 , is a common property of materials which are widely used for satellites , and which are well known as possible source of contamination . in appendix
[ subsubsec : contam monitor method ] , we explained the method to measure the thickness of contaminant on the ccd using g - band data .
the g - band intensity oscillation shown in figure [ fig : g - band 2nd bake ] is essentially expressed by equation ( [ eq : anti - refraction ] ) .
however , there are two differences between the calculated g - band intensity enhancement with the anti - reflection effect of equation ( [ eq : anti - refraction ] ) and the observed one that is characterized by the empirical model given by equation ( [ eq : empirical model ] ) . * in appendix [ subsubsec : measurement of contam thickness on ccd ] , from the observed amplitude of intensity enhancement in g - band , the refractive index of contaminant is estimated as @xmath1 1.4 . * in appendix [ subsubsec : x - ray analysis in phase 3 ] , the decrease in x - ray intensities caused by the accumulation of contaminant can be well explained by adopting a well - known material whose refractive index and density of contaminant are 1.5 and 0.986 g @xmath73 , respectively , as the contaminant material .
this difference may be explained with the possible hypothesis where the density of actually - accumulated contaminant is lower than the contaminant in its usual state ( as , _
e.g. _ , products ) , because the contaminant material is an organic compound and its accumulation was formed under vacuum deposition process but with much slower rate than standard ones in laboratories , and the accumulated contaminant most likely consists of sparsely - structured molecules ( rather than the dense crystalline structure ) .
hereafter , we call this state as sparse state .
the same trend as this hypothesis is seen in the case of c - poly filter as described in the last paragraph of appendix [ subsec : calibration of fpaf ]
. generally , the refractive index [ @xmath74 and density [ @xmath75 of material in sparse state are smaller than those ( @xmath76 and @xmath77 ) in the usual state .
if the above hypothesis holds , the method _ i _ ) gives the refractive index of contaminant in sparse state .
on the other hand , in method _ ii _ ) , we identified the refractive index and density in usual state from among the possible candidates of contaminant ( see table [ tbl : contaminant ] ) .
hence , the difference in the refractive indexes in methods _ i _ ) and _ ii _ ) is caused by the different state of the contaminant .
next , we consider the effect of this difference on the calibration results . on the basis of equation ( [ eq : osc ] )
, the smaller refractive index gives the thicker contaminant .
meanwhile , the smaller density gives the longer attenuation length [ @xmath78 in equation ( [ eq : t ] ) .
hence , the x - ray transmission calculated with @xmath79 and @xmath80 would be close to the transmission with @xmath76 and @xmath77 .
we note that even if the actual state of contaminant is not sparse , _
i.e. _ even if the density of actual contaminant is the same as the usual state , the difference in refractive indexes ( 1.4 and 1.5 ) is only 7% , and then the difference in the estimated thickness of contaminant on the ccd is also 7% .
since the maximum thickness of contaminant is about 3600 ( see figure [ fig : contam on ccd ] ) , the maximum error in the estimate is about 250 .
this error is smaller than the accumulated thickness of contaminant on the thinner fpafs ( see figure [ fig : contam on fpaf ] ) , and is negligibly small against the metal thicknesses of the thicker filters . hence , we conclude that , whether the hypothesis holds or not , the difference in refractive indexes does not significantly affect the results of xrt calibration , and adopt @xmath81 and @xmath82 g @xmath73 for the calibration . the second one is the decay of oscillation amplitude .
we consider that this is caused by the inhomogeneous accumulation rate of the contaminant . in appendix
[ subsubsec : spatial distribution ] , we demonstrated the accumulation rate of contaminant at the ccd center is larger than at the edges of the ccd .
this effect might appear in the 512 @xmath83 512 pixels data which is used to make the g - band intensity plot ( figure [ fig : g - band 2nd bake ] ) .
the difference of accumulation rate will result in an asynchronous phase in the intensity oscillation across the 512 @xmath83 512 pixels .
the g - band intensity plot is the average of such different oscillations which are sine curves with different phases .
because the phase difference becomes larger and larger as time passes , the oscillation amplitude of the g - band intensity enhancement decays . on the basis of equation ( [ eq : osc ] ) ,
the thickness of contaminant on the ccd is derived from only the refractive index of the contaminant and the period of the intensity oscillation .
hence , we expect that the refractive index of @xmath1 1.5 and the oscillation period in figure [ fig : g - band 2nd bake ] give the average thickness of contaminant on the ccd area of 512 @xmath83 512 pixels , which is reliable enough for the xrt calibration .
when we estimated the thickness of the contaminant accumulated on the ccd with the visible light intensity profile ( appendix [ subsubsec : measurement of contam thickness on ccd ] ) , we assumed a constant accumulation rate for the contaminant between two successive bakeouts .
however , actually , we see rapid accumulation of contaminant right after the ccd bakeout while for the rest of each period the rate is almost constant ( see right panel of figure [ fig:6th bakeout ] ) .
nevertheless , the assumption of constant accumulation for the entire period between two successive bakeouts is a valid one for the calculation of the xrt effective area , because the contaminants accumulated on the fpafs are much thicker than the difference between the actual and estimated thickness with the above assumption , _ i.e. _ such difference is negligible . in appendix [ subsec : contam analysis with x - ray ] , we calibrated the contaminant thickness on the fpafs with the observed x - ray data and dem in the quiet sun .
when we performed the calibrations described in this article , the dem model in the quiet sun derived by was the latest result available .
hence , we adopted their dem model for our analysis .
derived the dem of quiet sun from a data set taken with _
soho_/eit , _
soho_/cds , and _ trace _ on 1 may 1998 , while more recently analyzed the dem with 45 data sets observed with _ hinode_/eis in the period from january to april 2007 .
the profiles of dems studied in these two papers are very similar up to at least @xmath84 k in all cases .
we confirmed that the calibrated contaminant thicknesses on the fpafs with the dems from and are consistent with each other .
we found that there is remarkable similarity between the observed spatial distribution of the contaminant across the ccd and the simulated result by for the case where the contaminant was assumed to come from the direction in front of the ccd ( see figure [ fig : contam rate map ] ) .
meanwhile , we found that the contamination is triggered by the operational heater which warms the rear end of telescope tube up to about @xmath85c .
on the basis of these circumstantial evidences , we suspect that the contaminant most likely originates from somewhere in front of the ccd , in the telescope tube .
as we mentioned in appendix [ subsec : calibration of fpaf ] , we consider that the calibration of thick filters ( med - be , thick - al and thick - be filters ) with ground - based test data is not ideal , though the difference between calibrated and actual x - ray transmission of them should be small . for further calibration of thick filters ,
we need to take enough data sets where xrt observes intense x - ray sources , _
e.g. _ , active regions and flares , with thick filters .
calibration with on - orbit observation data is our future work .
also , in order to supplement our calibration , we plan to perform the following cross - calibrations : _ i _ ) between _ hinode_/xrt and _ hinode_/eis , and _ ii _ ) between xrt and _ goes13_/sxi , which is the grazing incidence x - ray telescope like xrt .
the authors thank members of the xrt team for useful discussions and comments .
we acknowledge d. brooks and h. warren for variable information on quiet sun and active region dems .
_ hinode _ is a japanese mission developed and launched by isas / jaxa , collaborating with naoj as a domestic partner , nasa and stfc ( uk ) as international partners .
scientific operation of the _ hinode _ mission is conducted by the _ hinode
_ science team organized at isas / jaxa .
this team mainly consists of scientists from institutes in the partner countries .
support for the post - launch operation is provided by jaxa and naoj ( japan ) , stfc ( uk ) , nasa , esa , and nsc ( norway ) .
chianti is a collaborative project involving the nrl ( usa ) , the universities of florence ( italy ) and cambridge ( uk ) , and george mason university ( usa ) .
we wish to express our sincere gratitude to the late takeo kosugi , former project manager of _ hinode _ ( _ solar - b _ ) at isas , who passed away in november 2006 . without his leadership in the development of _ hinode
_ , this mission would have never been realized .
we express our sincere gratitude to those who participated in , or supported , xrt end - to - end calibration measurement at msfc xrcf that was carried out for two weeks in may - june 2005 .
first of all , we are very much indebted to an extremely - talented team of people at xrcf led by c. reily and j. mccracken , including r. siler , e. wright , j. carpenter , j. keegley and g. zirnstein , and also m. baker , h. haight , b. hale , t. hill , b. hoghe , d. javins , j. norwood , h. rutledge , g. st . john , j. tucker , and d. watson . without their continuous and enormous support including preparation of ln2-cooled cold plate facing xrt radiators in the cryogenic chamber of xrcf , from early preparation phase of the xrcf experiment and throughout the experiment period
, the filter calibration at xrcf would have been totally impossible .
we are also grateful to people from sao who supported the xrcf measurement ; g. austin , w. podgorski , e. dennis , j. chappel , d. caldwell , w. martell , m. harris , m. cosmo , d. weaver , s. park and t. kent who participated in the experiment , and p. cheimets , j. bookbinder , j. boczenowski .
also a. sabbag of naval research laboratory is appreciated for his support during pre - shipment instrument check performed at sao .
k. kumagai and m. tamura of naoj , k. yaji of rikkyo u. , and k. kobayashi of msfc are greatly appreciated for preparing for , and participating in , the xrcf experiment . also , people from msfc , l. hill , j. owens , b. cobb , d. coleman , r. jayroe , t. perrin , d. schultz , a. sterling and c. talley are sincerely acknowledged for their various support throughout our stay in huntsville and also for our transportation to and from xrcf .
finally , the authors grateful to the referees and editor for their detailed and careful reviews . | the _ x - ray telescope _ ( xrt ) onboard the _ hinode _ satellite is an x - ray imager that observes the solar corona with unprecedentedly high angular resolution ( consistent with its 1@xmath0 pixel size ) .
xrt has nine x - ray analysis filters with different temperature responses .
one of the most significant scientific features of this telescope is its capability of diagnosing coronal temperatures from less than 1 mk to more than 10 mk , which has never been accomplished before . to make full use of this capability ,
accurate calibration of the coronal temperature response of xrt is indispensable and is presented in this article .
the effect of on - orbit contamination is also taken into account in the calibration .
on the basis of our calibration results , we review the coronal - temperature - diagnostic capability of xrt . |
classical or recurrent novae are typically characterized by a rapid optical increase of 6 magnitudes or more followed by a decline to quiescence over the next @xmath7 days @xcite .
they originate from an accreting binary system consisting of a white dwarf ( wd ) and a mass - losing late - type companion star .
novae are triggered by thermonuclear runaways ( tnr ) lasting @xmath8 s at the bottom of the accreted mass layer on the wd surface @xcite .
the tnr blows off the outer layer of the accumulated mass and causes an optically thick wind expanding up to @xmath9 .
it produces bright blackbody emission ( @xmath10 erg s@xmath5 , comparable to the eddington luminosity of a @xmath11 object ) at optical bands .
this optical nova phase lasts for @xmath12 days @xcite . at the same time
a blast wave , caused by a nova explosion in a dense circumstellar medium , sometimes produces shock - induced optically - thin hard x - ray emission lasting @xmath13 days , as observed in rs ophiuchi @xcite and v407 cyg @xcite , for example . after the wind stops
, the photosphere shrinks down to the wd surface ( @xmath14 km ) , and the blackbody temperature increases to @xmath8 ev , meaning the emission is in the soft x - ray energy range . this transient phase with soft x - ray emission
is called the super - soft source phase ( sss phase ) and it lasts about @xmath15 days @xcite .
when the nuclear burning stops , the sss phase ends .
novae are classified into speed classes by the decay time scale of their optical light curves @xcite .
faster novae show earlier turn - ons and shorter durations of the sss phase .
for example , the fastest nova , u sco , showed a turn - on of the sss phase at 10 days and had a duration of about 25 days @xcite .
in general , the evolution of classical / recurrent novae has been established , except for the early phase . at the time of the tnr , the very early and short emission ( a few hours )
is predicted to appear in the uv to soft x - ray range , called `` fireball phase '' @xcite .
however , no such signal has yet been observed due to the difficulty in detecting the abrupt short phenomenon appearing in these energy range .
monitor of all - sky x - ray image ( maxi ; * ? ? ?
* ) is an all - sky x - ray monitor , which is operated on the japanese experiment module , the exposed facility ( jem - ef ) on the international space station ( iss ) .
maxi carries two types of x - ray cameras : gas slit camera ( gsc ; * ? ? ?
* ; * ? ? ?
* ) and solid - state slit camera ( ssc ; * ? ? ?
* ; * ? ? ?
gsc and ssc have wide fields of view ( fovs ) of @xmath16 and @xmath17 , respectively , and they scan almost all of the sky every @xmath18 min utilizing the attitude rotation coupled with the iss orbital motion ( see fig . 1 in * ? ? ?
gsc covers the @xmath19 kev band using gas proportional counters , while ssc covers the @xmath20 kev band with the x - ray ccds .
maxi started its operation in orbit in 2009 august .
the maxi transient alert system @xcite was triggered on 2011 november 11 at 05:05:59 ut ( @xmath21 ) by a new bright soft x - ray source near the small magellanic cloud ( smc ; fig . [ fig : maxi_finding_chart]a ) .
: http://maxi.riken.jp/pipermail/new-transient/2011-november/000038.html ] we analyzed the data and reported the source position through an astronomer s telegram ( atel ; * ?
* ) and the grb coordinates network ( gcn ; * ? ? ?
* ) . at 0.44 days after @xmath22 ,
_ swift _ x - ray telescope ( xrt ; * ? ? ?
* ; * ? ? ?
* ) began follow - up observations @xcite with a tiling mode to cover the maxi error circle ( fig .
[ fig : maxi_finding_chart]d ; * ? ? ? * ) .
an uncataloged x - ray source was found within the maxi gsc error ellipse ( fig .
[ fig : maxi_finding_chart]d ; * ? ? ?
* ; * ? ? ?
* ) . within the _
swift _ xrt error circle , a single optical source is cataloged in usno - a2.0 , which was also reported as a source with a near - infrared excess ( i d 115 in * ? ? ?
the position is consistent with that of an optical counterpart observed by _
swift _ ultraviolet / optical telescope ( uvot ; * ? ? ?
* ) , @xmath23 , @xmath24 , with an estimated uncertainty of 0.5 arcsec ( 90% confidence ; fig .
[ fig : maxi_finding_chart]e ; * ? ? ?
the _ swift _ xrt spectra obtained after @xmath25 d were reported to be similar to the sss phase of novae @xcite .
further follow - up observations by _ swift _ and ground - based optical observations confirmed that this source is a binary system consisting of a wd and a be star at the distance of the smc ( @xmath26 kpc ; * ? ? ?
* ; * ? ? ?
@xcite analyzed the spectrum of the gsc scan at @xmath27 s , using the on - demand data products provided by the maxi team , and reported that the luminosity was @xmath28 erg s@xmath5 in the @xmath29 kev band ; this is one order of magnitude brighter than the eddington luminosity of a solar mass object .
to explain the huge outburst luminosity , @xcite proposed a model of the interaction of the ejected nova shell with the be star wind in which the wd is embedded .
here we present observations of maxi j0158@xmath0744 by maxi , _ swift _ and the small and moderate aperture research telescope system ( smarts ) in section [ sec : observation ] . the analysis and results of the maxi gsc and ssc data are described in sections [ sec : maxi / gsc ] and [ sec : maxi / ssc ] with the detailed spectral analysis for the third scan of maxi shown in sections [ sec : maxi / ssc spec m+2 ] and [ sec : maxi / ssc spec m+2 ph ] .
the upper flux limits before and after the maxi detection are given in section [ subsec : upper limit gsc ] while the analysis and results for the _ swift _ and smarts follow - up observations are presented in section [ sec : follow - up ] .
the historical observations of this source are described in section [ sec : history ] . in section [ sec : discussion ] , we interpret the results obtained by maxi , _ swift _ and smarts and discuss the emission mechanism of the very luminous soft x - ray transient detected by maxi . finally , we summarize this paper in section [ sec : summary ] .
maxi j0158@xmath0744 @xcite was first detected during a maxi gsc scan ( fig .
[ fig : maxi_finding_chart]b ) , centered at @xmath30 s within the @xmath31 s triangular transit response ( see fig . 9 in * ? ? ?
it was subsequently detected twice by the maxi ssc in scans at @xmath32 s and @xmath33 s ( fig .
[ fig : maxi_finding_chart]c ) .
hereafter , we designate the maxi scans by the mid - time of the scan transit , referred to @xmath22 .
subsequent gsc scans to date ( up to 2013 july 8) have failed to detect the source .
in addition , the source had not been detected in prior gsc scans , since maxi observations started on 2009 august 14 up to the previous scan at @xmath34 s. the maxi observations around @xmath22 are summarized in table [ table : maxi obs ] .
lccccc scan - id & scan time(start
end)(ut ) & @xmath35(s)@xmath36 & @xmath37(s)@xmath38 & detector & flux@xmath39 + m@xmath40@xmath41 & 2011 - 11 - 11 03:33:22 03:34:17@xmath42 & 55 & @xmath34 & gsc - h & @xmath43@xmath44 + m+0 & 2011 - 11 - 11 05:05:39@xmath45 05:06:34 & 55 & @xmath27 & gsc - h & @xmath46@xmath47 + m+1 & 2011 - 11 - 11 05:09:13 05:10:04 & 51 & @xmath32 & ssc - h & @xmath48@xmath47 + m+2 & 2011 - 11 - 11 05:27:09 05:28:00@xmath49 & 51 & @xmath33 & ssc - z & @xmath50@xmath51 + m+3 & 2011 - 11 - 11 06:37:56@xmath52 06:38:51 & 55 & @xmath53 & gsc - h & @xmath54@xmath44 + _ swift _ xrt performed follow - up observations from 0.44 days after @xmath22 ( see table 1 of * ? ? ?
_ swift _ uvot also observed the optical counterpart in image mode .
_ swift _ uvot grism observations were performed on 2011 november 19 ( + 8.23 days after @xmath22 ) and 2012 september 30 ( 324 days ) . a ground - based optical spectrum , with relatively high resolution ,
was obtained on 2012 may 19 ( 190 days after the @xmath22 ) with the rc spectrograph on the smarts / ctio 1.5 m telescope ; this is a long slit spectrograph oriented east - west @xcite . we used a 1 arcsec slit width and a loral 1k ccd for the detector .
on 2011 november 11 , the position of maxi j0158@xmath0744 was visible by three cameras of gsc - h ( gsc_2 , gsc_7 and gsc_8 ; * ? ? ? * ; * ? ? ? * ) .
one of these cameras ( gsc_2 ) was operated at the nominal high voltage ( @xmath55v ) , while the other two ( gsc_7 and gsc_8 ) were operated at the reduced voltage ( @xmath56v ) .
we analyzed the gsc event data version 1.0 or later , which included the data taken by cameras operated at the nominal and reduced voltages . in these versions ,
the position and energy responses of the anodes # 1 and # 2 were significantly improved from the previous versions ( 0.x ) .
we therefore used events taken from all anodes . to make light curves within the interval of the scan - id m+0 ( table [ table : maxi obs ] )
, we followed the method shown in @xcite . here
we selected events within 5 mm of the position coincident with this source along the anode wires , which corresponds to about 2@xmath57 on the sky .
the obtained light - curve data in energy bands of @xmath58 , @xmath59 , @xmath60 , and @xmath61 kev were fitted with a model consisting of a triangular transit response curve for a point source with a constant flux and a constant background .
the light curves are consistent with the model at the 90% confidence level , meaning that there was no significant variation of the source flux during that scan . for the spectral analysis , we removed the gsc_8 data due to its poor response in the soft energy band .
we selected a concentric circle and annulus centered at the target as the source and background regions , respectively .
the radius of the source circle was set to 1.8 deg .
the inner and outer radii of the background annulus were set to 2.2 and 3.5 deg , respectively . in both these regions , we excluded a circular region with a radius of 1.5@xmath57 centered at a near - by bright x - ray source , smc x-1 ( fig .
[ fig : maxi_finding_chart]b ) .
the spectrum and response files were made by the method described in @xcite .
the energy spectra obtained by the gsc scan - id m+0 are shown in fig .
[ fig : maxi spec ] ( left ) .
we rebinned the data with a minimum of 1 count per energy bin and applied cash statistics @xcite in the fit .
we used xspec v12.7.1 for the spectral analysis .
since the location of this source is near the smc , the interstellar absorption by the total galactic h i column density towards this source , @xmath62 , and optical extinction @xmath63 are expected to be small .
thus , we decided to fix them for the following x - ray and optical spectral analysis .
two plausible different @xmath62 values are obtained from the heasarc web site : @xmath64 @xmath65 by using lab map @xcite and @xmath66 @xmath65 by using dl map @xcite . the corresponding optical extinctions @xmath63 are derived to be 0.28 and 0.084 mag , respectively , by using the relation with the h i column density @xcite . on the other hand ,
the map of dust infrared emission @xcite suggests @xmath67 , which is closer to that from the dl map .
therefore , we decided to use the latter @xmath62 value , @xmath66 @xmath65 , for the interstellar absorption . in the following
, unabsorbed flux is corrected only for the interstellar absorption .
we fit the gsc x - ray spectrum with absorbed blackbody , power - law , thermal bremsstrahlung and mekal @xcite models from @xmath68 kev with @xmath62 fixed to @xmath66 @xmath65 ; the results are shown in table [ table : maxi spectrum ] .
the spectrum is statistically consistent with all the models . adopting the @xmath62 value of lab map increases the unabsorbed flux by 2% from that using dl map .
however , the difference in the spectral parameters and unabsorbed flux are negligibly small , when they are compared with the statistical uncertainty .
lcccccccc + pl & @xmath69 & & & & & @xmath70 & @xmath71 & 43.4 ( 60 ) + bb & & @xmath72 & @xmath73 & & & @xmath74 & @xmath75 & 51.8 ( 60 ) + tb & & @xmath76 & & @xmath77 & & @xmath78 & @xmath79 & 45.3 ( 60 ) + mekal & & @xmath80 & & @xmath81 & 0.1 ( fix ) & @xmath82 & @xmath83 & 44.6 ( 60 ) + mekal & & @xmath84 & & @xmath85 & @xmath86 & @xmath87 & @xmath88 & 44.2 ( 59 ) + + pl & @xmath89 & & & & & @xmath90 & @xmath91 & 20.5 ( 36 ) + bb & & @xmath92 & @xmath93 & & & @xmath48 & @xmath94 & 30.2 ( 36 ) + tb & & @xmath95 & & @xmath96 & & @xmath97 & @xmath98 & 22.5 ( 36 ) + mekal & & @xmath99 & & @xmath100 & 0.1 ( fix ) & @xmath101 & @xmath102 & 24.0 ( 36 ) + mekal & & @xmath103 & & @xmath104 & @xmath105(90% ) & @xmath106 & @xmath107 & 22.6 ( 35 ) + + pl & @xmath108 & & & & & @xmath109 & @xmath110 & 83.4 ( 69 ) + bb & & @xmath111 & @xmath112 & & & @xmath113 & @xmath114 & 84.2 ( 69 ) + tb & & @xmath115 & & @xmath116 & & @xmath117 & @xmath118 & 80.8 ( 69 ) + mekal & & @xmath119 & & @xmath120 & 0.1 ( fix ) & @xmath50 & @xmath121 & 80.8 ( 69 ) + mekal & & @xmath122 & & @xmath123 & @xmath124 & @xmath125 & @xmath126 & 79.4 ( 68 ) + after the first detection of maxi j0158@xmath0744 with the maxi gsc scan at @xmath27 s , maxi ssc detected this source twice as shown in table [ table : maxi obs ] . before @xmath27 s , maxi j0158@xmath0744 was below the ssc detection limit of @xmath127200 mcrab in each night scan . for the spectral analysis , we selected the source regions as shown in fig .
[ fig : maxi_finding_chart]c and reduced the ssc events in the same way as @xcite . while the second ssc detection at @xmath33 s was done at night in the orbit , the first ssc at @xmath32 s was done at the day - time . since the ssc day - time data were contaminated by the visible / infrared light from the sun @xcite , we have to be careful of the analysis of the @xmath32 s scan data .
we estimated the area suffering from the contamination based on the event distribution , which led to the decision that 63% of the source area was not suitable for the spectral analysis .
we thus used data from the remaining 37% area in the analysis .
the energy spectra obtained by the ssc scans at @xmath32 s and @xmath33 s are shown in fig .
[ fig : maxi spec ] ( middle , right ) . in the latter spectrum , since emission lines seemed to be present , we analyzed the ssc spectra with special care by following the method shown in `` low count spectra '' of the xspec wiki site .
to avoid losing information on emission lines due to the spectral binning , we did not bin the data and applied cash statistics @xcite to the fits .
we fit the source spectra with a model consisting of a source and a background component . here , the background model was analytically - described to approximate the actual background spectrum in @xmath128 kev .
it was constructed by averaging the two year ssc data weighted with geomagnetic cut - off - rigidity , because the background events are caused by charged particles and cosmic diffuse x - rays .
the details of the background spectrum are shown in @xcite .
we fit the spectra by absorbed blackbody , power - law , thermal bremsstrahlung and mekal @xcite models in @xmath128 kev with @xmath62 fixed to @xmath66 @xmath65 ( section [ sec : maxi / gsc ] ) and the results are shown in table [ table : maxi spectrum ] . for the mekal model we let
the abundance parameter ( hereafter , we abbreviate it to ` abund`.)}$ ] . here , @xmath129 = \log_{10}\left ( n_{\rm x } / n_{\rm h } \right)_{\rm source } - \log_{10}\left ( n_{\rm x } / n_{\rm h } \right)_{\rm sun}$ ] is a metallicity of a metal @xmath130 , where @xmath131 represents the number density of an element ( @xmath130 ) in a source ( @xmath132 ) . ]
be free or fixed to 0.1 ( a typical abundance of the smc ; * ? ? ?
for the first ssc spectrum ( @xmath32 s ) , the data are statistically consistent with all the models , while for the second , the free abundance mekal model is preferred . adopting the @xmath62 value of lab map increases the unabsorbed fluxes by up to 20% from those using dl map
however , differences in all the spectral parameters and unabsorbed fluxes are not significant ( less than 2.6 sigma level of the statistical uncertainty ) . as shown in table [
table : maxi spectrum ] , the initial x - ray outburst of maxi j0158@xmath0744 detected by maxi gsc and ssc was peaked at @xmath32 s. the peak luminosity was extraordinarily luminous , @xmath4 erg s@xmath5 in @xmath133 kev , which is two orders of magnitude larger than the eddington luminosity of a solar mass object . in the following subsections ,
we investigate the maxi ssc spectrum at @xmath33 s , where emission lines are apparent , with two scenarios : shock - induced emission and photospheric emission at the fireball phase ( see section [ sec : discussion ] ) .
to investigate the emission lines in the spectrum at @xmath33 s , we first fit the spectrum with models consisting of thermal bremsstrahlung continuum and gaussian lines , whose widths were fixed to be small against the detector energy resolution , 85 ev ( fwhm ) at 1.0 kev @xcite .
the best - fit parameters are summarized in table [ table : m+2 spectrum ] ( upper ) and the models are shown in fig .
[ fig : m002_various_fits ] ( abcd ) .
the results of the likelihood ratio tests @xcite in the last two rows of table [ table : m+2 spectrum ] ( upper ) indicate that the addition of the gaussian lines at the energies @xmath134 , @xmath135 and @xmath136 one by one improves the fits with a chance probability of 0.0044 , 0.084 and 0.085 , respectively .
the line at the energy @xmath134 is the most significant , and is inferred to be a resonance line of he - like neon ( 0.922 kev ) .
the other two lines are less significant than @xmath137 and no corresponding major lines exist at these energies . however , the line center energy @xmath135 may suggest a radiative recombination continuum of he - like neon ( 1.20 kev ) or lyman beta line of h - like neon ( 1.24 kev ) , and the @xmath136 may suggest a resonance line of he - like aluminum ( 1.60 kev ) or @xmath138 @xmath139 @xmath140 @xmath141 @xmath142 line of he - like magnesium ( 1.58 kev ) .
the neon emission line suggests that the initial bright outburst would have been produced by an optically - thin thermal emission mechanism , whose site was a region heated by the shock wave of a nova explosion as seen in some novae ( rs ophiuchi and v407 cyg ; * ? ? ?
* ; * ? ? ?
lcccc model & tb@xmath36 & tb + line@xmath38 & tb + 2 lines & tb + 3 lines + @xmath143@xmath144 ( kev ) & @xmath115 & @xmath145 & @xmath146 & @xmath147 + @xmath148@xmath41 ( @xmath149 @xmath150 ) & @xmath116 & @xmath151 & @xmath152 & @xmath153 + @xmath134@xmath154 ( kev ) & & @xmath155 & @xmath155 & @xmath155 + @xmath156@xmath157 ( kev ) & & @xmath158 & @xmath159 & @xmath160 + @xmath135@xmath161 ( kev ) & & & @xmath162 & @xmath162 + @xmath163 ( kev ) & & & @xmath164 & @xmath165 + @xmath136@xmath166 ( kev ) & & & & @xmath167 + @xmath168 ( kev ) & & & & @xmath169 + @xmath170-stat@xmath171(dof@xmath172 ) & 342.9(1723 ) & 332.1(1721 ) & 327.1(1719 ) & 322.2(1717 ) + @xmath173(@xmath174dof)@xmath175 & & 10.8(2 ) & 5.0(2 ) & 4.9(2 ) + @xmath176-value@xmath177 & & 0.0044 & 0.084 & 0.085 + model & bb@xmath178 & bb + line & bb + 2 lines & bb + 3 lines + @xmath143@xmath144 ( kev ) & @xmath111 & @xmath179 & @xmath180 & @xmath181 + @xmath182@xmath183 ( @xmath184 km ) & @xmath112 & @xmath185 & @xmath186 & @xmath187 + @xmath134@xmath154 ( kev ) & & @xmath155 & @xmath155 & @xmath155 + @xmath156@xmath157 ( kev ) & & @xmath188 & @xmath189 & @xmath190 + @xmath135@xmath161 ( kev ) & & & @xmath162 & @xmath162 + @xmath163 ( kev ) & & & @xmath191 & @xmath192 + @xmath136@xmath166 ( kev ) & & & & @xmath193 + @xmath168 ( kev ) & & & & @xmath194 + @xmath170-stat@xmath171(dof@xmath172 ) & 343.8(1723 ) & 326.7(1721 ) & 321.1(1719 ) & 318.2(1717 ) + @xmath173(@xmath174dof)@xmath175 & & 17.1(2 ) & 5.6(2 ) & 2.9(2 ) + @xmath176-value@xmath177 & & @xmath195 & 0.061 & 0.23 + [ cols="^,^ " , ] we next tried to fit the spectrum with more physically motivated models .
although it can be fit with an optically - thin thermal emission model ( mekal in xspec terminology ) with a temperature of @xmath196 kev ( table [ table : maxi spectrum ] ) , the model can not produce the observed strong he - like neon line . to reproduce the he - like neon ,
another optically - thin thermal component with a lower temperature ( about 0.1 kev ) is necessary .
thus , we examined a model consisting of two mekal models ( mekal@xmath197 + mekal@xmath198 ) , whose temperatures are @xmath199 kev in the lower component ( lt ) and @xmath196 kev in the higher component ( ht ) , respectively .
the best - fit result in table [ table : maxi spectrum ] shows that the ` abund ` of the mekal@xmath198 is consistent with that of the smc .
this conclusion , however , is not completely correct , because the fit included an energy range affected by the he - like neon line produced by the mekal@xmath197 component . to determine the ` abund ` of mekal@xmath198 , we fit the spectrum excluding the energy range @xmath200 kev , and then obtained an upper limit for the ` abund ` of 0.25 ( 90% confidence limit ) , which is consistent with that of the smc @xcite .
we thus decided to fix the ` abund ` of the mekal@xmath198 component to 0.1 .
for the mekal@xmath197 component , the observed strong he - like neon line suggests a large abundance for neon .
we postulate that the mekal@xmath197 component was produced in a reverse shocked region whose material was ejecta from the nova explosion ( see section [ sec : discussion ] ) .
to estimate the ` abund`s of the mekal@xmath197 component , we fit the spectrum with the mekal@xmath197 + mekal@xmath198 model assuming six combinations of the ` abund`s of neon and the other elements for mekal@xmath197 as shown in table [ table : vmekal ] . here , at the first step in the spectral fit , we let the temperature and emission measure of the mekal@xmath198 component freely vary .
when the 1@xmath201 error range of the temperature of the mekal@xmath197 was not constrained to less than 0.3 kev in the first step ( the first three cases of table [ table : vmekal ] ) , we fixed the temperature of mekal@xmath198 and then the emission measure of mekal@xmath198 to the best - fit values .
these best fit values were obtained by fitting the same spectrum , excluding the energy range @xmath200 kev , with a single mekal component and the ` abund ` fixed to 0.1 .
for the first case of table [ table : vmekal ] , the 1@xmath201 error range of the mekal@xmath197 could not be constrained to be less than 0.3 kev , even after both the temperature and emission measure of mekal@xmath198 were fixed .
figure [ fig : m002_various_fits ] ( efgh ) presents the difference in these spectral fits with respect to the neon ` abund`s of the mekal@xmath197 component . as a result ,
the neon ` abund ` of the mekal@xmath197 was suggested to be much higher than that of the smc ( table [ table : vmekal ] ) , and it indicates that the mekal@xmath197 component originates in ejecta from the nova . the unabsorbed flux in @xmath128 kev , assuming the mekal@xmath197 + mekal@xmath198 model with parameters shown in the fifth row of table [ table : vmekal ] ( the best - fit model ) , is @xmath202 erg s@xmath5 @xmath65 .
ll|ccccc & @xmath143(lt@xmath41 ) & @xmath148(lt ) & @xmath143(ht@xmath154 ) & @xmath148(ht ) & c - stat@xmath157 + ne & others & ( kev ) & ( @xmath149 @xmath150 ) & ( kev ) & ( @xmath149 @xmath150 ) & ( dof@xmath161 ) + @xmath203 & @xmath203 & 0.13@xmath166(unconstrained)@xmath171 & 11.5@xmath166(0.0 @xmath0 194)@xmath172 & @xmath204 ( fix)@xmath175 & @xmath205 ( fix)@xmath175 & 81.6 ( 69 ) + @xmath206 & @xmath206 & @xmath207 & @xmath208 & @xmath204 ( fix)@xmath175 & @xmath205 ( fix)@xmath175 & 80.6 ( 69 ) + @xmath209 & @xmath209 & @xmath210 & @xmath211 & @xmath204 ( fix)@xmath175 & @xmath205 ( fix)@xmath175 & 80.5 ( 69 ) + @xmath206 & @xmath203 & @xmath212 & @xmath213 & @xmath214 & @xmath215 & 74.8 ( 67 ) + @xmath216@xmath209 & @xmath203 & 0.13@xmath166(@xmath217)@xmath172 & @xmath218 & @xmath219 & @xmath220 & 73.8 ( 67 ) + @xmath209 & @xmath206 & @xmath212 & @xmath221 & @xmath222 & @xmath223 & 74.6 ( 67 ) + the initial bright outburst detected by maxi may also be explained by photospheric emission at the ignition phase of a nova explosion , the so - called fireball phase ( see section [ sec : discussion ] ) . in this scenario ,
the main continuum component in the spectrum of maxi ssc at @xmath33 s is blackbody emission , while the emission lines come from the optically - thin region surrounding the photosphere ( see fig .
[ fig : nova evolution ] in section [ sec : discussion ] ) .
thus , we fit the spectrum with models consisting of blackbody continuum and gaussian lines .
the results of the fit are shown in table [ table : m+2 spectrum ] ( lower ) .
the addition of the gaussian line at the energy @xmath134 significantly improves the fits with a chance probability of @xmath195 , while the other two lines are detected at @xmath224 .
the identifications of these lines are the same as in section [ sec : maxi / ssc spec m+2 ] .
the detection of the neon emission line suggests that the spectrum contains an optically - thin thermal emission component .
therefore , this spectrum could be explained by a composite model of a blackbody and a mekal with a temperature below 0.3 kev , and an exceptionally large neon abundance , similar to the two mekal models in section [ sec : maxi / ssc spec m+2 ] . when the ` abund`s of neon and the other elements are set to @xmath225 and 0.1 , respectively ,
the resultant best - fit spectral parameters are as follows : the temperature and emission measure of the mekal component are 0.14(@xmath226 ) kev ( see footnotes @xmath227 , @xmath228 and @xmath229 of table [ table : vmekal ] ) and @xmath230 @xmath150 , respectively .
the temperature and radius of the blackbody component are @xmath231 kev and @xmath232 km , respectively . in the scans at @xmath34
s and @xmath53 s ( table [ table : maxi obs ] ) , maxi j0158@xmath0744 was not detected by maxi gsc . to calculate the upper limits on these fluxes , we assumed the best - fit mekal@xmath197 + mekal@xmath198 model obtained by the maxi ssc scan at @xmath33 s ( section [ sec : maxi / ssc spec m+2 ] ) and the best - fit blackbody model ( the scan - id m+1 in table [ table : maxi spectrum ] ) . in the former model , the ` abund ` of the mekal@xmath198 and the mekal@xmath197
was fixed to 0.1 except for the neon ` abund ` in the mekal@xmath197 fixed to 10 ( table [ table : vmekal ] , the fifth row ) .
the 90% confidence - level upper limits on the unabsorbed flux in @xmath128 kev for these scans are @xmath233 and @xmath234 erg s@xmath5 @xmath65 in the two mekal model , and @xmath235 and @xmath54 erg s@xmath5 @xmath65 in the blackbody model .
in addition , in all five scans between the scan at @xmath53 s ( + 0.064 d ) and the start of the _ swift _ xrt follow - up ( i.e. at + 0.128 , + 0.192 , + 0.256 , + 0.320 , and + 0.385 days ) , maxi j0158@xmath0744 was not detected by maxi gsc .
the 90% confidence - level upper limit on the unabsorbed flux in the [email protected] kev band for this period was @xmath236 erg s@xmath5 @xmath65 , assuming the former model and @xmath237 erg s@xmath5 @xmath65 for the latter model .
we analyzed the same _ swift _ xrt archival data as listed in table 1 of @xcite , using _
swift _ software version 3.9 , released as part of heasoft 6.12 .
we extracted the source events from a circle with optimal radii ( @xmath238 ) and the background from an offset circular region of radius 142@xmath239 .
for the data on day 0.54 , we excluded events from the inner 5@xmath239 of the psf to avoid pile - up . in the spectral fit , we used the rmfs ( redistribution matrix files ) of swxpc0to12s6_20010101v013.rmf in pc mode and swxwt0to2s6_20010101v014.rmf in wt mode .
the arfs ( ancillary response files ) were generated by using the commands xrtexpomap ( to create the exposure maps ) and xrtmkarf .
we fit the _ swift _ xrt spectra with absorbed blackbody or mekal models with the intrinsic column @xmath62 allowed to vary .
these models include two absorption components : the interstellar absorption @xmath62 fixed at @xmath66 @xmath65 ( section [ sec : maxi / gsc ] ) and intrinsic absorption . in the mekal model ,
the ` abund ` was fixed to 0.1 , a typical ` abund ` of the smc ( * ? ? ?
* see also section [ sec : maxi / ssc spec m+2 ] ) .
the unabsorbed flux obtained by the blackbody fits are shown in figure [ fig : lc ] .
we also analyzed the _
uvot data obtained at the same time as _ swift _ xrt , using the _ swift _ software version 3.9 , released as part of heasoft 6.12 .
the image data of each filter , from each observation sequence , i.e. , with a given observation i d , were summed using uvotimsum .
however , for images taken within 2 days of the outburst ( @xmath240 , @xmath241 and @xmath242 bands ) individual exposures were long enough that summing was not necessary .
photometry of the source in individual sequences was derived via uvotmaghist , using an extraction region of radius 5@xmath239 and a suitable background region .
magnitudes are based on the uvot photometric system @xcite .
xspec compatible spectral files for the source were created using the same region with uvot2pha .
the @xmath241 band light curve is shown in figure [ fig : lc ] .
we calculated the absolute magnitude of the enhanced emission after extinction correction , where @xmath67 mag ( section [ sec : maxi / gsc ] ) , @xmath243 , @xmath244 @xcite , and the smc distance of @xmath26 kpc were assumed . here , we subtracted the flux in the plateau phase ( @xmath245 mag , average of @xmath241-band magnitudes from 11.65 days to 27.86 days . ) . by fitting it with a linear function
, we obtained the absolute magnitude of @xmath246 mag at 0.44 day and the speed class indicator parameter defined by the time to decline 2 mag from maximum @xcite , @xmath247 ( @xmath248 ) days .
this classifies the event as a `` very fast nova '' @xcite , assuming that the optical enhancement was due to the photospheric emission as in usual novae . to investigate the optical enhanced emission , we made a difference spectrum from the _ swift _ uvot photometry over the six filter bandpasses between day 1.5 and 149 , and we fit it with a blackbody model with fixed interstellar extinction ( section [ sec : maxi / gsc ] ) .
the blackbody temperature of @xmath249 k and the radius of @xmath250 cm were obtained at the best fit .
we also fit the spectrum from day 149 with the blackbody model , obtaining a temperature of @xmath251 k and a radius of @xmath252 cm .
the extrapolation of the best - fit blackbody spectrum towards the uv region is consistent with the uv flux obtained by galaxy evolution explorer ( galex ; * ? ? ?
* ) during the pre - outburst phase @xcite .
in addition , @xcite reported that the i - band flux returned to the pre - outburst level @xmath253 days after @xmath22 .
so , we can assume that the flux at day 149 contains only emission from the binary companion star .
the obtained temperature and radius are consistent with those of a b - type star , as shown in @xcite .
_ swift _ uvot grism spectra are close to that of an early b - type star .
no clear emission lines can be identified above the noise , as shown in @xcite .
we obtained three 200-s spectra of the optical counterpart of maxi j0158@xmath0744 in order to filter for cosmic rays .
we combine the three images , and extract the spectrum by fitting a gaussian in the spatial direction at each pixel .
wavelength calibration is accomplished by fitting a 3@xmath254 to 6@xmath255 order polynomial to the calibration lamp line positions .
the optical spectrum covers nearly the entire optical band ( @xmath256 ) at @xmath257 resolution .
there are clear emission lines of @xmath258 and @xmath259 , with equivalent widths of @xmath260 @xmath261 and @xmath262 @xmath261 , respectively .
we could not find any other significant emission or absorption lines above the noise level .
the smarts spectrum matches the ntt spectrum of @xcite , albeit with worse signal to noise . to investigate the activity of maxi j0158@xmath0744 before the discovery
, we searched for previous x - ray observations of the area including the target position .
this region was observed by the _
rosat _ all - sky survey , _ xmm - newton _ slew survey and maxi gsc .
the source was undetected in all these observations .
we calculated the upper limits on the unabsorbed fluxes in an energy range of @xmath128 kev , assuming the best - fit mekal@xmath197 + mekal@xmath198 model ( table [ table : vmekal ] , the fifth row ; outburst model ) and a typical spectrum in the sss phase observed by _ swift _
xrt , an absorbed blackbody with a temperature of 0.1 kev ( sss model ; * ? ? ?
the _ rosat _ all - sky survey covered this field , with an exposure of 775 s in total between 1990 september 22 and december 3 .
these data provide a pspc count rate upper limit of @xmath263 counts s@xmath5 ( 90% confidence limit ) over @xmath264 kev ; corresponding to @xmath265 and @xmath266 ergs @xmath65 s@xmath5 ( @xmath128 kev ) , assuming outburst model and sss model , respectively ( table [ table : historical x - ray ] ) . _
xmm - newton _ slewed over the source three times on 2006 november , 2007 october , and 2009 november .
we obtained epic pn count rate upper limits of 0.35 , 1.5 and 0.50 counts s@xmath5 ( @xmath137 level ) in @xmath267 kev , respectively .
the corresponding unabsorbed fluxes ( @xmath128 kev ) are shown in table [ table : historical x - ray ] .
we also analyzed the maxi gsc image in the 4@xmath010 kev band integrated for 7 months from 2009
september 1 to 2010 march 31 .
applying the same analysis procedure as used by @xcite , we obtain a 90% confidence - level upper limit of 0.10 mcrab .
it corresponds to an unabsorbed flux of @xmath268 ergs @xmath65 s@xmath5 ( @xmath128 kev ) , assuming outburst model .
ccccc date & telescope & exp.(s)@xmath36 & flux(outburst)@xmath38 & flux(sss)@xmath38 + 1990 - 09 - 22 1990 - 12 - 03 & _ rosat _ & 775 & @xmath265 & @xmath266 + 2006 - 11 - 01 10:03:35 & _ xmm - newton _ & 8.9 & @xmath269 & @xmath270 + 2007 - 10 - 28 11:57:17 & _ xmm - newton _ & 2.1 & @xmath271 & @xmath272 + 2009 - 11 - 30 23:41:46 & _ xmm - newton _ & 6.1 & @xmath273 & @xmath274 +
the x - ray transient maxi j0158@xmath0744 is characterized by ( 1 ) a soft x - ray spectrum with most of the x - ray photons being detected below 4 kev ( fig .
[ fig : maxi_finding_chart ] and fig .
[ fig : maxi spec ] ) , ( 2 ) a short duration ( between @xmath275 s and @xmath276 s ; table [ table : maxi obs ] ) , ( 3 ) a very rapid rise time ( @xmath277 s ; table [ table : maxi obs ] ) , and ( 4 ) a huge peak luminosity of @xmath4 erg s@xmath5 in the @xmath133 kev band recorded at the second maxi scan .
the unusually soft spectrum of the outburst is beyond astronomers expectations , because most short - lived luminous transient x - ray sources are hard x - ray emitters , e.g. gamma - ray bursts @xcite , soft gamma repeaters @xcite , super - giant fast x - ray transients @xcite , and x - ray bursts @xcite .
the discovery of maxi j0158@xmath0744 , together with supernova shock breakouts @xcite , suggests that the wide - field monitoring experiments in soft x - rays ( @xmath278 kev ) will open new discovery fields .
@xcite reported the optical spectra obtained by saao and eso , showing that the source is a be star .
in addition , they showed that the radial velocity of this source is consistent with the smc , which strongly supports that this source is located in the smc .
due to the similarity between the _
swift _ x - ray spectra and the sss phase of novae , they concluded that this source is a binary system consisting of a wd and a be star .
we also analyzed the sed of the optical counterpart and found that it is consistent with that of a b - type star . the optical spectrum taken by smarts showed clear emission lines of @xmath258 and @xmath259 , confirming the conclusion of @xcite .
we fit the _ swift _ xrt spectra with absorbed blackbody or mekal models .
neither model was strongly preferred from the statistics .
however , the evolution of the temperature of the mekal model shows an unexplained decrease at 3@xmath07 days , while the temperature and radius of the blackbody fits can be understood as being due to the shrinking photosphere . here ,
the radius decreased from @xmath279 km to @xmath8 km , while the temperature increased from @xmath280 kev to @xmath199 kev . therefore , we conclude that the spectra were basically blackbody - like , and hence can be identified as a super - soft x - ray phase of a nova .
this conclusion is the same as @xcite .
fits using wd atmosphere models @xcite would allow further insights , although a trial with the tbingen wd model failed to improve the fits , due to the computational upper limit for the temperature @xcite .
probably , more sophisticated spectral models like atmosphere models would improve the fits .
@xcite reported that adding a broad emission line at @xmath1270.7 kev and an absorption edge at @xmath281 kev improves the fit for the spectrum at @xmath282 d significantly , which also supports our interpretation that the early x - ray emission is due to the sss phase of novae .
the sss phase spectra from the first _ swift _ xrt follow - up observation at @xmath283 d and the simultaneous optical enhanced emission obtained by _ swift _
uvot are unusual in the evolution of standard novae .
nonetheless , if the optical enhanced emission is the photospheric emission of nova ejecta as in standard novae , the speed class indicator parameter of @xmath284 means that it is the second fastest nova next to u sco ( @xmath285 days ; * ? ? ? * ) and an equal record to v838 her ( @xmath286 days ; * ? ? ?
the peak absolute magnitude ( @xmath246 ) at 0.44 day in the first _ swift _ uvot observation is four magnitudes fainter than those of typical novae ( @xmath287 ) , expected from the mmrd ( maximum magnitude - rate of decline ) relation @xcite .
if this enhanced optical emission was photospheric emission from the nova ejecta , it suggests a low ejecta mass in this nova explosion . in the subsequent discussion below , we will show that the enhanced optical emission is not photospheric emission , however , the conclusion of the low ejecta mass remains correct .
optically - thin thermal x - ray emission of novae is usually explained by a shock heating mechanism at a blast wave produced by the nova explosion .
the recurrent nova , rs ophiuchi , in 2006 exhibited the most luminous optically - thin thermal x - ray emission ( @xmath288 erg s@xmath5 ) among novae that has ever been observed @xcite .
the luminosity of the x - ray outburst of maxi j0158@xmath0744 was , surprisingly , four orders of magnitude larger than this .
@xcite explained the luminosity of maxi j0158@xmath0744 by the shock heating mechanism , however , their explanation has difficulties as follows .
they tried to explain the observed luminosity of @xmath289 erg s@xmath5 at the time of the first gsc scan ( scan - id m+0 ; table [ table : maxi obs ] ) using equation ( 1 ) of @xcite , where the radius of the shock wave @xmath290 cm is assumed .
however , in order for the shock wave to expand to this radius within @xmath291 s ( table [ table : maxi obs ] ) , the shock wave velocity must be exceptionally large ( @xmath292 km s@xmath5 ) for novae .
in addition , if the velocity of the shock wave was such a large value , the temperature of the plasma thermalized by the shock wave becomes very high ( @xmath293 mev , where @xmath294 is the proton mass , and @xmath295 is the mean molecular weight for a typical smc abundance ( ` abund ` @xmath296 ; * ? ? ?
* ) . here ,
@xmath297 , @xmath298 , and @xmath299@xcite , where @xmath300 , @xmath301 , and @xmath302 are the atomic number , abundance , and relative atomic mass of the @xmath303-th element .
the abundance is defined by the ratio of the number densities of the @xmath303-th element and hydrogen ( @xmath304 ) . ] ) , which is contradictory to the observed soft spectrum of the outburst ( @xmath305 kev ) .
we further discuss the shock heating scenario , considering the very rapid rise time within @xmath306 s and the observed low temperature ( @xmath307 kev ; table [ table : maxi spectrum ] ) .
we set the onset time of the nova explosion @xmath308 between the last scan time of the scan - id m@xmath40 ( @xmath309 ; table [ table : maxi obs ] ) and the first scan time of the scan - id m+0 ( @xmath310 ; table [ table : maxi obs ] ) .
the elapsed time from @xmath308 to @xmath310 is @xmath311 s. we check whether the observed emission measure ( @xmath312 @xmath150 , where @xmath313 for the mekal model fit with free ` abund ` in table [ table : maxi spectrum ] and @xmath314 ; here @xmath315 and @xmath316 are the number densities of electrons and protons , respectively ) can be produced at @xmath317 by considering two simple geometries , a filled sphere and spherical shell . here , we assume the circumbinary space is filled with a fully - ionized electron - ion plasma with a constant density .
we assume the constant shock velocity @xmath318 , and then the distance ( @xmath319 ) of the shock wave front reaching from the surface of the wd at the time @xmath317 is @xmath320 where @xmath321 km s@xmath5 ( @xmath322 ) .
if the shape of the emission region is a filled sphere , the emission measure at @xmath317 is written as @xmath323 ( hereafter , we add `` @xmath324 '' for the physical value after the shock ) .
since we assume that the initial x - ray outburst is an optically - thin emission , the condition of the optical depth is @xmath325 , where @xmath326 is the thomson cross section . by removing @xmath327
, we obtain @xmath328 .
therefore , it is impossible to produce the observed emission measure at @xmath310 in the usual shock velocity ( @xmath329 ; * ? ? ?
* ; * ? ? ?
* ) , even at the speed of light . next ,
if the shape of the emission region is a spherical shell with a depth of @xmath330 , the emission measure at @xmath317 is written as @xmath331 .
the condition of the optical depth is @xmath332 . by removing @xmath327
, we obtain @xmath333 . using condition [ eq : r1 ] ,
the depth of the shell is limited to @xmath334 .
therefore , the emission region is a thin shell . on the other hand , removing @xmath330 and using condition [ eq : r1 ] , we obtain @xmath335 in this high density , the shock velocity can be estimated simply from the observed temperature of @xmath336 kev ( @xmath337 ; table [ table : maxi spectrum ] ) using the shock condition by @xmath338 , where @xmath339 is the temperature of the shock - heated gas in the down - stream region .
the radiative cooling time scale by free - free process is @xmath340 where @xmath227 , @xmath341 , and @xmath342 are planck constant , speed of light , and averaged gaunt factor , respectively , and we set @xmath343 @xcite . here , @xmath344 ( @xmath345 ) is 0.68 for a typical smc abundance . from condition [ eq : n_p shell ] , the @xmath346 is limited to @xmath347 s. since the cooling time scale is short , the width of the emitting shell is simply written as @xmath348 , where @xmath349 is the velocity of the post - shock region in the rest frame of the shock wave . using @xmath350 and equation [ eq : tcool ] , we derive the relation between @xmath351 and @xmath319 , @xmath352 using condition [ eq : r1 ] , the density is limited to @xmath353 by setting @xmath337 , @xmath313 , @xmath354 ( table [ table : maxi spectrum ] ) and assuming a typical smc abundance ( ` abund ` @xmath296 ) , we obtain @xmath355 .
this density is much larger than that in the stellar wind and even that in a circumstellar equatorial disk around a be star , typically @xmath356 @xmath150 @xcite .
therefore , we conclude that the shock heating scenario can not explain the soft x - ray outburst observed by maxi .
we thus instead propose another scenario to explain this outburst by invoking an extraordinary massive wd .
a more massive wd has a smaller radius @xcite , and thus a higher surface gravity leading to a higher pressure in the accumulated mass .
the nova explosion on a massive wd is triggered by less fuel , and thus it results in a short nova duration .
the observed sss phase of maxi j0158@xmath0744 started much earlier ( @xmath6 days ) and lasted a much shorter time ( @xmath127 one month ) than other fast novae @xcite .
the earliest turn - on of a sss phase observed so far was 10 days in u sco @xcite and @xmath357 days in one of 60 novae in m31 @xcite .
the extremely early sss phase of maxi j0158@xmath0744 is unexpected in models of novae on typical solar mass wds @xcite .
it suggests an unusually low ejecta mass in the nova explosion , and thus unusually massive white dwarf near the chandrasekhar mass .
it might even suggest a super - chandrasekhar mass .
indeed , according to theoretical models @xcite , wds can acquire super - chandrasekhar masses up to @xmath358 , if they rotate differentially . with this new perspective
, we propose to interpret the initial super - eddington x - ray outburst as an ignition phase of a nova just after the tnr , a fireball phase @xcite . in this process
, the thermal energy produced by the tnr is conveyed by the convection and released outside the envelope with a timescale of @xmath8 s , characterized by the half - lives of unstable nuclei ( fig .
[ fig : nova evolution ] ) . in novae on a white dwarf with a usual mass , transient soft x - ray emission ( @xmath359 kev ) for @xmath15
s just after the tnr is theoretically expected @xcite , but that has not been observed yet .
it is expected to reach about 10 times the eddington luminosity @xcite . for a very massive wd , we speculate that the tnr would produce more luminous x - ray emission with higher temperature due to a smaller amount of the envelope at the ignition phase of a nova . in this scenario ,
blackbody - like x - ray emission is expected at the ignition phase . in the spectral analysis ( table [ table : maxi spectrum ] ) , we obtained the radius of the photosphere to be @xmath360 km ( @xmath361 ) .
the rate of mass ejection ( @xmath362 ) can be estimated from this radius as follows . from the continuity equation for the distribution of ejecta around the wd , @xmath363 , where @xmath362 is a rate of mass ejection from the wd and constant in the radial distance ( @xmath364 ) , and @xmath365 is mass density , the number density of protons
is written by @xmath366 the optical depth condition is written by @xmath367 here , @xmath368 for large ` abund ` of neon .
therefore , @xmath362 is obtained to be @xmath369 on the other hand , the reaction rate of mass producing nuclear energy ( @xmath370 ) is related to the observed luminosity ( @xmath371 ) by @xmath372 , where @xmath373 and @xmath374 erg s@xmath5 ( table [ table : maxi spectrum ] ) .
then , @xmath375
g s@xmath5 .
therefore , the relation @xmath376 is obtained , which means that energy produced by the tnr at the bottom of accreted layer can escape as an x - ray photons efficiently with very small mass ejection , despite the super - eddington luminosity .
thus , there must be some sort of mechanism to realize the super - eddington luminosity with a small mass ejection .
we infer a convection just after the tnr @xcite for that mechanism , then we expect that future theoretical works of the tnr process , applied to the mass range near or over the chandrasekhar limit , will clarify this mechanism .
we also suspect that photon bubbles in highly magnetized atmospheres @xcite may work to solve this problem . according to @xcite , to produce the @xmath8 eddington luminosity with small mass ejection
, the magnetic pressure @xmath377 must be @xmath8 times larger than the gas pressure @xmath378 .
on the other hand , the gas pressure at the bottom of an accreted gas layer at an ignition of a nova is expected to be @xmath379 dyne @xmath65 @xcite .
then the magnetic field ( @xmath380 ) necessary for the @xmath8 eddington luminosity is @xmath381 g. interestingly , such highly magnetized wds with super - chandrasekhar masses ( @xmath382 ) are predicted theoretically @xcite .
since the tnr process is expected to last for @xmath8 s at the bottom of the accreted mass layer on the surface of wds @xcite , the rate of mass ejection probably peaked between the scans at @xmath32 s and @xmath33 s. it means that maxi scans at @xmath27 s and @xmath32 s observed the photospheric expansion phase ( b @xmath140 c in fig . 1 of * ? ? ? * ) , while the maxi scan at @xmath33 s observed the shrinking phase ( c @xmath140 d in the same figure ) .
the strong neon emission line at @xmath33 s suggests that there was an optically - thin thermal emission region surrounding the photosphere and filled with ejecta dredged - up from a massive o - ne wd .
such ejecta may have been provided by the previous photospheric expansion .
it must be noted that the existing models of the tnr do not predict this surrounding emission line region .
this observation provides us new physical details . in this scenario , the optical enhancement observed by _ swift _
uvot is no longer the usual photospheric emission of nova ejecta . since the optical decay seems correlated with the decay of sss x - ray emission ( fig .
[ fig : lc ] , middle and bottom ) , it can be explained by the reprocessed emission from the x - ray irradiated circumstellar disk of the be star .
it is justified by the fact that the size of the optical enhanced emission ( @xmath383 cm ; section [ sec : follow - up ] ) is comparable to the disk scale height @xcite .
monitor of all - sky x - ray image ( maxi ) discovered an extraordinarily luminous soft x - ray transient , maxi j0158@xmath0744 , near the small magellanic cloud ( smc ) on 2011 november 11 .
this source is a binary system consisting of a white dwarf ( wd ) and a be star at the distance of the smc .
maxi detected it in three scans at @xmath27 s , @xmath32 s and @xmath33 s after the trigger time .
the x - ray luminosity peaked on the second scan at @xmath4 erg s@xmath5 ( @xmath133 kev ) , which is two orders of magnitude brighter than the eddington luminosity of a solar mass object .
the spectrum of the third scan showed a he - like neon emission , suggesting that the emission contains an optically - thin thermal component and the wd is a massive o - ne wd . while the x - ray outburst could be considered as a kind of nova on the basis of the luminosity and the spectral evolutions , the huge peak luminosity and the rapid rise time ( @xmath384 s ) are difficult to explain by shock - induced emission , accepted for optically - thin thermal emission in nova explosions observed so far .
instead , we propose the scenario that the x - ray outburst is the direct manifestation of the thermonuclear runaway process at the onset of the nova explosion , the so - called fireball phase .
the super - eddington x - ray outburst and the subsequent very early super - soft source phase indicate a small ejecta mass , implying the underlying wd is unusually massive near the chandrasekhar limit , or possibly exceeding the limit .
we are grateful to the members of the maxi and _ swift _ operation teams .
we thank k. asano , i. hachisu , d. n. burrows , d. takei , s. r. kulkarni , y. maeda , and t. shigeyama for discussions and comments .
this research was partially supported by the ministry of education , culture , sports , science and technology ( mext ) , grant - in - aid no .
23740147 , 19047001 , and global - coe from mext `` the next generation of physics , spun from universality and emergence '' and `` nanoscience and quantum physics . ''
jpo , klp and npmk acknowledge financial support from the uk space agency .
jak acknowledges support from nasa .
this work was supported by the australian research council discovery projects funding scheme ( project number dp120102393 ) .
anders , e. & grevesse , n. 1989 , geochimica et cosmochimica acta , 53 , 197 band , d. et al .
1993 , , 413 , 281 begelman , m. c. 2001 , , 551 , 897 bohlin , r. c. , savage , b. d. & drake , j. f. 1978 , , 224 , 132 burrows , d. n. et al . 2005 , , 120 , 165 carrera , r. et al .
2008 , , 136 , 1039 cash , w. 1979 , , 228 , 939 das , u. & mukhopadhyay , b. 2012 , , 86 , 042001 della valle , m. & livio , m. 1995 , , 452 , 704 dickey , j. m. & lockman , f. j. 1990 , , 28 , 215 fujimoto , m. y. 1982 , , 257 , 752 galloway , d. k. , muno , m. p. , hartman , j. m. , psaltis , d. & chakrabarty , d. 2008 , , 179 , 360 gehrels , n. et al .
2004 , , 611 , 1005 hachisu , i. & kato , m. 2006 , , 167 , 59 hachisu , i. , kato , m. , saio , h. & nomoto , k. 2012 , , 744 , 69 henze , m. et al .
2011 , , 533 , a52 hilditch , r. w. , howarth , i. d. & harries , t. j. 2005 , , 357 , 304 hiroi , k. et al .
2011 , , 63 , s677 kalberla , p. m. w. et al . 2005 , , 440 , 775 kato , m. & hachisu , i. 1994 , , 437 , 802 kennea , j. a. et al .
2011 , proceedings of 4th international maxi workshop , arxiv:1101.6055 kennea , j. a. et al .
2011 , astron .
, 3758 kimura , m. et al .
2011 , astron .
3756 kimua , m. et al .
2012 , , accepted ( arxiv:1211.3737 ) krautter , j. 2008a , `` the super - soft phase in novae '' in `` rs ophiuchi ( 2006 ) and the recurrent nova phenomenon '' , asp conference series , vol .
401 , p.139 , ed .
a. evans , m. f. bode , t. j. obrien , and m. j. darnley , san francisco : astronomical society of the pacific .
li , k. l. et al .
2012 , , 761 , 99 , 13 matsuoka , m. et al .
2009 , , 61 , 999 mewe , r. , gronenschild , e. h. b. m. & van den oord , g. h. j. 1985 , , 62 , 197 mihara , t. et al .
2011 , , 63 , s623 morii , m. et al .
2011a , grb coordinates network , circular service , 12555 morii , m. et al .
2011b , , 63 , s821 morrissey , p. et al .
2005 , , 619 , l7 nakahira , s. et al .
2012 , , 64 , 13 nauenberg , m. 1972 , , 175 , 417 negoro , h. et al .
2010 , proceedings of a conference held october 4 - 8 , 2009 in sapporo , japan . ed .
by y. mizumoto , k. morita & m. ohishi .
asp conference series , 434 , san francisco : astronomical society of the pacific , 2010 .
, p.127 nelson , t. , donato , d. , mukai , k. , sokoloski , j. & chomiuk , l. 2012 , , 748 , 16 nishiyama , s. et al .
2007 , , 658 , 358 poole , t. s. et al . 2008 , , 383 , 627 rauch , t. et al .
2010 , , 717 , 363 roming , p. w. a. et al .
2005 , , 120 , 95 rybicki , g. b. & lightman , a. p. 1979
, `` radiative processes in astrophysics '' , john wiley & sons , inc . , u.s .
schlegel , d. j. , finkbeiner , d. p. & davis , m. , 1998 , , 500 , 525 schwarz , g. j. et al .
2011 , , 197 , 31 sguera , v. et al .
2006 , , 646 , 452 soderberg , a. m. et al .
2008 , , 453 , 469 sokoloski , j. l. , luna , g. j. m. , mukai , k. & kenyon , s. j. 2006 , , 442 , 276 starrfield , s. et al .
1998 , , 296 , 502 starrfield , s. , iliadis , c. & hix , w. r. 2008 , `` thermonuclear processes '' in `` classical novae '' , 2nd . edition , ed . m. bode & a. evans , cambridge univ .
press sugizaki , m. et al .
2011 , , 63 , s635 tomida , h. et al .
2011 , , 63 , 397 tsunemi , h. et al .
2010 , , 62 , 1371 van rossum , d. r. 2012 , , 756 , 43 , 11 warner , b. 1995 , `` novae in eruption '' in `` cataclysmic variable stars '' , cambridge univ .
press warner , b. 2008 , `` properties of novae : an overview . '' in `` classical novae '' , 2nd .
edition , ed .
m. bode & a. evans , cambridge univ . press .
walter , f. m. , stringfellow , g. s. , sherry , w. h. & field - pollatou , a. 2004 , , 128 , 1872 walter , f. m. , battisti , a. , towers , s. e. , bond , h. e. & stringfellow , g. s. 2012 , , accepted ( arxiv:1209.1583 ) waters , l. b. f. m. , van den heuvel , e. p. j. , taylor , a. r. , habets , g. m. h.
j. & persi , p. 1988
, , 198 , 200 woods , p. m. & thompson , c. 2006 , `` soft gamma repeaters and anomalous x - ray pulsars : magnetar candidates . '' in `` compact stellar x - ray sources '' . ed
by w. lewin & m. van der klis .
cambridge astrophysics series , no .
cambridge , uk : cambridge university press , 547 yoon , s .- c . & langer , n. 2004 , , 419 , 623 zorec , j. , arias , m. l. , cidale , l. & ringuelet , a. e. 2007 , , 470 , 239 | we present the observation of an extraordinary luminous soft x - ray transient , maxi j0158@xmath0744 , by the monitor of all - sky x - ray image ( maxi ) on 2011 november 11 .
this transient is characterized by a soft x - ray spectrum , a short duration ( @xmath1 s @xmath2 s ) , a very rapid rise ( @xmath3 s ) , and a huge peak luminosity of @xmath4 erg s@xmath5 in [email protected] kev band . with _
swift _ observations and optical spectroscopy from the small and moderate aperture research telescope system ( smarts ) , we confirmed that the transient is a nova explosion , on a white dwarf in a binary with a be star , located near the small magellanic cloud .
an extremely early turn - on of the super - soft x - ray source ( sss ) phase ( @xmath6 d ) , the short sss phase duration of about one month , and a 0.92 kev neon emission line found in the third maxi scan , 1296 s after the first detection , suggest that the explosion involves a small amount of ejecta and is produced on an unusually massive o - ne white dwarf close to , or possibly over , the chandrasekhar limit .
we propose that the huge luminosity detected with maxi was due to the fireball phase , a direct manifestation of the ignition of the thermonuclear runaway process in a nova explosion . |
in the early morning hours of june 1 , 2009 , air france flight af 447 , with 228 passengers and crew aboard , disappeared during stormy weather over the atlantic while on a flight from rio de janeiro to paris . upon receiving notification of the crash ,
the french bureau denqutes et danalyses ( bea ) pour la scurit de laviation and french search and rescue authorities organized an international search by aircraft and surface ships to look for signs of the plane and possible survivors . on the sixth day of this effort , the first debris and bodies were found 38 nm north of the aircraft s last known position .
that day a large portion of the galley was found along with other debris and some bodies .
figure [ figlastknownpos ] shows the aircraft s last known position , intended flight path and a 40 nm circle about the last known position .
analysis by the bea determined that the wreckage had to lie within 40 nm of the plane s last known position .
the aircraft was equipped with a flight data recorder and a cockpit voice recorder .
each of these recorders was fitted with an underwater locator beacon that activates an acoustic signal upon contact with water .
the bea initiated a search to detect these beacons .
the search was performed by two ships employing passive acoustic sensors supplied by the u.s . navy and operated by personnel from phoenix international .
the search began on june 10 , 2009 and lasted 31 days until the time when the batteries in the beacons were estimated to be exhausted .
the ships searched extensively along the intended flight path , but the beacons were not detected . next the bea began an active acoustic search with side - looking sonar to detect the wreckage on the ocean bottom .
this search took place in august 2009 south of the last known position in an area not covered by the passive acoustic search .
this search was also unsuccessful . after the unsuccessful search in 2009 , the bea commissioned a group of oceanographic experts to estimate the currents in the area at the time of the crash and to use these estimates along with the times and locations where the surface search had found bodies and debris in order to estimate the location of the wreckage . in @xcite the group recommended the rectangular search area north and west of the last known position shown in figure [ figollitraultsearcharea ] .
this rectangle is described as a @xmath0 confidence zone .
the group used available current measurements to make a number of estimates of the currents in the area of the wreck at the time of the loss . using these estimates
, they performed a backward drift on recovered debris and bodies to produce a number of trajectories ending at an estimated location of the crash .
the group removed trajectories that they felt were outliers .
a bivariate normal error was estimated for each of the remaining crash location estimates and used to produce a weighted mean with a bivariate normal error distribution .
this error distribution was used to compute a rectangle centered at the mean with a @xmath0 probability of `` containing '' the wreck location .
this rectangle guided the active acoustic search in april and may of 2010 .
the 2010 searches were performed by two teams .
navy and phoenix international team used a towed side - scan sonar system .
the woods hole oceanographic institute team used autonomous underwater vehicles with side - scan sonars and a remotely operated vehicle . after an unsuccessful search in the rectangle
, the teams extended their efforts to the south and west of the rectangle .
unfortunately , this search was also unsuccessful .
in july 2010 we were tasked by the bea to review all information about the loss of af 447 as well as the previous search efforts to produce a probability distribution ( map ) for the location of the underwater wreckage .
the probability maps that resulted from this process were used to guide the 2011 search .
on april 3 , 2011 , almost two years after the loss of the aircraft , the underwater wreckage was located on the ocean bottom some 14,000 feet below the surface .
on april 8 , 2011 , the director of the bea stated , `` this study @xcite published on the bea website on 20 january 2011 , indicated a strong possibility for discovery of the wreckage near the center of the circle .
it was in this area that it was in fact discovered after one week of exploration ''
subsequently , the flight data recorder and cockpit voice recorder were found , retrieved from the ocean bottom and flown to the bea in paris where the data in these recorders were recovered and analyzed .
this data provided crucial information for determining the cause of the crash . finding the wreckage also allowed the bea to return the bodies of many passengers and crew to their loved ones . in the sections below we describe the bayesian process used to compute these probability distributions .
bayesian analysis is ideally suited to planning complicated and difficult searches involving uncertainties that are quantified by a combination of objective and subjective probabilities .
this approach has been applied to a number of important and successful searches in the past , in particular , the searches for the _ uss scorpion _ @xcite and _ ss central america _ @xcite .
this approach is the basis for the u.s .
coast guard s search and rescue optimal planning system ( sarops ) that is used to plan coast guard maritime searches for people and ships missing at sea @xcite .
complicated searches such as the one for af 447 are onetime events .
we are not able to recreate the conditions of the crash 1000 times and record the distribution of locations where the aircraft hits the water . as a result
, definitions of probability distributions in terms of relative frequencies of events do not apply .
instead , we are faced with computing a probability distribution on the location of the wreckage ( search object ) in the presence of uncertainties and conflicting information that require the use of subjective probabilities .
the probability distribution on which the search is based is therefore a subjective one .
it is based on the analysts best understanding of the uncertainties in the information about the location of the search object . in an ideal situation ,
search effort is applied in an optimal fashion to maximize the probability of detecting the search object within the effort available .
the optimal search problem is a bayesian decision problem often based on a subjective probability distribution .
the basic optimal search problem can be stated as follows .
the search object is located in one of @xmath1 cells with @xmath2 being the probability the object is in cell @xmath3 .
we assume @xmath4 .
for each cell @xmath3 , there is a detection function @xmath5 where @xmath6 is the probability of detecting the object with effort @xmath7 given the object is in cell @xmath3 .
search effort is often measured in hours , which we will use for this discussion . a search allocation @xmath8 specifies the effort @xmath9 to be placed in cell @xmath3 for @xmath10 .
the probability of detection @xmath11 and cost @xmath12 for the search allocation @xmath8 are computed by @xmath13 suppose there are @xmath14 hours of search available .
the optimal search problem is to find an allocation @xmath15 for which @xmath16 and @xmath17 as the search proceeds and the search object is not found , the posterior distribution given failure of the search is computed and used as the basis for planning the next increment of search . even at this stage ,
subjective estimates of the detection capability of the sensors must often be used because of the lack of previous testing against the search object .
classical statistics has no formalism for approaching this type of decision problem .
in contrast , bayesian statistics and decision theory are ideally suited to the task .
in the analysis that we performed for the bea , we were not called upon to provide a recommended allocation of search effort but only to compute the posterior distribution for the location of the wreckage .
the approach taken for this analysis follows the model described in @xcite and @xcite .
the information about a complex search is often inconsistent and contradictory .
however , one can organize it into self - consistent stories or scenarios about the loss of the aircraft . within each scenario ,
the uncertainties in the information are quantified using probability distributions .
these distributions may be subjective if little or no data is available to estimate the uncertainties .
for each scenario , a probability distribution on target location is computed to reflect the uncertainties in the information forming the scenario .
this is typically done by simulation .
the resulting distributions are combined by assigning subjective probabilities to the scenarios and computing a weighted mixture of these scenario distributions to obtain the prior distribution .
the subjective probability assigned to a scenario reflects the analysts evaluation of the probability that the scenario represents what happened .
we approximated the continuous spatial distribution for the location of the wreckage by a discrete distribution represented by a set of @xmath18 point masses or particles @xmath19 for @xmath20 , where @xmath21 is the probability mass attached to particle @xmath22 .
the probabilities sum to 1 . in the case of a stationary search object
, @xmath23 is a latitude - longitude point . in the case of a moving object
, @xmath23 is a continuous space and time path over the time interval of interest . for visualization purposes ,
a grid of cells is imposed on the search space .
cell probabilities are computed by summing the probabilities of the particles in each cell , and the cells are color coded according to their probabilities . for the figures in this paper we used a black to white scale with black indicating the highest probability cells and white the lowest .
computation of the probability distributions described below was performed by a modified version of sarops .
the computation of the posterior distribution involves two basic steps , ( 1 ) computation of the prior ( before search ) distribution and ( 2 ) computation of the posterior distribution given the unsuccessful search . during flight
, a commercial aircraft sends messages via satellite containing maintenance and logistic information about the aircraft .
every 10 minutes it sends a gps position for the aircraft . the last known position , @xmath24n latitude@xmath25w longitude , for af 447 was sent at 02 hours 10 minutes and 34 seconds coordinated universal time on june 1 , 2009 . based on failure to receive any messages after 02 hours , 14 minutes and 26 seconds , the bea estimated that the plane could not have traveled more than 40 nm from its last known position before crashing into the ocean .
thus , we assumed the location of the wreckage was within the 40 nm circle centered at the last known position with probability 1 .
any probability distribution for the location of the wreckage that had probability outside this circle was truncated at the circle and renormalized to a probability distribution .
errors in gps positions typically have a standard deviation of roughly 10 m which is dwarfed by the other uncertainties in the location of the aircraft , so this error was not considered to be significant in our analysis .
the prior distribution @xmath26 on the location of the wreck was taken to be a mixture of three distributions , @xmath27 , @xmath28 and @xmath29 .
the distribution @xmath27 is uniform within the 40 nm circle , and @xmath28 is based on data from crashes that involved loss of control while a plane was at flight altitude .
@xmath29 is based on an analysis that drifted dead bodies found on the surface backward in time to possible crash locations . on the basis of discussions with analysts at the bea , we decided on the following subjective weights for these distributions ( scenarios ) , @xmath30 and @xmath31 so that @xmath32 in retrospect , it appears it would have been more appropriate to view @xmath29 as a likelihood function and multiply the distribution @xmath33 by @xmath29 to obtain the prior .
the distribution @xmath28 is based on an analysis of data from nine commercial aircraft accidents involving loss of control at flight altitude .
the analysis was performed by the russian interstate aviation group and the bea .
it showed that all impact points ( adjusted to the 35,000-foot altitude at which af 447 was cruising ) were contained within a circle of radius 20 nm from the point at which the emergency began .
these results were represented by a circular normal distribution centered at the last known position with standard deviation 8 nm along any axis .
we set @xmath28 equal to this distribution truncated at the 40 nm circle .
the @xmath29 scenario is the reverse - drift scenario .
the distribution for this scenario was computed using data on currents and winds to reverse the motion of recovered bodies back to the time of impact .
we used current estimates produced for bea @xcite and wind estimates from the u.s .
navy s operational global atmospheric prediction system model to perform the reverse drift . at daylight on june 1st , 2009 ,
french and brazilian aircraft began a visual search for survivors and debris from the wreck .
the first debris was found on june 6th , and more than 60 bodies were recovered from june 6th june 10th , 2009 .
there are two components of drift , drift due to ocean current and drift due to wind .
the latter is called leeway . to produce the reverse - drift scenario , we used positions and times at which bodies were recovered from june 610 .
we did not reverse - drift pieces of debris because we lacked good leeway models for them , whereas we could use the model in @xcite for bodies .
we used polygons to represent the positions of selected bodies recovered on each day from june 610 . for each polygon
16,000 positions were drawn from a uniform distribution over the polygon .
each position became a particle that was drifted backward in time subject to winds and currents in the following manner .
we used a 60-minute time step . for each step and each particle ,
a draw was made from the distributions on wind and current for the time and position of the particle in the manner described below .
the negative of the vector sum of current plus the leeway resulting from the wind was applied to the particle motion until the next time step .
the large uncertainties in the ocean currents at the time of the crash produced a distribution for the crash location that spread way beyond the 40 nm circle .
figure [ figrevdriftdistribution ] shows the reverse - drift distribution produced in this fashion and truncated at the 40 nm circle .
because of the large uncertainties in the currents , this scenario was given a lower weight than the other two that comprise the prior .
figure [ figpriordistribution ] shows the prior @xmath26 . for the prior distribution and the subsequent posteriors given failure to detect , we used @xmath34 points .
this section discusses simulation of winds and currents .
wind and current estimates are provided by the environmental community in the form of a grid of velocity vectors @xmath35 indexed by space and time where @xmath36 is the speed in the east west direction and @xmath37 is the speed in the north south direction .
we interpret these as mean values of the actual wind and current velocities and add a stochastic component by the method described below . to obtain @xmath35 for a wind or current at a time that corresponds to a grid time but for a spatial point that is not equal to one of the spatial grid points , we take the three closest spatial grid points and use a weighted average of the values at those points , where the weights are proportional to the inverses of the distances from the desired point to the chosen grid points
. most often we will need @xmath35 for times that are not equal to one of the time grid values .
to get @xmath35 in this case , we use the values as calculated above for the two closest times in the data and then linearly interpolate between these values . for every time step and every particle ,
the simulation perturbs the speeds @xmath36 and @xmath37 obtained from the data by adding a random draw from a normal distribution with a standard deviation of 0.22 kts for current speeds and 2.0 kts for wind speeds .
these draws are independent for @xmath36 and @xmath37 and from particle to particle , but for a given particle and speed the draws are correlated in time . specifically ,
if @xmath38 is the increment in time , measured in minutes , between two time steps , then the correlation is given by @xmath39 where @xmath40 is chosen so that @xmath41 .
there are two forces acting on a drifting particle : currents and winds .
the effect of current is straightforward . the particle s velocity due to the current is equal to the velocity of the current .
while it is reasonable to expect that a current of 3 knots will push an object at a speed of 3 knots , the same is not true for the wind .
drift due to wind ( leeway ) results from the sum of the force of the wind acting on the exposed surfaces of the object and the drag of the water acting on the submerged surfaces of the object .
the wind does not push an object at the wind s speed , and it often does not push an object exactly in the downwind direction .
there is typically a downwind and crosswind component of leeway .
the downwind component is in the direction the wind is blowing .
the crosswind component is perpendicular to the downwind component , and the direction of the crosswind leeway is not predictable . to account for this ,
the simulation switches between the two crosswind directions at exponentially distributed times as it is producing a particle path .
the magnitudes of the downwind and crosswind components are computed as follows . for a given particle and time
, we compute the wind velocity from the gridded wind data in the same fashion as for the ocean currents .
let @xmath42 be resulting wind speed expressed in meters per second . for a deceased person floating in the water
, we used the following model developed in @xcite to compute the mean downwind leeway @xmath43 and crosswind leeway @xmath44 measured in centimeters per second ( cm / s ) : @xmath45 we added a random component to the mean values computed above by adding the value of a draw from a normal distribution with mean 0 and standard deviation equal to the standard error computed for the regression used to estimate the mean downwind and crosswind leeway , respectively .
the time correlation of the random components of leeway was handled in the same way as for the ocean currents .
the posterior distribution was computed in four steps with each step accounting for an increment of unsuccessful search .
the result is the posterior distribution on the wreck location given failure of the search effort in 2009 and 2010 .
the steps , that is , increments of unsuccessful search , are listed below : 1 .
failure of the surface search to find debris or bodies for almost 6 days during june 16 , 2009 . 2 .
failure of the passive acoustic search to detect the underwater locator beacons on the flight data recorder and cockpit voice recorder in june and july 2009 . 3 .
failure of the active side - looking sonar search for the wreckage in august 2009 .
failure of the active side - looking sonar search for the wreckage in april and may of 2010 .
we use the following notation for the posterior distributions .
@xmath46 denotes the posterior given failure of search increment 1 ; @xmath47 denotes the posterior given failure of search increments 1 and 2 , and so on to @xmath48 which denotes the posterior given failure of search increments 14 .
it is @xmath49 that the bea used to plan the successful 2011 search .
these distributions were computed sequentially .
first @xmath50 was computed and used as the `` prior '' for computing @xmath47 .
then @xmath47 was used as the prior to compute @xmath51 and so on .
in this section we describe how we accounted for the four increments of unsuccessful search by computing the posterior distributions described above .
we represented the prior distribution @xmath26 by making @xmath34 independent draws from this distribution for the location of the wreck on the ocean bottom .
for the @xmath22th particle we set @xmath23 equal to the location of the @xmath22th draw and @xmath52 for @xmath20 . if an unsuccessful search takes place , we compute the probability @xmath53 that the search would have detected the search object if it were located at @xmath23 .
the posterior probabilities @xmath54 on the particles are computed using bayes rule as follows : @xmath55\\[-8pt ] & & \eqntext{\mbox{for } n=1,\ldots , n.}\end{aligned}\ ] ] the updated particles @xmath56 for @xmath20 provide a discrete approximation to the posterior given failure of the search .
if the search object is moving , then a stochastic motion model must be specified .
in addition to drawing the initial position , we make draws from the distributions specified by the motion model to create a continuous time and space path for the search object over the time interval of interest .
each particle is a path with a probability on it .
the set of particles is a discrete sample path approximation to the stochastic process describing the motion of the search object .
when a search takes place , we account for the motion of the particles and the sensors in calculating @xmath53 .
the posterior is again computed from ( [ eqposteriorparticle ] ) .
we decided to incorporate the effect of the almost 6 days of unsuccessful surface search as follows .
each point @xmath23 in the prior specifies a location on the ocean bottom . in the calculation of this distribution
, we assumed that when the aircraft crashed into the surface of the ocean it fell straight to the bottom some 14,000 feet below .
it is likely that there was some lateral motion as the wreckage drifted to the bottom , but we reasoned that this uncertainty was small compared to the other uncertainties in the problem , so we ignored it .
for each particle we constructed a path starting at the position of the particle projected up to the surface of the ocean .
we drifted the particle forward in time for six days using the wind and current drift model described in section [ secprior ] .
however , this time we used the drift vector itself rather than its negative .
this produced a path for each particle .
as with the reverse drift scenario , the leeway component of drift was based on that of a body in the water .
aircraft searches are reported in terms of sorties .
sorties described a sequence of straight line segments ( flight legs ) flown at specified times , speeds and altitudes .
typically the set of legs for a sortie covers a rectangular area on the ocean surface .
we used the detectability of the galley found on the 6th day of surface search as a surrogate for the detectability of the floating debris .
we further assumed that the detectability of the galley is equivalent to that of a four - man raft ( they are roughly the same size ) .
the coast guard has developed tables that provide estimates of the probability of detecting a four - man raft from an aircraft on one leg of a sortie as a function of the speed and altitude of the aircraft , range at the point of closest approach to the search object on the leg , and environmental variables such as visibility , cloud cover and sea state of the ocean .
for each particle and each leg of each sortie , we computed the range at closest point of approach and used the coast guard tables to compute the probability that the particle would fail to be detected on that leg .
we assumed an independent detection opportunity on each leg and multiplied the failure probability for each leg and sortie to obtain an overall probability of failure to detect for each particle .
search by ships was incorporated in a similar manner .
the result was the computation of the failure probability @xmath57 for the air and ship searches for each particle for the six days of unsuccessful search . because of the many uncertainties and approximations involved in the computation of the failure probabilities for this search , and because we felt that it was unlikely that the search would fail to detect any debris for almost six days under the assumptions we had made , we decided it was necessary to allow for the possibility that the search was ineffective for most of those six days for reasons unknown to us .
we made a subjective estimate that the search was ineffective ( failure probability 1 ) with probability 0.7 and effective [ failure probability @xmath57 ] with probability 0.3 .
as a result we set @xmath59 and computed the posterior probabilities @xmath60 on the path of the @xmath22th particle by ( [ eqposteriorparticle ] ) , which is also the posterior probability on the wreck location being equal to @xmath23 given failure of the six days of surface search .
thus , @xmath61 for @xmath20 yields the posterior distribution @xmath50 .
more details on the aircraft searches may be found in @xcite .
the aircraft was equipped with a flight data recorder and a cockpit voice recorder .
each of these recorders was fitted with an underwater locator beacon that activates an acoustic signal upon contact with water .
the batteries on the locator beacons were estimated to last for 40 days .
the passive acoustic search to detect these beacons lasted 31 days ending on 10 july 2009 .
the search was performed by two ships employing passive acoustic sensors supplied by the u.s . navy and operated by personnel from phoenix international .
based on a calculation involving the source level of the beacons and propagation loss through the water , we estimated the sensors to have probability at least @xmath62 of detecting the beacons within lateral range 1730 m. experience in past searches has shown that detection estimates based on manufacturers specifications and operator estimates tend to be optimistic .
thus , we put a maximum of 0.9 on estimates of sensor detection probabilities .
the search paths , which are shown in figure [ figsearchpath ] , were designed so that 1730 m would be the maximum lateral range to the nearest path for any point in the search region . in estimating the probability of detection for these sensors we accounted for the possibility that one or both of the beacons were destroyed in the crash .
based on survival data for these beacons obtained from previous crashes , we estimated a probability of 0.8 that a single beacon survived the crash .
if beacon survival is independent , then @xmath63 , the probability of detecting at least one of the beacons given they are within lateral range 1730 m of the sensor , equals @xmath64 if the beacons were mounted sufficiently close together to consider their chances of survival to be completely dependent , then the probability of detecting at least one beacon drops to @xmath65 .
we decided to use a weighted average of 0.25 for the independent and 0.75 for the dependent probabilities , yielding a detection probability of @xmath66 .
the ships tracks displayed in figure [ figsearchpath ] show the passive acoustic search paths , which were designed to cover the expected flight path of the aircraft .
we computed the posterior distribution @xmath47 by starting with @xmath50 as the prior and computing @xmath67 for each particle by the same method used for the aircraft search .
for each particle and path , we determined whether the path came within 1730 m of the particle .
if it did , the failure probability for that particle was multiplied by @xmath68 .
the resulting @xmath67 was used in ( [ eqposteriorparticle ] ) with @xmath69 to compute @xmath70 and the posterior distribution @xmath47 .
the bea employed a side - looking ( active ) sonar from the french research institute for exploration of the seas towed by the french research vessel _
pourquoi pas _ ?
to continue the search after the batteries on the beacons were estimated to have been exhausted .
this search took place in august 2009 in the white rectangle in row 24 in figure [ figsearchpath ] .
this region was chosen because it was suitable for search by side - looking sonar and had not been searched before .
we assumed a 0.90 probability of detection in the searched region .
this represents a conservative , subjective estimate of the probability of this sensor detecting a field of debris .
the detection of small items such as oil drums on the ocean bottom confirmed that the sensor was working well .
we computed the posterior after this search effort by setting @xmath71 this @xmath67 was used in ( [ eqposteriorparticle ] ) with @xmath72 to compute @xmath73 and the posterior distribution @xmath74 .
figure [ fig2010searcharea ] shows the regions searched during 2010 with active side - looking sonar .
the search began in the rectangular region recommended by @xcite inside the 40 nm circle and proceeded to the remainder of the areas shown in medium gray including a small rectangular region southwest of the last known position .
as with the previous active sonar search , we estimated that within these regions the search sensors achieved detection probability 0.9 .
we felt this was a conservative subjective estimate based on the careful execution of the search , the quality of sonar records and the numerous small articles detected during this search .
the search region was represented by a rectangular and a polygonal region .
as with the active sonar search in 2009 , we computed @xmath67 in the manner given in ( [ eqposterioraftersearch ] ) with the rectangular and polygonal search regions replacing the rectangle of the 2009 search .
this produced the desired posterior @xmath48 which accounts for all the unsuccessful search .
figure [ figposterior1234 ] shows the posterior after the unsuccessful searches in steps 14 . even though this posterior allows for the possibility that the beacons did not work , doubts about the beacons
compelled us to produce the alternate posterior shown in figure [ figfailedbeaconsposterior ] , which assumes the beacons did not function .
the location of the wreckage which is shown in this figure falls in a high probability area .
this posterior distribution seems remarkably accurate and raises the question of why the beacons were not detected .
the bea recovered one data recorder with the beacon attached .
testing by the bea showed that when the beacon was connected to a fully charged battery , it did not produce a signal .
this indicates that the beacons were damaged in the crash and did not function .
this would explain why the beacons were not detected by the passive acoustic search .
a better way to handle the doubts that we had about the beacons would have been to compute a joint distribution on beacon failure and wreck location
. the marginal distribution on wreck location would then be the appropriate posterior on which to base further search .
figure [ figfailedbeaconsposterior ] shows that the wreckage is located in a high probability area assuming the beacons failed to operate properly . because the wreckage is located in an area thoroughly covered by the passive search ( see figure [ figsearchpath ] ) , these results and the tests of the recovered beacon by the bea suggest that both beacons failed to actuate .
it appears that the likely failure of the beacons to actuate resulted in a long and difficult search .
the approach described in this paper used a careful and methodical consideration of all data available with their associated uncertainties , to form an analytic assessment of the highest likelihood areas for future search efforts .
the weighted scenario approach allowed inconsistent information to be combined with subjective weights that capture the confidence in the data .
the analysis of the detection effectiveness of each search component produced the bayesian posterior distributions shown in figures [ figposterior1234 ] and [ figfailedbeaconsposterior ] and formed a solid basis for planning the next increment of search .
in fact , the 2011 search commenced in the center of the distribution and quickly found the wreckage .
failure to use a bayesian approach in planning the 2010 search delayed the discovery of the wreckage by up to one year .
the success of the analysis described in this paper provides a powerful illustration of the value of a methodical , bayesian approach to search planning .
the full report of this work is available on the bea website in @xcite .
we thank joseph b. kadane for his helpful comments and suggestions . | in the early morning hours of june 1 , 2009 , during a flight from rio de janeiro to paris , air france flight af 447 disappeared during stormy weather over a remote part of the atlantic carrying 228 passengers and crew to their deaths . after two years of unsuccessful search ,
the authors were asked by the french bureau denqutes et danalyses pour la scurit de laviation to develop a probability distribution for the location of the wreckage that accounted for all information about the crash location as well as for previous search efforts .
we used a bayesian procedure developed for search planning to produce the posterior target location distribution .
this distribution was used to guide the search in the third year , and the wreckage was found with one week of undersea search . in this paper
we discuss why bayesian analysis is ideally suited to solving this problem , review previous non - bayesian efforts , and describe the methodology used to produce the posterior probability distribution for the location of the wreck .
, , |
density functional theory ( dft ) has proven itself a remarkably successful tool in condensed matter physics@xcite .
the foundations were laid in 1964 , when hohenberg and kohn ( hk)@xcite proved that the ground state energy of any quantum mechanical system could be described as a functional of the one - body density only .
subsequently , kohn and sham@xcite developed an orbital based method which could be applied to electronic systems , and mermin@xcite extended the hk proof to finite temperatures , opening up the possibility of using dft to calculate the free - energy of a statistical mechanical system .
since then many different practical methods have been developed to apply dft to electronic problems , with countless applications in condensed matter physics , chemistry , and biology@xcite .
examples of electronic structure techniques relevant to this paper are the _ ab initio _ molecular dynamics ( aimd ) method of car and parrinello@xcite and the orbital free _ ab initio _ molecular dynamics ( of - aimd ) scheme of madden and co - workers@xcite . in a parallel development ,
the finite temperature version of dft has been widely used to study classical systems , with myriad applications to phase - transitions and the theory of fluids ( in its broadest sense ) , see e.g.@xcite for reviews .
however , the classical and quantum versions of dft have rarely been combined the two fields have largely developed independently . in this paper
we will attempt to bring together these two strands of dft into one unified formalism , with the aim of applying them to liquid metals .
these are ideal systems for which to test a mixed quantum - classical dft approach since they can be viewed as binary mixtures of classical ions and quantum electrons . to begin
, we first define a two - component free - energy functional @xmath0 = f_i^{id}[\rho_i ] + f_e^{id}[\rho_e ] + f^{ex}[\rho_i,\rho_e],\ ] ] split into the usual way between the ideal free energies of the ions , @xmath1 , and the electrons , @xmath2 , and the excess free energy @xmath3 , which is a functional of both one - body densities @xmath4 and @xmath5 .
this unique intrinsic free - energy functional functional obeys the variational principle : @xmath6}{\delta \rho_\alpha({\bf r } ) } + \phi_\alpha({\bf r } ) = 0,\ ] ] where the @xmath7 are the external potentials which uniquely induce the one - body densities @xmath8 .
this principle is subject to the constraint that the integral of @xmath9 , the total number of particles of species @xmath10 .
we will apply this free - energy functional both to homogeneous and to inhomogeneous liquid metals .
in section [ section2 ] we use the dft formulation to derive the quantum ornstein zernike(qoz ) equations . from these
we can derive a set of integral equations , called the quantum hypernetted chain approximation(qhnc ) by chihara@xcite , which self - consistently solves for the ion - ion and ion - electron pair correlation functions in a homogeneous liquid metal .
the great simplification of the qhnc approach is that the original many - centre electronic problem is reduced to an effective one - centre problem , which is much easier to solve . as a specific application
, we study the ion - electron radial distribution function for a set of simple metals . in section [ section3 ]
we extend this dft formulation to inhomogeneous fluids . by using a simple version of this functional
, we compare the electron density profile at the liquid - solid interface of al to the recent simulations using of - aimd@xcite .
we end with some conclusions .
dft provides a nicely unified route to the correlation functions of homogeneous fluids . to begin
, we first show how to derive the qoz relations by defining the direct correlation function as the second functional derivative of the excess free - energy defined in eq .
( [ eq1.1 ] ) : @xmath11 where @xmath12 is the inverse susceptibility matrix of the full system , and @xmath13 is the inverse susceptibility matrix of the ideal system . these relationships between direct correlation functions and inverse susceptibilities are called the qoz equations , and they hold for an arbitrary number of components . if any components are classical , then the susceptibilities can be connected to the pair - correlations through the fluctuation - dissipation theorem@xcite : @xmath14 where the @xmath15 are the structure factors , defined in the usual way@xcite , and we have taken the homogeneous limit and the @xmath16 are the homogeneous limits of the densities @xmath8 .
specializing to the homogeneous limit for two components , with one component being classical ( the case at hand ) , the qoz equations reduce to a set of relationships between the direct correlation functions and the the ion - ion and ion - electron pair correlation functions as well as the electron - electron susceptibilities .
the latter are still quantum - mechanical , and can not be simply connected to electron - electron pair correlations . to make further progress
, one needs a way of solving the qoz relations for a liquid metal .
we sketch the derivation here , for more details we refer to reference@xcite and to the original work of chihara@xcite . as a first step , we use a quantum version of the percus trick@xcite to relate the _ homogeneous _ two - body pair - correlation functions to the one - body _ inhomogeneous _ density around a single classical particle fixed at the origin . for the ion - electron pair - correlation function
we fix an ion at the origin to find : @xmath17 where @xmath18 is the ( interacting ) valence electron density@xcite .
a similar relationship can be found for the ion - ion pair - correlation function , but this trick can not be used for the electron - electron pair - correlation function , since one can not `` fix '' an electron at one position . the next step is to solve for the interacting one - body electron - density . to do this , one can use the kohn - sham trick@xcite , namely that there exists a single - particle external potential @xmath19 which will induce in a _ non - interacting _ system the same one - particle density @xmath20 as is found in the full _ interacting _ system .
this external effective potential , felt by the non - interacting electrons or non - interacting ions , follows from the euler equations .
if one further makes a functional taylor expansion around the equilibrium homogeneous densities , then the following effective ion - ion and ion - electron potentials result : @xmath21 where the @xmath22 are the homogeneous limits of the direct correlation functions defined in eq .
( [ eq2.1 ] ) , and the percus trick was used to rewrite @xmath23 in terms of the correlation functions @xmath24
= @xmath25 .
the remaining higher order terms are lumped into the so - called _ bridge functions _
@xmath26 , well known in the theory of liquids@xcite .
again , we note that no such potential can be derived for the electron -electron correlations . with that caveat in mind ,
our formulation , derived from a dft approach , is still in principle exact . to make progress ,
however , some approximations must be made , in particular to treat the electron - electron correlations .
we follow chihara , who made a series of such approximations to derive the qhnc equations@xcite .
the most important one is : 1 . _
the valence electron correlations are treated in the jellium approximation_. i.e. the effect of the ion - ion and ion - electron correlations on the electron - electron correlations are ignored , and the electron - electron direct - correlation function is written as : @xmath27,\ ] ] where @xmath28 is the local field factor for the electron gas@xcite .
this has the great advantage that the qoz relations ( [ eq2.1 ] ) now reduce to only two coupled equations for the ion - ion and the ion - electron correlations . while this is the key approximation that makes the qoz relations tractable , it is also the most important and uncontrolled approximation in the qhnc scheme .
the fact that the effective pair potentials derived from a linear response scheme@xcite are very sensitive to the exact form of the local field factor also testifies to the importance of the approximations in eq .
( [ eq2.5 ] ) to the physics of a liquid metal .
+ qhnc also uses an implicit separation of the valence electrons from the core electrons .
this approximation is closely related to the jellium approximation discussed above , and is only good when the core and valence electron energy levels are well separated so that the ionic correlations do nt have a significant effect on the core states ; it is likely to break down near resonances .
+ the other approximations used in qhnc are much better understood , and are often encountered in either their electronic structure@xcite or liquid state theory context@xcite . in descending order of importance they are ( for a more detailed discussion see ref .
@xcite . ) : 2 . _ the local density approximation ( lda ) is used for the one - centre ion - electron problem_. this approximation is well known and understood in electronic - structure theory@xcite . although in principle this approximation is similar to approximation 1 for @xmath29 , since the exchange and correlation energy of an inhomogeneous electron gas are approximated by the local energy density of jellium , there are a number of reasons to believe the lda is rather accurate@xcite . various improvements to the lda
exist , but these are not thought to be very important for the simple metals we study .
the ion - electron bridge function @xmath30 is set to 0 , which is related to the hypernetted - chain approximation , well known in liquid state theory @xcite .
the ion - ion bridge function @xmath31 is approximated by the bridge - function of a hard - sphere reference state .
this is called the rhnc or mhnc approximation in liquid state theory@xcite , and is generally very accurate .
the bare ion - ion potential is taken to be purely coulombic . again , this is an approximation whose limitations are well understood@xcite . in summary
then , the qhnc approximation achieves the following very important simplification for the electronic problem : the original many - centre electronic problem has been reduced to an effective one - centre problem by replacing the direct effect of the ions with an effective external potential , given by eq .
( [ eq2.4 ] ) , that depends self - consistently on the ion - ion correlations .
this is depicted schematically in fig .
[ fig : qhnc - schematic ] .
the main advantages are that the ion - ion and ion - electron correlations emerge naturally and on the same footing and that the calculations are much more rapid than full _ ab initio _ simulations . by using dft to derive the qhnc equations ,
their origin and meaning becomes more transparent . for details on the ( non - trivial ) numerical implementation of the qhnc we refer to refs .
@xcite .
the ion - electron radial distribution function , defined as the conditional probability of finding a valence electron a distance @xmath32 away , given that there is an ion at the origin , can be written as : @xmath33where @xmath34 is the so - called `` pseudo - atom '' density , which , when superimposed according the ion - ion radial - distribution function @xmath35 , gives the correct value of the total electron - density . the ion - electron radial distribution function and the related pseudo - atom density , calculated with the qhnc approach ,
are shown in fig .
[ fig : gei ] for a series of simple metals . at short distances
the probability of finding an electron a distance @xmath32 away is equal to the pseudo - atom density , but further away the effects of the other surrounding pseudo - atoms kick in .
note that the qhnc is an all - electron calculation ; the oscillations near the core are correctly reproduced@xcite , in contrast to the traditional aimd approaches , which rely on pseudopotentials .
the ion - electron correlation functions calculated with the qhnc compare very well for the cases where aimd simulations are available .
however , it turns out that for most of the metals in our set a much simple linear response formalism also performs remarkably well@xcite .
the surprising accuracy of the linear response arises from a quantum interference effect between two length - scales : the fermi wave - vector and the core - size of the ions , which makes the non - linear response terms much less important than one might naively expect@xcite .
the accuracy of the qhnc also benefits from this effect . 0.1 cm the only metal in our set where the qhnc has difficulty is ga .
we found earlier@xcite that an ad - hoc change in the local field factor @xmath36 seems to improve matters .
we also found that the d - electrons are close to a resonance .
these two points indicate that approximation 1 , discussed in the previous section , begins to break down for ga .
in this section we explore the possibilities of studying inhomogeneous metals within a two - component dft formulation .
the advantage of treating metals directly as as electron - ion mixtures in this way is that we bypass the need for effective ( density dependent ) ion - ion potentials .
modern classical dft methods use input from the homogeneous liquid state@xcite ; the qhnc is ideally suited to provide such input for a two - component dft .
one of the simplest approximations for the dft of an inhomogeneous system , first proposed by ramakrishnan and yussouf(ry)@xcite , corresponds to truncating at second order a functional taylor expansion of the excess free - energy around the homogeneous phase : @xmath37 = - \sum_{\alpha } \sum_{\beta } \int d{\bf r } \int d{\bf r } ' \delta \rho_\alpha({\bf r } ) c_{\alpha \beta } ( \rho_\alpha^0 , \rho_\beta^0 ; |{\bf r } - { \bf r}'| ) \delta
\rho_\beta({\bf r}'),\ ] ] where @xmath38 - f_0^{ex}(\{\rho_\alpha^0\})$ ] , and the direct correlation functions @xmath39 are those of the homogeneous liquid phase . for metals ,
the qhnc provides @xmath40 and @xmath41 , while @xmath42 is fixed by the jellium approximation .
together , these correlation functions completely determine the excess free - energy of eq.([eq3.1 ] ) . to obtain the total free - energy ,
one then adds the ideal contributions of the electrons and the ions . for the ions , this ideal free energy functional
has the form : @xmath43 = \int d{\bf r } \rho_i({\bf r } ) \left\ { \ln \left [ \rho_i({\bf r } ) \lambda ^3 \right ] -1 \right\},\ ] ] but for the electrons the exact form is not known , and various approximations must be made . to preserve the advantages of the current dft approach
, one needs a kinetic energy functional .
a number of forms have been proposed ; some are remarkably accurate , see e.g. @xcite and references therein
. these kinetic energy functionals also vary in the ease with which they can be implemented in a full two - component dft calculation@xcite .
classical dft has frequently been applied to the freezing transition and the fluid - solid interface of simple fluids@xcite .
the two - component dft described above provides a basis upon which to build similar applications for metallic systems . as a first application
, we attempted to use the two - component ry formalism described above to calculate the freezing transition of al . a very similar approach was already implemented by one of us@xcite to study the freezing transition of h. however there a simpler semi - empirical prescription was used for @xmath41 . since @xmath44 does not have a core radius , the cancellation of higher order response terms found for simple metals@xcite does not hold .
this implies that the correlation functions and also the freezing transition of @xmath44 should be more difficult to treat accurately than would be the case for simple metals .
other properties of liquid @xmath44 , such as the metal - insulator transition@xcite are also much harder to treat .
partially for these reasons , we would initially expect that the simple metals are actually better systems on which to test rudimentary dft s .
the one - body densities of the solid phase are parameterized by sums over the reciprocal lattice vectors ( r.l.v . )
@xmath45 : @xmath46.\ ] ] for the ions a further , well established , simplification is to take the real - space density @xmath4 as a sum of gaussians , centred on the lattice sites , which implies @xmath47 $ ] . with this density parameterization and approximation , the full free - energy functional for the difference between the solid and the liquid state free energy , taken from eqs .
[ eq1.1 ] , [ eq3.1 ] and [ eq3.2 ] , reduces to : @xmath48 - \frac{3}{2 } - \ln 2\right ) + \delta f^{id}_e[\rho_e ] & - & \frac{1}{2 } \sum_{\{{\bf g}\ } } \ , ' ( \xi_i^{{\bf g}})^2 \rho_i^0 \hat{c}_{ii}(|{\bf g}| ) - \nonumber \\
\frac{1}{2 } \sum_{\{{\bf g}\ } } \ , ' ( \xi_e^{{\bf g}})^2 \rho_e^0 \hat{c}_{ee}(|{\bf g}| ) & - & \sum_{\{{\bf g}\ } } \ , ' \xi_i^{{\bf g } } \xi_e^{{\bf g } } \rho_e^0 \hat{c}_{ie}(|{\bf g}|).\end{aligned}\ ] ] here the @xmath49 are the fourier transforms of the direct correlation functions for the liquid state , and the primes mean that the term @xmath50 should be omitted in the sum .
the first term between brackets is the ideal free - energy of the ions ( [ eq3.2 ] ) , written out in terms of the gaussian parameter @xmath51 .
this free - energy can then be minimized through the variational principle ( [ eq1.2 ] ) to solve for the free - energy difference between a given solid crystal structure and the fluid phase . however , when we attempted this for al , with the direct correlation functions given by the qhnc , we found that the free energy of the solid did not have a stable minimum .
this is a common problem encountered when applying a dft to the freezing transition ( see e.g. ref @xcite ) : if the full direct correlation functions are used , then it is very hard to stabilize solid phase .
if instead one uses the dft for a hard - sphere reference system , adding the attractive interactions as a perturbation , very good agreement with experiment and simulation is recovered .
examples include phase diagram@xcite and the fluid - solid interface of the lj fluid@xcite , and a one - component treatment of the freezing of al@xcite , based on an effective pair potential .
this suggests that a successful 2-component dft treatment of the freezing transition also needs to pass via this route .
unfortunately this quickly reduces to a theory similar to the one - component dft , which means introducing the effective pair potentials we are trying to avoid in the first place .
another application of the two - component dft formalism is to examine the fluid - solid interface of al , for which besson and madden@xcite have recently completed extensive of - aimd simulations .
they found non - trivial behaviour , with the ion - ion density profiles decaying from an ordered structure to a smooth liquid structure across several atomic layers .
similarly , they found that the valence electron density also showed oscillations which decayed into the bulk liquid phase . in principle , a two - component dft would be ideally suited to study this problem , since the ionic and electronic density profiles come out on equal footing .
unfortunately , the difficulties encountered when trying to treat the freezing transition also make studying this problem very difficult .
instead , we tackle a simpler problem : given the ionic density profile , can we use our dft to calculate the electron density profile ?
if the ionic density profiles are given , which fixes the @xmath52 , then the electronic @xmath53 which satisfy the variational principle ( [ eq1.2 ] ) follow from eq .
( [ eq3.4 ] ) : @xmath54 where we have made an additional approximation , also made in @xcite , namely that the ideal electron contribution @xmath55 $ ] is expanded to second order in @xmath56 ( linear response ) .
this is consistent with the spirit of the ry dft approach , and greatly simplifies the minimization of the electronic degrees of freedom , since they are now linearly coupled to the ionic ones .
note that taking the lowest order approximation to the ion - electron direct correlation function , @xmath57 in eq .
( [ eq3.7 ] ) would be equivalent to a standard linear response calculation .
since we use the full @xmath41 , non - linear response terms are included , although one would need to use the full kinetic energy functional to be completely consistent with the ry dft approach of eq .
( [ eq3.4 ] ) .
the results of using eq.([eq3.7 ] ) are shown in fig .
[ fig : hong2 ] as a function of the distance @xmath58 along the interface .
( see ref .
@xcite for details of parameters ) .
the qhnc results are in fact less accurate than a simpler linear - response approximation which replaces @xmath59 , with an empty - core ashcroft pseudopotential@xcite . at first this
may seem surprising , but a careful look at the inset of fig .
[ fig : hong2 ] shows that the difference results from the fact that @xmath59 is considerably smaller at @xmath60 , where the strongest ionic scattering occurs .
this difference stems in part from the fact that the true pseudopotential in al is known to be highly non - local@xcite because of the @xmath61 character of the bonds , suggesting that a local formulation , such as the simple linear response dft we employed , will not work well together with the qhnc calculation which makes no local pseudopotential assumptions .
non - linear response effects are also expected to be more important at the highly inhomogeneous fluid - fluid interface .
in contrast to the qhnc , the pseudo - potential contains an empirically adjustable parameter which effectively includes non - local and non - linear effects . to achieve accurate results with a dft based on the qhnc
, one needs to treat the two approaches at comparable levels of approximation . as is often encountered in condensed - matter physics ,
mixing different levels of approximation does not work very well .
two ways of improving this would be * ( 1 ) * to use qhnc with the improved local pseudopotentials used in the ofmd@xcite or * ( 2 ) * to use a more sophisticated dft approach .
although far from complete or completely satisfactory , this first , and rather primitive , example of a two - component dft used for a liquid - solid interface suggests that our dft approach could in principle be fruitfully used to study fluid - solid interfaces in simple metals .
in conclusion , we have shown how dft provides a unified formalism from which to derive the qoz equations , and the related qhnc approach , first pioneered by chihara .
the qhnc reduces the many - centre electronic problem to an effective one - centre one .
this dramatically reduces the computational time needed to calculate ion - electron and ion - ion correlation functions .
it also has the advantage that no explicit use of effective pair potentials or pseudo - potentials is needed .
the most important approximation in the qhnc approach is to treat the electron - electron correlations as those of jellium .
it is not yet clear under which conditions this approximation begins to break down . with this caveat in mind ,
the other approximations entering the qhnc are reasonably well understood .
we used the qhnc to calculate the ion - electron radial distribution functions for a number of simple metals , finding very good agreement with ab - initio molecular dynamics calculations where these are available .
we have also discussed an exploratory application of the simple ry dft approach to liquid metals as two - component electron - ion mixtures . we found that describing the freezing transition suffers from a similar problem to that found for classical simple fluids : the solid phase does not develop a stable minimum if the full direct correlation functions are used as input .
this is rather disappointing .
slightly more ( partial ) success was found when we attempted to use the dft to describe the inhomogeneous electron density at a liquid - solid al interface .
although this result was only a partial application of the dft , since the ionic density profile came from the simulations , it does suggest that developing a full fledged dft could be very fruitful .
finally , it is clear that our rudimentary dft approach is not yet accurate enough , and could be improved in a number of ways .
we are attempting to generalize more sophisticated dft schemes like the mdwa@xcite or gela@xcite to the two - component ion - electron problem .
we are also studying the effect of using different local field factors and different kinetic energy functionals .
aal acknowledges support from the isaac newton trust , cambridge , and the hospitality of lydric bocquet at the ecole normale superieure in lyon , where some of this work was completed .
he thanks n.w .
ashcroft , p. madden , and m. sprik for illuminating discussions .
99 w. kohn , rev . mod .
phys . * 71 * , 1253 ( 1999 ) .
p. hohenberg and w. kohn , phys . rev .
* 136 * , b864 ( 1964 ) .
w. kohn and l. j. sham , phys .
140 * , a1133 ( 1965 ) .
n. d. mermin phys . rev . *
137 * , a1441 ( 1965 ) .
see e.g. _ density functional theory _ , edited by e. k. u. gross and r. m. dreizler ( plenum press , new york 1995 ) . ) r. car and m. parrinello ,
lett . * 55 * , 2471 ( 1985 ) .
m. pearson , e. smargiassi , and p. a. madden , j. phys . : condens
. matter , * 5 * , 3221 ( 1993 ) .
m. baus , j. phys .
condensed matter * 2 * , 2111 ( 1990 ) .
r. evans in d. henderson ed . ,
_ fundamentals of inhomogeneous fluids _ , ( marcel dekker , new york , 1992 ) . j. chihara , prog . theor .
phys . * 59 * , 76 ( 1978 ) .
b. j. jesson and p.a .
madden , j. chem .
phys . * 113 * , 5924 ( 2000 ) , _ ibid _ , * 113 * , 5935 ( 2000 ) .
r. kubo , rep .
* 19 * , 255 ( 1966 ) .
n. w. ashcroft and d. stroud , solid state physics * 33 * , 1 ( 1978 ) .
hansen and i. r. mcdonald , _ theory of simple liquids , 2nd ed .
_ , ( academic press , london ( 1986 ) ) .
j. a. anta and a. a louis , phys .
b * 61 * , 11400 ( 2000 ) . j. chihara , _ j. phys .
c. : solid state phys _ * 18 * , 3103 ( 1985 ) .
j. chihara , phys .
a * 40 * , 4507 ( 1989 ) ; m. ishitobi and j. chihara , _ j. phys .
: condens . matter _
* 4 * , 3679 ( 1992 ) .
j. k. percus , phys .
* 8 * , 462 ( 1962 ) ; , see also the appendix of j. chihara , j. phys . : condens . matter * 3 * , 8715 ( 1991 ) . in principle
one could also define @xmath62 for all the electrons , including the core - electrons , but since the core electron density changes very little , they are not plotted for the sake of clarity ) .
y. rosenfeld and n.w .
ashcroft , phys .
a * 20 * , 1208 ( 1979 ) . j. hafner , _ from hamiltonians to phase diagrams _
, ( springer verlag , berlin , ( 1987 ) ) . in this paper
we used the form described in s. ichimaru and k. utsumi , phys .
b. * 24 * , 7385 ( 1981 ) .
a. a. louis and n. w. ashcroft , phys .
* 81 * , 4456 ( 1998 ) ; see also , a. a. louis and n. w. ashcroft , j. non - cryst .
solids , * 250 - 252 * , 9 ( 1999 ) .
w.a . curtin and n.w .
ashcroft , phys .
. lett . * 59 * , 2385 ( 1987 ) .
a.r . denton and n.w .
ashcroft , phys .
a * 39 * , 426 ( 1989 ) .
ramakrishnan and m. yussouff , phys .
b * 19 * , 2775 ( 1979 ) .
a.d.j . haymet and d.w .
oxtoby , j. chem . phys . * 74 * , 2559 ( 1981 ) . h. xu and j .- p .
hansen , phys .
e * 57 * , 211 ( 1998 ) .
h. xu , j.p .
hansen , and d. chandler , europhys .
lett . * 36 * , 419 ( 1994 ) .
w. j. nellis , a. a. louis , n. w. ashcroft , phil .
a. * 356 * 119 ( 1998 ) .
a. de kuijper , w. l. vos , j.l .
barrat , j.p .
hansen , and j.a .
schouten , j. chem .
* 93 * , 5187 ( 1990 ) .
d.w . marr and a.p .
gast , phys .
e * 47 * , 1212 ( 1993 ) .
a.r . denton , g. kahl , and j. hafner , j. non - cryst .
solids * 250 - 252 * 15 ( 1999 ) .
j. a. anta , b. j. besson , and p. a. madden , phys .
b * 58 * , 6124 ( 1998 ) . j. a. anta and p. a. madden , j. phys . : condens . matter , * 32 * , 6099 ( 1999 ) .
g. a. de wijs , b. pastore , a. selloni , and w. van der lugt , phys .
lett . * 75 * , 4480 ( 1995 ) . | we combine techniques from quantum and from classical density functional theory ( dft ) to describe electron - ion mixtures . for homogeneous systems ,
we show how to calculate ion - ion and ion - electron correlation functions within chihara s quantum hypernetted chain approximation , which we derive within a dft formulation .
we also sketch out how to apply the dft formulation to inhomogeneous electron - ion mixtures , and use this to study the electron distribution at the liquid - solid interface of al .
pacs numbers:71.22.+i,61.10.-i,61.20.gy,61.12.bt |
it is well known that the properties of biological vesicles and membrane cells strongly depend on the their constituent blocks , usually composed of phospholipid bilayers with embedded proteins .
these molecules are in general anisotropic ( rod or plate - like ) and the demixed states usually have liquid - crystal symmetries , such as isotropic ( i ) , nematic ( n ) or biaxial nematic ( b ) symmetries . for certain conditions these complex mixtures of biomolecules phase separate , creating regions rich in different species and consequently changing the membrane curvature .
there is much experimental evidence of demixing transitions in monolayers and bilayers of mixed anisotropic biomolecules @xcite .
the adsorption of a large variety of mixtures of rod - like molecules in langmuir monolayers has been extensively studied both experimentally and theoretically .
many of these works focus on the chemical and thermodynamic conditions for which the monolayers become spatially heterogeneous , i.e. when the mixture demixes in different phases usually possessing liquid - crystal ordering @xcite .
finally many experiments on the adsorption of rod - like colloidal particles at the interfaces separating two immiscible fluids showed the propensity of these particles to self - assemble into clusters of different geometries @xcite .
the degree of adsorption of these particles at the interface and their relative orientation with respect to it strongly depend on their chemical compositions .
this in turn can modify their wetting properties and consequently the effective capillary forces acting between particles .
the resulting effect is the existence of anisotropy in the pair interaction potential which forces the particles to self - assemble into clusters @xcite .
when colloids with very different chemical properties are adsorbed at the interface they usually phase separate into phases with different composition of species .
a recent experimental work showed how demixing of adsorbed colloids strongly modifies their self - assembling properties @xcite .
all the systems discussed above share the following properties : ( i ) they are mixtures of anisotropic particles , ( ii ) the degrees of freedom of their centres of mass are strongly restricted , usually resulting in an effective two - dimensional fluid , and ( iii ) the particle axes can rotate in 3d but with certain restrictions which depend on the degree of particle adsorption on the monolayer , bilayer or interface .
the main motivation of the present work is the formulation of a very simple model for a binary mixture of anisotropic particles ( specifically a mixture of rods and plates ) which allows a detailed study of the conditions ( particle aspect ratios , degree of adsorption ) under which these mixtures demix into two different phases . to this purpose
we choose particles to have a board - like shape and interact through a hard - core repulsion .
also , for simplicity , we use the zwanzig approximation to account for the orientational degrees of freedom , which are restricted to be three .
finally , we will use a mean - field density functional ( df ) based on the fundamental - measure theory ( fmt ) , derived for the present model in the late 90 s @xcite and more recently implemented to calculate the phase diagrams of rod and plate board - like particles @xcite .
monolayers of one - component rods or plates were recently studied within this theory , considering uniaxial @xcite and biaxial @xcite particle geometries . in the latter work , phase diagrams were calculated as a function of a geometric parameter @xmath3 $ ] that measured particle shape , with @xmath4 for uniaxial rod and plate geometries , respectively .
one - component monolayers of prolate or oblate freely - rotating ellipsoids were also recently studied via the parsons - lee df and molecular - dynamics simulations @xcite .
in particular , the effect of orientational restriction of ellipsoids on the orientational properties of monolayers was studied .
the ground states of monolayers of hard ellipsoids interacting through a quadrupole pair - potential were recently found @xcite .
apart from the t - like configurations , three more particle orientations were predicted to be stable . here we extend our previous model @xcite to a binary mixture of uniaxial rods and plates .
board - like - shaped rods and plates are taken to be symmetric : although they have different eccentricities ( prolate and oblate ) , their volumes and their aspect ratios ( ratio between major and minor particle edge - lengths ) are taken to be the same .
this choice of shape geometries is motivated by their extensive use in studies of the biaxial - nematic ( b ) phases stability with respect to nematic - nematic ( n - n ) phase separation in binary @xcite and polydisperse @xcite mixtures of rods and plates .
we are interested in the effect of particle adsorption on the phase behaviour of the mixture , in an effort to elucidate ( i ) the propensity of the system to phase separate into different phases , ( ii ) the nature ( second vs. first order ) of the n - b transition when the adsorption strengths are changed , ( iii ) the relative stability of the b phase with respect to the n or other non - uniform phases , and ( iv ) the representative phase diagrams of the system ( calculated for certain selected values of model parameters ) . in general
we found a rich phase behaviour with the presence of two disconnected demixed n - b phase transitions , one located at low pressures , with a b phase rich in rods , and the other at very high pressures , with a b phase rich in plates ( although b - b demixing also occurs in a narrow range of pressures ) .
when plates are strongly adsorbed , the mixture exhibits a strong first - order phase transition .
different demixing scenarios depend on the relative values of plate and rod adsorption coefficients .
when adsorption of rods is large as compared to that of plates , the b - phase stability is greatly enhanced , the n - b transition is always of second order , and no demixing occurs . the article is organized as follows . in sec .
[ model ] we introduce the model and the theoretical tools used to perform the calculations . in sec .
[ results ] we summarize all the results obtained , with subsections presenting different mixtures with various relative adsorption strengths and particle aspect ratios . finally some conclusions are drawn in sec .
[ conclusions ] .
we use the zwanzig model for a binary mixture of prolate ( rods ) and oblate ( plates ) board - like particles with centres of mass lying on the @xmath5 plane and with main axes pointing along the @xmath6 directions .
the edge lengths of species @xmath7 ( @xmath8 for rods and @xmath9 for plates ) are represented through the tensor @xmath10 with @xmath11 the kronecker delta symbol , while @xmath12 and @xmath13 are the particle sizes parallel and perpendicular to the main particle axis , respectively .
therefore , particles are uniaxial parallelepipeds with a square section of area @xmath14 .
only symmetric mixtures will be studied , namely those composed of particles with the same volume ( which is set to unity , @xmath15 ) , and with aspect ratios @xmath16 and @xmath17 related by @xmath18 .
thus the edge lengths of the different species are calculated as @xmath19 , @xmath20 .
[ fig1 ] for a schematic representation of our model .
the theory used in the present calculations is the uniform limit of the fmt density functional obtained by applying the dimensional crossover property .
this feature allows to correctly transform the functional from 3d to 2d by assuming a 3d density profile , for species @xmath7 and orientation along the @xmath21-axis , of the form @xmath22 , i.e. imposing that the particle centres of mass are constrained to the flat surface perpendicular to @xmath23 . when this density is substituted into the 3d version of the excess free - energy functional , the resulting functional depends on the ( constant ) 2d number densities @xmath24 where @xmath25 is the total number density
, @xmath26 is the molar fraction of species @xmath7 , while @xmath27 is the fraction of species @xmath21 with main axes pointing along the @xmath21-direction .
obviously these quantities fulfill the following constraints : @xmath28 and @xmath29 or alternatively @xmath30 and @xmath31 , with @xmath32 the number density of species @xmath7 . the resulting excess free - energy density in reduced thermal units @xmath33 depends on the following weighted densities : @xmath34 and has the explicit form @xmath35 where @xmath36 is the boltzmann factor and @xmath37 the total area of the system .
we note that the weighted density @xmath38 is just the total packing fraction , @xmath39 , of the binary mixture . the ideal part
, @xmath40 is , as usual @xmath41 and the effective interaction between species @xmath7 , with projected area on the @xmath5 plane @xmath42 , and the surface , is accounted for by an external potential contribution to the free - energy density : @xmath43 note that this contribution is proportional to the projected particle areas , and that we allow for the possibility that the adsorption strengths @xmath44 be dependent on species @xmath7 . to find the equilibrium orientational properties of the fluid we minimize the total free - energy density @xmath45 with respect to the fractions @xmath27 .
these can be related to the uniaxial nematic order parameters , @xmath46 which measure the order about the direction perpendicular to the surface , and to the biaxial nematic order parameters , @xmath47 which measure the degree of biaxial order .
note that the factor @xmath48 in the definition of @xmath49 is necessary to take account of the orthogonality of the plate and rod main axes when their projections have the same orientations .
we have calculated the phase diagram by searching for possible demixing transitions through ( i ) the equality between chemical potentials of species @xmath7 , @xmath50 , and ( ii ) the equality between the pressures of the demixed phases .
the latter can be calculated as @xmath51 ( in reduced thermal units ) .
these calculations are equivalent to finding the coexisting molar fractions of the demixed phases through the double - tangent construction of the gibbs free - energy density , @xmath52 as a function of the molar fraction @xmath53 at constant value of @xmath54 .
this constraint allows us to find the total number density @xmath55 as a function of @xmath56 at fixed @xmath54 , while all the quantities @xmath27 should be calculated for the same values @xmath57 from the set of equations @xmath58 . to find the location of second - order phase transitions to biaxial phases we use a bifurcation analysis of the total free - energy density @xmath59 with respect to the ( small ) order parameters @xmath49 . for details of these calculations
see appendix [ append ] . in the rest of the manuscript we use dimensionless densities @xmath60 and pressures @xmath61 .
this section is devoted to the study of the phase diagram topologies as a function of the adsorption strengths @xmath62 and the aspect ratio @xmath0 of the mixture .
we firstly studied the monolayer of rods and plates with zero adsorption to the surface by setting @xmath63 .
the phase diagram obtained for a mixture with @xmath1 is shown in fig .
[ fig2](a ) in the pressure - composition plane . the dashed line , which departs at its lowest pressure from the left vertical axis ( @xmath64 ) , represents a continuous n - b phase transition . from low to high pressures , but below the n - b spinodal , the configuration of plates changes from a nearly equimolar composition of their three species to that in which the species with the largest ( square ) projected area , equal to @xmath65 , has the lowest composition while the other two species , with rectangular projected shapes of aspect ratio @xmath0 and surface area @xmath66 , have equal compositions .
the plate axes of the latter species are parallel to the surface but their rectangular sections are yet randomly oriented in 2d .
this phase is a planar n with a negative uniaxial order parameter which decreases with pressure .
the configuration of rods at low pressures always exhibits a preferential alignment perpendicular to the monolayer , resulting in a higher proportion of projected ( small ) squares of surface area @xmath67 .
the other two species of rods , having rectangular shapes , aspect ratio @xmath0 , and surface area @xmath68 , have again the same composition
. thus the uniaxial order parameter of rods is always positive and increases with pressure .
the authors of the recent work @xcite showed that the uniaxial order parameter @xmath69 of one - component rods on a surface increases linearly from zero as a function of @xmath70 .
as the pressure increases , the fraction of plates with their axes oriented parallel to the surface and that of rods oriented perpendicular to it become larger , and the uniaxial order parameters tend to @xmath71 and @xmath72 , respectively . at a certain pressure ( which depends on molar fraction )
, the @xmath5 orientational symmetry is broken in a continuous fashion , and the projected rectangular species for both , rods and plates , begin to align along a preferential direction , say the @xmath73 direction . from this pressure the b phase becomes stable and the b order parameters continuously increase from zero . see fig .
[ fig0 ] for a sketch of projected shape configurations for the different phases .
[ fig2 ] indicates that biaxial ordering is promoted by the plates : the n - b spinodal , departing from the left vertical axis ( @xmath74 ) , is a monotonically increasing function of @xmath56 and possesses an asymptote at @xmath75 .
the latter means that one - component monolayers of rods do not exhibit b ordering . from now on we refer to this spinodal as the _ plate _ n - b _
spinodal_. at very high pressures , most of the plate axes align parallel to the monolayer , while those of rods are perpendicular to it .
thus the system can be approximated by a 2d mixture of zwanzig hard rectangles ( of area @xmath66 and aspect ratio @xmath0 ) and parallel hard squares ( of area @xmath67 ) .
this 2d mixture demixes into two phases , each one rich in one of the species , as shown in fig .
[ fig2 ] . the critical point of the demixing transition is above the n - b spinodal , meaning that b - b coexistence takes place in some pressure interval . for pressures above the crossing point between the right demixing binodal and the n - b spinodal
the system phase separates into a b phase , rich in plates , and a n phase , rich in rods .
finally the dotted line in fig .
[ fig2 ] represents the spinodal instability of the uniform phases with respect to density modulations , which corresponds to the presence of stable non - uniform phases ( see appendix [ append1 ] for details on these calculations ) .
we can see that the n / b - b demixing is , except for a small interval of pressures , metastable with respect to transitions to non - uniform phases . fig .
[ fig2 ] ( b ) shows the phase diagram when the adsorption strengths are still relatively small : @xmath76 .
we can observe that the phase diagram topology is similar to that of the preceding case , except that now there appears a region , close to @xmath75 and bounded by a dashed line , where the b phase becomes stable .
from now on this spinodal will be called the _ rod _ n - b _ spinodal_. the total free - energy is lowered when the fraction of rods with main axes parallel the monolayer increases , since this is proportional to the projected areas . in turn , these rods exhibit two continuous n - b and b - n transitions : b ordering increases with pressure , reaches a maximum , then decreases and finally disappears altogether .
this reentrant behaviour of the b phase with pressure can be explained as follows .
when the loss in free - energy given by a preferential adsorption and further alignment of rods with projected rectangular shapes can not compensate the free - energy increase due to the large excluded volumes between rectangular projected species , as compared to those of small squared species , the most favored configuration of particles is that of rods pointing perpendicular to the monolayer .
as the total amount of rods lying on the surface becomes small a b - n transition takes place .
this behaviour was already found in monolayers of one - component zwanzig rods with @xmath77 and zero adsorption .
the values of @xmath78 were found to be 21.3 and 12 from the spatially continuous @xcite and discrete lattice models @xcite .
when the continuous orientational degrees of freedom are restored , this transition disappears @xcite .
however if particles can rotate freely except for a small solid angle with respect to the surface normal , the b phase again becomes stable @xcite .
thus in real situations when the surface / interface promotes a preferential adsorption of rods with their axes parallel to the surface the b phase will certainly become stable .
now we proceed to describing the phase behaviour of monolayers of rods and plates with relatively high adsorption strengths , specifically those with @xmath79 .
the phase diagram in the pressure - composition and total density - composition planes are shown in fig . [ fig3](a ) and ( b ) , respectively .
the most salient features that can be observed from the figure are : ( i ) the presence of strong n - b demixing at pressures located between two tricritical points , both lying on the rod - n - b spinodal which ends at @xmath75 , and ( ii ) the existence of a strong first order n - b transition departing from @xmath80 and ending in a tricritical point located at the plate - n - b spinodal .
panel ( b ) shows the density of the coexisting phases along all these binodals and spinodals .
we first describe the n - b demixing .
[ fig3](b ) shows that the values of the coexisting densities in the demixed phases are similar , with b being the densest phase , rich in rods ( two different pairs of coexisting densities are shown with circles and squares ) .
however the packing fraction has the opposite behaviour ( see the black line of fig .
[ fig4 ] ) : the phase with the highest packing fraction is the n phase , rich in plates . this behaviour ( nearly the same coexisting densities but very different composition ) is typical in entropy - driven demixing .
note that plates in both coexisting n and b phases have always a positive uniaxial order parameter along the binodals [ see fig .
[ fig5](b ) ] , so the fraction of projected large squares ( cross - section of plates ) is relatively high .
the projected rectangles corresponding to rods lying on the surface also have a relatively high fraction ( see the negative values of @xmath69 in ( a ) along the binodals , except for a region close to the upper tricritical point ) , as compared to that of the square projected areas ( when rods point perpendicular to the monolayer ) .
as usually occurs in entropy - driven demixing , the total excluded area between the rod - projected rectangles and the plate - projected squares is lowered if the demixed phases are rich in one of the species .
the high proportion of large squares is the reason behind the high packing fraction values of the coexisting n phase , as compared to that of the b phase .
it is interesting to note the highly non - monotonic behaviour of the packing fraction along the demixing binodals ( see fig . [ fig4 ] ) , a direct consequence of the dependence of @xmath39 not only on @xmath70 and @xmath56 but also on the order parameters @xmath81 .
finally fig .
[ fig6 ] shows the evolution of the biaxial order parameters along the demixing binodals .
the biaxial order of both species along the coexisting b phase rapidly increases and saturates to its highest value ( unity ) as the mixture gets away from the lower tricritical point . at some point
they invert their monotonicity and decrease to zero at the upper tricritical point . from the other side of the phase diagram ( @xmath80 ) , and for intermediate values of pressure ,
a strong first order n - b phase transition takes place , as can be inferred from the large coexisting density gap in fig .
[ fig3](b ) .
this transition ends in a tricritical point located at the plate - n - b spinodal , and is driven by the reorientation of plates .
note the positive and negative values of @xmath82 , corresponding to the coexisting n and b phases , respectively [ see fig .
[ fig5](b ) ] .
the rods are now mainly oriented perpendicular to the monolayer ( perfectly oriented in the n phase , and with a small degree of orientation in the b phase ) .
it is interesting to note that the coexisting molar fractions are now similar ( with the b phase slightly rich in plates ) , while the coexisting densities are very dissimilar ( b being the densest phase ) .
this transition is driven by a differential change in free - energy from a n phase with a high fraction of adsorbed larges squares ( plate axes perpendicular to the monolayer ) to a b phase with a high fraction of projected rectangles ( corresponding to plates with their axes lying on the monolayer and pointing along @xmath73 ) . when we follow a constant pressure path from the n to the b coexisting phases , the free - energy contribution corresponding to the external potential increases , while that coming from the entropic interaction part is lowered ( because the particle excluded areas decrease ) .
the differential change in the total adsorbed area of particles is huge , so the transition becomes strongly first order .
again the coexisting packing fractions are inverted : that of the b phase is lower ( see the gray solid curve in fig . [ fig4 ] ) . at very high pressures
the same n / b - b demixing transition ending in a critical point takes place [ similar to the cases @xmath63 and @xmath83 .
finally we calculated the instability of uniform phases with respect to non - uniform density modulations . the pressures and densities
at which these instabilities occur are plotted as a function of @xmath56 in fig .
[ fig3](a ) and ( b ) , respectively .
we can see that the lower tricritical point is located above this curve , suggesting that all demixing transitions are metastable with respect to transitions or demixing between non - uniform phases .
when the molar fraction of plates with axes perpendicular to the surface is high due to their large surface adsorption , their square cross - sections may crystallize in a simple square lattice at a certain pressure .
if pressure is increased beyond this value , plates will reorient their axes parallel to the monolayer and consequently crystal ordering could be destabilized with respect to a uniform or nonuniform phase exhibiting b ordering . on the other side of the phase diagram , where the molar fraction of rods is high and the b phase is stable , the most likely scenario is that of b - smectic or b - columnar phases , where the projected rectangles exhibit two - dimensional smectic or columnar arrangements .
these open questions should be settled out by performing df minimisation with respect to non - uniform density profiles , @xmath84 , and search for possible coexistences , a formidable task that we leave for future studies . in this section
we study the effect of adsorption asymmetry on the phase behaviour of monolayers of rods and plates . to this purpose
we have chosen the adsorption strengths as @xmath85 , i.e. rods are more strongly adsorbed on the surface than plates .
to better compare the results obtained with those described in the preceding sections , we again set the aspect ratio to @xmath1 .
the phase diagram for this mixture is plotted in fig .
[ fig7 ] in the ( a ) pressure - composition and ( b ) density - composition variables .
the main features we can extract from these results are : ( i ) when the rods are strongly adsorbed on the surface , the lower part of the rod - n - b spinodal meets the plate - n - b spinodal at intermediate compositions , creating a monotonic , fully connected spinodal curve over the whole composition interval .
( ii ) there is a lower tricritical point located on this curve , above which a demixing transition occurs .
( iii ) the demixing transition coalesces with the strong first - order transition driven by plates at higher pressures ( the one ending at @xmath80 ) .
( iv ) the upper part of the rod - n - b spinodal joins the right part of the plate - n - b spinodal , creating an island of n phase stability .
( v ) the entropic n / b - b demixing at high pressure remains invariant , which confirms the fact that the system behaves like a 2d mixture of squares and rectangles .
[ fig8 ] shows the strong non - monotonic behaviour of the packing fraction along the coexistence binodals , with the presence of a large loop .
( vi ) the spinodal for the uniform phase instability with respect to density modulations is located again below the tricritical point . also ,
packing fraction inversion ( with respect to density ) does occur , with n being the densest phase . figs . [
fig9](a ) and ( b ) show the uniaxial order parameters along the binodals , with a b phase of rods and plates having axes lying on the monolayer , and a n phase with plate axes pointing perpendicular to it .
interestingly , the rods in n phase are oriented parallel to the surface , although to a lesser degree .
the insets show the completely saturated ordering of particles ( perpendicular and parallel to the surface for rods and plates , respectively ) along the boundaries limiting the island of n phase stability . finally fig .
[ fig10 ] shows the biaxial order parameters along the coexisting binodals , which have the usual behaviour : a rapid increase as the system gets away from the tricritical point , and then saturation to perfect biaxial ordering .
[ fig11](a ) shows the phase diagram of a mixture with @xmath1 , but with larger asymmetry in their adsorption strengths , @xmath86 .
now the plates are slightly adsorbed on the surface , while rods are strongly adsorbed .
we can see that : ( i ) the demixing and first - order n - b transitions ( present in the @xmath87-mixtures ) are substituted by continuous transitions , with the n - b spinodal now being a monotonically decreasing function of @xmath56 .
thus the region of b stability is greatly enhanced .
( ii ) the island of n - phase stability and the n / b - b demixing at high pressure remain as before .
[ fig11](b ) shows the phase diagram of the mixture @xmath88 , i.e. plates and rods are strongly and slightly adsorbed , respectively .
note the presence of a strong first - order n - b transition driven by the desorption of plates at intermediate pressures .
as before , this transition ends in a tricritical point located on the plate - n - b spinodal .
the b phase , rich in rods , is again stable inside a island bounded by the rod - n - b spinodal .
no demixing was found in this part of the phase diagram , with the n - b transition being of second order .
the n / b - non - uniform - phase spinodals are now discontinuous and located above the b phase of rods .
the last study concerns the phase behaviour of mixtures with higher aspect ratios , in particular those with @xmath90 and 40 . as shown in fig .
[ fig12 ] the phase - diagram topologies are similar , but there is an important difference : now the lower tricritical point is always located below the spinodal instability to non - uniform phases . thus there is always a range of pressures , which increases with @xmath0 , for which demixing into a n phase and a b phase rich in rods is stable . in figs .
[ fig12 ] ( a ) and ( b ) the phase diagrams for @xmath90 and @xmath91 , panel ( a ) , and @xmath92 , panel ( b ) , are shown .
note that the former mixture is symmetric with respect to adsorption .
the upper boundary of the rod - n - b spinodal meets the n - b - plate spinodal at intermediate compositions .
the phase behaviour includes : n - b demixing between two tricritical points ( the lower one departing from the rod - n - b spinodal ) , and a strongly first - order n - b transition driven by the desorption of plates , beginning at @xmath80 and ending in a tricritical point located at the plate - n - b spinodal .
this point and the upper critical point of the demixing transition are now very close to each other .
an island of n - phase stability exists at high pressure as a consequence of the coalescence between the right part of the plate - n - b spinodal and the upper part of the rod - n - b spinodal .
finally , n / b - b demixing of the effective two - dimensional mixture of squares and rectangles at high pressure is also present .
the phase diagram topology for the second mixture studied , that with @xmath93 , is very similar to the for @xmath1 described before .
again there is an important difference , namely the existence of stable n - b demixing in some pressure range .
also the b phase , rich in rods , is stable in a rather large region of the phase diagram .
we end this section by showing the phase diagram of a mixture with @xmath94 and @xmath76 , in figs . [ fig12](c ) and ( d ) .
we concentrate on some details of the phase diagram topology not found before : ( i ) the presence of a lower critical point ( at low pressures and close to the lower n - b tricritical point ) above which b - b demixing takes place in a rather small range of pressures [ see inset of panel ( d ) ] .
( ii ) the two distinct tricritical points , located close to each other in previous cases [ e.g. the case @xmath90 and @xmath91 ] now coalesce into a single azeotropic point [ see panel ( d ) ] .
we have systematically studied the phase behaviour of mixtures of rods and plates adsorbed on a monolayer . in our model , the particle centres of mass are taken to freely move on the surface , while particles can rotate in 3d within the restricted - orientation , zwanzig approximation .
adsorption of the particle surfaces on the monolayer is mimicked through an attractive external potential proportional to the area of the particle surface contact , while the strengths of this interaction , @xmath62 , depend on the species type .
rods and plates were taken to be symmetric , i.e. with the same volume and same aspect ratio @xmath0 .
a fmt - based df , adapted to the present constrained geometry , was minimised , and phase diagrams of mixtures with @xmath1 , 20 and 40 and different values of @xmath62 were calculated .
the main results can be summarized as follows .
( i ) when both adsorption strengths @xmath96 are zero or small , rods and plates orient perpendicular and parallel to the surface , respectively .
the degree of orientation continuously increases with pressure , and at some value ( which depends on the composition ) a second - order n - b transition occurs , at which the plate axes orient along a director lying on the surface .
although rods also exhibit biaxial ordering , this transition is governed by the orientational symmetry breaking of plate axes , and consequently the plate - n - b spinodal is a monotonically increasing function of @xmath56 . (
ii ) at some value of @xmath97 and , starting at the @xmath75 vertical axis at low pressures , there appears an island of b phase stability enclosed by a rod - n - b spinodal which is disconnected from the plate - n - b spinodal .
the n - b transition is of second order and is governed by the alignment of rods with axes on the monolayer . as pressure increases rods prefer to align perpendicular to the surface , and biaxial ordering disappears .
( iii ) when rods and plates are symmetrically adsorbed on the monolayer ( @xmath98 ) and the strengths are relatively large , two tricritical points appear on the rod - n - b spinodal ; between these two points n - b demixing takes place , with demixed phases rich in one of the components .
the b and n coexisting phases are mostly populated by rods and plates , respectively . also , there is a strongly first - order n - b phase transition with a large density gap starting at the @xmath80 vertical axis .
this transition ends in a tricritical point located on the plate - n - b spinodal , and is driven by the desorption of the largest cross - section of the plates ( corresponding to axes perpendicular to the monolayer ) .
finally at very high pressures , when the degree of order is high , the system effectively becomes a two - dimensional mixture of squares ( the smallest projected section of rods ) and rectangles ( the smallest projected sections of plates ) , which demix into a b phase , rich in plates , and a n phase , rich in rods .
this demixing transition ends in a critical point above which there exists a rather narrow b - b demixing region .
( iv ) when the adsorption of particles is very asymmetric and @xmath99 , the lowest boundary of the rod - n - b spinodal connects with the left part of the plate - n - b spinodal , forming a monotonically - decreasing spinodal over the whole range of compositions .
the n - b transition is always of second order .
also the upper boundary of the rod - n - b spinodal connects with the right part of the plate - n - b - plate spinodal , forming an island of n stability .
the highest pressure n / b - b demixing remains invariant .
( v ) when the adsorption of particles is very asymmetric and @xmath100 , the strongly first - order n - b transition governed by desorption of plates and their b ordering is present up to high molar fractions , while the n - b transition , governed by orientation of rods , is of second order and the n - b demixing disappear .
the region of stability of the b phase ( enclosed by the rod - n - b spinodal ) is reduced as @xmath97 becomes smaller .
we also calculated the spinodal instability of uniform phases with respect to density modulations with different symmetries ( smectic , columnar or crystalline ) .
we found that a b phase rich in rods is stable over a relatively large interval of pressures , while the strong n - b phase transition is always metastable .
for @xmath90 and 40 there exists a range of pressures for which n - b demixing is stable .
of course n / b - b demixing at high pressure , ending in a lower critical point , is also metastable .
we note that demixing regions could be made wider if we chose shape - asymmetric mixtures ( different volumes and/or different aspect ratios ) , and consequently the regions of n and b phase stability could be modified .
even the orientational symmetries of the demixed phases could be different for asymmetric mixtures as shown in theoretical calculations of freely - rotating rod - plate onsager mixtures @xcite .
we are confident that the results presented in this work will be qualitatively similar if we remove the restricted orientation approximation and consider the free rotation of particle axes .
computer simulations of binary mixtures adsorbed on a flat monolayer could confirm this conclusion .
mc simulations of 2d mixtures of rods on a lattice show an interesting phase behavior @xcite . when the aspect ratio of the longer rods is 7 there
exist two i - n and n - i transitions as the density of longer rods is increased while that of the shorter rods is fixed bellow a certain critical density .
this behavior resembles that of the present rod - plate mixture for which two n - b and b - n transitions take place at fixed composition as the pressure is increased and the adsorption strengths of rods is high enough .
thus it would be interesting to perform dft calculations on mixtures of adsorbed rods to find the differences and similarities between monolayers and strictly 2d hard rod mixtures .
also interesting are the similarities that the present system shares with the phase behaviour of monolayers of biaxial particles studied in @xcite . in that case the phase diagrams , in the density - biaxial parameter plane , present a n - b spinodal , completely analogous to that of the present system , if we replace the biaxial parameter by molar fraction . moreover , by increasing the aspect ratio of biaxial rods in the previous study , one obtains an island of b phase stability ; a similar effect is found in the present study when adsorption strength is increased . above a particular value for the largest aspect ratio of the biaxial particles , this island coalesces with the n - b spinodal , again a behaviour similar to the one found in the present study if the adsorption strength is increased beyond some critical value .
additionally , we may ask ourselves how particle biaxiality would affect the present phase behaviour .
if particles are biaxial , we would expect the n - b spinodal to shift to higher pressures , favouring the stabilisation of non - uniform phases . on the other hand
, biaxiality would also reduce the island of b stability in the rod - rich part of the phase diagram .
finally it would also be possible that particle biaxiality reduced demixing gaps , because particle projections become similar with biaxiality .
the constrained minimisation of the free - energy density with respect to the variables @xmath27 gives the following set of equations : @xmath101 where @xmath102 using the definition of biaxial order parameter @xmath49 from ( [ biaxial ] ) , we arrive at @xmath103 expanding ( [ itera1 ] ) with respect to @xmath49 ( obviously the functions @xmath104 depend on @xmath105 ) up to first order gives a system of two equations which has a nontrivial solution only if the number density is such that @xmath106,\end{aligned}\ ] ] where @xmath107 is the solution of ( [ itera ] ) for @xmath108 , and the functions @xmath104 are calculated from ( [ uno])-([tres ] ) , with all the weighted densities depending only on @xmath107 , as @xmath109 and @xmath110,\nonumber\\&&\\ & & n_2=\eta=\rho\sum_s x_s\left[\kappa^{s/3}+\gamma_{sz}\left(\kappa^{-2s/3}-\kappa^{s/3}\right)\right ] \end{aligned}\ ] ] thus we need to solve a system of two non - linear equations to find the equilibrium values of @xmath107 , and consequently the number density @xmath25 at which a bifurcation from the n to the b phase occurs .
we have calculated the spinodal instability of uniform phases with respect to spatial inhomogeneities through the divergence of the structure factor .
the fourier transform of the direct correlation functions , calculated through the second functional derivative of the functional , reads @xmath111 where @xmath112 implies symmetrisation with respect to the pair of indexes @xmath113 , and the weighted densities @xmath114 correspond to the stable uniform n or b phases , with orientational order parameters calculated from the minimisation of the corresponding free energies .
the fourier transforms of the weighting functions are @xmath115 where @xmath116 and @xmath117 .
we define the @xmath118 structure factor matrix @xmath119 where @xmath120 and @xmath121 if @xmath9 , and @xmath122 and @xmath123 if @xmath8 ( with the corresponding relabelling @xmath124 and @xmath125 for @xmath56 , @xmath73 and @xmath23 , respectively ) . evaluating the determinant of this matrix
@xmath126 $ ] at the wave vectors @xmath127 or @xmath128 ( corresponding to inhomogeneities along or perpendicular to the nematic director , respectively ) , we found the corresponding values at the spinodal instabilities , @xmath129 and @xmath130 , as the values for @xmath131 and @xmath70 where the absolute minimum of @xmath132 as a function of @xmath131 becomes zero for the first time .
this is equivalent to solving the pair of equations @xmath133 100 p. sharma , a. ward , t. gibaud , m. f. hagan , and z. dogic , nature * 513 * , 77 ( 2014 ) .
p. a. rice and h. m. mcconnell , proc .
usa * 86 * , 6445 ( 1989 ) .
j. p. hagen and h. m. mcconnell , bba - biomembranes * 1329 * , 7 ( 1997 ) .
s. l. keller , w. h. pitcher iii , w. h. huestis , and h. m. mcconnell , phys .
rev . lett . * 81 * , 5019 ( 1998 ) .
s. l. keller , and h. m. mcconnell , phys .
lett . * 82 * , 1602 ( 1999 ) .
s. baukina , e. mendez - villuendas , and d. peter tielemann , j. am .
soc . * 134 * , 17543 ( 2012 ) j. zanghellini , f. wodeli , and h. h. von grnberg , j. theor . biol . * 264 * , 952 ( 2010 ) .
h. e. ries jr . , and d. c. walker , j. coll .
imp u tok * 16 * , 361 ( 1961 ) .
r. k. smith , s. m. reed , et .
al . , j. phys .
b * 105 * , 1119 ( 2001 ) .
t. martynkski , r. hertmanowski , and d. bauman , liq .
* 28 * , 2001 .
h. mller , r. zentel , a. janshoff , and m. janke , langmuir * 22 * , 11034 ( 2006 ) .
i. kalashnikova , h. bizot , p. bertoncini , b. cathala , and i. capron , soft matter * 9 * , 952 ( 2013 ) .
d. kim , w. d. kim m. s. kang , s. h. kim , d. c. lee , nano letters * 15 * , 714 ( 2015 ) .
e. kim , k. stratford , r. adhikari , and m. e. cates , langmuir * 24 * , 6549 ( 2008 ) .
s. r. vora , b. bognet , h. s. patanwala , f. chinesta , a. w. k. ma , langmuir * 31 * , 4663 ( 2015 ) .
g. b. davies , t. krger , p. v. coveney , j. harting , and f. bresme , adv .
* 26 * , 6715 ( 2014 ) .
a. n. sekar iyengar , a. datta , and m. s. jakani , phys . rev .
e * 92 * , 032907 ( 2015 ) .
j. a. cuesta and y. martnez - ratn , phys .
* 78 * , 3681 ( 1997 ) .
y. martnez - ratn , phys .
e * 69 * , 061712 ( 2004 ) y. martnez - ratn , s. varga , and e. velasco , j. chem . phys . * 140 * , 204906 ( 2014 ) .
m. gonzlez - pinto , y. martnez - ratn , e. velasco , and s. varga , phys .
* 17 * , 6389 ( 2015 ) .
s. varga , y. martnez - ratn , e. velasco , g. bautista - carbajal , and g. odriozola , phys .
phys . * 18 * , 4547 ( 2016 ) .
t. heinemann , m. antlanger , m. mazars , s. h. l. klapp , and g. kahl , j. chem .
144 * , 074504 ( 2016 ) .
r. van roij and b. mulder , j. phys .
ii france * 4 * , 1763 ( 1994 ) .
s. varga , a. galindo , and g. jackson , phys .
e * 66 * , 011707 ( 2002 ) .
s. varga , a. galindo , and g. jackson , j. chem . phys . *
117 * , 10412 ( 2002 ) . y.
martnez - ratn , and j. a. cuesta , phys .
lett . * 89 * , 185701 ( 2002 ) . y. martnez - ratn , and j. a. cuesta , j. chem
. phys . * 118 * , 10164 ( 2003 ) .
m. oettel , m. klopotek , m. dixit , e. empting , t. schilling , and h. hansen - goos , arxiv:1605.02903 ( 2016 ) . h. h. wensink , g. j. vroege , and h. n. w. lekkerkerker , phys .
e * 66 * , 041704 ( 2002 ) .
j. kundu , j. stilck , and r. rajesh , epl * 112 * 66002 ( 2015 ) . | we theoretically study the phase behaviour of monolayers of hard rod - plate mixtures using a fundamental - measure density functional in the restricted - orientation ( zwanzig ) approximation .
particles can rotate in 3d but their centres of mass are constrained to be on a flat surface .
in addition , we consider both species to be subject to an attractive potential proportional to the particle contact area on the surface and with adsorption strengths that depend on the species type .
particles have board - like shape , with sizes chosen using a symmetry criterion : same volume and same aspect ratio @xmath0 . phase diagrams were calculated for @xmath1 , 20 and 40 and different values of adsorption strengths . for small adsorption strengths the mixtures exhibit a second - order uniaxial nematic - biaxial nematic transition for molar fraction of rods @xmath2 . in the uniaxial nematic phase the particle axes of rods and plates are aligned perpendicular and parallel to the monolayer , respectively . at the transition ,
the orientational symmetry of the plate axes is broken , and they orient parallel to a director lying on the surface . for large and equal adsorption strengths the mixture demixes at low pressures into a uniaxial nematic phase , rich in plates , and a biaxial nematic phase , rich in rods .
the demixing transition is located between two tricritical points .
also , at higher pressures and in the plate - rich part of the phase diagram , the system exhibits a strong first - order uniaxial nematic - biaxial nematic phase transition with a large density coexistence gap . when rod adsorption is considerably large while that of plates is small , the transition to the biaxial nematic phase is always of second order , and its region of stability in the phase diagram considerably widens . at very high pressures the mixture can effectively be identified as a two - dimensional mixture of squares and rectangles which again demixes above a certain critical point .
we also studied the relative stability of uniform phases with respect to density modulations of smectic , columnar and crystalline symmetry . |
one of the effects of dependence in a time series is that extremes tend to cluster .
this has applied implications to risk contagion over time but is also mathematically interesting and the challenge is to precisely relate the dependence structure to the clustering . for markov dependence ,
how do we describe exceedance clusters ?
point processes powerfully describe extremal behaviour of certain time series . under appropriate conditions on marginal distributions and rapid decay of dependence as a function of time lag for the process @xmath2 ,
the exceedance point process @xmath3 defined by @xmath4 \times(a,\infty ] \bigr ) = \ # \ { j \leq sn : x_j > ab_n \}\ ] ] converges weakly to a poisson limit as @xmath5 , where @xmath6 is a threshold sequence .
this leads to a number of results on asymptotic distributions of large order statistics and exceedances of an extreme level .
such results have been developed in a variety of contexts by @xcite .
more specific results exist for regularly varying processes @xcite , regenerative processes @xcite , and markov chains @xcite .
distributions of functionals of such point processes have been considered in @xcite . for stationary processes , the dependence structure causes extremes to occur in clusters .
the clustering is often summarized using the extremal index @xmath7 introduced by leadbetter et al .
@xcite , which is related to the asymptotic mean cluster size . to obtain a point process convergence result
, authors often employ the big block / little block technique and mixing conditions , such as leadbetter s @xmath8 ( see @xcite ) , to split the process into approximately independent and identically distributed blocks . with an appropriate choice of block size , extremes within one such block belong asymptotically to the same cluster . under an assumption controlling the extremal behaviour within each block , such as via the distribution of the number of exceedances
, @xmath3 generally converges to a limiting compound poisson process , where the compounding at each time point approximates the clustering within each block . for markov chains , the _ tail chain _
is an asymptotic process that models behaviour upon reaching an extreme state ; see @xcite .
point process results for stationary markov chains employ the tail chain to specify the compounding in the limit process . under markov dependence , the within - block behaviour
is determined merely by conditions on the marginal distribution and the transition kernel .
basrak and segers @xcite extended the tail chain model to general multivariate regularly varying stationary processes .
rootzn @xcite focuses on regenerative processes , which split naturally into _
cycles_. in this case , the within - block condition is replaced by an assumption on the extremal behaviour over a cycle .
the main difference is that the cycles are of random but finite length , whereas the block size increases deterministically with @xmath9 .
in particular , rootzn shows that convergence of the sequence of processes counting the number of exceedances depends on the asymptotics of the distribution of the cycle maximum as well as the marginal distribution .
we combine these two approaches to derive the weak limit of @xmath10 when @xmath11 is a positive recurrent markov chain .
such chains display a regenerative structure and in the limit , @xmath12 is approximated by a process consisting of clusters of points stacked above common time points , each corresponding to a separate regenerative cycle .
the heights of the points in each cluster are determined by an independent run of the tail chain .
this paper requires some distributional results for the tail chain process that were derived in @xcite and section [ secextrmch ] offers a summary of necessary facts .
we focus on the case of heavy - tailed marginals , but believe our approach could be extended to accommodate more general marginal distributions .
we review notation and relevant concepts . in general ,
bold symbols represent vectors or sequences and for @xmath13 , write @xmath14 . [ cols= " < , < " , ] for a space @xmath15 which is locally compact with countable base ( for example , a subset of @xmath16^d$ ] ) , a sequence of measures @xmath17 converges vaguely to @xmath18 ( written @xmath19 ) if @xmath20 as @xmath21 for any @xmath22
. the vague topology on @xmath23 is metrizable by the vague metric , @xmath24 , i.e. , @xmath25 iff @xmath26 . see @xcite for further details .
a distribution @xmath27 on @xmath28 has a regularly varying tail with index @xmath29 , denoted @xmath30 , if there exists @xmath31 such that @xmath32 \qquad\mbox { as } t \rightarrow\infty,\ ] ] where @xmath33=x^{-\alpha}$ ] for @xmath34 .
the function @xmath35 is called a _ scaling function_.
if @xmath36 is a ( homogeneous ) markov chain and @xmath37 is a markov transition kernel , we write @xmath38 to mean that the dependence structure of @xmath39 is specified by @xmath37 , i.e. , @xmath40 = k(x,\cdot ) , \qquad n=0,1,\ldots.\ ] ] also , @xmath41 $ ] specifies the initial distribution @xmath42 = \mu$ ] ( abbreviate @xmath43 ) , and @xmath44 denotes expectation with respect to @xmath45 .
let @xmath46 be a markov chain on @xmath28 with transition kernel @xmath37 .
the tail chain is a finite - dimensional approximation to the chain @xmath39 used to study the limit of @xmath10 given by ( [ eqppnn ] ) . building on theory developed in @xcite , @xcite presents the tail chain approximation in terms of a related process known as the _ extremal component _ of @xmath39 , an approach we follow here .
suppose the transition kernel @xmath37 is in the domain of attraction of a distribution @xmath47 ( denoted @xmath48 ) which means @xcite , @xmath49 \qquad \mbox{as } t\rightarrow\infty.\ ] ] taking @xmath50 i.i.d . random variables with common distribution @xmath47 , set @xmath51 , @xmath52 with @xmath53 and write @xmath54
. the _ tail chain associated with @xmath47 _
@xcite is @xmath55 given by @xmath56 thus @xmath57 is a multiplicative random walk and @xmath58 is an absorbing barrier for @xmath57 , accessible if @xmath59 . an _
extremal boundary _ for @xmath39 is a function @xmath60 satisfying @xmath61 , such that @xmath62\bigr ) { \longrightarrow}g \bigl(\{0\}\bigr)\qquad \mbox{as } t\rightarrow\infty,\ ] ] for any non - negative function @xmath63 .
such a function always exists if @xmath64 ( * ? ? ?
* section 3.2 ) .
if @xmath65 is an extremal boundary , any function @xmath66 with @xmath67 for @xmath68 is also an extremal boundary .
if @xmath69 , then @xmath70 is a convenient choice .
given an extremal boundary for @xmath37 , the _ extremal component _ of @xmath39 is the process @xmath39 prior to @xmath39 crossing below the scaled extremal boundary and identically @xmath71 afterwards .
the first downcrossing occurs at time @xmath72 and the extremal component is the process @xmath73 defined by @xmath74 starting from an extreme level @xmath75 , the extremal boundary separates extreme states from non - extreme states for the scaled process @xmath76 .
the _ tail chain approximation _ is the following ( * ? ? ?
* theorem 3.3 ) . for @xmath77 , @xmath78 , @xmath79 , @xmath80 { \rightarrow}{\mathsf{p}}_u \bigl [ ( t_1,\ldots , t_m ) \in\cdot \bigr ] \qquad\mbox{on } [ 0,\infty]^m \qquad\mbox{as } t\rightarrow\infty.\ ] ] so , the tail chain maps extreme states onto @xmath81 and contracts non - extreme states to the point @xmath58 .
note @xmath82 .
if the finite - dimensional extremal behaviour of @xmath39 is completely accounted for by the extremal component , then the tail chain approximation ( [ eqecfddconv ] ) extends from @xmath83 to @xmath39 .
when is this the case ?
say that @xmath37 satisfies the _ regularity condition _ if for any non - negative function @xmath84 , @xmath85 \qquad\mbox{as } t\rightarrow\infty.\ ] ] equivalent forms of ( [ eqregcond ] ) exist in ( * ? ? ?
* section 4 ) , and a relatively easy - to - check sufficient condition is given in terms of update functions .
if either ( a ) @xmath86 is an extremal boundary ; or ( b ) @xmath37 satisfies the regularity condition ( [ eqregcond ] ) , then for @xmath78 , we strengthen ( [ eqecfddconv ] ) to ( * ? ? ?
* theorem 4.1 ) , @xmath87 { \rightarrow}{\mathsf{p}}_u \bigl [ ( t_1,\ldots , t_m ) \in\cdot \bigr ] \qquad\mbox{on } [ 0,\infty]^m \qquad\mbox{as } t\rightarrow \infty.\ ] ] the conditional approximation ( [ eqecfddconv ] ) requires that the initial state become extreme . combining ( [ eqecfddconv ] ) with a heavy tailed initial distribution
makes @xmath83 have an unconditional distribution that is regularly varying ( in a sense to be discussed ) with a limit measure determined by the tail chain .
depending on assumptions , convergences take place on @xmath88\times[0,\infty]^m$ ] or the larger space @xmath89^{m+1}\setminus\{{\boldsymbol0}\}$ ] .
[ thmecjrv ] let @xmath39 be a markov chain on @xmath28 with @xmath77 , and suppose @xmath90 , with @xmath91 with scaling function @xmath35 .
on @xmath92 define the measure @xmath93,\ ] ] and extend this to a measure @xmath94 on @xmath95 by defining @xmath96 and @xmath97 . for any @xmath79 ,
the following convergences take place as @xmath98 .
a. [ parta ] in @xmath99^m\times[0,\infty])$ ] , @xmath100^m\times[0,\infty ] \bigr ] \,\mathop { \longrightarrow}^v\,\mu\bigl ( ( \cdot)\cap ( 0 , \infty]^m\times[0,\infty ] \bigr),\ ] ] and in @xmath101 @xmath102 \,\mathop { \longrightarrow}^v\,\mu(\cdot).\ ] ] if either ( ) @xmath103 ; ( ) @xmath104 is an extremal boundary ; or ( ) @xmath37 satisfies the regularity condition ( [ eqregcond ] ) , then ( [ eqecjrv ] ) can be strengthened to @xmath105 \,\mathop{\longrightarrow}^v\ , \mu(\cdot ) , \qquad\mbox{in $ { \mathbb{m}_{+}}({\mathbb{e}}_\sqsupset)$.}\ ] ] b. [ partb ] in the bigger space @xmath95 , we have @xmath106 \,\mathop { \longrightarrow}^v\,\mu^*(\cdot ) \qquad\mbox{in } { \mathbb{m}_{+}}\bigl({\mathbb{e}}^ * \bigr)\ ] ] if and only if @xmath107 \,\mathop { \longrightarrow}^v\ , \bigl({\mathsf{e}}\xi_1^ { \alpha }
\bigr)^j { \nu_\alpha}(\cdot ) \nonumber \\[-8pt ] \\[-8pt ] & & \quad\mbox{in } { \mathbb{m}_{+}}(0,\infty],\qquad j=1 , \ldots , m . \nonumber\end{aligned}\ ] ] part ( a ) requires that the first observation is large and with added conditions , part ( b ) removes this requirement .
regardless of whether ( [ eqecmargrv ] ) holds , the limit is always a lower bound on the tail weight of @xmath108 , since @xmath109 \geq\mu\bigl((0,\infty ] \times[0,\infty]^{j-1 } \times ( x,\infty ] \times[0,\infty]^{m - j } \bigr ) = \bigl({\mathsf{e}}\xi_1^ { \alpha } \bigr)^j { x}^{-\alpha}\ ] ] by ( [ eqecjrv ] ) and lemma [ lemintmom ] below .
markov s inequality , ( [ eqecjrv ] ) and a moment condition : @xmath110 = 0,\qquad j=1,\ldots , m,\ ] ] imply ( [ eqecmargrv ] ) .
see @xcite . here is a formula that helps evaluate @xmath111 in ( [ eqlimmeas ] ) on certain sets .
[ lemintmom ] for a random variable @xmath112 , define the measure @xmath113 $ ] on @xmath114 ^ 2 \setminus\{(0,0)\}$ ] .
we compute @xmath115 \times(y,\infty ] \bigr ) = { y}^{-\alpha } { \mathsf{e}}\bigl [ y^\alpha{\mathbf{1}_{\{y > yx^{-1}\ } } } \bigr ] - { x}^{-\alpha } { \mathsf{p}}\bigl [ y > yx^{-1 } \bigr].\ ] ] in particular , @xmath116 \times(y,\infty ] ) = { y}^{-\alpha } { \mathsf{e}}y^\alpha$ ] .
we obtain @xmath117\times(y,\infty ] \bigr ) & = & \int_{[0,x ] } { \nu_\alpha}(ds ) { \mathsf{p}}\bigl[y > y s^{-1 } \bigr ] = \int_{[{x}^{-\alpha},\infty ] } ds\ , { \mathsf{p}}\bigl[y > y s^{1/\alpha } \bigr ] \\ & = & \int_{[{x}^{-\alpha},\infty ] } ds\ , { \mathsf{p}}\bigl[{y}^{-\alpha } y^\alpha > s \bigr]\end{aligned}\ ] ] by change of variables . applying fubini s theorem ,
this becomes @xmath118 } \bigl(s-{x}^{-\alpha } \bigr ) { \mathsf{p}}\bigl[{y}^{-\alpha } y^\alpha\in ds \bigr ] = { y}^{-\alpha } { \mathsf{e}}\bigl [ y^\alpha{\mathbf{1}_{\{y > yx^{-1}\ } } } \bigr ] - { x}^{-\alpha } { \mathsf{p}}\bigl [ y > yx^{-1 } \bigr].\ ] ] letting @xmath119 , this quantity converges to @xmath120 by monotone convergence .
we give conditions on the extremal component which enable an informative point process limit result by controlling the positive portion of the extremal component , the random vector of random length @xmath121 .
the conditions imply restrictions on the behaviour of the tail chain @xmath57 .
we study a positive recurrent chain @xmath39 by splitting it into regenerative cycles and analyzing its extremal properties via the extremal components of the cycles . for regenerative processes ,
asmussen @xcite and rootzn @xcite point out the connection between point process convergence and the asymptotic distribution of cycle maxima .
informed by this approach , we consider when the distribution of the maximum over the extremal component has a regular variation property .
the limit measure of this regular variation can be used to compute an extremal index for @xmath39 @xcite .
here is a condition that controls the persistence of non - zero values of the extremal component : @xmath122 = 0 \qquad\mbox{for all } a , \delta > 0.\ ] ] note @xmath123 .
compare this condition with ( * ? ? ?
* condition 4.1 ) and ( * ? ? ?
* equation ( 3.1 ) ) , which are formulated in terms of block sizes . condition ( [ condeclrd ] ) is a tightness condition that complements the finite - dimensional convergences ( [ eqecjrv ] ) .
section [ secmch ] gives simpler sufficient conditions depending on whether @xmath124 or @xmath69 .
condition ( [ condeclrd ] ) requires that the chain drift back to the non - extreme states after visiting an extreme state and makes non - extreme states recurrent and the tail chain transient .
[ propcondeclrd ] let @xmath125 be markov on @xmath28 with initial distribution @xmath126 satisfying @xmath91 with scaling function @xmath127 , so that ( [ eqecjrv ] ) holds .
if @xmath39 satisfies condition ( [ condeclrd ] ) , then @xmath128 = 0,\qquad a>0,\ ] ] and @xmath129 as @xmath130 in probability and almost surely and therefore , @xmath131 = 1.\ ] ] so the tail chain is transient and the additive random walk @xmath132 satisfies @xmath133 .
the tail chain @xmath57 and @xmath39 live on the same state space @xmath28 but for @xmath57 , @xmath58 is a special boundary state which represents the collection of non - extreme states of @xmath39 under the tail chain approximation .
proof of proposition [ propcondeclrd ] observe from ( [ eqecjrv ] ) , as @xmath134 , @xmath135 & { \longrightarrow } & \int_{(1,\infty ] } { \nu_\alpha}(dx ) { \mathsf{p}}_x \bigl[\sup_{m\leq j\leq r } t_j > 1 \bigr ] \\ & = & \int_{(1,\infty ] } { \nu_\alpha}(dx ) { \mathsf{p}}\bigl[\sup_{m\leq j\leq r } \xi ( j ) > x^{-1 } \bigr ] . \ ] ] therefore , by monotonicity and then monotone convergence , @xmath136 & \geq&\lim_{r \rightarrow\infty } \int _ { ( 1,\infty ] } { \nu_\alpha}(dx ) { \mathsf{p}}\bigl [ \sup_{m \leq j \leq r } \xi(j ) >
x^{-1 } \bigr ] \\ & = & \int_{(1,\infty ] } { \nu_\alpha}(dx ) { \mathsf{p}}\bigl [ \sup_{j \geq m } \xi(j ) > x^{-1 } \bigr ] \\ & = : & \int_{(1,\infty ] } { \nu_\alpha}(dx ) f_m(x).\end{aligned}\ ] ] condition ( [ condeclrd ] ) implies that @xmath137 } { \nu_\alpha}(dx ) f_m(x ) \rightarrow0 $ ] as @xmath138 . we claim that @xmath139 for any @xmath34 .
suppose instead that @xmath140 for some @xmath141 .
since the @xmath142 are all increasing in @xmath143 , we have @xmath144 for @xmath145 . but
this implies that @xmath146 } { \nu_\alpha}(dx ) f_m(x ) \geq\liminf_{m\rightarrow\infty } \int_{(1 { \vee}x_0,\infty ] } { \nu_\alpha}(dx ) f_m(x ) \geq c { \nu_\alpha}(1 { \vee}x_0,\infty ] > 0\ ] ] by fatou s lemma , contradicting condition [ condeclrd ] .
therefore , @xmath147 \rightarrow0 $ ] as @xmath148 for all @xmath34 , establishing ( [ eqcondeclrdtch ] )
. condition ( [ condeclrd ] ) assumes the first observation exceeds @xmath149 which is in the spirit of ( [ eqecjrv ] ) . for translating the stronger convergence of unconditional distributions ( [ eqecjrvfull ] ) in the bigger space @xmath89^{m+1}\setminus\{{\boldsymbol0}\}$ ] to point process convergence
, we will require an additional assumption : @xmath150 \\[-8pt ] & & \quad\lim_{\delta\downarrow0 } \limsup_{t\rightarrow\infty } t{\mathsf{p}}\bigl [ { x_0}/{b(t ) } \leq\delta , \sup_{j\geq m_0 } { x_j^{(b(t))}}/{b(t ) } > a \bigr ] = 0 \qquad\mbox{for all } a > 0 .
\nonumber\end{aligned}\ ] ] analogously to ( [ eqecmargrvmom ] ) , by markov s inequality a moment condition is sufficient for condition ( [ condeclrdmrv ] ) : @xmath151 = 0 , \qquad\mbox{for some $ \varepsilon > 0$}.\ ] ] condition ( [ condeclrdmrv ] ) implies a uniform bound on the @xmath152th moment of the tail chain states .
[ pro2.2 ] let @xmath125 be a markov chain on @xmath28 with initial distribution @xmath126 satisfying @xmath91 , whose extremal component satisfies ( [ eqecjrvfull ] ) on @xmath153 .
if @xmath39 satisfies condition ( [ condeclrdmrv ] ) then , @xmath154 under ( [ eqcondeclrdtchmom ] ) , we necessarily have @xmath155 since @xmath156 recalling ( [ eqecmargrv ] ) , the marginal tails of the extremal component can not be heavier than the tail of @xmath126 .
proof of proposition [ pro2.2 ] under ( [ eqecjrvfull ] ) , @xmath157 { \longrightarrow}\int_{(0,\infty ] } { \nu_\alpha}(dx ) { \mathsf{p}}\bigl [ \sup_{m_0 \leq j \leq r } \xi(j ) > x^{-1 } \bigr ] = { \mathsf{e}}\bigl(\sup_{m_0 \leq j \leq r } \xi(j)^\alpha \bigr)\ ] ] by lemma [ lemintmom ] .
therefore , @xmath158 \geq\lim_{r\rightarrow\infty } { \mathsf{e}}\bigl(\sup_{m_0 \leq j \leq r } \xi ( j)^\alpha \bigr ) = { \mathsf{e}}\bigl(\sup_{j \geq m_0 } \xi(j)^\alpha \bigr).\ ] ] furthermore , by condition ( [ condeclrdmrv ] ) , for some @xmath159 , @xmath160 \\ & & \quad\leq\limsup_{t\rightarrow\infty } t{\mathsf{p}}\bigl [ x_0 \leq \delta b(t ) , \sup_{j\geq m_0 } { x_j^{(b(t ) ) } } > { b(t ) } \bigr ] + \limsup_{t\rightarrow \infty } t{\mathsf{p}}\bigl [ x_0 > \delta{b(t ) } \bigr ] < \infty\end{aligned}\ ] ] showing that @xmath161 which is enough for ( [ eqcondeclrdtchmom ] ) . under both conditions ( [ condeclrd ] ) and ( [ condeclrdmrv ] ) , we derive the tail behaviour of the maximum of the extremal component of @xmath39 .
[ proprvecmax ] let @xmath125 be a markov chain on @xmath28 with initial distribution @xmath126 satisfying @xmath91 , whose extremal component satisfies both ( [ eqecjrvfull ] ) on @xmath153 . if @xmath39 satisfies conditions ( [ condeclrd ] ) and ( [ condeclrdmrv ] ) , then @xmath162 \,\mathop{\longrightarrow}^v\,c \cdot{\nu_\alpha}(\cdot ) \qquad \mbox{in } { \mathbb{m}_{+}}(0,\infty],\ ] ] where @xmath163 + { \mathsf{e}}[\sup_{j \geq1 } \xi(j)^\alpha{\mathbf{1}_{\{\sup_{j \geq1 } \xi(j ) > 1\ } } } ] $ ] . for @xmath34 , we have @xmath164 = [ \sup_{j\geq 0 } x_j^{(b(t))}/b(t)>x]$ ] . for @xmath79
, we have on the one hand , by ( [ eqecjrvfull ] ) , @xmath165 & \geq & \lim_{t\rightarrow\infty } t{\mathsf{p}}\bigl[\sup_{0 \leq j < m } { x_j^{(b(t))}}/{b(t ) } > x \bigr ] \\ & = & { x}^{-\alpha } + \int_{[0,x ) } { \nu_\alpha}(du ) { \mathsf{p}}_u \bigl [ \sup_{1 \leq j < m } t_j > x \bigr],\end{aligned}\ ] ] from which , letting @xmath138 , @xmath166 \geq{x}^{-\alpha } + \int _ { [ 0,x ) } { \nu_\alpha}(du ) { \mathsf{p}}_u \bigl [ \sup_{j \geq1 } t_j > x \bigr].\ ] ] on the other hand , for @xmath159 we have @xmath167 \leq t{\mathsf{p}}\biggl [ \frac{{x_0}}{b(t ) } > \delta , \sup_{j \geq m } \frac { x_j^{(b(t))}}{b(t ) } > x \biggr ] + t { \mathsf{p}}\biggl[\frac{{x_0}}{b(t ) } \leq\delta , \sup_{j \geq m } \frac { x_j^{(b(t))}}{b(t ) } > x \biggr].\ ] ] given @xmath168 , by condition ( [ condeclrdmrv ] ) , we may choose @xmath169 small enough that @xmath170 < \varepsilon/2,\ ] ] where @xmath171 is from condition ( [ condeclrdmrv ] ) .
condition ( [ condeclrd ] ) permits the choice @xmath172 so large that @xmath173 < \varepsilon/2.\ ] ] therefore , @xmath174
< \varepsilon$ ] for @xmath175 , and so @xmath176 & < & \lim_{m\rightarrow\infty } \limsup_{t\rightarrow \infty } t{\mathsf{p}}\bigl [ \sup_{0 \leq j < m } { x_j^{(b(t))}}/{b(t ) } > x \bigr ] + \varepsilon \\ & = & { x}^{-\alpha } + \int_{[0,x ) } { \nu_\alpha}(du ) { \mathsf{p}}_u \bigl [ \sup_{j \geq1 } t_j > x \bigr ] + \varepsilon.\end{aligned}\ ] ] combine this with ( [ eqpfeclb ] ) , and apply formula ( [ eqintmom ] ) for @xmath177\times(x,\infty])$ ] to complete the proof .
we now derive the limit of the exceedance point process @xmath3 defined in ( [ eqppnn ] ) , where @xmath46 is a markov chain on @xmath28 with transition kernel @xmath64 .
we write @xmath178 using the notation @xmath179 to denote the measure assigning unit mass at the point @xmath143 and @xmath3 is a random element of @xmath180)$ ] , the space of radon point measures on @xmath181 $ ] , endowed with the topology of vague convergence @xcite .
if @xmath39 is positive recurrent , it is a regenerative process ( ( * ? ? ?
* section vii.3 ) , @xcite ) so the sample path of @xmath39 splits into identically distributed cycles between visits to certain set .
the extremal properties of @xmath39 are determined by extremal behaviour of the individual cycles .
this approach has been developed for markov chains by rootzn @xcite , as well as for queues by asmussen @xcite .
our approach introduces the tail chain approximation to describe the extremal behaviour of the regenerative cycles using their extremal component .
consider the case where @xmath39 has a positive recurrent atom @xmath182 . for positive recurrent chains
, atoms can be constructed by several methods if no natural atom exists .
see , e.g. , ( * ? ? ?
* chapter 6 ) or ( * ? ? ?
* chapter i.5 ) .
an atom is a set such that for a probability distribution @xmath126 on @xmath28 , @xmath183 = 1 \qquad\mbox{for } y \geq0,\ ] ] and @xmath184 is the first hitting time of @xmath182 .
positive recurrence means that @xmath185 where @xmath186 denotes expectation with respect to @xmath126 considered as the initial distribution of @xmath187 .
under ( [ eqatom ] ) , the sample path of @xmath39 splits into i.i.d .
cycles between visits to @xmath182 , as follows .
define the times @xmath188 , @xmath189 recursively according to @xmath190 thus , the sequence @xmath191 gives the indices when @xmath39 is in @xmath182 , and @xmath192 for @xmath193 .
the values @xmath194 are the number of steps @xmath39 takes outside of @xmath182 between visits to @xmath182 .
the cycles end by visits to @xmath182 ; cycles are the random elements @xmath195 of the space of finite sequences @xmath196 .
the strong markov property implies @xmath197 are independent , and @xmath198 are identically distributed . in particular , for @xmath199 , @xmath200 = { \mathsf{p}}\bigl [ \bigl\ { ( x_{s_{k-1 } } , \ldots , x_{s_{k-1 } + \tau^a_k } ) ; \tau^a_k \bigr\ } \in \cdot \bigr ] = { \mathsf{p}}_h \bigl [ \bigl\{(x_0,\ldots , x_{{\tau_a } } ) ; { \tau_a}\bigr\ } \in\cdot \bigr].\ ] ] furthermore , @xmath201 is a renewal process , with @xmath202 by ( [ eqposrec ] ) . applying the cycle decomposition , we may now write ( [ eqn : nn ] ) as @xmath203 as a family of random elements in @xmath180)$ ] , @xmath204 is asymptotically negligible .
[ propppneg ] assuming ( [ eqposrec ] ) and @xmath205 , @xmath206 , the null measure , in @xmath180)$ ] .
let @xmath207)$ ] with support in @xmath208\times[m,\infty]$ ] for integers @xmath209 .
it is sufficient to verify that @xmath210 \rightarrow0 $ ] , for any @xmath211 .
we have as @xmath5 , = 6.5pt = 6.5pt @xmath212 & = & { \mathsf{p}}\biggl [ \sum _ { j=0}^{s_0 - 1 } f \biggl(\frac{j}{n } , \frac{x_j}{b_n } \biggr ) > \gamma \biggr ] \\[-2pt ] & \leq&\sum_{m=0}^r { \mathsf{p}}\biggl [ \sum _ { j=0}^{m } f \biggl ( \frac{j}{n } , \frac{x_j}{b_n } \biggr ) > \gamma , { \tau_a}= m \biggr ] + { \mathsf{p}}[{\tau_a } > r ] , \end{aligned}\ ] ] and @xmath213 & \leq&(r+1 ) { \mathsf{p}}\biggl [ \sum _ { j=0}^{r } f \biggl(\frac{j}{n } , \frac { x_j}{b_n } \biggr ) > \gamma \biggr ] \\[-2pt ] & \leq&(r+1 ) { \mathsf{p}}\bigl [ \sup_{0 \leq j \leq r } x_j \geq b_n m \bigr ] { \longrightarrow}0.\end{aligned}\ ] ] choosing @xmath214 to make @xmath215 $ ] arbitrarily small , the result follows .
lemma [ propppneg ] and slutsky s theorem means that the asymptotic behavior of @xmath12 and @xmath216 are the same .
we obtain a weak limit for @xmath216 using the tail chain approximation discussed in section [ secextrmch ] , provided that a cycle s extremal behaviour is adequately described by its extremal component . as usual ,
assume @xmath64 , @xmath217 and suppose @xmath65 is an extremal boundary for @xmath39 .
we require a mild assumption that the atom @xmath182 be a bounded subset of the state space @xmath28 , @xmath218 as would usually be the case .
fix @xmath199 .
the number of steps needed by the scaled process in the @xmath219-th cycle to cross below the extremal boundary is @xmath220 the extremal component of the @xmath219-th cycle is @xmath221 . without loss of generality ,
we suppose the extremal component of a cycle is a subset of the complete cycle .
to see this , observe from the definition of @xmath222 and @xmath223 , without loss of generality , @xmath224 = 1.\ ] ] indeed , ( [ eqatombdd ] ) implies that @xmath225 $ ] some @xmath226 .
define @xmath227 and @xmath228=1 $ ] ; we claim further that we may suppose @xmath229=1 $ ] .
if @xmath230 for all @xmath231 , then this follows directly . otherwise , verify that @xmath232 is also an extremal boundary for @xmath37 ( see the remarks after ( [ eqextrbdry ] ) ) , and the corresponding downcrossing time satisfies @xmath233=1 $ ] .
therefore , for @xmath199 , @xmath234 = { \mathsf{p}}_h \bigl [ \bigl\{(x_0,\ldots , x_{{\tau(t)}-1 } ) ; { \tau(t ) } , { \tau_a}\bigr\ } \in \cdot \bigr]\ ] ] and @xmath235 are independent , since each is a function of @xmath236 .
these facts suggest we approximate @xmath237 by a point process whose observations consist of the extremal components of iid copies of the chain @xmath39 started from @xmath90 .
this approximation is facilitated by additional notation .
let @xmath238 be i.i.d .
copies of the markov chain @xmath38 with respect to @xmath239 ; that is the initial distribution of each chain is fixed to be @xmath126 . define @xmath240 and for @xmath193 , form the extremal component @xmath241 of the @xmath219th chain .
thus with respect to @xmath242 , @xmath243 for @xmath193 , with the tilde differentiating the times @xmath244 defined on the @xmath219th process @xmath245 from the cycle times @xmath246 defined on @xmath39 .
recall @xmath222 is also defined on @xmath39 .
next , generate an i.i.d .
family of tail chains by letting @xmath247 be i.i.d .
copies of the process @xmath248 , recalling the notation around ( [ eqtch ] ) . additionally , put @xmath249 , the first time the @xmath219th tail chain hits @xmath250 .
use the convention @xmath251 ; for example , @xmath252 a.s .
, @xmath193 , if @xmath69
. finally , let @xmath253 be a poisson random measure on @xmath180)$ ] , independent of the @xmath254 , with mean measure a product of lebesgue measure on the time axis @xmath28 and pareto measure @xmath255 ( given by @xmath256 = { x}^{-\alpha}$ ] ) on the observation axis @xmath257 $ ] .
recall @xmath152 is the tail index of @xmath258 .
the point process consisting of the observations @xmath259 , spaced in time according to the renewal times @xmath188 , converges to a cluster poisson process which is basically @xmath260 with a time scaling and compounded in the second coordinate according to the i.i.d .
tail chains @xmath254 .
this result is basic to analyzing the asymptotic behavior of @xmath12 in ( [ eqn : nn ] ) .
[ thmppconv ] let @xmath39 be a markov chain on @xmath28 with transition kernel @xmath77 , and initial distribution @xmath261 , such that @xmath262 in @xmath263 $ ] , where @xmath264 and @xmath265 .
the renewal process @xmath188 is defined in ( [ eqrenewaltimes ] ) , with mean interarrival time @xmath266 given by ( [ eqdefq ] ) . with the notation introduced in the previous paragraphs , we have the following with respect to @xmath267 .
a. if @xmath39 satisfies condition ( [ condeclrd ] ) , then given @xmath159 , in @xmath180)$ ] , as @xmath5 , @xmath268 b. suppose @xmath269 satisfies ( [ eqecjrvfull ] ) as well as both conditions ( [ condeclrd ] ) and ( [ condeclrdmrv ] ) . then in @xmath180)$ ] , as @xmath5 , @xmath270 section [ secppconv ] ( p. ) contains the proof . paralleling the discussion in section [ secextrmch ]
, we have two results depending on the strength of the conditions
. the weaker assumptions of part ( a ) yield a result that selects cycles starting from an exceedance . part ( b )
does not have to do such cycle selection .
the points of the limit process are arranged in stacks above common time points @xmath271 .
the heights of the points in each stack are specified by an independent run of the tail chain starting from @xmath272 . if @xmath124 , then the @xmath273 are i.i.d .
geometric random variables with parameter @xmath274 , so all stacks have finite length . if @xmath103 , then @xmath275=1 $ ] for each @xmath219 . in this case ,
condition ( [ condeclrd ] ) is necessary to ensure that @xmath276 is radon , by forcing the tail chain to drift towards 0 as in ( [ eqtchtransient ] ) .
the process @xmath277 retains only those stacks of @xmath276 whose initial value exceeds the threshold @xmath169 . because there are an infinite number of @xmath272 in any neighbourhood of 0 , dispensing with the restriction in @xmath169
requires that not too many of the @xmath278 are large .
this translates to the condition @xmath279 , provided by condition ( [ condeclrdmrv ] ) . to analyze @xmath12 in ( [ eqn : nn ] )
, we approximate @xmath237 in ( [ eqnndecomp ] ) by @xmath280 in ( [ weenie2 ] ) , provided the extremal component adequately describes extremal behaviour within each cycle .
if the extremal boundary is not identically zero , behavior between the end of the extremal component and the end of the cycle is not be captured by the tail chain and we require that such observations do not significantly influence extremal properties . to guarantee a result analogous to part ( a ) above
, we require , @xmath281 = 0 \qquad \mbox{for all } a , \delta > 0,\ ] ] and for a result analogous to part ( b ) above , we require , @xmath282 = 0 \qquad\mbox{for all } a > 0.\ ] ] with these conditions , the point process @xmath3 converges to the limit @xmath276 , and the distribution of the cycle maximum behaves as if it has a regularly varying tail .
[ thmppconvmc ] let @xmath39 be a markov chain on @xmath28 with transition kernel @xmath77 .
suppose that @xmath37 has a positive recurrent bounded atom in the sense of ( [ eqatom ] ) , ( [ eqposrec ] ) , and ( [ eqatombdd ] ) .
define the renewal process @xmath188 with mean interarrival time @xmath266 as in ( [ eqrenewaltimes ] ) and ( [ eqdefq ] ) and assume further that @xmath283 in @xmath263 $ ] , where @xmath31 . with respect to @xmath267 , the following hold .
a. if @xmath39 satisfies conditions ( [ condeclrd ] ) and ( [ condcyc ] ) , then given @xmath159 , @xmath284 in @xmath180)$ ] , as @xmath5 , where @xmath277 is defined in ( [ weenie1 ] ) .
b. suppose additionally that @xmath39 satisfies ( [ eqecjrvfull ] ) as well as conditions ( [ condeclrd ] ) , ( [ condeclrdmrv ] ) and ( [ condcycmrv ] ) . recall @xmath276 from ( [ weenie2 ] ) .
then , in @xmath180)$ ] , as @xmath285 , @xmath286 and furthermore , the distribution of the cycle maximum has a regularly varying tail , @xmath287 \,\mathop { \longrightarrow}^v\,c \cdot{\nu_\alpha}(\cdot ) \qquad\mbox{in } { \mathbb{m}_{+}}(0 , \infty],\ ] ] where @xmath288 + { \mathsf{e}}\bigl[\sup_{j \geq1 } \xi(j)^\alpha{\mathbf{1}_{\{\sup_{j \geq1 } \xi(j ) > 1\ } } } \bigr].\ ] ] \(a ) first , note that @xmath289 , where @xmath290 hence , by lemma [ propppneg ] it remains to show that @xmath291 . split @xmath292
according to the times @xmath293 : @xmath294 the equality holds on the set @xmath295 , which has probability 1 by ( [ eqcycletimes ] ) . because of ( [ eqcycdistrib ] ) and the independence of the @xmath296 , we have @xmath297 for each @xmath9 , and @xmath298 by proposition [ thmppconv](a ) . by slutsky
s theorem , the result follows if @xmath299 , so we show@xmath300 = { \mathsf{p}}\biggl [ \sum_{k=0}^\infty\sum _ { j={\tau_{k+1}(b_n)}}^{\tau^a_{k+1 } } f \biggl(\frac{s_k+j}{n } , \frac{x_{s_k + j}}{b_n } \biggr ) { \mathbf{1}_{\{x_{s_k } \geq\delta b_n\ } } } > \gamma \biggr ] { \longrightarrow}0,\ ] ] for any @xmath301)$ ] and @xmath302 . let @xmath303 have support in @xmath208\times[m,\infty]$ ] for integers @xmath209 .
the previous probability is bounded by @xmath304 \\ & & \quad { } + { \mathsf{p}}\biggl [ \sum_{k=2rn}^\infty\sum _ { j={\tau_{k+1}(b_n)}}^{\tau^a_{k+1 } } f \biggl(\frac{s_k+j}{n } , \frac{x_{s_k + j}}{b_n } \biggr ) { \mathbf{1}_{\{x_{s_k } \geq\delta b_n\ } } } > 0 \biggr].\end{aligned}\ ] ] observe that the second term is at most @xmath305 = { \mathsf{p}}[{s_{2rn}}/{2rn } \leq1/2 ] \rightarrow0 $ ] as @xmath5 , since @xmath306 a.s . , and @xmath307 by ( [ eqdefq ] ) .
the first term is bounded by @xmath308\\ & & \quad { } \leq2r n { \mathsf{p}}_h \biggl [ \frac{x_0}{b_n } \geq \delta , \sup_{{\tau(b_n ) } < j
< { \tau_a } } \frac{x_j}{b_n } \geq m \biggr],\end{aligned}\ ] ] which vanishes as @xmath5 by condition ( [ condcyc ] ) .
\(b ) recalling the decomposition ( [ eqnndecomp ] ) , by lemma [ propppneg ] it is sufficient to show that @xmath309 .
this follows by a similar argument as in part ( a ) .
write @xmath310 then @xmath311 by proposition [ thmppconv](b ) , and condition ( [ condcycmrv ] ) implies that @xmath312 .
next , we show ( [ eqrvcycfull ] ) . in light of ( [ eqcycletimes ] ) , we have @xmath313 - t{\mathsf{p}}_h \biggl[\sup_{0 \leq j < \tau(b(t ) ) } \frac{x_j}{b(t ) } > x \biggr ]
\leq t{\mathsf{p}}_h \biggl[\sup_{\tau(b(t ) ) < j < { \tau_a } } \frac{x_j}{b(t ) } > x \biggr ] { \longrightarrow}0\ ] ] under condition ( [ condcycmrv ] ) . recalling that @xmath314 { \longrightarrow}c { x}^{-\alpha}\ ] ] as @xmath98 by proposition [ proprvecmax ] ( p. ) , where @xmath226 is as in ( [ eqrvcycmaxconst ] ) , completes the proof
. setting @xmath315 , rootzn shows ( * ? ? ?
* theorem 3.2 ) that ( [ eqrvcycfull ] ) implies @xmath316 { \longrightarrow}\exp \bigl(-cq^{-1}{x}^{-\alpha } \bigr),\qquad x>0,\ ] ] where @xmath226 is given by ( [ eqrvcycmaxconst ] ) , and @xmath266 is the mean interarrival time ( [ eqdefq ] ) .
hence , in the stationary case , @xmath317 is the extremal index of the process @xmath39 ( ( * ? ? ?
* section 2.2 ) , @xcite ) . on the other hand , for stationary regularly varying markov chains with @xmath64 satisfying a condition analogous to condition ( [ condeclrd ] ) , it is known ( * ? ? ?
* remark 4.7 ) , @xmath318 = { \mathsf{p}}\bigl [ \sup_{j \geq1 } \xi(j ) \leq1 \bigr ] - { \mathsf{e}}\bigl[\sup_{j \geq1 } \xi(j)^\alpha{\mathbf{1}_{\{\sup_{j \geq1 } \xi(j ) \leq1\ } } } \bigr ] = c - { \mathsf{e}}\bigl(\sup_{j \geq1 } \xi(j)^\alpha \bigr),\ ] ] where @xmath319 supported on @xmath320 , independent of @xmath321 .
hence , for a stationary markov chain @xmath39 satisfying the assumptions of theorem [ thmppconvmc](b ) , the extremal index is given by @xmath322 we now consider simplifications of the above conditions . if @xmath103 , we can replace @xmath323 with @xmath39 in the finite - dimensional convergence ( [ eqecjrv ] ) when @xmath126 has a regularly varying tail , meaning that the tail chain approximation completely describes the extremes of the chain @xmath39 in a finite dimensional sense . however , @xmath103 also implies that for any @xmath324 , as @xmath98 , @xmath325 { \longrightarrow}1\ ] ] ( see ( * ? ? ?
* proposition 5.1(d ) ) ) meaning that , as the initial observation becomes more extreme , it takes longer for @xmath39 to return to @xmath182 to complete the cycle .
hence , for condition ( [ condeclrd ] ) to hold , we need a condition that ensures @xmath39 eventually drifts away from extreme states : @xmath326 = 0 \qquad\mbox{for all } a>0.\ ] ] [ propsufflrdz ] suppose @xmath327 with @xmath103 and positive recurrent bounded atom @xmath182 , and @xmath90 with @xmath91 .
if @xmath39 satisfies condition ( [ condsufflrdz ] ) , both conditions ( [ condeclrd ] ) and ( [ condcyc ] ) hold and consequently , the convergence ( [ eqppconvxdfull ] ) takes place . we first show that @xmath328 = 0 \qquad\mbox{for all } a , \delta>0.\ ] ] indeed , for @xmath329 , we have , @xmath330 & \leq&\int_{[\delta , c ] } t { \mathsf{p}}_h \biggl [ \frac{x_0}{b(t ) } \in du \biggr ] { \mathsf{p}}_{b(t)u } \biggl [ \sup_{m \leq j < { \tau_a } } \frac{x_j}{b(t ) } > a \biggr ] \\ & & { } + t{\mathsf{p}}_h \biggl[\frac{x_0 } { b(t ) } > c \biggr].\end{aligned}\ ] ] furthermore , for @xmath331 , @xmath332 \leq{\mathsf{p}}_{b(t)u } \biggl [ \sup_{m \leq j
< { \tau_a } } \frac { x_j}{b(t)u } > \frac{a}{c } \biggr ] \leq\sup_{s \geq b(t)\delta } { \mathsf{p}}_{s } \biggl [ \sup_{m \leq j < { \tau _ a } } \frac{x_j}{s } >
\frac{a}{c } \biggr].\end{aligned}\ ] ] hence , by condition ( [ condsufflrdz ] ) , @xmath333 \leq{\nu_\alpha}[\delta , c ] \cdot0 + { \nu_\alpha}(c,\infty ] = { c}^{-\alpha}.\ ] ] letting @xmath334 establishes ( [ eqpflrdz ] ) . as ( [ eqcycletimes ] ) implies that @xmath335 , condition ( [ condeclrd ] ) follows . to verify condition ( [ condcyc ] ) , argue that @xmath336 & \leq & { \mathsf{p}}_h \biggl[\frac{x_0}{b(t ) } > \delta,\tau \bigl(b(t ) \bigr ) \leq m-1 \biggr ] \\ & & { } + { \mathsf{p}}_h \biggl [ \frac{x_0}{b(t ) } > \delta , \bigvee _ { j = m}^ { \tau_a-1 } \frac{x_j}{b(t ) } > a \biggr],\end{aligned}\ ] ] of which the first term vanishes as @xmath98 because of ( [ eqcyctimesgz ] ) ( see ( * ? ? ?
* proposition 5.1(d ) ) ) .
appeal to ( [ eqpflrdz ] ) and let @xmath138 to complete the proof .
condition ( [ condsufflrdz ] ) is a condition on the transition kernel @xmath37 ; this is best discussed by recalling ( see @xcite for discussion ) that a transition kernel @xmath77 has an update function @xmath337 of the form @xmath338 where @xmath339 and @xmath340 for @xmath341 with @xmath342=1 $ ] and we can represent @xmath37 as @xmath343.\ ] ] take @xmath344 , i.i.d .
copies of @xmath345 , and write @xmath346
. for @xmath347 let @xmath348 denote the @xmath214-step update function , i.e. , @xmath349 $ ] , and @xmath350 . by iteration , @xmath351 thus condition ( [ condsufflrdz ] ) requires both @xmath352 as in ( [ eqtchtransient ] ) , and also an asymptotic stochastic boundedness condition on @xmath353 .
alternately , one could give criteria for condition ( [ condsufflrdz ] ) using mean drift conditions for @xmath39 or @xmath354 @xcite .
in this case , ( [ eqecfddconv ] ) implies @xmath355 { \longrightarrow}{\mathsf{p}}\bigl[\tau^ * = m \bigr ] , \qquad m\geq1,\ ] ] where @xmath356 is a geometric random variable with parameter @xmath274 .
hence , the tail chain terminates after a finite number of steps .
if either @xmath357 is an extremal boundary , or @xmath37 satisfies the regularity condition ( [ eqregcond ] ) ( p. ) , theorem [ thmecjrv ] assures us that convergence ( [ eqecjrv ] ) holds for @xmath39 with respect to @xmath267 , and condition ( [ condeclrd ] ) follows directly since @xmath358 & \leq&\limsup_{t\to\infty}t { \mathsf{p}}\bigl [ x_0>\delta b(t ) , \tau \bigl(b(t ) \bigr)\geq m \bigr ] \\ & = & \int_\delta^\infty\nu_\alpha(dx){\mathsf{p}}\bigl[x \xi(m)>0 \bigr ] \to0\qquad(m\to\infty).\end{aligned}\ ] ] the regularity condition ( [ eqregcond ] ) extends to any finite number of steps ; that is , iterates of @xmath37 also satisfy the condition .
however , unless @xmath357 is an extremal boundary , we need the regularity condition to hold uniformly over the whole cycle of random length @xmath359 to prevent @xmath39 from returning to an extreme state within the same cycle , after crossing below the extremal boundary . condition ( [ condcycreg ] ) given next accomplishes this .
( note that even if @xmath360 is an extremal boundary for @xmath37 , we are using an extremal boundary @xmath65 chosen to satisfy ( [ eqcycletimes ] ) . )
@xmath361 = 0 \qquad\mbox{whenever } u_t = u(t ) \rightarrow0,\qquad a>0.\ ] ] recalling the update function form ( [ equpdfunc ] ) , the regularity condition ( [ eqregcond ] ) holds if the function @xmath362 is bounded near 0 for each @xmath363 in a set of probability 1 ( * ? ? ?
* proposition 4.1 ) .
condition ( [ condcycreg ] ) is a stronger boundedness restriction on @xmath362 near 0 . alternatively ,
when @xmath37 satisfies the regularity condition ( [ eqregcond ] ) , condition ( [ condcycreg ] ) may be viewed as a restriction on @xmath359 , since then @xmath364 = 0 \qquad\mbox{whenever } u_t = u(t ) \rightarrow0,\ ] ] it is sufficient for ( [ condcycreg ] ) .
this follows from the decomposition @xmath365 \leq{\mathsf{p}}_{tu_t } [ \tau_a > m ] + { \mathsf{p}}_{tu_t } \bigl [ \sup_{1 \leq j \leq m } { x_j } > { t}a \bigr],\ ] ] with ( [ psychot ] ) controlling the first right - hand term and ( [ eqregcond ] ) controlling the second .
suppose @xmath327 with @xmath366 and @xmath91 and @xmath39 has a positive recurrent , bounded atom @xmath182
. then @xmath39 satisfies ( [ condeclrd ] ) with respect to @xmath267 . moreover , if either 1 .
@xmath357 is an extremal boundary for @xmath37 , or 1
. condition ( [ condcycreg ] ) holds , then condition ( [ condcyc ] ) holds with respect to @xmath267 and thus convergence ( [ eqppconvxdfull ] ) takes place . first , note that by ( * ? ? ?
* proposition 5.1(d ) ) , as @xmath98 , @xmath367 & \leq & t{\mathsf{p}}_h \biggl [ \frac{x_0}{b(t ) } > \delta , \tau\bigl(b(t)\bigr ) > m \biggr ] \nonumber \\[-8pt ] \\[-8pt ] & \to&{\delta}^{-\alpha } \bigl(1-g\bigl(\{0\}\bigr ) \bigr)^m . \nonumber\end{aligned}\ ] ] since @xmath366 , the right side of ( [ eqpfgzpos ] ) vanishes as @xmath138 , establishing condition ( [ condeclrd ] ) .
next , to analyze condition ( [ condcyc ] ) , consider the case where @xmath357 is an extremal boundary , and write @xmath368 . for any @xmath369 , @xmath370 & \leq&\sum _ { r=1}^m t { \mathsf{p}}_h \bigl[{x_0}/{b(t ) } > \delta , \tau\bigl(b(t)\bigr)= r , \tau_0 > r \bigr ]
\nonumber \\ & & { } + t{\mathsf{p}}_h \bigl[{x_0}/{b(t ) } > \delta , \tau\bigl(b(t)\bigr ) > m \bigr],\end{aligned}\ ] ] which is obtained by splitting according to whether @xmath371 or the complement and using the fact that @xmath372 and @xmath373 implies @xmath374 . for a typical term in the sum , @xmath375 & \leq & t{\mathsf{p}}_h \bigl[{x_0}/{b(t ) }
> \delta,\tau_0 > r \bigr ]
\\ & & { } - t { \mathsf{p}}_h \bigl[{x_0}/{b(t ) } > \delta
, \tau\bigl(b(t)\bigr ) > r \bigr ] \\ & { \longrightarrow}&{\delta}^{-\alpha } \bigl(1-g\bigl(\{0\}\bigr)\bigr)^r - { \delta}^{-\alpha } \bigl(1-g\bigl(\{0\}\bigr)\bigr)^r = 0,\end{aligned}\ ] ] the convergence following from ( * ? ? ?
* proposition 5.1(d ) ) , since both @xmath65 and @xmath376 are extremal boundaries . the right most term in ( [ eqn : merd ] ) is handled as in ( [ eqpfgzpos ] ) .
finally analyze condition ( [ condcyc ] ) when condition ( [ condcycreg ] ) holds . for any @xmath369 , we have @xmath377 \\ & & \quad\leq\sum_{r=1}^m t { \mathsf{p}}_h \bigl[{x_0}/{b(t ) } >
\delta , \sup_{r < j < { \tau_a } } { { x_j}}/{b(t ) } > a , \tau\bigl(b(t)\bigr)= r \bigr ] \\ & & \phantom{\quad\leq } { } + t{\mathsf{p}}_h \bigl[{x_0}/{b(t ) } > \delta , \tau\bigl(b(t)\bigr ) > m \bigr],\end{aligned}\ ] ] and @xmath378 \\ & & \quad= \int_{(\delta,\infty ] \times(y(b(t)),\infty]^{r-1 } \times [ 0,\infty ] } t{\mathsf{p}}_h \bigl [ ( { x_0 } , { { \boldsymbol{x}}_r } ) /b(t ) \in d(x_0,\boldsymbol x_r ) \bigr ] h_t(x_r ) , \nonumber\end{aligned}\ ] ] where @xmath379\ } } } ( x ) { \mathsf{p}}_{b(t)x } \bigl [ \sup_{1 \leq j < { \tau_a } } { x_j}/{b(t ) } > a \bigr].\ ] ] we claim that @xmath380 whenever @xmath381 . indeed ,
if @xmath78 , then @xmath382 for large @xmath383 such that @xmath384
. otherwise , @xmath385 , and @xmath380 by condition ( [ condcycreg ] ) .
therefore , the integral converges to 0 by combining lemmas 8.2 and 8.4 with theorem 3.2 from @xcite .
applying ( [ eqpfgzpos ] ) completes the proof .
if the finite - dimensional distributions of @xmath269 are jointly regularly varying ( in the sense of ( [ eqecjrvfull ] ) with @xmath39 replacing @xmath323 ) , we obtain a point process limit for @xmath39 under a condition analogous to condition ( [ condeclrdmrv ] ) : _ there exists @xmath386 such that _ @xmath387 = 0 \qquad\mbox{for all } a>0.\ ] ] suppose @xmath327 has a positive recurrent , bounded atom @xmath182 , and @xmath91 .
assume further that , with respect to @xmath267 , @xmath39 is regularly varying in the sense of ( [ eqecjrvfull ] ) , with @xmath39 replacing @xmath323 , and satisfies condition ( [ condcyc ] ) . under condition ( [ condsufflrdfull ] ) ,
both conditions ( [ condeclrdmrv ] ) and ( [ condcycmrv ] ) hold with respect to @xmath267 .
recalling @xmath388 yields ( [ condeclrdmrv ] ) .
next , given @xmath159 , write @xmath389 & \leq&{\mathsf{p}}_h \biggl[\frac{x_0}{b(t ) } > \delta , \sup_{\tau(b(t ) ) < j < { \tau_a } } \frac{{x_j}}{b(t ) } > a \biggr ] \\ & & { } + { \mathsf{p}}_h \biggl[\frac{x_0}{b(t ) } \leq \delta , \sup_{1 \leq j < { \tau_a } } \frac{{x_j}}{b(t ) } > a \biggr].\end{aligned}\ ] ] condition ( [ condcyc ] ) makes the first right side term go to @xmath250 as @xmath390 and for the second term we have , @xmath391 \\ & & \quad\leq\limsup_{t\rightarrow\infty } { \mathsf{p}}_h \biggl[\frac { x_0}{b(t ) } \leq \delta , \sup_{1 \leq j < m_0 ' } \frac{{x_j}}{b(t ) } > a \biggr ] + \limsup_{t\rightarrow\infty } { \mathsf{p}}_h \biggl[\frac{x_0}{b(t ) } \leq \delta , \sup_{m_0 ' \leq j < { \tau_a } } \frac{{x_j}}{b(t ) } > a \biggr ] \\ & & \quad= \mu^ * \bigl([0,\delta ] \times{[{\boldsymbol{0}},\boldsymbol { a}]}^{\mathrm{c } } \bigr ) + \limsup_{t\rightarrow\infty } { \mathsf{p}}_h \biggl[\frac{x_0}{b(t ) } \leq \delta , \sup_{m_0 ' \leq j < { \tau_a } } \frac{{x_j}}{b(t ) } > a \biggr]\end{aligned}\ ] ] where @xmath392 . letting @xmath393 , the first term vanishes by ( [ eqecjrvfull ] ) , since @xmath394 (
* theorem 5.1 ) .
the second term is taken care of by condition ( [ condsufflrdfull ] ) .
we now rephrase theorem [ thmppconvmc ] in terms of our new conditions .
[ thmppconvg ] let @xmath39 be a markov chain on @xmath28 with transition kernel @xmath77 such that @xmath37 has a positive recurrent bounded atom in the sense of ( [ eqatom ] ) , ( [ eqposrec ] ) , and ( [ eqatombdd ] ) .
the initial distribution @xmath126 has a regularly varying tail and satisfies @xmath395 in @xmath263 $ ] .
assume for any @xmath396 that @xmath397 is regularly varying in @xmath398^{m+1 } \setminus\{{\boldsymbol{0}}\})$ ] , @xmath399 \,\mathop{\longrightarrow}^v\,\nu_\alpha(dx_0 ) { \mathsf{p}}_{x_0 } \bigl [ ( t_1,\ldots , t_m ) \in d \boldsymbol x_m \bigr]\ ] ] and that condition ( [ condsufflrdfull ] ) holds with respect to @xmath267 .
a. if @xmath69 , and @xmath37 satisfies condition ( [ condsufflrdz ] ) , then @xmath400 \bigr ) \qquad\mbox{as } n\rightarrow\infty.\ ] ] b. if @xmath124 , and either @xmath357 is an extremal boundary for @xmath37 , or @xmath37 satisfies condition ( [ condcycreg ] ) , then @xmath401 \bigr ) \qquad\mbox{as } n\rightarrow\infty,\ ] ] where the @xmath402 are i.i.d .
geometric random variables with parameter @xmath274 .
recall that @xmath403 are i.i.d .
copies of a markov chain @xmath38 with heavy tailed initial distribution @xmath126 satisfying @xmath404 in @xmath405 $ ] where @xmath31 .
the extremal boundary downcrossing time by @xmath406 is @xmath407 .
let @xmath408 be i.i.d .
copies of the multiplicative random walk @xmath409 .
the hitting time of @xmath250 by @xmath410 is @xmath411 .
a prm on @xmath181 $ ] with mean measure @xmath412 , independent of the @xmath408 is @xmath413 and @xmath188 is the renewal process given by ( [ eqrenewaltimes ] ) with finite mean interarrival time @xmath266 . for convenience ,
write @xmath414 and @xmath415 .
proof of proposition [ thmppconv ] ( a ) first , recall that , under our assumptions @xmath77 and @xmath126 having a regularly varying tail , the convergence ( [ eqecjrv ] ) takes place for the chain @xmath39 on the space @xmath416 \times[0,\infty]^m$ ] , with limit measure @xmath111 given by ( [ eqlimmeas ] ) .
this implies that ( * ? ? ?
* corollary 6.1 , p. 183 ) , @xmath417 in @xmath418 .
since @xmath419 is prm on @xmath420 ( ( * ? ? ?
* proposition 5.3 , p. 123 ) ) , a mapping argument ( ( * ? ? ? * proposition 5.2 , p. 121 ) ) implies @xmath421 in @xmath422 , by ( [ eqlimmeas ] ) .
so we can rewrite ( [ eq : prmlim ] ) in @xmath423 , @xmath424 second , we rescale the time axis to place points at the epochs @xmath425 .
( see @xcite . )
the counting function for the points @xmath188 is @xmath426 $ ] and @xmath427 } $ ] is the left continuous inverse process .
define @xmath428 , so that @xmath429 and @xmath430 is a random element of @xmath431 , the subspace of non - decreasing elements of @xmath432 . by the strong law of large numbers , with probability @xmath433 , @xmath434}{n } \frac{s_{[nt]}}{[nt ] } { \longrightarrow}t\cdot q , \qquad t\geq0,\ ] ] so @xmath435 in @xmath436 .
we transform time points using the mapping @xmath437 given by @xmath438 applying ( * ? ? ?
* proposition 3.1 , p. 57 ) to ( [ eqpfppvecconv ] ) , we have @xmath439 in @xmath440 . since @xmath441 is a.s .
continuous at @xmath442 ( lemma [ lemcontvagueconv ] , p. ) , the continuous mapping theorem gives in @xmath418 , @xmath443 now stack the components of @xmath444 above the time point @xmath445 . to make functionals continuous , it is necessary to compactify the state space by letting @xmath446\times[0,\infty]^m$ ] . define the restriction functional @xmath447 by @xmath448 . from ( * ? ? ? * proposition 3.3 ) , @xmath449 is almost surely continuous at @xmath450 provided @xmath451 = 1 $ ] and since @xmath452 due to @xmath453 , the a.s .
continuity is verified .
therefore , in @xmath454 , the restricted version of ( [ eq : torestrict ] ) is @xmath455 .
define the stacking functional @xmath456)$ ] by @xmath457 or for @xmath458 , @xmath459)$ ] , @xmath460 .
given such @xmath303 with support in @xmath208\times[0,\infty]$ ] , @xmath461 a positive integer , the function @xmath462 , since it is clearly non - negative continuous , and @xmath463 outside of @xmath208\times{\lambda_\delta}$ ] .
the continuity of @xmath464 is clear : given @xmath465 in @xmath454 , we have @xmath466 .
consequently , in @xmath467)$ ] , @xmath468 now adjust the sum over @xmath469 to replace @xmath470 with @xmath471 . from ( [ eqwcppnvst ] ) ,
we readily get , @xmath472 \bigr ) \bigr ) { \rightarrow}{\hat{\eta}}\bigl ( \cdot\cap \bigl([0,\infty ) \times ( 0,\infty ] \bigr ) \bigr ) \qquad\mbox{in } { \mathbb{m}_{p}}\bigl([0,\infty ) \times ( 0,\infty ] \bigr)\ ] ] by noting that any @xmath301)$ ] extends to @xmath473)$ ] with @xmath474 for @xmath475 . moreover , recalling @xmath476 and @xmath477 , the first hitting times of 0 by @xmath478 and @xmath254 respectively , put @xmath479 using this notation , the convergence ( [ eqpfpptrunc ] ) becomes , in @xmath180)$ ] , @xmath480 equation ( [ eqwcppnst ] ) allows spreading the stacks of @xmath481 in time and we verify @xmath482 \bigr).\ ] ] this follows from slutsky s theorem if @xmath483 , where @xmath24 is the vague metric on @xmath484)$ ] ; it suffices to show that @xmath485 \rightarrow0 $ ] for any @xmath486)$ ] and @xmath211 . for such @xmath303 with support in @xmath208 \times[m,\infty]$ ] , for @xmath209 positive integers , we have @xmath487 \\ & & \quad{}= { \mathsf{p}}\biggl [ \biggl|\sum_{k=0}^\infty\sum _ { j=0}^{{\sigma_{k+1}(b_n)}-1 } f \biggl(\frac{s_k+j}{n } , \frac{{x_{kj}}}{b_n } \biggr ) { \mathbf{1}_{\{\frac{{x_{k0}}}{b_n}\geq\delta\}}}\\ & & \hphantom{\quad{}= { \mathsf{p}}\biggl [ \biggl| } - \sum _ { k=0}^\infty\sum_{j=0}^{{\sigma_{k+1}(b_n)}-1 } f \biggl(\frac{s_k}{n } , \frac { { x_{kj}}}{b_n } \biggr ) { \mathbf{1}_{\{\frac{{x_{k0}}}{b_n } \geq\delta\ } } } \biggr| > \gamma \biggr ] \\ & & \quad{}\leq{\mathsf{p}}\biggl [ \sum_{k=0}^\infty\sum _ { j=1}^{{\sigma_{k+1}(b_n)}-1 } \biggl| f \biggl ( \frac{s_k+j}{n } , \frac{{x_{kj}}}{b_n } \biggr ) - f \biggl ( \frac{s_k}{n } , \frac{{x_{kj}}}{b_n } \biggr ) \biggr| \\ & & \phantom{\quad\leq{\mathsf{p}}\biggl [ \sum_{k=0}^\infty\sum _ { j=1}^{{\sigma_{k+1}(b_n)}-1 } } { } \times\epsilon _ { ( \frac{s_k}{n } , \frac { { x_{kj}}}{b_n } ) } \bigl([0,r]\times[m , \infty ] \bigr ) { \mathbf{1}_{\ { \frac{{x_{k0}}}{b_n}\geq\delta\ } } } > \gamma \biggr].\end{aligned}\ ] ] since @xmath303 is uniformly continuous , given @xmath488 , there exists @xmath489 such that @xmath490 whenever @xmath491 .
for @xmath9 so large that @xmath492 , we have @xmath493\times[m,\infty ] \bigr ) { \mathbf{1}_{\ { \frac{{x_{k0}}}{b_n}\geq\delta\ } } } \\ & & \quad < \rho\cdot{\tilde{\eta}_n}\bigl([0,r]\times[m,\infty ] \bigr),\end{aligned}\ ] ] implying that @xmath494 & \leq&\limsup_{n\rightarrow\infty } { \mathsf{p}}\bigl[{\tilde{\eta}_n}\bigl([0,r]\times[m , \infty ] \bigr ) \geq\gamma\rho^{-1 } \bigr ] \\ & \leq&{\mathsf{p}}\bigl[{\tilde{\eta}}\bigl([0,r]\times[m,\infty ] \bigr ) \geq \gamma \rho^{-1 } \bigr]\end{aligned}\ ] ] by ( [ eqwcppnst ] ) .
so ( [ eqwcppnstj ] ) follows by letting @xmath495 .
finally , we remove the restriction in @xmath369 on the stacks .
recall the definitions of @xmath496 and @xmath497 from proposition [ thmppconv ] . to apply a slutsky argument ( e.g. , @xcite , theorem 3.5 , p. 56 ) , we show , for @xmath211 , @xmath498 = 0 \quad\mbox{and}\quad\lim_{m\rightarrow\infty } \limsup_{n \rightarrow\infty } { \mathsf{p}}\bigl [ d_v \bigl(\tilde { \eta}^*_n , { \eta_n}\bigr ) >
\gamma \bigr ] = 0.\ ] ] let @xmath301)$ ] with support @xmath208\times[m,\infty]$ ] . taking @xmath499 , we write @xmath500 hence , @xmath501 & \leq&{\mathsf{p}}\biggl [ \sum _ { k=0}^\infty\sum_{j = m+1}^{\infty } f\bigl(qt_k , i_k{\xi_{k}(j)}\bigr ) { \mathbf{1}_{\{\delta\leq i_k < a\ } } } > \gamma/2 \biggr ] \\ & & { } + { \mathsf{p}}\biggl [ \sum_{k=0}^\infty\sum _ { j = m+1}^{\infty } f\bigl(qt_k , i_k{\xi_{k}(j)}\bigr ) { \mathbf{1}_{\{i_k \geq a\ } } } > \gamma/2 \biggr ] \\ & = & a+b.\end{aligned}\ ] ] writing @xmath502 for @xmath503 , term @xmath182 is bounded by @xmath504 \times[\delta , a ) \bigr ) \sum_{j = m+1}^{\infty } { \mathbf{1}_{\{{\xi_{k}(j ) } > \frac ma\ } } } \biggr\ } > 0 \biggr ] \\ & & \quad\leq{\mathsf{p}}\bigl[\zeta'_m \bigl([0,r / q ] \times [ \delta,\infty ] \times(m/{a } , \infty ) \bigr ) > 0 \bigr],\end{aligned}\ ] ] where , since @xmath254 are i.i.d . and independent of @xmath260 , in @xmath180 \times[0,\infty])$ ] , @xmath505 \bigr).\ ] ] therefore , @xmath506 \times[\delta,\infty ] \times(ma^{-1 } , \infty ) ) > 0 ] = 1-\exp\{-\lambda\}$ ] , where @xmath507 \cdot{\nu_\alpha}[\delta,\infty ] \cdot{\mathsf{p}}\bigl [ \sup_{j \geq m+1 } \xi(j ) > ma^{-1 } \bigr ] \\ & = & r q^{-1 } { \delta}^{-\alpha } { \mathsf{p}}\bigl [ \sup_{j \geq m+1 } \xi(j ) > ma^{-1 } \bigr ] \\ & { \longrightarrow } & 0\end{aligned}\ ] ] as @xmath138 by ( [ eqcondeclrdtch ] ) , a consequence of condition ( [ condeclrd ] ) . for term @xmath508 , we have the bound @xmath509\times[a,\infty ] \bigr ) > 0 \bigr ] = 1-\exp \bigl \{-{\mathsf{e}}\zeta \bigl([0,r / q ] \times[a,\infty ] \bigr ) \bigr\ } = 1-\exp \bigl \{-rq^{-1}{a}^{-\alpha } \bigr\}.\ ] ] letting @xmath510 establishes the first limit in ( [ eqpfconvtog ] ) . to prove the second limit in ( [ eqpfconvtog ] ) , observe that @xmath511 & = & { \mathsf{p}}\biggl[\sum_{k=0}^\infty\sum _ { j = m+1}^{{\tilde{\tau}_{k+1}(b_n)}-1 } f \biggl(\frac { s_k+j}{n } , \frac{{x_{kj}}}{b_n } \biggr ) { \mathbf{1}_{\{\frac{{x_{k0}}}{b_n}\geq\delta\ } } } > \gamma \biggr ] \\
& \leq&{\mathsf{p}}\biggl[\sum_{k=0}^{2rn-1 } \sum _ { j = m+1}^{\infty } f \biggl(\frac{s_k+j}{n } , \frac{{x_{kj}}^{(b_n)}}{b_n } \biggr ) { \mathbf{1}_{\{\frac{{x_{k0}}}{b_n}\geq\delta \ } } } > 0 \biggr ] \\ & & { } + { \mathsf{p}}\biggl[\sum_{k=2rn}^\infty\sum _ { j = m+1}^{\infty } f \biggl(\frac { s_k+j}{n } , \frac{{x_{kj}}^{(b_n)}}{b_n } \biggr ) { \mathbf{1}_{\{\frac{{x_{k0}}}{b_n}\geq\delta \ } } } > 0 \biggr].\end{aligned}\ ] ] the first term is bounded by @xmath512 \leq2rn { \mathsf{p}}\biggl[\frac{x_0}{b_n } \geq\delta , \sup_{j \geq m+1 } \frac{{x_j}^{(b_n)}}{b_n } \geq m \biggr],\ ] ] and @xmath513 = 0\ ] ] by condition ( [ condeclrd ] ) .
the second term is at most @xmath305 = { \mathsf{p}}[{s_{2rn}}/{2rn } \leq1/2 ] \rightarrow0 $ ] as @xmath5 , since @xmath514 a.s . , and @xmath307 by ( [ eqdefq ] ) .
this establishes ( [ eqpfconvtog ] ) , completing the proof of part ( a ) .
\(b ) this amounts to removing the restrictions in @xmath169 , under the additional assumptions ( [ eqecjrvfull ] ) and condition ( [ condeclrdmrv ] ) .
we proceed via a slutsky argument showing that for any @xmath211 , @xmath515 = 0 \quad\mbox{and } \quad \lim_{\delta\rightarrow0 } \limsup_{n \rightarrow\infty } { \mathsf{p}}\bigl [ d_v \bigl ( { \eta_n } , \eta^*_n \bigr ) > \gamma
\bigr ] = 0.\ ] ] let @xmath301)$ ] with support @xmath208\times[m,\infty]$ ] , and note that @xmath516 hence , writing @xmath517 , and @xmath518 , we have @xmath519 & \leq & { \mathsf{p}}\biggl [ \sum _ { k=0}^\infty{\epsilon_{(t_k , i_k ) } } [ 0,r / q ] \times(0,\delta ) \sum_{j=0}^{\infty } { \mathbf{1}_{\{i_k { \xi_{k}(j)}\geq m\ } } } > 0 \biggr ] \\ & \leq & { \mathsf{p}}\bigl [ \zeta ' \bigl([0,r / q ] \times(0,\delta ) \times[m , \infty ] \bigr ) > 0 \bigr].\end{aligned}\ ] ] the @xmath254 are i.i.d . and independent of @xmath520 ,
so @xmath521 on @xmath522 \times [ 0,\infty]$ ] with @xmath523\ ] ] by ( * ? ? ?
* proposition 5.6 , p. 144 ) .
therefore , @xmath524 \times(0,\delta ) \times [ m,\infty ] ) > 0 ] = 1-\exp\{-\lambda\}$ ] , where by lemma [ lemintmom ] , @xmath525 \leq r q^{-1}{m}^{-\alpha } { \mathsf{e}}\bigl [ \sup_{j\geq1 } \xi(j)^\alpha\cdot{\mathbf{1}_{\ { \sup_{j\geq1 } \xi(j ) > m \delta^{-1}\ } } } \bigr].\ ] ] apply ( [ eqcondeclrdtchmom ] ) and dominated convergence as @xmath393 to get @xmath526 and hence the first limit in ( [ eqconvtogrv ] ) . for the second limit in ( [ eqconvtogrv ] )
, we have @xmath527 & = & { \mathsf{p}}\biggl [ \sum_{k=0}^\infty\sum _ { j=0}^{{\tilde{\tau}_{k+1}(b_n)}-1 } f \biggl(\frac { s_k+j}{n } , \frac{{x_{kj}}}{b_n } \biggr ) { \mathbf{1}_{\{\frac{{x_{k0}}^{(b_n)}}{b_n } < \delta\ } } } > \gamma \biggr ]
\\ & \leq&{\mathsf{p}}\biggl[\sum_{k=0}^{2rn-1 } \sum _ { j=0}^{{\tilde{\tau}_{k+1}(b_n)}-1 } f \biggl(\frac{s_k+j}{n } , \frac{{x_{kj}}}{b_n } \biggr ) { \mathbf{1}_{\{\frac{{x_{k0}}^{(b_n)}}{b_n } < \delta\ } } } > 0 \biggr ] \\ & & { } + { \mathsf{p}}\biggl[\sum_{k=2rn}^\infty\sum _ { j=0}^{{\tilde{\tau}_{k+1}(b_n)}-1 } f \biggl(\frac { s_k+j}{n } , \frac{{x_{kj}}}{b_n } \biggr ) { \mathbf{1}_{\{\frac{{x_{k0}}^{(b_n)}}{b_n } < \delta\ } } } > 0 \biggr].\end{aligned}\ ] ] as above , the second term is at most @xmath305 \rightarrow0 $ ] as @xmath5 .
the first term is bounded by @xmath528 \\ & & \quad\leq2rn { \mathsf{p}}\biggl [ \frac{x_0^{(b_n)}}{b_n } < \delta , \sup_{j\geq1 } \frac{x_j^{(b_n)}}{b_n }
\geq m \biggr ] \\ & & \quad\leq2r n{\mathsf{p}}\biggl[\frac{x_0^{(b_n)}}{b_n } < \delta , \sup_{1 \leq j \leq m_0 }
\frac{x_j^{(b_n)}}{b_n } \geq m \biggr ] + 2r n{\mathsf{p}}\biggl[\frac { x_0^{(b_n)}}{b_n } < \delta , \sup_{j \geq m_0 + 1 } \frac{x_j^{(b_n)}}{b_n } \geq m \biggr ] \\ & & \quad= a_n(\delta ) + b_n(\delta),\end{aligned}\ ] ] with @xmath171 as in condition ( [ condeclrdmrv ] ) . for @xmath529 , we have by ( [ eqecjrvfull ] ) , @xmath530 \times { \bigl([0,m ) ^{m_0 } \bigr)}^{\mathrm{c } } \bigr ) = 0.\ ] ] for the second , @xmath531 by condition ( [ condeclrdmrv ] ) .
this establishes ( [ eqconvtogrv ] ) .
for completeness , the following lemma notes the continuity of the map @xmath441 defined in ( [ eqpftimech ] ) .
see also @xcite .
[ lemcontvagueconv ] the mapping @xmath532 given by @xmath533 is continuous at @xmath534 whenever the function @xmath535 is continuous .
\(a ) suppose @xmath536 in @xmath431 ( with respect to the skorohod topology ) , where @xmath141 is continuous , and @xmath537 in @xmath538 .
let @xmath539 with support contained in @xmath208\times b$ ] .
we show that @xmath540 . for @xmath541 ,
write @xmath542 .
the @xmath543 are supported on @xmath544)\times b$ ] , and @xmath544 ) = [ 0,x^{\leftarrow}_n(r)]$ ] , where @xmath545 is the right - continuous inverse of @xmath546 .
we now argue that the @xmath543 , @xmath541 , have a common compact support . indeed , we have @xmath547 pointwise , so @xmath548 .
thus , for large @xmath9 , @xmath549 \times b \subset[0,x^{\leftarrow } _ 0(r)+1 ] \times b$ ] ; without loss of generality , @xmath550 \times b))= 0 $ ] . furthermore , @xmath551
uniformly : suppose @xmath552 .
then @xmath553 since @xmath141 is continuous , and so @xmath554 by the continuity of @xmath303 .
consequently , @xmath555 by ( * ? ? ?
* lemma 8.2(b ) ) .
s.i . resnick and d.
zeber were partially supported by aro contract w911nf-10 - 1 - 0289 and nsa grant h98230 - 11 - 1 - 0193 at cornell university . | the tail chain of a markov chain can be used to model the dependence between extreme observations . for a positive recurrent markov chain , the tail chain aids in describing the limit of a sequence of point processes @xmath0 , consisting of normalized observations plotted against scaled time points . under fairly general conditions on extremal behaviour
, @xmath1 converges to a cluster poisson process .
our technique decomposes the sample path of the chain into i.i.d .
regenerative cycles rather than using blocking argument typically employed in the context of stationarity with mixing . |
the lhcb detector @xcite at the large hadron collider is dedicated to study the physics in the decay of b - flavoured and other heavy hadrons . at the nominal luminosity of @xmath0 at the location of lhcb and a production cross section of @xmath1 at 14tev proton - proton collisions , @xmath2 pairs
are expected to be produced annually .
the modest luminosity requirement for lhcb can be met very early during lhc operation , long before the high luminosity runs and later maintained adjusting the lhc optics when larger luminosities are delivered to the other multi - purpose experiments at the lhc .
lhcb is designed as a single arm forward spectrometer adapted to the angular distribution of the pairs which are produced predominantly at low polar angles as shown in figure [ fig : lhcb ] .
the detector covers an angular region from about 10mrad to 300(250)mrad in the bending ( non - bending ) plane of the magnet .
lhcb uses a warm dipole magnet that delivers an integrated magnetic field of 4tm .
there are two ring imaging cherenkov detectors , rich1 and rich2 located before and behind the magnet , respectively .
these two detectors that use three different radiator materials , aerogel , @xmath3 and @xmath4 , give excellent @xmath5-k separation in a large momentum range of 2 - 100 gev / c . the vertex detector ( velo ) and the full silicon tracking station ( tt - station ) before the magnet and the three tracking stations behind the magnet
provide the tracking system of lhcb .
the stations behind the magnet are divided into the silicon strip inner tracker and the straw tube outer tracker .
the calorimeter system consists of a scintillator pad detector and preshower ( spd / ps ) , a `` shashlik type '' electromagnetic calorimeter , and the fe plus scintillating tile hadronic calorimeter .
the detector is completed with a muon system that uses multi - wire proportional chambers and triple - gem s in the very first muon station .
lhcb is using a level-0 hardware trigger that reduces the initial interaction rate of 16mhz down to 1mhz using transverse energy measured in the calorimeters and the two highest @xmath6 muons seen in the muon chambers .
after a level-0 trigger , the whole detector is read out and the information used in the high level trigger ( hlt ) where in several stages of reconstructing the events , the rate is reduced to 2khz at which the events are stored .
commissioning activities have started in 2007 with the first individual sub systems and safety devices being tested in parallel with the final installation of subdetectors which finished in spring 2008 .
first commissioning activity made use of internal test pulses to test the functionality of the readout and of detector elements .
sending test pulses and receiving the detector responses allows testing the full control and readout chain of the sub - detectors .
dead or noisy channels are located ( and eventually fixed / replaced where possible ) and consistency checks also allow in certain cases to track down possible cabling swapping and alike . from known cable lengths and delays measured using test pulses the initial settings for the time alignment were set for the different sub - detectors . despite the horizontal geometry of the lhcb detector and a very low rate under 1hz of reasonably horizontal cosmic particles within the 250mrad of the horizontal acceptance , over a million cosmic events were recorded in lhcb .
first these rare cosmic events were used to commission the level-0 trigger , aligning calorimeters and later the muon chambers in time to one another as done end of 2007 .
special low thresholds were used in the calorimeter to be able to trigger on mips together with a loose coincidence between ecal and hcal which in the end results in a trigger rate of about 10hz .
a cosmic particle showing a nice `` track '' in the calorimeters is shown in figure [ fig : cosmic ] .
internal time alignment of the various calorimeter cells using the known pulse - shape from test - beam measurements is reaching a precision of about 3ns .
the muon trigger also was used for cosmic events using a coincidence between two stations omitting the vertex constrained used in the real experiment s trigger .
the muon stations were also time aligned using these cosmic events as shown in the left part of figure [ fig : cosmic ] .
time alignment with cosmic muons in lhcb needs to disentangle the forward from the backward going particles . which is nicely displayed in figure [ fig : cosmic ] , showing the timeing for forward and backward particles .
the outer tracker and inner tracker stations located in front of the calorimeter stations ( separated only by the rich2 detector ) were commissioned using the cosmic particles triggered by either the calorimeters or the muon detector . despite the small surface of the inner tracker
, some clear cosmic events could be seen as `` coincidences '' of hits in neighbouring silicon layers within a detector box .
this allowed for a very first rough time alignment of the inner tracker with the calorimeters , using known cable lengths for the internal time alignment . very helpful for this exercise was the possibility in lhcb to trigger on and read out up to 15 consecutive events .
this allowed to read out a window of @xmath77 events around an event triggered by the calorimeters and find the corresponding coincidences in the inner tracker .
two tracks with hits in all three inner tracker stations were found in over a million cosmic triggers .
while commissioning with cosmic events is not feasible for the vertex detector and the tt - station `` far away '' from the calorimeters , we profited here from synchronisation tests in the injection lines done by the lhc machine group . during these tests ,
the injection beam of the sps accelerator was dumped on a beam stopper ( ted ) located about 300 m downstream of lhcb .
this resulted in a large number of about 10gev muons in lhcb and the first tracks seen in the velo .
careful analysis of these data allowed also for first spatial alignment studies of the velo , and some of those results are shown in figure [ fig : velott ] .
it shows the hit residuals as a function of the strip pitch which increases with radius of the radial strips .
the plot shows the performance achieved during the test - beam measurements and the expected performance for binary resolution without any charge sharing between the strips . the current measured resolution after this first alignment step was already well inbetween these two lines . of the velo as a function of the strip pitch which changes with sensor position .
the left plot shows the residual of tracks reconstructed in the velo and extrapolated to one of the tt - station layers.,title="fig:",scaledwidth=28.0% ] of the velo as a function of the strip pitch which changes with sensor position .
the left plot shows the residual of tracks reconstructed in the velo and extrapolated to one of the tt - station layers.,title="fig:",scaledwidth=38.0% ] the tt - station consists of only four individual layers , too little to make stand alone tracking .
however , once can extrapolate the velo tracks and look at the hit residuals in the different layers of the tt - station .
this results in very nice peaks with a widths of about 500@xmath8 and small overall shifts of the order of some hundred @xmath8depending on the layer . for one layer
this is shown in figure [ fig : velott ] .
the spread expected from simple extrapolation of the tracks would be of the order of 300@xmath8due to the resolution in the velo .
this shows a very good initial alignment of the two sub - detectors w.r.t each other as well as a first hint of the internal positioning accuracy of the tt - station modules .
the large number of hits recorded in the silicon trackers during these `` ted - events '' also allowed for a very accurate time alignment of the detectors within a few ns , not achievable with the very limited number of cosmic events .
a reconstructed charge distribution in the inner tracker and a pulseshape derived from varying the relative timing between the beam and the readout clock is shown in figure [ fig : pulseshape ] .
commissioning in lhcb had started as early as 2007 . making extensive use of internal generated test pulses but also cosmic ray events despite the horizontal geometry of lhcb .
very useful proved the beam induced events when the injection beam was dumped in the injection line some 300 m behind the detector which in the end allowed us to get first time alignment and spatial alignment of basically all sub - detectors .
the long shutdown that followed after the lhc incident on september 19 is being used for maintenance and further improvement on the overall efficiency of the detector .
the time is also being used for more tests of the detector readout at the full trigger rate of 1mhz , data processing and writing .
the latter has been successfully tested to speeds up to 1.9khz , close to the design of 2khz . for the year
2009 run , the full hlt computer farm will be available which will allow to take data at full lhcb rate . | lhcb has been installed by spring 2008 , followed by intensive testing and commissioning of the system in order to be ready for first data taking . despite the horizontal geometry of the lhcb detector it was possible to collect over one million useful cosmic events that allowed a first time alignment of the sub - detectors .
moreover events from beam dumps during the lhc synchronisation tests provided very useful data for further time and spacial alignment of the detector . here
we present an overview of our commissioning activities , the current status and an outlook on the startup in 2009 . |
graphene is a very promising table - top laboratory to indirectly probe some of the fundamental mysteries of nature @xcite .
the low energy regime of its @xmath3 electrons is very well described by an effective theory that shares many of the features of a massless dirac quantum field theory ( qft ) in the presence of a background spacetime . in order for graphene to keep its promises ,
we need to have full control of what sort of fields are there and what they represent in a field theory language . in the vast literature on the gauge fields of graphene ( see @xcite , and @xcite for a recent review ) , there are a variety of proposals , sometimes practically valuable for the applications to condensed matter physics , but most of the time unsatisfactory for probing fundamental physics .
the landscape of proposals ranges from @xmath4 monopolelike gauge fields in the case of graphene membranes with intrinsic curvature ( the inflated graphene buckyballs of @xcite ) to a concurrence of a spin - connection field and a @xmath5 field , in the case of purely strained graphene @xcite ( although sometimes non - abelian fields are evoked in this case as well @xcite ) . even in the simplest case , that is purely strained graphene
, there is some confusion : does the spin connection arising from straining graphene give physical effects or not ? and , what is the interpretation of the @xmath5 field from a fundamental point of view ? in this paper , we try to clarify part of this confusion , having in mind to come back to the full scenario of gauge fields in graphene in a forthcoming work .
our principal conclusions are the following : ( 1 ) when only strain is present , spin connection plays no role ; only the @xmath5 pseudogauge field gives physical effects .
( 2 ) when intrinsic curvature is present , spin connection plays a crucial role , to the point that , for small strain , it is spin connection that dictates most of the physics . in addition , we also clarify why some confusion might arise in relation to the non - abelian structure ( that is the lorentz group ) behind the weyl ( abelian ) pure gauge field @xcite . to do that
, we proceed as follows .
we start off with the fully relativistic approach , and proceed by seeing that the weyl field associated to strain ( that is what the spin connection reduces to in this case ) is a pure - gauge field , that can never give a physical effect , unless non - trivial topology is involved . since we are in a pure strain configuration ( elastic deformations ) , and we only focus on local properties ( our samples here have infinite size ) , non - trivial topology of the weyl field is excluded _ a priori_.
later we show , for the first time explicitly , that this fully relativistic approach is equivalent to the standard ( non - fully relativistic , static hamiltonian ) approach .
we do so by showing that all the terms one obtains through the latter method are there also in the former . having the equivalence , this closes the debate on the physical effects of the spin - connection for pure strain . on the other hand ,
a crucial issue arises for the meaning of weyl symmetry itself .
in fact , the weyl field is a pure gauge _ also _ when the membrane has nonzero constant ( hence non - trivial ) intrinsic curvature , and we know that physical effects emerge from there @xcite .
how can we reconcile this , on general grounds , with the previous result ?
we have solved the issue by first finding the equations that the conformal factor of the @xmath0-dimensional theory has to satisfy , in the case of zero curvature ( pure strain ) and in the case of constant curvature .
this is interesting on its own right , as we have found a nontrivial generalization to @xmath0 dimensions of the classic liouville equation of two - dimensional differential geometry of surfaces , and of @xmath6-dimensional conformal field theories ( cfts ) .
we then realize that the key difference between the flat and curved case , although they both give standard weyl invariance , lie in the absence or presence , respectively , of constraints on the @xmath0-dimensional spacetime . therefore , in the curved case , the theory effectively lives in @xmath6 dimensions where weyl symmetry is anomalously broken .
this is an interesting classical manifestation of a quantum weyl anomaly .
the paper flow is as follows : in section [ top - down ] we recall the results of previous work that we need here , and set the notation . in section [ sigmasigma ]
we find the connection between spatial and spatiotemporal conformal factors that allow us to conclude that the spin - connection can not give physical effects for pure strain .
we focus the last part of this section on the nonzero constant curvature case .
we leave to section [ contact ] to show explicitly how the standard hamiltonian and strain tensor language can be recovered from the fully relativistic formalism . after this _
top - down _ approach , in the section [ bottom - up ] , we use the tight - binding condensed matter hamiltonian description , and comment on the peculiarities of the only gauge field that can describe strain .
this latter road , that we call _ bottom - up _
, is actually necessary because , this kind of gauge field could not be guessed within standard qft in curved spacetime . in the concluding section
we also offer some scenarios of fundamental physics that this peculiar gauge field could help to realize .
as is well known ( see , e.g. , @xcite ) , the physics of the large wavelength @xmath3 electrons of graphene , at half - filling , can be efficiently encoded within the dirac massless two - dimensional _
hamiltonian_. in our approach , we include time to make it fully relativistic , although with the speed of light @xmath7 traded for the fermi velocity @xmath8 ( see , e.g. , @xcite ) ; hence , we start from the @xmath0-dimensional action [ flatactionweyl2 + 1 ] a = i v_f d^3 q | ^a _ a , where @xmath9 are the flat spacetime coordinates , @xmath10 are the usual dirac matrices in three dimensions , and we expand around only one of the two dirac points , as we shall focus on strain . as shown in @xcite , the metric @xmath11 can also describe strain ; hence , we shall use the customary dirac action in that curvilinear background .
as also pointed out in @xcite , we recall that although the system is @xmath0 dimensional , the riemann tensor @xmath12 , @xmath13 , has only one independent component , proportional to the gaussian curvature , @xmath14 .
furthermore , surfaces of zero or constant @xmath14 , make the metric flat or conformally flat , respectively , and both cases can be treated at once within a formalism that uses [ metriccf2 + 1 ] g _ ( q ) = e^2 ( q ) _ , where @xmath15 , and the information about the metric being flat or not is encoded in the conformal factor @xmath16 . the coordinates where the metric can be explicitly written in a conformally flat fashion , [ cfcoordinates ] q^(t , x , y ) , are , in general , different from the original coordinates @xmath17 . therefore , the natural candidate action to describe strained graphene is ( for a while we shall use @xmath18 ) [ actionweyl2 + 1 ] a = i d^3 q | e_a^^a ( _ + _ ) , where @xmath19 , with @xmath20 $ ] , the lorentz generators , @xmath21 is the inverse of the three dimensional vielbein @xmath22 ( the dreibein ) , @xmath23 is the spin connection , and being in @xmath0 dimensions , we can write the spin connection in the one - index notation [ omega2 + 1 ] ^a_= ^a b c _ b c . here , @xmath24 is the non - abelian index of the local lorentz group so(2,1 ) , and @xmath25 is the vector index on the spacetime manifold . in this formalism , the index @xmath24 , a tangent spacetime index , plays the role of an internal / gauge index .
the metric can always be written in more suitable coordinates @xmath26 , where @xmath27 is the same time for both coordinate systems , and @xmath28 are the _ spatial isothermal _ coordinates of the surface @xmath29 the single scalar function @xmath30 identifies the surface / graphene membrane .
other isothermal coordinates can be found , say @xmath31 , but the function identifying the surface is always the same could be written as @xmath32 with conformal factor @xmath33 being @xmath34 $ ] and @xmath35 . on the other hand , using the coordinates @xmath36 , and @xmath37 , @xmath38 , with @xmath39 , we recover the standard metric of the sphere @xmath40 . ] : [ intrinsicsigma ] ( ( , ) , ( , ) ) = ( , ) .
when the surface described by @xmath30 has zero or constant curvature , the two metrics and both describe the same spacetime , although with different coordinates , g _ ( q ) = e^2 ( q ) _
= g _ ( ) ; hence , their gaussian curvature can not differ . due to the simple structure of the metric in the two coordinate frames , it is easy to compute the gaussian curvature & = & - e^- 2 [ ^2 ] + & = & e^- 2 where @xmath41 , and @xmath42 . clearly , the two conformal factors are related , @xmath43 : if we know @xmath44 , we can write @xmath45 , and then knowing @xmath46 , we obtain @xmath43 . nonetheless , we have the general equations that @xmath16 has to satisfy for the two cases , for @xmath47 , [ eqforsigmaflat ] = - ( _ a ) ( ^a ) , which corresponds to @xmath30 harmonic functions ( i.e. , solutions of @xmath48 ) , and for @xmath49constant [ eqforsigmacurved ] = - ( _ a ) ( ^a ) + e^ 2 , which corresponds to @xmath30 liouville functions ( i.e. , solutions of @xmath50 ) .
let us focus on the flat case which , since it corresponds to pure strained graphene , is the one of interest here . in this case
, besides the obvious constant solution of , @xmath51 , we also have [ sigmaflat ] _ flat = - ( t^2 - x^2 - y^2 ) + c .
the constant @xmath52 could be set to zero , but we shall keep it , to later compare with the curved case . for the conformal factor , the metric reads @xmath53 hence , nothing constrains the norm of vectors just as for the minkowski case [ normflat ] q ^2 = g _ q^q^= that means that @xmath54 truly describes a flat three - dimensional spacetime .
in order to see explicitly that is in fact the minkowski spacetime , take the following change of coordinates : @xmath55 on these coordinates the line element is @xmath56 , showing that the singularities appearing in the metric are simply coordinates singularities due to our choice of nonstandard coordinates . as a result of that , when we use @xmath57 in the spin connection of [ aomegacf ] _ _ a b = ^c_(_c a ^_b - _ c b ^_a ) _ , we can safely use @xmath58 as a proper dreibein , because it indeed connects the tangent space with a flat manifold . here
@xmath59 , and @xmath60 , and we used the result that in three dimensions @xmath61 . therefore the action for the metric is @xmath62 and when the dirac field is properly transformed [ psiweyl ] = e^- ( q ) , where @xmath63 refers to the minkowskian flat spacetime @xmath64 [ see ] , the action is simply [ flataction2 + 1 ] a = i d^3 q | ^a _ a , which refers to the background metric @xmath64 , which in turn is the unstrained situation .
strain is gone altogether .
it has no physical effects .
another way to see this , is to stop at the action , i.e. , _ before _ implementing the transformation of the spinor as in , and consider the gauge field @xmath65 .
this gauge field is itself a pure derivative , hence it can not produce any physical effect through its field strength f_a
b = _ a _ b - _ b _ a = ( _ a _ b - _ b _ a ) = 0 , as can be seen also by explicitly computing @xmath66 and @xmath67 with @xmath57 in .
the result is zero @xmath68 .
therefore , from here , we see that the very well - known pseudomagnetic field ( and , for what matters , even a putative pseudoelectric field ) induced by pure strain , can not be accounted for by the spin - connection / weyl pure - gauge field .
let us now comment on @xmath65 . as seen , this is a pure gauge field associated to the local weyl transformations and .
indeed , the weyl field transforms as w_w_- _ .
the reason why we do not have here the full weyl gauge field , @xmath69 , but only its pure gauge part , is due to the local weyl invariance of ; see @xcite .
like any other weyl field , @xmath65 is an abelian gauge field .
abelian gauge fields are those routinely used in the graphene literature , @xcite and @xcite . on the other hand
, @xmath65 carries information on the non - abelian structure of the local lorentz group that is encoded in the spin connection .
this information is in the tangent space index `` @xmath24 '' of @xmath65 ; see , and discussion in between . indeed , ( local ) weyl transformations , in general , rephrase spacetime scaling as an internal transformation .
non - abelian gauge fields have also appeared in various discussions on the gauge field approach to strained graphene @xcite .
the above mentioned properties are common to the full weyl gauge field @xmath69 ; hence , they hold also for theories that do not have local weyl invariance .
the extra property of @xmath65 is that it is @xmath70 ; i.e. the true degree of freedom is just one scalar , @xmath16 , the one related to the two - dimensional spatial phonon , @xmath30 , of the graphene membrane , as shown before .
these facts are fully transparent in the coordinates @xmath54 in .
when the coordinates are different , the three aspects of this gauge field(i ) scalar nature , ( ii ) abelian field , and ( iii ) non - abelian lorentz structure get mixed together , and they may appear , in the standard coordinates / frames used in graphene , as originating from different gauge fields , as we show in section [ contact ] . apparently , all seems clear : when @xmath71 , which should describe pure strain , no physical effects can be described by the qft in the curved spacetime approach .
nonetheless , we also saw that when the gaussian curvature of the membrane is constant , a similar procedure could be applied ; hence , this seems to lead to conclude that also in that case as well , there is no physical effect .
however , as we shall now show , this is not the case .
when one solves , e.g. , for negative curvature , @xmath72 , one obtains [ sigmacurved ] _ curved = - ( t^2 - x^2 - y^2 ) + r , and evidently the associated @xmath73 and @xmath74 are zero .
nonetheless , this time the metric reads @xmath75 and the light - cone becomes essentially singular due to the change of signature , typical of horizons in black hole spacetimes .
hence , the light - cone is a proper killing horizon , as shown already when considering hawking effects reproduced on graphene ; see , e.g. , @xcite . let us now see what happens to the length of a position vector , [ normcurved ] q ^2 = g _
q^q^= r^2 > 0 . that is a dramatic difference with the previous case (
amusingly , the difference comes about because of the different multiplicative constant factors : @xmath76 for the flat vs @xmath77 for the curved case ) . in the curved case
the coordinates obey a constraint ; hence , the effective theory lives in one dimension less , with the coordinates that we call @xmath78 , @xmath79 .
the behavior of the flat and curved case is depicted in fig .
[ newspacetimes ] . furthermore
, the @xmath58 in the expression for the spin - connection is not the dreibein anymore : the manifold index @xmath25 now refers to a true curved space .
all of this makes us conclude that the action that corresponds to the curved case ( in the negative curvature case ) is ^(2)_eff i d^2 p | ^(_+ _ ) , where everything has been dimensionally reduced to the two dimensions and , in particular , the metric is the induced curved metric : g^(2 ) _ ( p ) = _ a b . as well known , this two - dimensional theory has a quantum weyl anomaly . hence , interestingly , through the classical constraints we have a manifestation of a quantum weyl anomaly . the weyl ( trace ) anomaly is known to be in one - to - one correspondence with the hawking radiation @xcite .
this is an interesting road to pursue for an alternative computation of the hawking phenomenon on graphene as the one present in the literature @xcite .
to make contact with the literature focused on the phenomenology of graphene ( see , e.g. , @xcite ) we need first to move from the action in the @xmath54 coordinates to the hamiltonian in the @xmath80 coordinates , keeping the curvature radius @xmath81 finite .
for @xmath82 , the @xmath54 coordinates are t & = & e^t / r [ coord1 ] + x & = & e^t / r e^ ( ) [ coord2 ] + y & = & e^t / r e^ ( ) [ coord3 ] to write the hamiltonian in the @xmath83 coordinates , we have to consider the action , legendre - transform it = dtdd()=dt dd , and follow the steps presented in the appendix a. this gives , for @xmath84 , [ h2phiavf(x ) ] h = - i d d ^ ( ^i _ i^ j _ j + v_f + v_f ^i a_i ) where @xmath85 are the pauli matrices , we have re - introduced @xmath86 and @xmath8 , and [ phiexplicit ] = e^2 _ y [ aexplicit ] a_1 = - , a_2 = - , [ fermiveltensor ] _ i^ j = v_f ( cc v_1 & v_1 + v_2 & v_2 + ) , with v_1 & = & + v_1 & = & - + v_2 & = & - + v_2 & = & - . the computations here were carried on for the @xmath16 in , for which we can present the coordinates - ; hence , the expressions for @xmath87 , @xmath88 and @xmath89_i^{\ ; \ ; \ ; \tilde j}$ ] depend of that choice .
nonetheless , even though the detailed expression of those quantities change for @xmath16 in , the structure of the hamiltonian remains the same for the case of interest of pure strain .
we see here that , through this top - down method , we were able to reproduce all the terms except one that customarily appear in the literature of strained graphene ; see , e.g. , @xcite .
the latter is the one gauge field that gives unambiguous physical effects , and that couples to the spinors with an imaginary factor ( an instance that , on its own right , is an indication that such field can not be a weyl field , see , e.g. , @xcite ) . in the next section
we shall extensively comment on this field . before moving to that ,
let us get one step closer to the language usually adopted in the graphene literature .
in fact , the former expressions are written in the language of conformal factors @xmath30 for the membrane and in isothermal coordinates whereas the usual approach employs strain tensors @xmath90 ( where , as customary , the @xmath91 measures the departure from the unstrained position ) , and cartesian coordinates . although we started off with a fully relativistic formalism , due to the structure of the metric , everything of the previous expressions necessarily depends only on the spatial coordinates .
henceforth , we can focus on the spatial metric only and make the customary ansatz that , in cartesian coordinates @xmath92 , [ guij ] g_i j ( x , y ) _ i j + 2 u_i j . on the other hand , this metric is related through a coordinate change to the one in [ legesonsigma ] g_i j ( q ) = _ k l e^2 ( ) = l_i j e^2 ( ( q ) ) .
this needs to be considered in its infinitesimal form , i.e. , @xmath93 , so that @xmath94 , which gives , at first order , @xmath95 ; hence , [ gutildeij ] g_i j ( x , y ) ( _ i j + 2 _ i j ) ( 1 + 2 ) . comparing the two expressions and , we obtain the wanted link between conformal factor , cartesian strain tensor , and isothermal strain tensor u_i j _
i j + _ i j .
near the dirac points ( with our choice of the basis vectors in fig .
[ honeycomb ] , @xmath96 ) , we can describe the behavior of the @xmath3 electrons of graphene with the following dirac hamiltonian , @xmath97 where @xmath98 are two - component dirac spinors , @xmath99 , and @xmath100 , with @xmath101 the usual pauli matrices , all emerging from the honeycomb lattice structure of fig .
[ honeycomb ] . for strained graphene ,
the slope of the fermi cone is changed and the dirac points are shifted @xcite , @xmath102 , where @xmath103 and @xmath104 , so that the hamiltonian is now @xmath105 where @xmath106 and @xmath2 , or @xmath107 where @xmath108 .
this is the usual pseudogauge field interpretation of the strain effect in the graphene sheet @xcite .
the third rank anisotropic tensor @xmath109 , contracted with the strain tensor @xmath110 , is defined in appendix b , and will be analyzed in the following , as this will help to clarify the peculiar ( anisotropic ) nature of this gauge field . for homogeneous strain , the unstrained fermion @xmath111 is related to the strained one @xmath98 via a phase transformation @xmath112 the modification of the hamiltonian in the presence of an inhomogeneous strain is computed in @xcite . in this case
becomes @xmath113 where we follow the dirac prescription @xcite , @xmath111 is the unstrained solution . as a consequence of time - reversal symmetry , each dirac point
is charged with opposite sign .
we can select a point @xmath114 as a reference strain , i.e. , a point where the strain effect @xmath115 could be gauged away .
the phase acquired in could be interpreted as the circulation from the zero strain region @xmath114 to the strained one @xmath116 .
we stress on the fact that the solution is _ formal _ : once the curl of @xmath117 is non - zero , the integral in is path - dependent ( is a non - reversible process ) .
once we realize that the effect of the inhomogeneous strain is to shift the momenta of the fourier modes in a space - dependent way , we can interpret the situation as the @xmath118 fermion is in presence of an _ effective _ gauge field @xmath119 , which could give us a non - zero effective magnetic field if the curl of @xmath119 is non - zero , giving rise to the characteristic landau levels .
this gauge field , frequently called in the literature `` pseudo gauge field '' , is the one we announced in the preceding part of this paper .
clearly , it could not have guessed from the qft in curved space description .
we can follow the trail of the non - trivial behaviour of @xmath119 as the result of the contraction of the strain tensor @xmath110 with the third rank tensor @xmath109 .
this tensor is nonisotropic , which is zero in two dimensions @xcite . ] and its presence is due to the triad @xmath120 specific to the structure of graphene , which is built from two sub - lattices @xcite .
the nontriviality resulting from this contraction could be seen in a simple example @xcite .
consider the deformation vector @xmath121 , where @xmath122 is the maximum value of the strain and @xmath123 is the length of the graphene sample , as represented in fig .
[ strained - graphene ] .
this is a nonsingular vector field , in the sense that its curl and divergence is zero , so it is an irrotational vector field without any source or sink .
however , we notice here that the vector field @xmath119 resulting from the contraction of the corresponding strain tensor with @xmath109 is a clockwise rotational vector field with constant curl , as we can see in fig . [ pseudo - gauge - field ] .
this means that not only does the displacement vector @xmath124 matter but also the _ orientation _ of this vector with respect to the near - neighbor vector triad @xmath125 .
so , the @xmath109 carries some memory of the lattice structure ( short - distance behavior ) , even in the continuum limit ( large - distance behavior ) . to see more concretely that this is the case
, suppose that we have two unstrained graphene sheet samples .
the honeycomb orientation of both samples are such that one of them , say it sample @xmath126 , has the vector @xmath127 of the triad parallel to the @xmath128-axis , as in fig .
[ honeycomb ] , while sample @xmath129 has vector @xmath130 parallel to the @xmath131-axis , see fig .
[ graphene - samples ] .
the orientation of the triad are both the same .
a direct computation shows that the @xmath109 tensor are different for sample @xmath126 and sample @xmath129 .
now let us apply to both samples the same strain deformation vector shown in fig .
[ strained - graphene ] . for the sample @xmath126 ,
the pseudogauge vector field gives us a constant pseudomagnetic field @xmath132 , while for the sample @xmath129 the pseudo - magnetic field is zero .
so , even though in the continuum limit the two samples only differ by their orientation , hence should be indistinguishable by isotropy , the pseudo - magnetic field keeps track of the honeycomb orientation at the lattice level .
a detailed study of the so - called `` memory tensors '' in the graphene honeycomb and kagom lattices is found in @xcite .
once the curl of @xmath117 is not zero , equation is not single - valued , as we can see if we take a loop around the origin .
we are assuming that the variation of the strain tensor @xmath110 is very small so , following the lines of @xcite , we can envisage a process to extract a physical meaning of the solution , even if it is not single - valued .
consider a small planar box on the graphene sheet ( but very large compared with the lattice length @xmath133 ) situated at @xmath134 ( see fig .
[ geometric - phase ] ) . in the case of unstrained graphene ( @xmath135 ) , the solutions for the @xmath3-electrons have the form are operators and the wave packets are these operators applied to the vacuum . ]
now , the idea is to strain the graphene sheet in such a way that in the box the strain is homogeneous ( constant shift @xmath117 of the dirac point as in the previous section ) and the associated magnetic field is almost zero in that region .
the solution of the @xmath3 electrons in the box could be written using the dirac prescription @xmath137 we can see that @xmath118 in is single valued in @xmath138 and @xmath134 locally . now let the box be transported around a loop @xmath52 .
after completion of the loop , there will be a geometrical phase change that can be computed using berry s formula @xcite @xmath139 taking into account that the @xmath118 are normalized , we end up with the result @xmath140 where @xmath141 is the magnetic flux enclosed by the loop .
when only strain is present , the only gauge field that has unambiguous physical effects , is the one just discussed in section [ bottom - up ] .
the structure we saw there , is reminiscent of the gauge field arising in the aharonov - bohm effect , although strictly speaking , in order to have this effect , we need a magnetic flux line crossing the loop @xmath52 and a zero magnetic field outside the line .
this is also reflected in the fact that the flux obtained in is not constant and depends on the area enclosed by the loop .
however , it could be possible to imagine a strain vector concentrated in some region of the graphene and assumed to be zero outside this region . then ,
this procedure , recover the aharonov - bohm result .
the aharonov - bohm effect is not new in the graphene literature and some examples of strain vectors were proposed in order to see this effect with particular procedures , see for instance @xcite .
this gauge field could not have guessed from a top - down approach based on the standard qft in curved spacetime .
in fact , we saw here that if we define standard classical functions on graphene membrane the only fields emerging are the metric and the spin connection , which do not reproduce the behavior of this pseudogauge field .
this @xmath5 field can be put in correspondence with quantum field theoretical structures , such as the quantum anomalies . to see it ,
one just needs to realize that the origin of such gauge field is entirely quantum mechanical , and related to the fact that a `` translation '' in configuration space @xmath142 , that is the straining of the graphene membrane , is necessarily associated with a `` translation '' in momentum space @xmath143 .
hence , in the first quantization language of the wave functions , those operations are carried on by quantum operators @xmath144 that , for the very meaning of quantum mechanics ( that is the heisenberg uncertainty principle ) need to obey [ anomaly ] u(t_1 , b_1 ) u(t_2 , b_2 ) = e^(_1 _ 2 - _ 2 _ 1 ) u(t_2 , b_2 ) u(t_1 , b_1 ) , that are recognized , in the second quantization language , as instances of the nontriviality of the quantum field theoretical vacuum , and in turn of the quantum anomaly @xcite .
therefore , our simple table - top laboratory , can help explore yet another arena of fundamental physics , that is the deep meaning of the quantum anomaly .
on the other hand , this pseudogauge field , for the use of graphene we have in mind , is intriguing also because , as explained in the paper , it is a memory of the lattice structure ( that is the physics of the wavelengths comparable to @xmath133 , the lattice spacing ) in the continuum limit ( that is the physics of the large wavelengths ) .
that means that it is a relic at low energy of the high energy behavior of the system .
this is yet another reason why we can not reproduce this field from a top - down approach , because in such approach the isotropy of the graphene membrane is tacitly assumed .
effects of this kind would be of paramount importance to bring high energy theoretical constructions under the control of experiments .
one example that comes to mind is the standard model extension of @xcite , where tensorial fields that are relics of the lorentz invariant high energy string theory combine with the fields of the standard model ( sm ) and their derivatives within lorentz violating terms that have the form @xmath145 where @xmath146 . in our `` graphene universe '' @xmath147 . we plan to explore these scenarios , and to address the full variety of possible gauge fields in graphene , namely those arising beyond the pure strain limitation , in later work
a. i. acknowledges the czech science foundation ( gar ) , contract no . 14 - 07983s , for support , and
warmly thanks the centro de estudios cientficos ( cecs ) of valdivia in chile , for the kind hospitality while some of this work was carried on .
was partially supported from fondecyt grant 1140155 .
to be specific , let us focus on the case on negative constant gaussian curvature , although similar formulae must hold the pure strain / fully flat case .
the coordinates are t & = & e^t / r + x & = & e^t / r e^ ( ) + y & = & e^t / r e^ ( ) and @xmath148 to write the hamiltonian in the @xmath83 coordinates , we have to consider the action = i dt dx dy e^2 .
we have to write this in the @xmath80 coordinates .
we need the jacobian [ jacobian ] = - e^3t / r the three terms _ t & = & - = - + _ x & = & = e^ ( ) + _ y & = & = e^ ( ) and to re - express the derivatives , e.g. , @xmath149 , etc . , where , as usual , @xmath150 , @xmath151 , etc .
then we use = dtdd()=dt dd , from which we can read off the hamiltonian @xmath84 .
the final expression is h & = & - i v_f d d ( - ) [ a2 ] where we have reintroduced @xmath86 and the fermi velocity @xmath8 .
note that nothing depends on @xmath27 , as it must be .
the formula above for @xmath152 gives a field of the type @xmath153 in = e^2 _ y , the @xmath154 and @xmath155 fields in the non - derivative terms of and ( here we identify @xmath154 and @xmath155 according the pauli matrices @xmath156 and @xmath157 ) a_1 = - , a_2 = - , and the space - dependent , inhomogeneous fermi velocity tensor _
i^ j = v_f ( cc v_1 & v_1 + v_2 & v_2 + ) with v_1 & = & + v_1 & = & - + v_2 & = & - + v_2 & = & - therefore , obtaining the expression we have in the main text .
we start with the effective function of the hopping energy @xmath158 respect to intercarbon distance @xcite , @xmath159,\ ] ] where @xmath160 ev is the equilibrium hopping energy and @xmath161 , with @xmath162 , are the variation of the basis vectors @xmath163 . in order to deal with inhomogeneous strain ,
we consider the expansion of up to the first derivative of the strain tensor @xmath164 , i.e. , @xmath165,\ ] ] where with some abuse of notation , @xmath166 stands for the @xmath167 component of the vector @xmath163 , the indices @xmath168 are contracted ( dummy indices ) and @xmath169 . the tight - binding hamiltonian , in the second quantization formalism , could be written as @xmath170 where @xmath171 stands for the sublattice @xmath172 ( see fig .
[ honeycomb ] ) and , in order to have a hermitian hamiltonian , we added the complex conjugate of the first term .
using the following expressions for the fourier transforms @xmath173 we can write the hamiltonian in the fourier space as @xmath174 where we used the symmetry property of @xmath110 and defined @xmath175 as @xmath176 now , we expand the hamiltonian around one dirac point @xmath177 . for the sake of simplicity , we expand around @xmath178 , such that @xmath179 .
we can work a little bit more in the matrix content of as @xmath180 where in the last equality we made use of the identity , the formula is @xmath181 . ]
@xcite @xmath182 at this point , it will be useful to show the following identities @xcite @xmath183 @xmath184 @xmath185 @xmath186 the tensor @xmath187 , defined in the third identity , is an invariant under the discrete @xmath188 rotations .
this tensor is very important in the discussion about the anisotropy of the strain - induced gauge field . for our choice of basis vectors @xmath189 ,
its only nonzero components are @xmath190 .
we also note that the other three tensors are all isotropic . making use of all this ,
doing some standard algebra and going back to the configurations space via the anti - fourier transforms of @xmath24 , @xmath191 and @xmath192 , we end up with the following hamiltonian for the inhomogeneous strain @xmath193 where @xmath194 is the celebrated space - dependent fermi velocity @xcite , @xmath195 is the pseudogauge field , and @xmath196 is a connection - like coefficient .
the fermi velocity in unstrained graphene is @xmath197 . | we show that , when graphene is only subject to strain , the spin connection gauge field that arises plays no measurable role , but when intrinsic curvature is present and strain is small , spin connection dictates most the physics .
we do so by showing that the weyl field associated with strain is a pure gauge field and no constraint on the @xmath0-dimensional spacetime appears . on the other hand ,
for constant intrinsic curvature that also gives a pure - gauge weyl field , we find a classical manifestation of a quantum weyl anomaly , descending from a constrained spacetime .
we are in the position to do this because we find the equations that the conformal factor in @xmath0-dimensions has to satisfy , that is a nontrivial generalization to @xmath0-dimensions of the classic liouville equation of differential geometry of surfaces .
finally , we comment on the peculiarities of the only gauge field that can describe strain , that is the well known _ pseudogauge field _ @xmath1 and @xmath2 , and conclude by offering some scenarios of fundamental physics that this peculiar field could help to realize . |
since the famous experiments of taylor @xcite in long cylinders many experimental and theoretical investigations of the interesting phenomena of arising and evolution of taylor vortices were done .
the main part of these studies was devoted to cases of long cylinders .
it was done under the assumption that a sufficiently long cylinder and periodic boundary conditions will emulate the infinitely long cylindrical region well .
in addition such regions are convenient for theoretical investigations .
the experimental investigations of short cylinders started much later and gave surprising results ( see , for example , benjamin and mullin @xcite ) , which indicated that the zone and the magnitude of the influence of the caps of the cylinders is much bigger and takes effect on the type of motion in the whole region .
it was assumed that close to the caps of the cylinders the boundary layer is directed from the outer to the inner cylinder as a result of the existence of an ekmans boundary layer .
however , the experiments gave a more complicated picture of the motions .
there exists also an atypical layer in opposite direction , i.e. from the inner to the outer cylinder connected withe the existence of the anomalous mode. the first mathematical model which took the influence of the cylinder caps into account was represented in the work of schaeffer @xcite .
this model is not very consistent and needs improvements .
the role of the anomalous mode was studied later numerically and experimentally in the works of cliffe et al @xcite , @xcite .
the transient features of the circular couette flow were investigated in laboratory and numerical experiments ( using the commercial program _ nekton _
( fluent gmbh ) by neitzel et al @xcite .
the results of the numerical experiments were close to the laboratory experiments . in case of short cylinders with an aspect ratio @xmath2 ( relation of annulus span to the annulus gap width , defined in section [ main ] ) usually only one pair of taylor vortices
arises . in general , this case is considered to be not really interesting due to the poorness of the patterns to be expected .
but in the experiments of aitta et al @xcite a series of interesting flow patterns for short cylinders was found .
a non - equilibrium tricritical points occurs when a forward bifurcation becomes a backward bifurcation . in the taylor - couette problem
the tricritical point can be observed for @xmath3 in the classical case of a pure rotation of the inner cylinder without axial flow as demonstrated in @xcite and later in @xcite .
the system of two vortices is symmetrical if the rotation velocity does not exceed a critical velocity @xmath4 if the velocity of the flow is larger , @xmath5 , then one vortex grows at the expense of the other one . in the case of the forward bifurcation
the symmetry is broken continuously and in the case of the backward bifurcation - abruptly . in the phase transition language
this is a bifurcation of the second order or first order correspondingly @xcite .
the continuous transition was observed for @xmath6 and the abrupt one for @xmath7 where @xmath8 an exhaustive study of the influence of the finite length effects is done for the classical arrangement of two concentric cylinders with a very small aspect ratio @xmath9 $ ] and a relatively large annulus with a radius ratio ( relation between the radii of inner and outer cylinders , see section [ main ] ) of @xmath10 for moderate reynolds numbers @xmath11 $ ] in the work of furukava , watanabe , toya & nakamura @xcite in the case when the inner cylinder is rotating .
the numerical investigations were verified by experiments .
the three main flow patterns were found which correspond to a normal two - cell mode , an anomalous one - cell mode and a twin - cell mode . a steady case as well as an unsteady mode of
the fully developed flow different from wavy taylor - couette flow were obtained .
there exists a series of experimental and theoretical works devoted to the linear stability of the taylor - couette problem with implosed axial effects .
the first such analysis was done for the case of axisymmetric disturbances in a narrow gap @xcite , @xcite . in the work of marques and lopez @xcite the linear stability of the flow in the annulus between two infinitely long cylinders , driven by a constant rotation and harmonic oscillation in the axial direction of the inner cylinder
was analyzed using floquet theory .
the axial effects in the taylor - couette problem were studied in the work @xcite on two different flows . in the first one the axial effect
is introduced by an inertial axial sliding mechanism between the cylinders and in the other one via an imposed axial pressure gradient . in all cases
the annular gap is much larger as in our case and the radius ratio is @xmath12 the stabilizing effect of a periodical axial flow was studied in the works @xcite , @xcite and @xcite . in the work of lueptow
@xcite the stability problem of a taylor - couette flow with axial and radial flow was studied for a relatively large radius ratio @xmath13 and pressure - drive axial flow in the whole annulus . in all these studies
it was found that the axial flow stabilized the taylor - couette flow and enables a lot of super critical states with a rich variety of patterns .
nevertheless the theory of the taylor - couette flow is by far from being complete , so for instance the stability problem for the system of taylor vortices and a classical couette flow by influencing control of the cylinder caps .
it is also a little bit astonishing that the exact analytical solution for the couette flow profile with boundary conditions on cylinder caps was obtained only recently in the work of wendl @xcite . his results
expose a strong influence of the caps on the flow profile .
the difference between the two profiles with and without finite - length effects is approximately a logarithmic function of the aspect ratio @xmath14 for a wide range of radius ratios @xmath15 .
the solution is given in form of a slowly convergent series which contains trigonometric and bessel functions . to give a qualitative analysis of instabilities one takes an exact solution as a base flow to linearize the navier - stokes equations .
this type of solutions is not convenient to pursue the topic further towards analytical studies of the instabilities . in the present work we use the theory of small perturbations to study the instability of the laminar couette flow in a short cylinder .
the geometry of an annular gap and the boundary conditions are given in the section [ main ] .
our idea was to get an analytical formula which is simpler than that in @xcite but which gives us an approximation of the couette flow with finite - length effects which is as good as that found in @xcite . in the section [ sero ]
we describe this solution and discuss the region of applicability of the done approximation .
the solution obtained for a short cylinder is well suitable as a base flow for a comprehensive study of instability effects .
this study is represented in the section [ loc ] .
the onset of the instability is predicted using the method of local analysis , in which the equations governing the flow perturbation are linearized . for the very reason that the appearance of an instability depends not only on the base flow but also on the amplitude and the type of the flow perturbation two different types of perturbations
are studied in this section .
we consider the sensitivity of the couette flow under small perturbations corresponding to an axial flow which is the most important for industrial applications .
we take the case of a very narrow annulus @xmath16 which is typical for bearing or sealing systems and in contrary to general expectation we get a rich family of flow states .
we proved that the axial flow stabilizes the taylor - couette flow also in case of very short cylinders and narrow annular gap with axial flow directed as a thin layer along the surface of the inner rotating cylinder .
the elaborate series of the numerical experiment was done with help of the fluent 5.5/6.0 .
the detailed description of the grids used and others numerical facilities is given in the section [ sec7 ] .
the change from one laminar flow to another or a transition from a laminar flow to turbulence is a function of the base flow considered as well as of the type of small perturbations .
we found a rich family of typical patterns on large intervals of reynolds numbers .
the results of the numerical experiments are discussed in the section [ resul ] .
the study of the couette - taylor problem for the parameters defined below was stimulated by the industrial project @xcite . in many technical systems the problem of transmitting energy and momentum is solved using shafts and corresponding bushes with or without radial sealing . in a typical situation one is concerned with a small taylor - couette annulus , narrow slits in between cylinders and with axial and tangential oscillations or with a weak axial flow along the surface of the inner cylinder . in general , these systems are very complicated and it is necessary to study more simple problems in order to isolate a particular mechanism and to develop and test hypotheses governing their behavior .
we separate here only one of the most typical cases because of its technological importance and , from a fundamental point of view , for its rich dynamics and try to describe all possible flow states in this case .
we consider the motion of the fluid in an annulus between two cylinders as represented in fig .
[ k2 ] . the geometry of the annulus .
the radii of the inner and outer cylinders are equal to @xmath17 correspondingly , the annulus span is denoted by @xmath18 the annular gap width is @xmath19 the corresponding geometrical parameters of the slits are denoted by @xmath20 and @xmath21 other notations are explained in the text.,width=377 ] the annulus dimensions are the gap width @xmath22 , the annulus span @xmath23 , the radius of the inner cylinder @xmath24 , the of an outer cylinder @xmath25 .
the corresponding dimensionless geometric parameters are the aspect ratio @xmath26 and the radius ratio @xmath27 .
we use these notations during the whole paper . for many sealing or bearing systems @xmath2 and @xmath28
we fixe in all our calculations @xmath29 , @xmath30 we assume that the inner cylinder to turn at the constant rate @xmath31 the outer cylinder is at rest , the axial flow has a constant velocity @xmath32 and we suppose @xmath33 the equations of fluid motion and boundary conditions will be rendered dimensionless using @xmath34 as a unit for length .
the radius of the inner cylinder will be denoted by @xmath35 in dimensionless units which will be used in all sections of this paper .
the entrance way and the outgoing way for the axial flow are very thin cylindrical slits with the gap width @xmath36 and the annulus span of slits denoted by @xmath37 the slits are long and narrow in comparison to the main annulus i.e. @xmath38 @xmath39 the ends of the slits narrow conical to the lower order of scale that means the most narrow part of the slit has the width @xmath40 which is very small , @xmath41 . a more detailed picture of the slits is given in fig .
[ 2dgrid ] .
the studied flow states are very sensitive to the kind of perturbations .
so we try to construct a situation which is most adapted to technical facilities and imitate the sealing arrangements with conical tapering slits .
it is characteristically for quite all technical applications that the sealing medium is oil with non vanishing viscosity .
we will consider in our work an incompressible viscous fluid in the cylindrical annulus .
the classical case of the couette flow of an incompressible viscous fluid in an annulus between two cylinders was investigated analytically in @xcite . in comparison to the famous work of taylor the author studied cylinders of a finite length and got a much more complicated formula which describes the influence of the cylinder caps on the flow . as follows from experiments of kageyama ( @xcite ) if the reynolds number grows the influence of the ekmans layer drops along the cylinder caps . as a result
the profile of the solutions differs more and more from the pure couette solution because of the arising of secondary flows in the annulus .
we will consider the flow in the annular gap with the width @xmath34 the span @xmath18 and radius of inner cylinder @xmath24 as on the fig .
[ k2 ] without slits .
we suggest now that the caps of the cylinders contact the inner cylinder and that there is no axial flow in the annulus .
the equations of the fluid motion and the boundary conditions will be rendered dimensionless using @xmath42 and @xmath43 as units for velocity and the pressure .
the scaling of the velocities and of other quantities will change from section to section . due to the axial symmetry two of the components of the velocity vector @xmath44 vanish in this case and the velocity vector reads @xmath45 where @xmath46 is the azimuthal component .
the cylindrical navier - stokes equations can be reduced in this model to one equation , @xmath47 the boundary conditions can be rendered dimensionless as @xmath48,~r\in[0,1 ] .
\label{bcv}\\\ ] ] we make an additional scaling in @xmath49direction by @xmath14 to reach the more convenient situation that the both variables @xmath50 and @xmath51 lie in the interval @xmath52.$ ] the considered case was solved by wendl @xcite using the fourier transformation .
the solution has the following form @xmath53 with @xmath54 . here
@xmath55 and @xmath56 are modified bessel functions .
this solution differs from taylor - couette s simple linear solution especially strong in cases of short cylinders but remains an exact solution for long cylinders .
the series in ( [ sov ] ) converges very slowly and it is difficult to use it for the intended analytical studies .
this was the reason that we looked for simpler analytical representations of this solution or for some good approximated solution .
we study the case of short cylinders , i.e. @xmath57 with a narrow gap , @xmath16 , and @xmath58 under these assumptions and taking into account @xmath59 $ ] we can simplify the initial equation ( [ eqv ] ) by replacing @xmath60 by the constants @xmath61 for the simplified equation which has constant coefficients we obtain the solution in closed analytical form , @xmath62 for @xmath59 $ ] and @xmath63 $ ] . this is a convenient form of the approximated solution for the fundamental mode of the couette flow profile which we will use in the following .
the agreement between the approximated and the exact solutions is very good as can be seen from fig .
[ prof2 ] .
the profiles @xmath46 of the exact ( dashed line ) , the approximated solutions ( continous line ) and numerical solution ( dash - dotted line ) on @xmath49levels equal to @xmath64 with @xmath65 as functions of @xmath66 the numerical solution was found with program fluent 5.5/6.0 in axisymmetric case for @xmath67 , title="fig:",width=264 ] a ) . @xmath68 the profiles @xmath46 of the exact ( dashed line ) , the approximated solutions ( continous line ) and numerical solution ( dash - dotted line ) on @xmath49levels equal to @xmath64 with @xmath65 as functions of @xmath66 the numerical solution was found with program fluent 5.5/6.0 in axisymmetric case for @xmath67 , title="fig:",width=264 ] b ) .
@xmath69 the approximated solution ( [ appsov ] ) is stable for small reynolds numbers @xmath70 and can be used as a base state in a stability analysis .
we note an earlier attempt by vladimirov @xcite to find an approximative solution which takes the influence of the caps into account , which , however , differs much more from the exact solution than ours .
we consider the axisymmetrical stability problem of the taylor - couette flow .
we assume that the velocity vector @xmath71 which describes the fluid motion in the annular gap like on the fig .
[ k2 ] can be represented as a sum of two vectors @xmath72 and @xmath73 @xmath74 where @xmath75 describes the fundamental mode of the couette flow and @xmath76 a perturbation of this mode .
we assume also that the pressure @xmath77 can be also represented as sum @xmath78 under the following assumptions @xmath79 where @xmath80 denotes the spatial gradient of the corresponding velocity , the system of the navier - stokes equations can be linearized . the linearized navier - stokes equations ( lnse ) in the axisymmetrical case
take the form @xmath81 where @xmath82 is the laplace operator , @xmath83 the reynolds number is @xmath84 and @xmath85 denotes the viscosity of the fluid .
the dimensionless variables @xmath86 describing the flow in the cylindrical annulus belong to the intervals @xmath87 , ~~\ z\in\left [ 0 , 1 \right ] . \label{bonrz}\ ] ] here we use the dimensionless variables as introduced in the section [ main ] and additionally we use the value @xmath88 and @xmath89 to render time and velocities dimensionless too .
we leave the same notations for the velocities in the dimensionless case as above .
it means that now @xmath90 and from the conditions ( [ assumplin ] ) @xmath91 follows .
let s introduce a stream function @xmath92 as usual in the axisymmetrical case @xcite @xmath93 in terms of the stream function @xmath92 and the azimuthal velocity @xmath94 the lnse ( [ lnse ] ) turns into a system of two coupled equations @xmath95 we remark that the value @xmath96 is an azimuthal vorticity @xmath97 defined by @xmath98 the basic taylor - couette flow is an axial inhomogeneous flow in the case of the short cylinder because of the cylinder caps influence .
the inhomogeneity leads to additional terms which make a difference between the system ( [ lnse ] ) and the corresponding linearized navier - stokes equations for an infinitely long cylinder .
now we will prove that the additional terms causes the propagation of dispersive waves in the taylor - couette flow .
we remember that in the case of a short cylinder with very narrow gap we have @xmath99 under these assumptions we can simplify the system of equations ( [ azim ] ) by replacing the coefficients @xmath100 ( note @xmath59 $ ] ) by the constants @xmath101 the operators @xmath102 and @xmath103 can be replaced by the multiplicative operators @xmath104 or , in the dimensionless case , by @xmath105 because of the relations ( [ bonrz ] ) and ( [ ger ] ) .
furthermore we assume @xmath106 and neglected @xmath107 in the system ( [ azim ] ) .
we collect now the leading terms of the system ( [ azim ] ) only and obtain a simplified system of equations @xmath108 where no - slip boundary conditions @xmath109 on the inner surface @xmath110 of the annulus are adopted . in the system ( [ upr ] ) we denoted @xmath111 , and the constant @xmath112 is small , @xmath113 . the standard , straightforward way to investigate the stability problem would consist in finding the exact solution of the eigenvalue problem corresponding to the system of equations ( [ upr ] ) . then the perturbations @xmath114 can be represented as a series over the eigenfunctions .
alternatively , the system could be solved numerically whereas a spatial discretization of the problem may be accomplished by a solenoidal galerkin scheme . in our case
these procedures may be accomplished numerically only and this is outside the scope of the present work . nevertheless we can give good qualitative characteristics of the flow in the annulus if we use a local stability analysis .
this method was first developed in the analysis of an inhomogeneous complex astrophysical flow @xcite,@xcite .
it give us the possibility to establish the main physical processes of the considered fluid flow system @xcite , @xcite .
+ the main idea of the local stability analysis method also named a method of infinitesimal perturbations can be described as follows .
we suppose that the amplitude and the gradients of the main state ( the fundamental mode of the couette flow in our case ) is preserved under the perturbations .
this means we keep the values @xmath115 and @xmath116 unaltered or assume that they are constant .
for all components of the perturbation vector @xmath117 and their derivatives taylor series are used which are truncated after the first two terms . from the physical point of view
we assume that the perturbation scale is much smaller than the typical scale of the axial inhomogeneity in the taylor - couette system .
this is the locality condition .
as usual we represent the perturbation as a plain wave , @xmath118 where @xmath119 and @xmath120 are the amplitudes of the waves .
the vector of wave numbers @xmath121 defines the spatial direction of the plane wave propagation .
the locality condition requires that @xmath122 in ( [ planew ] ) , @xmath123 has the dimension of a frequency and in the stability analysis it is called the time increment ( or simply increment ) . in the case of an intrinsic instability of the investigated system @xmath123 will be positive .
this means that the amplitude of the plane wave will infinitely increase with time .
this is the unstable case . the substitution of the plane wave ( [ planew ] ) into the linearized system ( [ upr ] ) results in a system of algebraic equations for the increment @xmath123 , @xmath124 where @xmath125 this system can be solved explicitly and we obtain two solutions for the increment @xmath123 , @xmath126 as a consequence , in the system two different regimes of motion can exist . the motion in the annulus is strong inhomogeneous in @xmath49direction because of the influence of the cylinder caps .
let us now look for local increments on the middle surface @xmath127 on three different lines @xmath128 for @xmath129 the increment does not have any imaginary part and on this line the system does not show any oscillations . in this case the real part of the increment is represented in fig . [ inkre1 ] . on the lines @xmath130 the increment has an imaginary part and the system has oscillations with frequencies which are directly proportional to the gradient of the undisturbed state
. , width=257 ] , width=257 ] , width=257 ] for @xmath131 we have two real solutions for the increment and , consequently , two branches of non - oscillating solutions .
+ the two branches of the increment given by eq.([incr ] ) may have also an oscillating character .
for @xmath132 we deal with two oscillating solutions corresponding to waves of the inertial type . if additionally @xmath133 holds than these two oscillating branches have an intersection point . in the intersection point
@xmath134 we have to do with a time decreasing solution .
on the other hand the perturbations with @xmath135 have the possibility to exchange energy .
the study of the stability problem of the couette flow was done up to now for arbitrary small perturbations .
let s consider the effects of an axial flow near the surface of the inner cylinder .
this flow can be generated by the pair of taylor vortices or an axial flow through the narrow slits in the caps of the cylinders near the surface of the inner cylinder as shown in figure [ k2 ] . in order to simplify the study we suggest that the axial velocity @xmath136 has a nonzero value @xmath137 on the surface of the inner cylinder .
additionally we suggest that the boundary condition is homogeneous in @xmath138direction , i.e. @xmath139 holds . here
we can neglect the dependence on the @xmath140coordinate .
the constant constituent of the axial velocity distribution can be removed by an galilean transformation @xmath141 the boundary condition @xmath142 provides in the narrow gap approximation the linear velocity profile in the @xmath143direction @xmath144 now we can rewrite the lnse ( [ lnse ] ) in the form we will simplify this system of equations .
we remind that we investigate the influence of small perturbations and from the conditions ( [ assumplin ] ) @xmath145 or @xmath146 follows immediately .
we expand the coefficients in the system into a series with respect to the parameter @xmath147 and keep the main terms only .
the approximated system for the stream function @xmath92 and for the azimuthal velocity @xmath94 has the structure @xmath148 as in the general case in the previous point [ gsubs ] we will study the stability problem under small local perturbations . so the perturbation scale is much smaller than a typical scale of the inhomogeneity in the system .
this locality condition implies to the inequalities @xmath149 in the same way we assume that the values @xmath150 and @xmath116 change very slowly in comparison to the perturbations so that they can be represented by taylor series truncated after the first two terms . as usual
we assume that the perturbation @xmath72 has the form of a wave with time dependent amplitudes and with a time dependent wave number in @xmath151direction , @xmath152 where @xmath92 is the stream function ( [ streamf ] ) and @xmath94 is the azimuthal velocity .
the substitution of the formulae ( [ planet2 ] ) into the system of equations ( [ axisys2 ] ) leads to the simplified system of equations @xmath153 the solution of this system can be found for all @xmath50 if the condition @xmath154 is fulfilled because only two terms in the system ( [ timedep ] ) have coefficients which depend on the variable @xmath66 the equation ( [ keq ] ) can be easily solved and we obtain @xmath155 the radial component @xmath156 of a wave number vector @xmath157 increases or decreases linearly in dependence of the sign of the axial component @xmath158 of the wave number vector . in another words
we can explain the action of the axial shear as a change of the radial scale .
the time dependence of the wave number @xmath159 leads to an algebraic non - exponential evolution which can be described by the system of equations @xmath160 perturbations as studied here were considered in fluid dynamics since lord kelvin , but it has been realized only recently that these perturbations can play a trigger role in the subcritical turbulence transition .
these untypically properties of the flow resulting from the fact that the corresponding lnse operator is non - self - adjoint for shear flows such as a plane couette flow @xcite , @xcite , a poiseuille flow @xcite , a counter - rotating taylor - couette flow @xcite , @xcite , a boundary layer @xcite , a kepler rotation @xcite .
apart from the modes of the discrete spectrum there exist plenty of modes of the discontinuous spectrum .
the algebraic time evolution of these modes can achieve large amplitudes which can reach hundredfold of the initial values of amplitudes .
another remarkable property of these modes is their trait to redistribute energy and momentum between different types of perturbations @xcite , @xcite . to give the qualitative description of an impact from an axial flow on the stability of the couette flow we will find the increment for a fixed instant time .
analogously to ( [ planew ] ) we have the representation @xmath161 for the time dependent amplitudes of the wave ( [ planet2 ] ) . if we virtually hold unaltered the time - dependence of @xmath162 we can consider the stability analysis local in time .
the substitution of the formula ( [ inkax ] ) into the system of equations ( [ cylnse ] ) leads to an algebraic equation for the increment .
we get now two solutions of this system for the increment @xmath123 @xmath163 for small @xmath164 the dispersion curves have the same form as considered in the previous section ( without any axial flow ) . in the intersection area
@xmath165 we have now @xmath166 this dispersion equation is similar to the equation for the rossby waves for inhomogeneous rotation of the earth or othetrs systems .
the wave number @xmath167 is no longer fixed as in the previous case ( [ planew ] ) but it has a time dependent value . as a consequence ,
the point @xmath168 or @xmath169 will move along the branches ( similar to the branches on the fig .
[ inkre ] - fig .
[ inkre1 ] ) with time .
chagelishvili and chkhetiani @xcite have proved that under the influence of the shear the flow frequencies of the rossby waves and inertial waves will be modified so that they take the same values .
this circumstance leads to the energy exchange between different modes in the flow .
the same mechanism of the energy exchange is characteristic for the system ( [ cylnse ] ) .
suppose that we start the motion near the intersection point .
one mode is unstable and starts to grow while the second mode will dissipate . in the intersection
point their frequencies are very close to each other and a resonance interaction may take place . in this case
the dissipative mode loses the main part of its energy .
thus , we have found a mechanism for the instability which may be responsible for transport and exchange of energy and angular momentum .
if we analyze the solutions of the system ( [ cylnse ] ) we notice that axial flow leads to a strong differentiation in the energy growth of the perturbations in dependence on the axial level .
this dependence is represented for two reynolds numbers in fig .
. this asymmetry can be the seed for the symmetry breaking in the taylor - couette system in short cylinders .
, title="fig:",height=143 ] a ) @xmath170 , title="fig:",height=147 ] b ) @xmath171 we can assume that the same mechanism work also without any axial flow in the system .
consider two symmetrical counter rotating taylor vortices in the annulus .
then we have in the annular gap a radial distribution of the axial velocity component which is oppositely directed in different parts of the annular gap .
if now a small perturbation amplified the circulation in one of the vortices for a small interval of time then radial gradients of the axial velocity will show up in different parts of the gap .
therefore the degree of growth of perturbations in the continuous spectrum will be different in both parts of the gap and the faster growing perturbation can trigger an irreversible process of symmetry breaking .
the classical taylor - couette problem with pure rotation of one or both cylinders posses @xmath172 and @xmath173 symmetries .
the implementation of an axial flow immediately destroyes the @xmath172 mirror symmetry .
the axial perturbation flow amplifies in - homogeneously different intrinsic modes in the studied states as we proved in the previous section .
it results in a reach variety of super critical stationary and quasi periodical states .
this symmetry breaking for short cylinders can be considered as a generation of certain helicity states .
the helicity is a topological characteristic of the flow and it is the second inviscid integral invariant for the euler equations .
the helicity is defined by @xmath174 dv,\ ] ] where the vector @xmath175 defines the vorticity of the flow .
the helicity is a pseudo scalar because of @xmath176 it is well known that the helicity is an important quantity in the study of dynamics and stability of complicated vortex structures and flows .
the states with maximum of helicity correspond to the states with minimum of energy @xcite , @xcite , @xcite . in the same way as we get the evolution equation for the energy of the system @xmath177 ^{2}dv\ ] ] we obtain the following evolution equation for the helicity , @xmath178 _ { \mathbf{n}}\left ( \frac{\mathbf{u}^{2}}% { 2}-\frac{p}{\rho}\right ) ds-\nu\mathop{\displaystyle \int}\limits_{v}\left [ \nabla\times\mathbf{u}\right ] \cdot\left [ \nabla\times\left [ \nabla \times\mathbf{u}\right ] \right ] dv , \label{evolhel}\ ] ] where @xmath179 is the unit normal vector orthogonal to the boundary surface @xmath180 for the inviscid fluids the right hand side of equation ( [ evolhel ] ) reduces to the first term .
the helicity conserves if the normal component of the vorticity on the boundary surface disappears , @xmath181 _ { \mathbf{n}}=0
$ ] , or if we have a potential flow .
let @xmath110 include a bounded volume of an ideal fluid with @xmath182 and @xmath183 then the helicity will be conserved by the motion . for very small viscosity
@xmath184 the helicity is no longer conserved but it changes very slowly because of the negligibleness of the second term in ( [ evolhel ] ) . as a consequence , for large reynolds numbers we can treat the helicity as an approximately conserved quantity . in a classical taylor - couette system with mirror symmetric boundary conditions we expect an even number of counter - rotating vortices .
the growth of anomalous modes and the phenomenon of appearing of time - dependent states can be considered as a mirror symmetry breaking process and a generation of states with a definite sign of the helicity .
let us represent the helicity in the cylindrical coordinate system and reduce the formula to the case of short cylinders .
the components of the vorticity @xmath185 are @xmath186 using the expressions ( [ vort ] ) we obtain for the helicity @xmath187 in the axisymmetrical case the convenient representation @xmath188 for a narrow annular gap with @xmath189 and @xmath190 we can approximate the formula for the helicity by @xmath191 the axial flow generates a constant stream of helicity and angular momentum through the annulus .
it is an important source of the symmetry breaking in the system .
major advances in computer technologies and numerical techniques have made possible to propose an alternative or at least a complementary approach to the classical and analytical techniques used in laboratory experiments .
computational fluid dynamics becomes part of the design process .
but discontinuities or deep gradients lead to computational difficulties in the classical finite difference methods , although for the finite element methods often a suitable variational principle is not given .
therefore , we consider the finite volume methods , which are based on the integral form over a cell instead of the differential equation . rather than a pointwise approximation at the grid points , we break the domain into _ grid cells _ and approximate the total integral over each grid cell .
it is actually the _ cell average _ , which equals this integral divided by the volume of the cell .
we consider the axial symmetric case as well as the full 3d flow in the annulus .
the numerical method allows to choose a so - called segregated or a coupled solver .
the simulation process consists of ( see e.g. @xcite ) * division of the domain into discrete control volumes ( construction of the computational grid ) , * integration of the governing equations ( in axial - symmetric or in full 3d form ) on the individual control volumes to construct algebraic equations for the discrete variables ( velocity , pressure , temperature and other parameters ) , * linearization and solution of the algebraic system to yield updated values of the variables .
both the numerical methods of the segregated and the coupled solvers employ a similar finite volume process , but the approach used to linearize and solve the equations is different .
the fluent 5.5/6.0 @xcite simulation and post - processing program was applied for the numerical calculation of the flow .
the unsteady 2d axisymmetric or 3d version of the navier - stokes equations is solved as described using finite volume discretization on a structured quadrilateral grid ( fig .
[ 2dgrid ] ) .
2d structural grid for the annulus and the slits with 7690 quadrilateral cells.,width=181 ] the finite volume method has a first order upwind scheme and constant relaxation factors , while the simple procedure for the pressure - velocity coupling is applied .
the grid is generated with the help of gambit program and characterized by a large difference ( 100 times ) between boundary edges .
but it was possible to save the grid structural and quadrilateral .
the smallest inflow and outflow faces of the grid are divided uniformly by 20 nodes , whereas the nodes in the bigger geometry of the axial flow are a more concentrated near the boundary because some vortices can arise there if the reynolds number is large @xcite .
the major volume of the annular gap is divided 50x70 nodes , uniformly in the axial direction and linear condensing in the radial direction .
as next , we set up a 2d axisymmetric model , using a rotating reference frame .
the time step is selected to be @xmath192 of a single rotation .
it helps to determine the flow motion state in the quarters of the rotation circle ( @xmath193 ) .
each time step includes 30 iterations or fewer if the solution converges with the residual of order @xmath194 .
the plane of the 3d grid that corresponds to 2d axisymmetric model have to be coarser , because of the large diameter of the inner cylinder and therefore of the large number of cells in the projection on the @xmath195-plane .
these are only 10 nodes on the inflow and outflow faces in the conical tapering of the slits and 25x30 in the annulus of the slits ( fig . [ 3dgrid ] ) .
3d grid with 1087040 hexahedral cells.,width=359 ] it results in 1087040 hexahedral cells , that is close to the limit for a computation and post - processing on a single processor in an acceptable time .
another , more complicated 3d simulation would be possible using parallel computing @xcite .
such a program is used to being written with the help of message passing interface ( mpi ) @xcite or some other specific approaches @xcite .
but this is the aim of further investigations and developments .
some test simulations showed , that the 3d case is not sensitive enough to determine the very small scale instabilities correctly .
although , the results of the 3d and 2d axisymmetric simulations on the same grid are equivalent .
it was verified using the same coarser grid for the 2d case or by solution of the stationary problem without instabilities . a finer grid for the 2d axisymmetric case , by contrast
, does not give more substantial changes to the behavior of the unstable flow .
so , the 2d axisymmetric case ( fig . [ 2dgrid ] ) with 7690 quadrilateral grid cells is chosen and can be considered without loss of accuracy for all computations in that geometry .
in numerical experiments which simulated the flow in the annular gap as in fig . [ k2 ] and more detailed in fig . [ 2dgrid ]
both axial symmetrical and 3d cases were studied .
3d calculations had been done prepared for a large number of examples .
but it became apparent that no torsion or other remarkable 3d effects arose in the system for @xmath196 .
therefor we used for further numerical investigations axisymmetrical case which saved a lot of time and computer capacity . as a graphical representation of the results in 3-d cases on a paper sheet
does not show the interesting region in sufficient detail so that we represent only one fig .
[ 3dri ] which gives an impression , that the processes under investigation are indeed mostly axisymmetrical .
3d simulation of the two symmetrical stationary vortices .
only one of the vortices is represented to avoid an overload of the left figure .
the counter lines of the velocity magnitude are given in 3d case on the left and in cross sections on the right right by @xmath197 , where @xmath198 @xmath199 @xmath200 @xmath201 @xmath202,width=283 ] 3d simulation of the two symmetrical stationary vortices .
only one of the vortices is represented to avoid an overload of the left figure .
the counter lines of the velocity magnitude are given in 3d case on the left and in cross sections on the right right by @xmath197 , where @xmath198 @xmath199 @xmath200 @xmath201 @xmath202,title="fig:",width=98 ] 3d simulation of the two symmetrical stationary vortices .
only one of the vortices is represented to avoid an overload of the left figure .
the counter lines of the velocity magnitude are given in 3d case on the left and in cross sections on the right right by @xmath197 , where @xmath198 @xmath199 @xmath200 @xmath201 @xmath202,title="fig:",width=94 ] 3d simulation of the two symmetrical stationary vortices .
only one of the vortices is represented to avoid an overload of the left figure .
the counter lines of the velocity magnitude are given in 3d case on the left and in cross sections on the right right by @xmath197 , where @xmath198 @xmath199 @xmath200 @xmath201 @xmath202,title="fig:",width=79 ] in all investigations the inner cylinder rotates with the same azimuthal velocity @xmath203 on the surface of the inner cylinder .
we used this value to render other velocities dimensionless .
the weak axial flow is directed on all figures from left to right .
the entrance and outgoing slits in the ends of conical tapering have the radius ratio @xmath204 we found that there exist four different flow states with typical patterns in dependence on the relation between azimuthal and axial reynolds numbers .
the azimuthal reynolds number @xmath205 is defined by @xmath206 and the axial reynolds number by @xmath207 we studied a pure rotation and then switched on an axial flow to investigate the sensitivity of the flow patterns to small axial perturbations .
the pattern structure is very sensitive to the appearance of an axial flow .
we looked for very weak axial flow with @xmath208 nevertheless the pattern structure changed immediately under the influence of the axial flow .
we studied the states distribution in the regions @xmath209 $ ] and @xmath210.$ ] the results for the most interesting part are that for @xmath211 $ ] and @xmath212 $ ] and they are represented on the schematic diagram fig .
[ tab1 ] . the pattern distribution in dependence on the azimuthal @xmath205 and axial @xmath213 reynolds numbers .
two taylors vortices in the annulus are denoted by @xmath214 , one stable vortex by @xmath215 , natural oscillation states of two vortices by @xmath216,width=529 ] let us consider the pure rotation case without any axial flow .
the most probable state for a short cylinder with an aspect ratio @xmath217 can be described as follows .
the nearly quadratic cross section of the annulus is filled with two counter rotating stable taylor vortices as represented on the fig .
[ stabt ] .
we proved that this structure remains mirror symmetric for a large region of azimuthal reynolds numbers , @xmath218.$ ] the investigated states are denoted in fig.[tab1 ] by boxes , @xmath214 , on the horizontal axes .
all of them have quite the same structure - two stationary taylor vortices in the annulus up to very high azimuthal reynolds numbers .
an enlargement of the azimuthal reynolds numbers leads to the second state in the annulus .
for very large azimuthal reynolds numbers @xmath219 the intrinsic instabilities give rise to perturbed states with a lot of small vortices .
this flow pattern can be interpreted as a region with a fine - grained structure overlaid with a large scale structure - three taylor - like vortices ( see fig . [ fine ] ) . a lot of fine - grained structures of small vortices overlaid by large scale vortices . here
@xmath222 @xmath223 @xmath224,width=170 ] consider now the onset on an axial flow with relatively small velocity .
the existence of the axial flow can be considered as a small perturbation of the base flow with pure azimuthal velocity . in the region of the azimuthal reynolds numbers
@xmath225 the axial flow promoted one of the counter rotating taylor vortices in appropriate direction .
as result the annulus will be filled also for very small axial amplitudes of the axial flow with one stable vortex ( see fig . [ alone ] ) .
we see the abrupt change of the patterns structure from two stable taylor vortices to one by switching on an axial flow .
this is the third pattern structure observed in this system .
it is denoted by @xmath226 on the fig.[tab1 ] . in the region on the reynolds numbers
@xmath227 @xmath228 there exists a branch point in which three possible flow states meet .
for all axial reynolds numbers @xmath229 and @xmath230 it is the first branch for which is characterized by one stable vortex in the annulus ( fig.[alone ] ) . one large stationary vortex
filling out nearly the whole annulus ( denoted on the fig.[tab1 ] by @xmath215 ) . here
@xmath233 @xmath234 @xmath235,width=170 ] for @xmath236 and for the small axial reynolds numbers @xmath237 a second branch with two stable taylor vortices appears ( fig.[stabt ] ) . for @xmath238 and @xmath239 the third branch appears with two quasi periodically natural oscillating vortices ( fig.[unsteady ] ) .
the natural quasi periodic oscillations of a two vortex state in the annulus ( denoted on the fig .
[ tab1 ] by @xmath240 ) in four different moments of dimensionless time . here is @xmath241
@xmath242 @xmath243 @xmath244 @xmath245 , title="fig:",width=90 ] a ) @xmath246 the natural quasi periodic oscillations of a two vortex state in the annulus ( denoted on the fig . [ tab1 ] by @xmath240 ) in four different moments of dimensionless time . here is @xmath241 @xmath242 @xmath243 @xmath244 @xmath245 , title="fig:",width=90 ] b ) @xmath247 ( minimum energy ) the natural quasi periodic oscillations of a two vortex state in the annulus ( denoted on the fig .
[ tab1 ] by @xmath240 ) in four different moments of dimensionless time . here is @xmath241
@xmath242 @xmath243 @xmath244 @xmath245 , title="fig:",width=90 ] c ) @xmath248 the natural quasi periodic oscillations of a two vortex state in the annulus ( denoted on the fig . [ tab1 ] by @xmath240 ) in four different moments of dimensionless time . here is @xmath241
@xmath242 @xmath243 @xmath244 @xmath245 , title="fig:",width=90 ] d ) @xmath249 ( maximum energy ) this branch representes non stationary states . the energy and momentum of the system show also typical quasi periodical oscillations . the time dependence of the energy is given on the fig
[ autooscil1 ] .
the time dependence of the angular momentum is of the same type .
we rendered the energy dimensionless using its value for small reynolds numbers , @xmath250 in fig .
[ autooscil1 ] we also see that maximal values for the energy are smaller in the perturbation state as in the non perturbed flow .
the time dependence of the total energy @xmath251 for the flow by @xmath252 @xmath253,width=377 ] we will distinguish two different types of energy in the system - the rotation or toroidal energy , defined by @xmath254 and the energy caused by axial and local radial flow in the system @xmath255 named also poloidal energy . correspondingly with reynolds numbers the toroidal energy is three order larger as the last one .
but the intrinsic instabilities in the system lead to steady energy exchange between different motion types .
this exchange we demonstrate by a parametric representation in the plane of both energies in fig .
[ autooscil2 ] .
the parametric representation of the energy exchange between @xmath256 and @xmath257 in the plane of both energies by @xmath258 @xmath259.,width=377 ] the flow field between differently shaped bodies of revolution were studied experimentally or theoretically rarely .
the experimental work devoted to the case which is somewhat analogously to the motion in the conical slits at the caps of cylinders in our case was done by wimmer @xcite . for a rotating cylinder in a stationary cone he proved that due to the non constant gap different types of motion can coexist in the gap . by the growth of the reynolds numbers the first vortices appear at the location of the largest gap size and
are deformed , having larger axial extension .
the numerical studies in our geometry confirm this statement as can be seen on the fig . [
the convenient asymptotic representation for the exact steady solution of the taylor - couette problem was found in the case of closed short cylinders @xmath260 ( section [ sero ] ) . the found approximated solution ( [ appsov ] )
will be an exact solution for infinitely large viscosity or for @xmath261 . for
finite reynolds numbers a flow in a thin layer along the caps of the cylinder can arise .
we estimated the radial and axial components of its velocity . at first blush in the case of the strong nonlinear motion
the estimation @xmath262 @xmath263 must hold .
this follows from the comparison of the nonlinear terms in the navier - stokes equations .
but the numerical computations for the short cylinders with @xmath264 and large reynolds numbers lead to another estimation , @xmath265 @xmath266 which can be obtained from the navier - stokes equations in the case of very large viscosity . in fig .
[ tab2 ] we represent the radial component of the dimensionless velocity only . for the other component
the pattern distribution is quite the same .
the distribution of the radial component of the dimensionless velocity @xmath267 in the taylor - coutte flow for different states in dependence on the reynolds numbers @xmath205 and @xmath268 two taylors vortices in the annulus are denoted by @xmath214 , one stable vortex by @xmath215 , natural oscillation state of two vortices by @xmath216,width=529 ] from the estimations it follows that the azimuthal component of the velocity has the main influence on the stability of the system .
consequently , we can use for our qualitative stability analysis the found approximated solution . the stability analysis done in section [ loc ]
showed that the inhomogeneity of the base flow in the radial direction as well as the axial flow cause a perturbation of the rossby waves type .
we proved that the @xmath269symmetry breaking due to the axial flow in the taylor - couette flow is in the same time the helicity symmetry breaking ( section [ helica ] ) .
the states with non - zero helicity can accrue in the system ( section [ resul ] ) .
the energy oscillations in the case of the unsteady pattern with natural oscillations as in fig .
[ autooscil1 ] correspond to the helicity oscillations in the system.this can be seen from fig .
[ unsteady ] .
the weak axial flow in an annulus between the rotating inner cylinder and the fixed outer cylinder has several important engineering applications in particular if the axial flow is directed as a thin layer along the surface of the rotating cylinder .
we studied this case and proved that the axial flow stabilizes the motion and that in a system with very narrow annulus different super critical states are possible ( section [ resul ] ) .
we remark that experiments in short cylinders with an aspect ratio of @xmath2 without any axial flow showed a poor variety of states @xcite , @xcite .
but for smaller aspect ratio all the states which we described in our case can be observed .
this behavior of the taylor - couette system can be interpreted as follows .
the influence of the weak axial flow along the surface of the inner cylinder can be considered as a change in the geometry , i.e. , as a shortening of the axial length of the cylinder .
this interpretation is supported by simple estimations . for technical applications
it is important to describe not only the velocity field but also the pressure distribution in the device .
the pressure differences in the annulus coming from the vortex motion can be estimated as @xmath270 of the pressure difference corresponding to the axial flow .
the authors are grateful m. v. babich , s.k .
matveev , v. m. ponomarev , a.i .
shafarevich and g. windisch for interesting and fruitful discussions .
the work was kindly supported by the bmbf- project grant number pim3cb under the leadership of s. pickenhain . | we analyze the stability of a cylindrical couette flow under the imposition of a weak axial flow in case of a very short cylinder with a narrow annulus gap .
we consider an incompressible viscous fluid which is contained in the narrow gap between two concentric short cylinders , where the inner cylinder rotates with constant angular velocity . the caps of the cylinders have narrow tubes conically tapering to super narrow slits which allow for an axial flow along the surface of the inner cylinder .
the approximated solution for the couette flow for short cylinders was found and used for the stability analysis instead of the exact but bulky solution .
the sensitivity of the couette flow to general small perturbations and to the weak axial flow was studied .
we demonstrate that perturbations coming from the axial flow cause the propagation of dispersive waves in the taylor - couette flow .
the coexistence of a rotation and of an axial flow requires to study in addition to the energy and the angular momentum also the helicity of the flow .
the approximated form for the helicity formula in case of short cylinders was derived .
we found that the axial flow stabilizes the taylor - couette flow .
the supercritical flow includes a rich variety of vortical structures including a symmetric pair of taylor vortices , an anomalous single vortex and quasi periodic oscillating vortices .
pattern formation was studied at large for rated ranges of azimuthal and axial reynolds numbers .
a region where three branches of different states occur was localized .
numerical simulations in 3d and in axisymmetrical case of the model flow are presented , which illustrate the instabilities analyzed .
* interaction of a rotational motion and an axial flow in small geometries for a couette - taylor problem * l. a. bordag@xmath0 , o. g. chkhetiani@xmath1 , m. frhner@xmath0 and v. myrnyy@xmath0 + @xmath0 _ fakultt mathematik , naturwissenschaften und informatik + brandenburgische technische universitt cottbus + universittsplatz 3/4 , 03044 cottbus , germany _
+ @xmath1_space research institute , russian academy of sciences , + profsoyuznaya 84/32,117997 moscow _
+ key words and phrases : taylor - couette flow , stability under axial flow , pattern structures + pacs numbers : 47.32.-y , 47.20.-k , 47.54.+r + |
einstein , podolsky , and rosen ( epr ) showed that quantum mechanics is incomplete , in the sense that not every element of reality has a counterpart inside the theory @xcite .
epr proposed the following criterion to identify an element of reality : `` if , without in any way disturbing a system , we can predict with certainty ( i.e. , with probability equal to unity ) the value of a physical quantity , then there exists an element of physical reality corresponding to this physical quantity '' @xcite . in practice
, nondisturbance can be guaranteed when the measurements are performed on distant systems .
predictions with certainty are possible for states having perfect correlations .
a quantum state @xmath2 has @xmath3 perfect correlations when there are @xmath3 different observables @xmath4 such that @xmath5 .
thirty years after epr s paper , bell proved that there is a irresoluble conflict between epr s elements of reality and quantum mechanics @xcite .
all - versus - nothing ( avn ) proofs are the most direct way to reveal this conflict .
an avn proof is based on a set of @xmath6 perfect correlations of a specific quantum state .
the name `` all - versus - nothing '' @xcite reflects one particular feature of these proofs : if one assumes epr elements of reality , then @xmath7 of these perfect correlations lead to a conclusion that is the opposite of the one obtained from a subset of the other @xmath8 perfect correlations .
if all @xmath6 correlations are essential to obtain a contradiction ( i.e. , if the contradiction vanishes when we remove one of them ) , then the avn proof is said to be critical .
the first avn proof was obtained by heywood and redhead @xcite .
however , the most famous avn proof is greenberger , horne , and zeilinger s ( ghz ) @xcite .
the first bipartite avn proof with qubits is in refs . @xcite .
the first bipartite avn proof with qubits and using only single - qubit measurements is in refs . @xcite .
the interest of the case in which the parties are restricted to perform single - qubit measurements is motivated by the practical difficulty of making general @xmath9-qubit measurements ( @xmath10 ) when the qubits are encoded in different degrees of freedom of the same particle .
recently , several @xmath0-qubit @xmath1-particle states ( @xmath11 ) having perfect correlations have been experimentally prepared , for instance , 4-qubit two - photon @xcite , 6-qubit two - photon @xcite , 6-qubit four - photon @xcite , 8-qubit four - photon @xcite , and 10-qubit five - photon graph states @xcite . for these @xmath0-qubit @xmath1-particle states ,
a natural problem is the following : consider @xmath1 distant parties ; party @xmath12 can perform single - qubit measurements on particle @xmath12 , particle @xmath12 contains @xmath13 qubits ( @xmath14 ) ; which @xmath0-qubit @xmath1-particle states allow @xmath1-partite avn proofs ?
this problem has been solved for the case of @xmath15 particles / parties @xcite . in this article
we address the problem for an arbitrary number @xmath1 of particles / parties .
the article is organized as follows : in sec .
[ sec:2 ] , we show that there is an equivalence between pure states allowing avn proofs and graph states .
this will simplify the task of finding all inequivalent @xmath0-qubit @xmath1-partite avn proofs .
an @xmath1-partite avn proof is specific for an @xmath0-qubit @xmath1-particle graph state when there is no graph state with fewer qubits satisfying the same correlations . in sec .
[ sec:3 ] we discuss the requirements of an @xmath1-partite avn proof to be specific for an @xmath0-qubit @xmath1-particle graph state , and describe a method to decide whether a given @xmath0-qubit @xmath1-particle graph state allows a specific @xmath1-partite avn proof .
we apply this method to decide whether some @xmath0-qubit @xmath1-particle graph states recently prepared in the laboratory allow @xmath1-partite avn proofs . as supplementary material @xcite , we provide a computer program to decide whether a given @xmath0-qubit @xmath1-particle graph state allows a specific @xmath1-partite avn proof . in sec .
[ sec:4 ] we solve the following problem : given an @xmath0-qubit graph state , what is the minimum number @xmath1 of parties that allows a specific @xmath1-partite avn proof . as supplementary material
( see footnote 1 ) , we provide a computer program to obtain , given an @xmath0-qubit graph state , all distributions between @xmath1 parties and all distributions between a minimum number of parties which allow avn proofs .
the solution of the previous problem allows us to obtain all inequivalent distributions allowing avn proofs since any distribution obtained from one allowing a specific avn proof by giving qubits that originally belong to the same party to new parties will also allow an avn proof . as supplementary material
, we provide all inequivalent distributions between a minimum number @xmath1 of parties allowing specific @xmath1-partite avn proofs for all @xmath0-qubit graph states of @xmath16 qubits . for all inequivalent distributions between a minimum number @xmath1 of parties allowing specific @xmath1-partite avn proofs for all @xmath0-qubit graph state of @xmath16 qubits . ]
an avn proof requires an @xmath0-qubit quantum state distributed between @xmath1 parties .
this state has a set of perfect correlations between the results of single - qubit measurements .
these correlations must satisfy two requirements .
first , they must allow us to define @xmath1-partite epr s elements of reality .
this means that every single - qubit observable involved in the avn proof must satisfy epr s criterion of elements of reality ( i.e. , its value can be predicted with certainty using only the results of single - qubit measurements on distant particles ) .
second , they must lead to a contradiction when epr s criterion of elements of reality is assumed .
therefore the conclusion of an avn proof is that if the quantum predictions are correct , observables which satisfy epr s condition can not have predefined results since it is impossible to assign them values which simultaneously satisfy the perfect correlations predicted by quantum mechanics .
perfect correlations are necessary to establish elements of reality and to prove that they are incompatible with quantum mechanics . therefore the states which allow avn proofs must be simultaneous eigenstates of a sufficient number of commuting @xmath0-fold tensor products of single - qubit operators . indeed ,
the following observations lead us to the conclusion that without loss of generality , we can restrict our attention to a particular family of states .
two different single - qubit operators @xmath17 and @xmath18 on the same qubit can not commute .
a necessary condition to make @xmath0-fold tensor products be commuting operators is to choose @xmath17 and @xmath18 to be anticommuting operators .
therefore , in an avn proof , all single - qubit operators corresponding to the same qubit must be anticommuting operators .
the maximum number of anticommuting single - qubit operators is three .
therefore , without loss of generality , we can restrict our attention to a specific set of three single - qubit anticommuting operators on each qubit , for example the pauli matrices @xmath19 , @xmath20 , and @xmath21 .
this leads us to the concept of stabilizer state .
an @xmath0-qubit stabilizer state is the simultaneous eigenstate with eigenvalue @xmath22 of a set of @xmath0 independent commuting elements of the pauli group ( i.e. , the group , under matrix multiplication , of all @xmath0-fold tensor products of @xmath23 , @xmath24 , @xmath25 and the identity @xmath26 ) .
the @xmath0 independent elements are called stabilizer generators and generate a maximally abelian subgroup , the stabilizer group of the state @xcite . the @xmath27 elements of the stabilizer group are the stabilizing operators and provide all the perfect correlations of the stabilizer state .
a further simplification is possible since any stabilizer state is local clifford equivalent ( i.e. , equivalent under the local unitary operations that map the pauli group to itself under conjugation ) to a graph state @xcite .
therefore the problem of which @xmath0-qubit pure states and distributions of qubits between the parties allow @xmath1-partite avn proofs is reduced to the problem of which @xmath0-qubit graph states and distributions allow @xmath1-partite avn proofs .
a graph state @xcite is a stabilizer state whose generators can be written with the help of a graph .
@xmath28 is the @xmath0-qubit state associated with the graph @xmath29 , which gives a recipe both for preparing @xmath28 and for obtaining @xmath0 stabilizer generators that uniquely determine @xmath28 . on one hand , @xmath29 is a set of @xmath0 vertices ( each representing a qubit ) connected by edges ( each representing an ising interaction between the connected qubits ) . on the other hand , the stabilizer generator @xmath30 is obtained by looking at the vertex @xmath12 of @xmath29 and the set @xmath31 of vertices which are connected to @xmath12 and is defined by @xmath32 where @xmath33 , @xmath34 , and @xmath35 denote the pauli matrices acting on the @xmath12th qubit .
@xmath28 is the only @xmath0-qubit state that fulfills @xmath36 therefore the stabilizer group is @xmath37 where @xmath38 denotes a subset of @xmath39 .
the stabilizing operators provide all the perfect correlations of @xmath28 : @xmath40 graph states associated with connected graphs have been exhaustively classified .
there is only 1 two - qubit graph state ( equivalent to a bell state ) , only 1 three - qubit graph state ( equivalent to a ghz state ) , and 2 four - qubit graph states ( equivalent to a ghz and a cluster state ) , 4 five - qubit graph states , 11 six - qubit graph states , 26 seven - qubit graph states @xcite , and 101 eight - qubit graph states @xcite .
the perfect correlations of any graph state associated with a connected graph of three or more vertices lead to contradictions with the concept of elements of reality when each qubit is distributed to a different party @xcite .
however , the first problem consists of finding whether these contradictions are specific to a given distribution of a graph state or , on the contrary , they can be obtained with a graph state of fewer qubits .
for example , take the four - party avn proof based on the following four perfect correlations of the distribution of the four - qubit fully connected graph state @xmath41 ( a four - qubit ghz state ) in which each qubit belongs to a different party : @xmath42 this is an example of an avn proof which is nonspecific for the state @xmath41 , the reason being that neither the contradiction nor the definition of the elements of reality involved in this contradiction requires any choice from the party which has the fourth qubit .
this party only has to measure @xmath43 and broadcast the result .
the only role of the result of @xmath43 is to guarantee that @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath48 , and @xmath49 are elements of reality in a four - party scenario .
however , the contradiction occurs for any result of @xmath43 .
it occurs because the following equations can not be simultaneously satisfied : @xmath50 the particular value of @xmath43 is irrelevant .
the same contradiction can be obtained using the perfect correlations of a three - qubit fully connected graph state @xmath51 ( a three - qubit ghz state ) distributed between three parties .
the next example illustrates that whether an avn proof is specific can depend on the way in which the qubits are distributed between the parties .
consider the avn proof based on the following four correlations of the four - qubit linear cluster state @xmath52 associated with the graph where qubit @xmath22 is connected to qubit @xmath53 , which is connected to qubit @xmath54 , which is connected to qubit @xmath55 : @xmath56 if the qubits are distributed so that each qubit goes to a different party , then the avn proof is not specific since the party who has the fourth qubit does not need to make any choice , neither for the contradiction nor for the definition of the elements of reality .
the contradiction @xmath57 can be obtained from the perfect correlations of a three - qubit linear cluster state @xmath58 associated with the graph where qubit @xmath22 is connected to qubit @xmath53 , which is connected to qubit @xmath54 . however , if qubits @xmath22 and @xmath55 belong to alice , and qubits @xmath53 and @xmath54 belong to bob , then the only way to guarantee that , for example , @xmath46 is an element of reality in this scenario ( i.e. , that its result can be predicted using only the results of measurements on alice s side ) is by using the following perfect correlation of the @xmath52 : @xmath59 therefore the party who has qubit @xmath55 must choose between at least two measurements . to sum up
, an avn proof is specific for a given distribution of a graph state when at least two observables of all the qubits are involved .
since the additional correlations needed to define the elements of reality can ( together with those already used for the contradiction ) involve additional contradictions , it is appropriate that the observables needed to guarantee that other observables are elements of reality ( like @xmath60 and @xmath43 in the previous example ) are themselves elements of reality . therefore hereinafter we will focus on avn proofs in which at least two of the observables of all the qubits are elements of reality .
it can be easily seen that when two pauli observables , for example , @xmath33 and @xmath34 , are elements of reality , then the third pauli observable , @xmath35 , is also an element of reality .
therefore we shall focus only on those graph states and distributions in which the three pauli observables of each and every one of the qubits are elements of reality .
the next problem is , given a distribution of an @xmath0-qubit graph state between @xmath1 parties , how to decide whether it is one in which all single - qubit pauli observables are elements of reality .
for that purpose , it is useful to note that the @xmath27 perfect correlations ( i.e. , stabilizing operators ) of an @xmath0-qubit graph state can be classified in four classes : \1 .
there are @xmath61 stabilizing operators ( i.e. , a quarter of the stabilizing operators of the graph state ) that allow us to predict @xmath33 from the results of measurements on other qubits : those that are products of the stabilizer generator @xmath30 [ defined in eq .
( [ grule ] ) ] , an even number ( hereinafter `` even '' includes zero ) of @xmath62 with @xmath63 , and an arbitrary number ( hereinafter `` arbitrary number '' includes zero ) of @xmath64 with @xmath65 and @xmath66 .
there are @xmath61 stabilizing operators that allow us to predict @xmath34 from the results of measurements on other qubits : those that are products of @xmath30 , an odd number of @xmath62 with @xmath67 , and an arbitrary number of @xmath64 with @xmath65 and @xmath68 .
there are @xmath61 stabilizing operators that allow us to predict @xmath35 from the results of measurements on other qubits : those that are products of an odd number of @xmath62 with @xmath63 and an arbitrary number of @xmath64 with @xmath65 and @xmath69 .
there are @xmath61 stabilizing operators that contain @xmath70 : those that are products of an even number of @xmath62 with @xmath67 and an arbitrary number of @xmath64 with @xmath65 and @xmath68 .
each particle can carry more than one qubit .
it is therefore convenient to denote as @xmath71 the set of qubits which are in the same particle as qubit @xmath12 .
the previous classification of the stabilizing operators is useful in the following sense : given the distribution of an @xmath0-qubit graph state between @xmath1 parties , @xmath33 is an element of reality if and only if there exists a stabilizing operator of the graph state which satisfies the following two requirements : ( 1 ) it does not contain @xmath62 for all @xmath72 but contains an even number of @xmath64 with @xmath73 and ( 2 ) it contains @xmath30 and an even number of @xmath74 with @xmath75 .
for instance , consider the four - qubit linear cluster state @xmath52 associated with the graph where qubit @xmath22 is connected to qubit @xmath53 , which is connected to qubit @xmath54 , which is connected to qubit @xmath55 , distributed such that alice has qubits 1 and 4 and bob has qubits 2 and 3 .
the question is , is @xmath44 an element of reality ?
this is equivalent to the question , is there a stabilizing operator such that it does not contain @xmath76 [ since @xmath77 but contains an even number ( necessarily zero ) of @xmath78 [ since @xmath79 and @xmath80 and an even number ( necessarily zero ) of @xmath81 [ since @xmath82 ?
the answer is yes ; the only stabilizing operator with these properties is @xmath83 .
similarly , @xmath34 is an element of reality if and only if there is a stabilizing operator satisfying ( 1 ) and the following condition : ( 3 ) it contains @xmath30 and an odd number of @xmath74 with @xmath75 . finally , @xmath35 is an element of reality if and only if there is a stabilizing operator satisfying ( 1 ) and the following condition : ( 4 ) it does not contain @xmath30 but contains an odd number of @xmath74 with @xmath75 . to decide whether a specific distribution allows a specific avn proof , we first focus on qubit @xmath12 and test whether @xmath33 and @xmath34 are elements of reality .
if either is not an element of reality , then the distribution does not allow a specific avn proof .
if both are elements of reality , then we test whether @xmath84 and @xmath85 of qubit @xmath86 are elements of reality , and so on for all the qubits .
if all @xmath33 and @xmath34 are elements of reality , then the distribution allows a specific avn proof . indeed , there are simple cases where it can easily be seen that a distribution does not allow an avn proof .
for example , if more than @xmath87 qubits are carried by the same particle , for qubits of the particle with more than @xmath87 qubits , either requirement ( 1 ) is incompatible with ( 2 ) , or ( 1 ) is incompatible with ( 3 ) and ( 4 ) .
an alternative proof will be provided in sec . [ sec:4 ] . if there is a qubit @xmath12 such that @xmath88 ( i.e. , if in the graph representing the state , qubit @xmath12 is connected only to qubits of the same particle ) , requirement ( 1 ) is incompatible with requirements ( 3 ) and ( 4 ) .
as supplementary material ( see footnote 1 ) , we provide a computer program to decide whether a given @xmath0-qubit @xmath1-particle graph state allows a specific @xmath1-partite avn proof . as an example of the application of these rules , it is interesting to discuss whether some recently prepared 6-qubit two- and four - particle states allow specific avn proofs , assuming the natural scenario in which each party has one particle
between two and four particles .
each gray area represents a particle .
( a ) corresponds to the four - photon state prepared in ref .
( b ) corresponds to the two - photon state prepared in ref .
@xcite . in ( a ) ,
not all single - qubit pauli observables are epr elements of reality , and therefore no avn proof is possible .
( b ) and ( c ) allow avn proofs . ]
figure [ fig1 ] contains several possible distributions of a six - qubit linear cluster state @xmath89 .
figure [ fig1](a ) represents the four - photon @xmath89 prepared in ref .
this distribution does not allow a specific avn proof since qubit 1 is connected only to qubit 2 and qubit 6 is connected only to qubit 5 .
figure [ fig1](b ) represents the two - photon @xmath89 prepared in ref .
this distribution satisfies all the requirements thus allows a specific avn proof .
indeed , fig .
[ fig1 ] ( b ) represents the only bipartite distribution of the six - qubit linear cluster state which allows a specific avn proof @xcite .
some distributions of @xmath89 in four particles allowing avn proofs can be trivially obtained from fig .
[ fig1](b ) by splitting qubits that belong to the same particle into several particles .
for instance , a distribution allowing a specific avn proof is illustrated in fig .
[ fig1](c ) .
it can be easily seen that there is no distribution in four particles which allows a specific avn proof which can not be obtained from the distribution in fig .
[ fig1](b ) .
figure [ fig2 ] contains several possible distributions of a six - qubit @xmath24-graph state @xmath90 .
figure [ fig2](a ) represents the four - photon @xmath90 prepared in ref .
this distribution does not allow a specific avn proof since qubit 1 is connected only to qubit 2 and qubit 5 is connected only to qubit 4
. figures .
[ fig2](b)(d ) represent distributions of @xmath90 between four particles allowing specific avn proofs .
in the previous section , we have seen that @xmath90 admits specific avn proofs when their qubits are suitably distributed between four particles .
the question is whether @xmath90 admits specific avn proofs when it is distributed between three particles or less , or more generally speaking , the question is , given an @xmath0-qubit graph state , what is the minimum number of parties @xmath1 which allows @xmath1-partite avn proofs specific for this state ?
the following definition will be useful for solving this problem .
let us define the reduced stabilizer of particle @xmath17 s qubits as the one obtained from the stabilizer of the original state by replacing the observables on all other particles qubits with identity matrices .
_ proof : _ suppose that particle @xmath94 carries qubits @xmath95 , where @xmath96 is the maximum number of qubits carried by any particle , and that particle @xmath97 carries qubits @xmath98 .
if @xmath44 , @xmath99 , @xmath45 , @xmath100 are elements of reality , then the reduced stabilizer of @xmath94 s qubits must contain moreover , the reduced stabilizer of @xmath94 s qubits must contain all possible products of eqs .
( [ eq1])([eq7 ] ) , that is , all possible variations with repetition of the four elements @xmath26 , @xmath23 , @xmath24 , and @xmath25 , choosing @xmath102 , which are @xmath103 .
a similar reasoning applies to the three pauli matrices of each and every one of @xmath97 s qubits .
therefore the reduced stabilizer of @xmath97 s qubits must also contain all possible products of given an @xmath0-qubit graph state , @xmath96 restricts the possible minimum numbers of particles and the possible numbers of qubits per particle .
given @xmath0 , table [ table1 ] presents the possible minimum numbers of particles and the corresponding possible distributions .
other possible distributions are already contained between those in table [ table1 ] , but in those cases the number of particles is not the minimum .
.[table1]possible distributions of an @xmath0-qubit graph state between a minimum number @xmath1 of particles .
for instance , ( 2,2,1 ) denotes a distribution of @xmath108 qubits between @xmath109 particles such that particles 1 and 2 have two qubits each and particle 3 has one qubit . [ cols="^,^,^",options="header " , ] m. hein , w. dr , j. eisert , r. raussendorf , m. van den nest , and h. j. briegel , in _ quantum computers , algorithms and chaos _ , edited by g. casati , d.l .
shepelyansky , p. zoller , and g. benenti ( ios press , amsterdam , 2006 ) . | all - versus - nothing ( avn ) proofs show the conflict between einstein , podolsky , and rosen s elements of reality and the perfect correlations of some quantum states . given an @xmath0-qubit state distributed between @xmath1 parties ,
we provide a method with which to decide whether this distribution allows an @xmath1-partite avn proof specific for this state using only single - qubit measurements .
we apply this method to some recently obtained @xmath0-qubit @xmath1-particle states .
in addition , we provide all inequivalent avn proofs with less than nine qubits and a minimum number of parties . |
we start with the schrdinger equation in @xmath1 at a fixed positive energy @xmath2\psi(x,\alpha)=0\qquad\text{in } \mathbb{r}^3\ ] ] whose scattering solution takes the form @xmath3 where @xmath4 is the direction of the incident wave and @xmath5 is the scattering amplitude .
the following general theorem is due to ramm @xcite .
@xmath6 scattering amplitude @xmath7 , @xmath8 ( arbitrary small open subsets of @xmath9 ) determine @xmath10 uniquely in the function class @xmath11 , @xmath12 .
the present treatment is restricted to spherically symmetric potentials , that is @xmath13 . in this case
we have the partial wave expansion of the wave function : @xmath14 where @xmath15 satisfies the radial schrdinger equation ( see below ) with the appropriate boundary conditions . for
spherically symmetric potentials the scattering amplitude takes a similar expansion form , namely @xmath16 the phase shifts , @xmath17 can be restricted to @xmath18 , since from the experimental point of view they are undetermined to an additive factor of @xmath19 , @xmath20 .
then it is apparent that the knowledge of the phase shifts is equivalent to that of the scattering amplitude .
however not all the phase shifts ( as considered in the theorem of loeffel @xcite ) are needed for unique reconstruction of the potential .
the following result is due to horvth and ramm @xcite . the phase shifts @xmath21 determine the potential uniquely in the class @xmath22 if the mntz condition holds for @xmath23 : @xmath24 this condition is almost necessary in the sense that in the class @xmath25 with @xmath26 the set @xmath27 with @xmath28 is not enough to recover the potential uniquely .
@xmath29 .
it is because of this result that one may look for a potential when a part of the scattering phase shifts are known .
the phase shift @xmath30 appears in the asymptotic form of the regular solution of the schrdinger equation , which is defined in the following manner : @xmath31,\\ \varphi_\ell(r)=\frac{r^{\ell+1}}{(2\ell+1)!!}+o(r^{\ell+1}),\qquad r\to0,\\ \varphi_\ell(r)=b_\ell\sin(r-\frac{\ell\pi}{2}+\delta_\ell)+o(1),\qquad r\to\infty\end{aligned}\ ] ] with @xmath32 being a constant . note that for the @xmath33 zero potential we have @xmath34 , where @xmath35 $ ] and @xmath36 and @xmath37 is the @xmath38th order bessel function of the first kind .
the povzner - levitan representation in the fixed energy problem @xcite has long been used , however the existence and uniqueness of a @xmath39 transformation kernel have been proven rigorously only recently @xcite for potential @xmath40 independent of @xmath41 and satisfying @xmath42 for @xmath43 we have a goursat - type problem ( equivalent to the schrdinger equation ) : @xmath44 now , in analogy with the fixed@xmath41 problem a gelfand - levitan - marchenko - type integral equation is written up for @xmath43 : @xmath45 since @xmath43 is unique this integral equation must have a unique solution for @xmath43 to make sense .
furthermore , for consistency with the original schrdinger equation , @xmath46 must satisfy @xmath47 and through the integral equation @xmath48 is maintained .
@xmath46 is a good candidate for approximation since its differential equation can be satisfied trivially .
for instance if the angular momentum expansion @xcite @xmath49 is imposed then for the @xmath50 functions we have only the potential - independent restrictions of @xmath51 and the information is only stored in the @xmath52 expansion coefficients .
in the framework of the method proposed by cox and thompson @xcite one takes the following separable form for the @xmath50 functions @xmath53 which is also the green s function of the @xmath33 radial schrdinger equation for the @xmath41th partial wave : @xmath54\gamma_\ell(r , r')=\delta(r - r');\ ] ] and the summation in @xmath46 runs only over a finite set @xmath55 of @xmath41 s : @xmath56 containing the riccati - bessel functions , connected to the bessel and neumann functions by @xmath57 for solving the glm - type equation ( [ glm ] ) we use the separable ansatz @xmath58 with a finite set @xmath59 of `` shifted angular momenta '' satisfying @xmath60 and @xmath61 . the use of such an ansatz can be motivated by the following result ascertained from @xcite .
if @xmath62 , @xmath63 is imposed on @xmath43 and the @xmath64 numbers are restricted to @xmath65 then @xmath66 where @xmath67 is a one - to - one mapping .
also , this form provides a reasonably easy way to solve the glm - type integral equation .
in fact it makes a system of algebraic equations instead of the integral equation , namely @xmath68 the parameters of the set @xmath59 can be obtained from the phase shifts @xmath69 through the transformation equation ( [ pl ] ) which in terms of the @xmath64 s takes the form @xmath70 taking both equation ( [ pl1 ] ) and ( [ glmalg ] ) for large @xmath71 s , i.e. at @xmath72 we get the system of nonlinear equations @xcite connecting @xmath69 to @xmath59 : @xmath73 with @xmath74_{\ell l}[m_{\cos}^{-1}]_{l\ell'}e^{\pm i(\ell-\ell')\pi/2}\qquad \ell\in s,\ ] ] @xmath75 after some investigation of this nonlinear system of equations one can conclude that for a given set of phase shifts it yields an infinity of solutions , each giving rise to a potential .
it was our result in @xcite that one can select a physical potential ( perhaps uniquely ) with the aid of a consistency check .
this check is first revisited in sec . 4 and then it is applied for the special cases of one and two dimensions , @xmath76 .
before continuing with the consistency check it is worthwhile to further study the ct method particularly concerning the potential one can obtain by applying it .
we address the case when only a finite number of input phase shifts are used to construct the potential .
first , we show that the potential is generally not compactly supported nor is of long - range . to see this define the functions @xmath77 by the limit @xmath78 we infer from @xcite that the asymptotic coefficients @xmath79 assume the form @xmath80 where @xmath81 s and @xmath82 s are constant depending on all the elements of both @xmath55 and @xmath59 : @xmath83 @xmath84 this form in turn implies that @xmath85 and thus @xmath86 which means that the potential generally falls of like an inverse power of two
. one can give a necessary condition for the potential to decrease more rapidly then @xmath87 .
one only needs @xmath88 the quantites @xmath4 and @xmath89 are given by @xmath90 alternatively , @xmath91 can be written as a series in @xmath41 ( this formula does not hold for @xmath43 with @xmath92 ) , @xmath93 which can readily be seen from the glm equation ( [ glm ] ) taken at @xmath94 and the ct formula ( [ ctg1 ] ) substituted for @xmath46 : @xmath95,\end{aligned}\ ] ] where the formula ( [ pl ] ) for @xmath96 has appeared .
this allows for the alternative conditions @xmath97 involving the expansion coefficients , the input phase shifts and the normalization constants of the partial wave functions .
these new results may serve as useful tools to check numerical results or incorporated into a solution method they might provide a way to control the undesirable oscillations of the inverse potential . at this point
we shortly discuss the applicability of the ct method .
the inverse potential is finite at the origin starting with a zero derivative ( see e.g. @xcite ) and possesses also a finite first moment @xmath98 . at fixed scattering energy
the ct procedure is thus a particularly successful method for ( re)constructing inverse potentials .
it was used to recover interaction potentials ( see e.g. @xcite ) for various physical system .
furthermore it is also possible to generalize the method to recover potentials having coulomb tail e.g. by replacing the riccati - bessel function by coulomb wave functions @xcite . in real - life scattering experiments one encounters cases when the colliding particles are zero spin bosons .
then it is well - known that only the even numbered partial waves contribute to the scattering amplitude because of the symmetry of the wave function . in such cases with the ct method it is possible to use only the experimentally available phase shift data corresponding to even partial waves .
it is then reasonable to ask for the value of phase shifts provided by the ct potential for the odd partial waves .
it turns out that these odd phase shifts are zero . to see this
let us derive the phase shifts of the ct potential which can be obtatined from eq .
( [ pl1 ] ) taken at @xmath72 : @xmath99 now the trigonometric form ( [ trigo ] ) is substituted for @xmath100 .
there are two special cases that we consider : + i ) the set @xmath55 of input angular momenta consists of only even numbers , + ii ) @xmath55 consists of odd numbers .
+ in the first case we have @xmath101 @xmath102 while in the second @xmath103 @xmath102 .
this is deductible from eqs .
( [ al ] ) and ( [ bl ] ) which assume the forms @xmath104 for the cases i ) and ii ) respectively .
one can assume @xmath105 and @xmath106 , since the matrix @xmath107 with elements @xmath108 is invertible ( it is a cauchy matrix ) and can not be singular unless @xmath109 which is excluded by assumption .
we thus have @xmath101 @xmath102 for i ) and @xmath103 @xmath102 for ii ) .
now @xmath110 which implies @xmath111 in other words the ct phase shifts of the opposite parities are exactly zero .
notice that the ct method allows the construction of potentials which are transparent for half the partial waves ( being even or odd in parity ) .
applications suggest that if dealing with input partial wave data of one parity the performance of the ct method with the same number of input diminishes compared to the case when data with both parity is employed .
therefore our sum rules derived in the previous subsection can be extremely useful to improve performance by supressing oscillations of the potential .
also note that one of our sum rules simplify ( due to @xmath112 in the even case and @xmath113 in the odd case @xcite ) to @xmath114 while the other becomes @xmath115 and @xmath116 for the even and the odd case , respectively .
the kernel @xmath46 only makes sense if the integral equation is uniquely solvable with it . as eq .
( [ glm ] ) can be viewed as a fredholm type integral equation of the second kind for fixed @xmath71 , viz . if @xmath117 and @xmath118 we have @xmath119 according to fredholm s alternative the fredholm determinant thereof must be nonzero for all fixed @xmath120 .
when using the ct ansatz the fredholm determinant of the integral equation becomes the determinant of the system of the algebraic equations ( [ glmalg ] ) , @xmath121_{\ell l}\right\}.\ ] ] it is easy to see that for @xmath122 one gets @xmath123 thus the potential is not in @xmath124 .
while for @xmath125 we have @xcite @xmath126 one can conclude the following @xcite . if and only if @xmath127 on @xmath120 we get a unique solution of the glm - type integral equation in @xmath128 and from that an inverse potential in @xmath124 . from @xcite
we know that @xmath127 on @xmath120 is not the case for arbitrary choice of @xmath55 and @xmath59 : e.g. there it was shown , that if @xmath129 and @xmath130 then @xmath131 at some @xmath120 . with the help of theorem 4.1 we can convince ourselves that a particular ct inverse potential obtained numerically from arbitrary data is integrable or not .
if it is so then that potential will generate the input phase shifts . also , using the theorem it is possible to determine the admissible set of @xmath64 numbers for given @xmath41 s . in the one-@xmath41 case
we get a straightforward admissible set of @xmath64 s , however in higher dimensions the formulation becomes extremely involved . in this case
@xmath132 , @xmath133 the function @xmath134 must be examined carefully .
this was performed in @xcite and thus it shall not be discussed here in detail .
we only give the key idea of the proof : the wronskian @xmath135 is nonzero on @xmath136 if and only if the constituent bessel functions ( @xmath137 and @xmath138 ) are interlaced .
this is deductible from the observation @xmath139 we could prove the next theorem of broader interest @xcite the positive zeros of the bessel functions @xmath140 , @xmath141 , @xmath142 , @xmath143 , @xmath144 , @xmath145 for nonnegative orders , @xmath146 are interlaced according to the inequalities @xmath147 if and only if @xmath148 ( otherwise @xmath149 for some @xmath150 and @xmath151 for some @xmath152 ) .
the following gives the admissible set @xmath153 .
@xmath154 has no roots on @xmath155 , that is at @xmath156 the glm - type equation is uniquely solvable for the ct method with @xmath132 and @xmath157 if and only if @xmath158 .
@xmath159 , @xmath160 is supposed .
this result allows us to choose _ uniquely _ from the solutions of the system of equations ( [ nl ] ) as the solution for one input phase shift is @xmath161 which for @xmath162 $ ] in conjunction with the previous theorem yields @xmath163 at low energies it may happen that only one partial wave contributes mostly to the scattering amplitude ( e.g. for some partial wave resonances ) .
it is worthwhile to look at the phase shifts yielded by the ct inverse potential at the one - phase - shift level to get a sense of the quality of the inversion procedure .
deductible from eq .
( [ pl1 ] ) is the following formula @xmath164 which specifies the phases of the ct potential for @xmath129 .
if e.g. the phase shift is restricted to describe an attractive potential ( i. e. @xmath165 ) the bound @xmath166 can be found using eq .
( [ 1shph ] ) .
this result assures the proper reproduction of the phase shifts by the ct potential . to illustrate these results fig .
1 shows synthetic test potentials corresponding to @xmath167 , @xmath168 obtained by the ct method .
in addition to the @xmath0 potential an inconsistent one is also shown where @xmath169 in eq .
( [ ct1s ] ) is chosen to be other than zero . as indicated before we get a non - integrable potential .
[ fig1 ] , @xmath168 with @xmath170 ( solid line ) and @xmath171 ( dashed line).,scaledwidth=80.0% ] already in this case the fredholm determinant becomes complicated , which is also apparent from it s integral representation @xmath172 therefore instead of the analytical treatment we determine the admissible set of @xmath64 s numerically for some choices of @xmath41 s .
our numerical method was to check the determinant at the points of a fine lattice on the @xmath173@xmath174 quarter plane of @xmath175 whether it has any zeros on @xmath176 , where @xmath177 is a large number chosen to be great enough for @xmath178 with a very small @xmath179 .
this can be done since every @xmath180 in any dimensions @xmath181 has only a finite number of zeros , since the constituent wronskians all tend to constants at large @xmath71 distances , viz .
@xmath182+o\left(\frac{1}{r}\right),\quad r\to\infty.\ ] ] on fig .
2 the admissible sets of @xmath183 pairs are depicted for the particular choices @xmath184 and @xmath185 .
[ fig2 ] in the next example ( fig .
3 ) we calculated some possible @xmath183 pairs for a given @xmath186 .
we note that only one of them is inside the permitted domain and could only find a single solution of the system of non - linear equation that is permitted by the consistency condition . again
the @xmath0 and an inconsistent potential is shown .
[ fig3 ] , with phase shifts calculated from a woods - saxon potential . @xmath187
( solid line ) and @xmath188 ( dashed line ) .
also , the original potential , @xmath189^{-1}$ ] , yielding the phase shifts ( @xmath190 , @xmath191 ) is depicted ( dotted line).,scaledwidth=90.0% ]
after recalling some crucial results of the inverse scattering problem for the schrdinger equation at fixed scattering energy we reviewed a particular constructive scheme , the cox - thompson ( ct ) method for recovering scattering potentials responsible for a finite set of given phase shifts .
we have analyzed the asymptotic properties of the ct potentials and from that we have obtained a sum rules for the expansion coefficients ( eq .
( [ sumr1],[sumr2 ] ) ) .
then we considered ct potentials corresponding to specific quantum statistical characters of the colliding pattern . if the scattered particles are zero spin bosons ( e.g. oxygen atoms or carbon nuclei ) the ct potentials are constructed from even phase shifts being accessible for measurements .
we have shown that in this case the ct potentials in addition to reproducing the input even phase shifts give exactly zero phases for the odd partial waves .
then we turned to the consistency check .
it has been shown that for the potential to be integrable ( or more precisely , to belong to the function class @xmath0 ) a condition must be fulfilled for the intermediate quantities @xmath64 of shifted angular momenta of the ct method .
we discussed the case when only one input phase shift is used in which circumstance we have explicit uniqueness .
the two - phase - shift case has also been discussed .
the condition for the @xmath64 numbers to produce an integrable potentials has been obtained through a numerical calculation for two particular cases and the results have been illustrated in fig.3 . | the inverse scattering problem of the three - dimensional schrdinger equation is considered at fixed scattering energy with spherically symmetric potentials .
the phase shifts determine the potential therefore a constructive scheme for recovering the scattering potential from a _
finite _ set of phase shifts at a fixed energy is of interest .
such a scheme is suggested by cox and thompson @xcite and their method is revisited here .
also some new results are added arising from investigation of asymptotics of potentials and concerning statistics of colliding particles .
a condition is given @xcite for the construction of potentials belonging to the class @xmath0 which are the physically meaningful ones .
an uniqueness theorem is obtained @xcite in the special case of one given phase shift by applying the previous condition .
it is shown that if only one phase shift is specified for the inversion procedure the unique potential obtained by the cox - thompson scheme yields the one specified phase shift while the others are small in a certain sense .
the case of two given phase shifts is also discussed by numerical treatment and synthetic examples are given to illustrate the results . besides the new results this contribution provides a systematic treatment of the ct method . |
a lot of research in the field of information retrieval aims at improving the _ quality _ of search results .
search quality might for instance be improved by new scoring functions , new indexing approaches , new query ( re-)formulation approaches , etc . to make a scientific judgment of the quality of a new search approach , it is good practice to use so - called benchmark test collections , such as those provided by trec @xcite .
the following steps typically need to be taken : 1 .
the researcher codes the new approach by adapting an experimental search system , such as lemur @xcite , pf / tijah @xcite , or terrier @xcite ; 2 .
the researcher uses the system to create an inverted index on the documents from the test collection ; 3 .
the researcher puts the queries to the experimental search engine and gathers the top @xmath0 search results ( a common value for trec experiments is @xmath1 ) ; 4 .
the researcher compares the top @xmath0 to a golden standard by computing standard evaluation measures such as mean average precision .
in our experience , step 1 , actually coding the new approach , takes by far the most effort and time when conducting an information retrieval experiment .
coding new retrieval approaches into existing search engines like lemur , pf / tijah and terrier is a tedious job , even if the code is maintained by members of the same research team .
it requires detailed knowledge of the existing code of the search engine , or at least , knowledge of the part of the code that needs to be adapted .
radical new approaches to information retrieval , i.e. , approaches that need information that is not available from the search engines inverted index , require reimplementing part of the indexing functionality .
such radical new approaches are therefore not often evaluated , and most research is done by small changes to the system . in his wsdm keynote lecture , dean
@xcite describes how mapreduce @xcite is used at google for experimental evaluations .
new ranking ideas are tested off - line on human rated query sets similar to the queries from trec . running such off - line tests
has to be easy for the researchers at google , possibly at the expense of the efficiency of the prototype .
so , it is okay if it takes hours to run for instance 10,000 queries , as long as the experimental infrastructure allows for fast and easy coding of new approaches .
a similar experimental setup was followed by microsoft at trec 2009 : craswell et al .
@xcite use dryadlinq @xcite on a cluster of 240 machines to run web search experiments .
their setup also sequentially scans all document representations , providing a flexible environment for a wide range of experiments .
the researchers plan to do many more to discover its benefits and limitations .
the work at google and microsoft shows that sequential scanning over large document collections is a viable approach to experimental information retrieval .
some of the advantages are : 1 .
researchers spend less time on coding and debugging new experimental retrieval approaches ; 2 .
it is easy to include new information in the ranking algorithm , even if that information would not normally be included in the search engine s inverted index ; 3 .
researchers are able to oversee all or most of the code used in the experiment ; 4 .
large - scale experiments can be done in reasonable time .
we show that indeed sequential scanning is a viable experimental tool , even if only a few machines are available . in section [ sec : approach ] we describe the mapreduce search system .
sections [ sec : results ] and [ sec : conclusion ] contain experimental results and concluding remarks .
mapreduce is a framework for batch processing of large data sets on clusters of commodity machines @xcite .
users of the framework specify a _
mapper _ function that processes a key / value pair to generate a set of intermediate key / value pairs , and a _ reducer _ function that processes intermediate values associated with the same intermediate key .
the pseudo code in figure [ tab : pseudocode ] outlines our sequential search implementation .
the implementation does a single scan of the documents , processing all queries in parallel .
[ cols= " < " , ] anchor text extraction on all english documents of clueweb09 takes about 11 hours on our cluster .
the anchor text representation contains text for about 87 % of the documents , about 400 gb in total .
a subsequent trec run using 50 queries on the anchor text representation takes less than 30 minutes .
our linear search system implements a fairly simple language model with a length prior without stemming or stop words .
it achieves expected precision at 5 , 10 and 20 documents retrieved of respectively 0.42 , 0.39 , and 0.35 ( mtc method ) , similar to the best runs at trec 2009 @xcite .
figure [ fig : time ] shows how the system scales when processing up to 5,000 queries , using random sets of queries from the trec 2009 million query track .
reported times are full hadoop job times including job setup and job cleanup averaged over three trials .
processing time increases only slightly if more queries are processed .
whereas the average processing time per query is about 35 seconds per query for 50 queries , it goes down to only 1.6 second per query for 5,000 queries . for comparison ,
the graph shows the performance of `` lemur - one - node '' , i.e. , lemur version 4.11 running on _
one fourteenth _ of the anchor text representation on a single machine . a distributed version of lemur searching the full full anchor text representation would not do faster
: it would be as fast as the slowest node , it would need to send results from each node to the master , and to merge the results .
lemur - one - node takes 3.3 seconds per query on average for 50 queries , and 0.44 seconds on average for 5,000 queries .
the processing times for lemur were measured after flushing the file system cache .
although lemur can not process queries in parallel , the system s performance benefits a lot from receiving a lot of queries .
lemur s performance scales sublinearly because it caches intermediate results .
still , at 5,000 queries lemur - one - node is only 3.6 times faster than the mapreduce system . for experiments at this scale , the benefits of the full , distributed lemur are probably negligible .
the idea to use sequential scanning of documents to research new retrieval approaches is certainly not new : we know of at least one researcher who used sequential scanning over ten years ago for his thesis @xcite . without high - level programming paradigms like mapreduce , however , efficiently implementing sequential
scanning is not a trivial task , and without a cluster of machines the approach does not scale to large collections .
lin @xcite used hadoop mapreduce for computing pairwise document similarities .
our implementation resembles lin s brute force algorithm that also scans document representations linearly .
our approach is simpeler because our preprocessing step does not divide the collection into blocks , nor does it compute document vectors .
a faster turnaround of the experimental cycle can be achieved by making coding of experimental systems easier .
faster coding means one is able to do more experiments , and more experiments means more improvement of retrieval performance .
we implemented a full experimental retrieval system with little effort using hadoop mapreduce . using 15 machines to search a web crawl of 0.5 billion pages ,
the proposed mapreduce approach is less than 10 times slower than a single node of a distributed inverted index search system on a set of 50 queries .
if more queries are processed per experiment , the processing times of the two systems get even more close .
the code used in our experiment is open source and available to other researchers at : http://mirex.sourceforge.net
many thanks to sietse ten hoeve , guido van der zanden , and michael meijer for early implementations of the system .
the research was partly funded by the netherlands organization for scientific research , nwo , grant 639.022.809 .
we are grateful to yahoo research , barcelona , for sponsoring our cluster .
n. craswell , d. fetterly , m. najork , s. robertson , and e. yilmaz . at trec 2009 : web and relevance feedback tracks . in _ proceedings of the 18th text retrieval conference ( trec)_. nist special publication 500 - 278 , 2009 .
brute force and indexed approaches to pairwise document similarity comparisons with mapreduce . in _ proceedings of the 32nd international acm sigir conference on research and development in information retrieval _ , 2009 .
y. yu , m. isard , d. fetterly , m. budiu , u. erlingsson , p. kumar , and j. currey . : a system for general - purpose distributed data - parallel computing using a high - level language . in _ proceedings of the 8th symposium on operating system design and implemention ( osdi ) _ , 2008 . | we propose to use mapreduce to quickly test new retrieval approaches on a cluster of machines by sequentially scanning all documents .
we present a small case study in which we use a cluster of 15 low cost machines to search a web crawl of 0.5 billion pages showing that sequential scanning is a viable approach to running large - scale information retrieval experiments with little effort .
the code is available to other researchers at : http://mirex.sourceforge.net |
a property of critical importance for complex networks is their resilience to damage or attack @xcite
. one fascinating demonstration of the underlying complexity of these systems is that compromise in structure can be substantial even after the loss of a very small number of nodes @xcite .
the repercussions to the network from this structural compromise can be immense , usually resulting in complete loss of communication among the surviving nodes and therefore a complete destruction of the intended network functionality . because of the obvious importance for practical applications , the robustness of a network s structure to damage has continuously remained the focus of intensive research in the network science literature @xcite .
even though we now have a thorough understanding of complex network disintegration , the inverse process of ` healing ' a network is much less understood .
is there a direct relationship between healing structural features and restoring function ?
what does it really take for a network to restore its functionality after losing some of its nodes ? while there are various types of healing , one realistic approach is to consider the case in which disabled nodes have been permanently removed and can not be resurrected , but the surviving nodes can still generate new links to other surviving nodes . of course , a node could trivially create as many new links as possible , but in many realistic settings the cost and time associated with establishing a new link can be high . we then need to optimize the network functionality given the constraints of rebuilding costs @xcite . as a natural first case
, we explore the scenario in which optimal network functionality is achieved under the structural condition that all the nodes are connected in one large cluster . in this way
, every node can reach any other node in the network , even if the connecting path may be long .
this reflects systems such as communication networks @xcite electrical grids @xcite , air traffic @xcite and shipping routes @xcite , etc .
additional requirements can be imposed , for example the maximum path between any two nodes can be bounded @xcite , or entirely different cases for structure and function can be explored ( e.g. function is optimized when the network structure has a particular degree distribution , or is highly modular , etc . ) , but as a first approach our only metric here is the size of the largest cluster . a basic , but fundamental insight is that healing can be employed with one of two possible goals in mind .
the first goal can be to fix the structure in a way that the topology remains as close to the original network as possible .
for example , descriptions of some social groups have revealed particular degree distributions and one natural inclination in healing a damaged social network might be to restore edges by focusing on this aspect of repair .
however , degree distribution itself is unlikely to support social function . for features such as social identity or support from a close group of friends , the structural features of modularity and clustering are more important .
therefore , if we are attempting repair in order to restore function , we may decide to ignore entirely the impact to degree distribution from repair , instead adding back links that restore ( for example ) local clustering coefficient . since damage can alter both structural and functional features , it is easy to consider repairing structure since regaining initial structure should provide restored initial function .
critically , however , it may not be that one and only one structural feature can support the original function .
the main question is therefore to discover efficient strategies that can restore network functionality through the addition of new links . in a recent work ,
redundant ( dormant ) links were considered to be activated after damage and it was shown that they can restore functionality in infrastructure trees @xcite .
similarly , healing in interdependent networks was shown to prevent cascading failures @xcite . in a different approach
, nodes were allowed to spontaneously recover and become active again , leading to an interesting behavior of phase - flipping where the network switches between high - activity and low - activity modes @xcite . under different definitions of network function in which a global minimum traffic flow must be maintained and each node has a particular flow capacity that must be redistributed in case of damage , the network functionality can be restored by mitigation strategies @xcite .
one critical and under - discussed feature of damage , especially in inhomogeneous complex networks , is that the impact from the loss of a single link on overall function can differ drastically depending on where that link occurs in the local and global structure .
loss of a single link that uniquely connects two otherwise disconnected components of a network will have a greater impact to functions ( such as the one we study ) that rely on the size of the largest cluster than would the loss of one of the three links that make up a triangle .
while obvious , it is nevertheless important and profound that we consider healing as a process that should focus on areas of the network in which the most damage has been done to the function , rather than to the structure . in the case of our largest connected component example , this means we should focus more effort towards creating new edges that restore connectivity to nodes that are most direly impacted in their ability to relay messages . phrased in this way , it may seem as though the only option is to analyze global network structure in order to understand which nodes are most critical to restoring function .
such a global algorithm could potentially locate the optimal solution by mapping the current state of the network and adding only the links that are missing to restore functionality . however , this centralized planning may not always be feasible , since it may cost a lot in terms of advance planning , communication between the central authority and the individual nodes , the time that it takes to transmit all relevant information , the possibility that communication takes place through the network itself and the central node may have been functionally disconnected from this critical communication , as well as any combination of the above factors which may limit the ability of constructing one central plan and communicate it to all the surviving nodes .
luckily , there is no need for this type of centralized global analysis .
we proceed to propose a self - organizing healing algorithm that exploits this heterogeneity feature of damage while relying only on local information accessible to each node , and with low cost from the construction of new links .
since one of our goals will be to limit the amount of centralized information that will be necessary before repair to the network can begin , we propose purely local definitions of damage that can be assessed by each node . to make this distinction clearly , in this work optimal ` healing ' will refer to adding new links so that every node can reach any other node in the system , and the term ` self - healing ' indicates that individual nodes decide on their own whether they need to create new links or not . we here demonstrate that our self - healing model can restore connectivity function very efficiently in two different ways .
firstly , we show effective restoration of function with the creation of very few new links relative to the number lost to damage .
we demonstrate the effectiveness of the method through the size of the healed largest component , the number of nodes that need to form new links , and the changes in modularity compared to the original network .
we further show this is true even if we impose cost constraints on the length of new links ( i.e. the number of links in the shortest path required to connect the two nodes without the addition of the new link ) .
secondly , we show that a null healing model ( in which the same number of links are constructed randomly ) achieves a drastically lesser restoration of function .
we analyze the efficiency of our self - healing algorithm over various network topologies and in cases of both random - node - removal damage and when damage is targeted to affect only specific , high - structural - impact nodes .
based on this ` uneven damage to function ' perspective , we show that relatively inexpensive , rapid healing may be relatively easy to achieve , even in the absence of global information . to formalize the problem , we consider the simplest possible case of an isolated complex network which undergoes a loss of a fraction of its nodes , resulting in a possible loss of large - scale connectivity . to mitigate this loss of function , each node
can then generate one new link in an attempt to restore connectivity . as mentioned above , this is a trivial problem without any additional constraints because connecting to random nodes would result in a structure similar to an erdos - renyi network . in our work ,
we study what conditions are necessary and sufficient to restore connectivity under the constraints that : a ) establishing a new link is costly and that cost is borne by the node initiating the link , b ) the new links are as short as possible ( based on the path distances of the undamaged network ) , and c ) the decision is purely local , so that a node does not know anything about the network state , except for its own neighborhood .
we show that these conditions can be easily met through a simple local - decision algorithm , based on the number of surviving neighbors .
there is no need to transfer any information between nodes , and the only requirement is that a node can know if a neighbor within a given distance remains alive or not .
the first step in identifying efficient healing algorithms is to determine the state of the network immediately after the removal process .
this topic has not received attention in the existing literature , which instead mainly focuses only on the properties of the largest connected cluster . for the purpose of a healing algorithm
, we consider all the clusters that remain in the system after a fraction @xmath0 of system nodes have been removed . a possible measure that can quantify the extent of damage is the minimum distance between these clusters , @xmath1 , which we define as the minimum possible distance from a node in the cluster to any other node in any other cluster .
since all clusters are disconnected , this distance is calculated in terms of the original network . in practice , we renormalize the damaged structure in the following way ( fig . [
fig1]a - d ) : every cluster is substituted by one super - node .
all these super - nodes are obviously isolated .
we restore all the removed nodes and links among them .
the links that were connecting removed nodes with any node contributing to a super - node result in a link between the super - node and the removed node ( we also remove any double links ) . in this renormalized network , we can then calculate how close the super - nodes are to each other . obviously , this distance is bounded by the minimum value of @xmath2 .
notice also that this distance is different from the distance between a random node in the cluster and other clusters , since it is solely determined by the one node in the cluster which is closest to another cluster .
( color online ) cluster distance after damage .
( a ) a toy network .
( b ) a number of nodes are removed from the network .
( c ) the remaining clusters in the network .
( d ) we consider each cluster to be one unit .
the distance between the clusters , @xmath1 , is measured according to the distances of cluster nodes in the original network .
( e ) the change of inter - cluster distance , @xmath1 , as a function of removed nodes , @xmath0 , for model and real networks after random node removal .
( f ) the change of inter - cluster distance , @xmath1 , as a function of removed nodes , @xmath0 , for model and real networks after targeted node removal . ] intuitively , the inter - cluster distance is small in centralized networks and is large in sparse , extended networks .
for example , in a star network where all nodes are connected to a central node , the distance @xmath1 can never be larger than 2 independent of how many nodes we remove . in a one - dimensional lattice
this distance can become very large , especially when @xmath0 becomes large and the gap among surviving clusters increases rapidly .
our focus on inter - cluster distance is to enable estimation of the most efficient length for the healing links .
if , for example , typical inter - cluster distances were @xmath3 then it would be pointless to add links much shorter than that .
similarly , since one of our goals is to minimize the cost , calculating @xmath1 allows us to avoid constructing unnecessarily long links .
we studied the two typical cases of node removal in complex networks : random removal and targeted attack . in fig .
[ fig1]e we show the average @xmath4 over all clusters as a function of the fraction of randomly removed nodes , @xmath0 . in model networks , such as the two - dimensional square lattice , erdos - renyi , and scale - free networks ,
the value of @xmath5 remains close to 2 until roughly @xmath6 .
when we remove a larger percentage of the nodes the inter - cluster distance increases up to @xmath7 . in real networks , on the other hand , the average distance never increases significantly , independently of the value of @xmath0 . in targeted attacks
we remove the nodes in decreasing order of their degree .
as expected , the targeted attack leads to increasing damage and consequently the inter - cluster distance increases in most of the networks ( fig .
[ fig1]f ) , with the exceptions of the email network and the square lattice .
the email network includes many strong hubs and , as a consequence , node distances are very short , which also reflects on the inter - cluster distances . for lattices ,
a targeted attack is the same as random removal , because the degree distribution is a delta function .
since er networks also have a narrow degree distribution , one might expect that the inter - cluster distance in er networks would also be independent of the attack strategy . in fig .
[ fig1 ] , we find that the opposite is true .
the average distance @xmath8 increases rapidly after @xmath6 , and at @xmath9 it reaches the highest value we observed , @xmath10 .
this is a result of the absence of hubs . when the majority of the highest - connected nodes have been removed , the small remaining clusters consist only of low degree nodes , which have a low probability of being close to each other since there are no hubs to centralize the network .
so , in general , we observe that the isolated clusters are relatively close to each other , and the inter - cluster distances remain for the largest part close to the minimum value of @xmath11 , at least for random removals , while in very few cases does it reach @xmath12 or more under targeted attacks .
the basic idea of the self - healing method is that each node monitors the fraction , @xmath13 , of its neighbors that remain alive .
if this fraction falls below a given threshold @xmath14 , e.g. @xmath14=50% , then the neighbor attempts to establish a new link .
this link can be at a maximum distance of @xmath15 , where distances are always measured as path length in the original network .
if there is no node within this distance , then the node abandons its attempt to build a new connection .
importantly , in agreement with the above findings on inter - cluster distance , we show that even if the maximum distance is the smallest possible , i.e. @xmath15=2 , large - scale connectivity is still restored even though the node has no information about whether or not the new link will connect to the largest cluster , or even if it will connect to a different cluster than the one it currently belongs to .
this process can be dynamic , so that nodes do not even need to be aware that there is an attack , as long as they can monitor how many of their links lead to live nodes , in which case they can respond immediately and attempt to create a new link .
this real - time healing is more effective than if the nodes can only attempt to heal once an attack is completed ( e.g. surprise attacks , or cases in which new link construction is risky until an attack has concluded ) and obviates the need for rigorous definitions of global attacks vs local damage . here ,
we show that the self - healing algorithm is efficient , even in this worst - case scenario , where the nodes do not have the capacity to respond until the attack is over .
( color online ) demonstration of the self - healing algorithm .
( a ) the initial structure of a toy network .
( b ) three nodes , indicated by green ( light gray ) , are removed from the network along with all their links .
( c ) the nodes which have lost half or more of their neighbors ( blue nodes , empty symbols ) will seek new connections .
( d ) these nodes will choose to connect to a random surviving node , if its distance @xmath16 in the original a network is at most @xmath15 .
here we use @xmath17 .
( e ) ( left panel ) network of airport connections in usa .
( middle panel ) a targeted attack removes the 20% most connected airport nodes , and results to a disconnected network .
the removed nodes are shown in green ( light gray ) .
( right panel ) the nodes that have lost more than 50% of their neighbors ( shown in blue , dark gray ) try to establish new links within distance @xmath17 . the new links ( red lines ) restore the large - scale network connectivity . ]
the order parameter , which quantifies the large - scale connectivity , is the fraction of the live nodes that belong in the largest connected component , denoted by @xmath18 after the attack , and by @xmath19 after the healing process .
the implementation of this model includes three steps ( fig .
[ fig2]a - d ) , which are as follows : 1 .
a fraction @xmath0 of the @xmath20 nodes initially in the system is removed from the network .
the size of the remaining largest connected cluster , @xmath18 , is expressed as a fraction of the remaining nodes .
2 . each of the remaining nodes decides independently if it needs a new link , depending on whether the fraction of its remaining neighbors @xmath21 exceeds a given threshold @xmath14 ( for any given node , @xmath22 is the number of neighbors after damage , and @xmath23 is the original number of neighbors of this node ) .
we denote the fraction of the surviving nodes that need healing , i.e. those whose degree has fallen below the threshold , as @xmath24 .
3 . these @xmath25 nodes attempt to find a new neighbor within a distance @xmath15 .
if such a neighbor is found , the nodes establish the new link .
the fraction of nodes that succeed to establish a new link is @xmath26 .
these new links then result to a new large cluster that includes a fraction of @xmath27 [ @xmath28 nodes .
the size of the largest cluster , @xmath18 , indicates the extent of damage to the structure immediately after the nodes removal .
a value close to @xmath29 means that the remaining nodes are isolated in small clusters and they can not access each other .
the plot of @xmath18 as a function of @xmath0 represents the well - studied percolation process of the largest cluster as we increase the number of removed nodes .
the plot of @xmath30 can be used to estimate the efficiency of the healing process .
after the self - healing process , the largest cluster becomes @xmath31 , since we have added links in the structure .
the difference between the two sizes is a measure of the process success . as a real - world example
, we can demonstrate how the airport network in usa can be influenced by attacks and healing ( fig .
[ fig2]e ) .
the unperturbed network ( left panel ) is quite dense and contains a large number of connections , condensed in many hub nodes .
after an intentional attack on 20% of the most connected nodes ( middle panel ) , 80% of the nodes are still functional but are barely connected to other nodes , since most of the traffic was directed through the hubs .
the original network contains 332 nodes , so after the removal of 66 nodes we are left with 266 nodes .
the largest cluster size connects only 30 of these nodes ( @xmath32 ) , and the surviving nodes are either completely isolated or belong to very small clusters .
after we apply the self - healing algorithm ( right panel ) , the connectivity is restored and the largest cluster grows to 235 nodes ( @xmath33 ) .
of course , the network becomes less centralized as hubs are removed and local connections are added during healing , so that the network topology changes , but the important part is that the functionality is restored and the network becomes navigable again .
now , it is possible to reach practically any node , independently of the origin . in this example
it is also easy to see that the new connections have to remain close to the healing node , since the cost of long - range connections may be prohibitively high .
the most interesting cases are those where the damage destroys the large - scale connectivity ( i.e. @xmath18 tends to 0 ) .
the question now becomes how easy it is to bring @xmath30 as close to 1 as possible .
therefore , the main quantities to compare are the values of @xmath34 vs @xmath35
. an ideal process would correspond to a plot of values as close to the horizontal @xmath36 line as possible .
this means that , independent of the initial damage , the self - healing process guarantees complete connectivity .
the worst - case scenario , on the other hand , is a @xmath37 line along the diagonal , indicating that there is no benefit from the self - healing process . in fig .
[ fig3 ] we show the evolution of @xmath18 and @xmath19 as a function of the removed percentage of nodes , p , in a random removal process .
the results for lattices , erdos - renyi , and scale - free networks are well known and we recover the typical curves for @xmath18 @xcite . the self - healing process results in significantly larger critical points for lattice and er networks .
this delay indicates that even though a network has been fully disintegrated , our healing algorithm can reconstitute one large cluster of size @xmath19 among the surviving nodes .
this improvement is obvious even for large threshold values , such as @xmath14=0.75 , when a node delays adding a new link until it has lost more than 75% of its neighbors .
as we relax this threshold to @xmath14=0.5 the largest cluster reaches higher values over a wider @xmath0 interval . lowering this threshold
even more did not produce noticeable improvement .
therefore , in the following analysis we fix this value to @xmath14=0.5 .
( color online ) fraction of surviving nodes in the largest cluster as a function of the removed nodes , @xmath0 .
the panels correspond to : ( a ) two - dimensional square lattice , ( b ) erdos - renyi network with average degree @xmath38 , ( c ) random scale - free network with degree exponent @xmath39 , and ( d ) random scale - free network with degree exponent @xmath40 .
the network size in all cases was @xmath41 nodes .
black ( lower ) lines indicate the cluster size @xmath18 , after the initial removal of nodes .
the red ( middle ) lines represent the cluster size after the healing process , @xmath42 , where a node decides to search for new connections if it has lost at least @xmath43 of its original neighbors .
the blue ( upper ) line shows the healed cluster size , @xmath44 , for a corresponding threshold of @xmath45 .
the healing distance in all cases is @xmath17 . ]
each node seeks new connections within a maximum distance @xmath15 , and for the results seen in fig .
[ fig3 ] we used the minimum possible distance , i.e. @xmath15=2 . as we increase the radius of possible new connections the connectivity becomes easier to achieve for a number of reasons .
first , there are more options because the node can create a link , even if all second - neighbors have been removed .
second , the effect of long - range shortcuts is known to favor large - scale connectivity @xcite . for example , if the attack results in small , isolated clusters , then short - range shortcuts will tend to remain within the same cluster and can not assist in bridging between different clusters . on the other hand , long - range shortcuts will have a much higher probability of bridging otherwise unconnected parts of the network because they can choose nodes that are far from the immediate neighborhood of their initiator @xcite .
the problem with this approach , of course , is that the associated cost of long links may become significantly higher .
it is important , therefore , to know if the minimum possible cost is enough to restore the network structure . as we will demonstrate below , in the absence of cost restrictions , higher values of @xmath15 improve the repair results , with the optimum case being that @xmath15 is unrestricted and the node can randomly choose any other surviving node in the system
this is a trivial theoretical result , because long - range links can connect isolated clusters which are far from each other in case of extended damage .
importantly , however , beyond this trivial result , we show that the minimum value of @xmath15=2 is already sufficient to restore network structure .
we therefore focus our discussion by reporting only results for this worst - case scenario .
( color online ) ( a ) snapshots of the healing process in a square two - dimensional lattice , after removing @xmath46 of the nodes .
the nodes that survived the attack are shown in gray color , with black nodes indicating the largest cluster after a fraction , @xmath47 , of the surviving nodes have generated a new link .
the healing links are shown in green ( light gray on black nodes ) .
( b ) evolution of the largest cluster size during healing , as a function of the percentage of surviving nodes , @xmath47 , that establish a new link . from top to bottom , the fraction of removed nodes increases in steps of 0.1 , from @xmath48 to @xmath9 ( or to the value indicated on the plot ) .
the starting point of each curve at @xmath49 corresponds to the largest cluster size immediately after the removal of @xmath0 nodes .
the end point of each curve indicates the final cluster size after the end of the healing process . ] in fig .
[ fig4 ] we present many examples of the self - healing process , for both model and real - world networks .
a lattice structure offers the simplest example of self - healing . in fig .
[ fig4]a we present a toy example , in a lattice where @xmath46 of the nodes have been removed . as we add the links to a fraction fs of the nodes that have lost more than half of their neighbors , the size of the largest cluster increases and presents essentially a typical second order percolation transition .
this behavior is verified in the first panel of fig .
[ fig4]b , where the large - scale connectivity is easily restored for values of @xmath50 .
the second - order transition is observed in all the model networks , but only when the values of @xmath0 reach close to their critical values pc . for @xmath51
, the initial largest cluster at @xmath49 extends over a non - zero fraction of the remaining nodes and , notably , there is no transition in these cases .
the linear increase of the largest cluster size indicates that the new links attach small clusters to the spanning cluster , and there is never any other cluster of significant size .
the absence of a transition is observed in all real networks , even when the initial largest cluster is very small .
for instance , at @xmath9 , the largest cluster size of the co - authorship network starts at less than @xmath52 and with healing it increases almost linearly up to @xmath53 .
notice that in these plots the process does not necessarily go up to @xmath54 , since not all nodes need , or can find , new connections according to the rules of the algorithm .
we then compare the value of the cluster size at the end of healing , @xmath34 , with the cluster size before healing , @xmath35 , for different values of @xmath0 ( fig .
[ fig5 ] ) . in lattices , we know that close to the threshold @xmath55 the system is in the critical phase so that a small number of links are enough to restore connectivity .
however , in the plot we can see the large improvement that healing brings to the tolerance of the system . even though the network disintegrates very rapidly as we remove more nodes and increase @xmath0 to @xmath56 ,
the largest cluster after self - healing practically includes all remaining nodes , so that @xmath57 . moving from @xmath58
to @xmath59 causes a much greater damage which can not be restored via our self - healing process and @xmath34 abruptly drops to @xmath60 . in practice ,
the self - healing algorithm has managed to delay the location of the critical point from @xmath55 to @xmath61 .
this simple behavior of either a fully connected or fully disconnected cluster is rather idiosyncratic of the highly organized configuration of nodes in the lattice , which is not found in random models and real - world networks , as we show below .
( color online ) how the largest cluster size changes after healing , @xmath19 , as a function of the largest cluster size before healing , @xmath18 .
results are presented for model and real - world networks .
the percentage of nodes removed is represented by the color of a node , increasing from right to left from @xmath48 to @xmath9 .
each scattered point corresponds to a different realization of the process in each network .
the efficiency of the healing process is demonstrated by the fact that all the points tend to be far from the diagonal , which would indicate no improvement , and remain close to the maximum efficiency value of @xmath62 . ]
the erdos - renyi network ( with average degree @xmath38 ) follows a similar pattern where removing 70% of the nodes results to isolated clusters , @xmath63 . after healing with our algorithm the largest cluster size encompasses a large part of the remaining nodes , @xmath64 .
this striking difference indicates that the remaining clusters remain close to criticality and relatively close to each other , since they can be merged when nodes add new local links with @xmath15=2 .
similar improvement is found for the case of scale - free networks with different degree exponents .
an important difference between random model networks and real - world networks is usually the level of organization .
for instance , typical empirical networks exhibit higher degrees of clustering and modularity , features that are missing in the randomly built networks @xcite .
nevertheless , this organization can potentially drastically impact the outcome of the healing process .
in the examples that we studied ( fig . [ fig5 ] ) we see that self - healing is equally , if not more , efficient in restoring the largest cluster in empirical networks . in certain cases ,
e.g. the friendship social network in irvine or the email exchange network , practically all the surviving nodes belong to the same cluster after healing , even though the starting network state is highly disintegrated , i.e. @xmath65 . of particular importance is the result for the network of airport connections , because it is a spatially embedded structure and we can approximate spatial distance with network distance , at least to demonstrate the principle that longer jumps may be very costly to implement .
the example highlights the importance of small @xmath15 values .
if an airport closes down then the traffic needs to be redirected to another airport in the general area , but it is impractical if this distance remains unrestricted .
additionally , in this example connectivity is crucial for the network to be functional ( one needs to be able to reach any destination ) . under the conditions of our self - healing algorithm , the large - scale connectivity was almost certain in all our simulations , with more than 80% of the nodes belonging in the largest cluster .
the large fluctuations , especially in the extreme case where 90% of the nodes have been removed , are the result of the small network size , which includes 332 nodes .
this demonstrates that healing may become unpredictable for very small networks , where one or two links may be enough to connect the small number of clusters .
the percentage of nodes that find themselves with fewer neighbors than the threshold increases with the number of nodes that are removed .
these nodes do not necessarily manage to find a new neighbor within the distance @xmath15=2 , either because all these nodes have been removed or because there is already a connection . as we show in fig .
[ fig6 ] , all nodes in lattice and in model random networks manage to find a new neighbor as long as less than 50% of the initial nodes have been removed . as the extent of damage increases , a larger percentage of nodes need new connections , but less than half of those manage to do that . on the contrary , in real networks
all nodes manage to find new connections almost independently of the extent of damage in the network .
( color online ) fraction of the surviving nodes that successfully established new connections , @xmath47 , as a function of the fraction of nodes , @xmath24 , that attempted to establish new connections following the removal of a percentage @xmath0 of nodes .
the value of @xmath0 increases from left to right .
the points that are closer to the diagonal indicate a higher percentage of success . ]
even though functionality is restored , since the majority of nodes can now communicate with each other through the structural repair that reconnected a large cluster , it is not clear what impact restoration of this particular structural feature may have on other topological / structural features .
we therefore study the effect on the modularity of the healed network compared to the modularity of the undamaged network as an example of a potentially important structural feature which is not necessarily correlated to the size of the largest cluster .
we use the standard definition for modularity @xmath66 , as the fraction of links between modules , minus the expected number of links within these modules for a random graph with the same node degree distribution @xcite . for fig .
[ fig6 ] we report the maximum possible value of @xmath66 , using the algorithm from ref . @xcite . in fig .
[ fig7 ] we show changes in modularity after a random removal of @xmath67 nodes and the application of the self - healing algorithm . a common trend in both model and real networks
is that modularity increases considerably with increasing @xmath0 .
this shows that the healing effect tends to create significantly stronger modules , and nodes tend to become more connected within the same network area .
of course , this is the result of the short - range links in the presented simulations , where @xmath15=2 . in practice
, the healing algorithm replaces the links that are removed with local links , enhancing thus the modular character of the network .
the only instances where modularity decreases are at large values of @xmath0 in the model networks .
this result can be explained by fig .
[ fig6 ] where only a small percentage of nodes manage to find new connections and therefore the form of the resulting structure is not very different than the damaged structure .
( color online ) effect of the self - healing process on the network modularity , as a function of the removed nodes .
the modularity at @xmath68 corresponds to the original network modularity . for @xmath69
, we consider the modularity of the network after healing with @xmath17 .
] the efficiency of our method can be tested against a random insertion of links under a similar set of constraints , as for the self - healing algorithm .
for example , we can add the same number of links at the same distance , but random nodes are chosen to create new links , rather than those who have lost most of their neighbors . in this way , we can detect if the crucial factor is the number of edges or the particular features of the nodes that select to establish new connections .
even if our model results did not exceed those of random insertion , our method is based on a self - organized algorithm that provides a simple local decision scheme , where there is no need for a central authority to coordinate actions around the network .
still , we demonstrate below that our method , which we tested in a number of model and real - world networks , significantly exceeds the results of random decisions .
we assess the effectiveness of the algorithm by comparing our results to such a null model . in this way
, we can determine if there is a benefit to choosing which nodes create new links based on @xmath14 .
for each realization of nodes removal with the self - healing algorithm we calculated how many new links were introduced in the system , resulting to @xmath19 .
we then added the same number of links to the damaged network , but this time the link originated from a random node instead of the node that had lost more than a fraction @xmath45 of its neighbors .
these new links were established within a distance @xmath15=2 , so that the total cost remained the same in both cases .
the result in the top row of fig .
[ fig8]a shows that the self - healing algorithm performed better in all cases compared to this null model .
the reason is that when a random node decides to establish a new link , it is possible that many of its neighbors still survive and the new link does not bring together isolated parts of the network .
this is also why when the removed percentage of nodes is small , e.g. 10 - 20% , there is no increase in the largest cluster size , unless we use the nodes that have lost most of their neighbors .
we also studied the effect of @xmath15 on the results ( fig .
[ fig8 ] ) .
as expected , we find that as we increase the maximum distance for new connections , the repair process becomes more efficient .
as shown in the plot for the self - healing algorithm , @xmath70 is already enough to completely restore connectivity among all the remaining nodes , practically under any conditions .
the optimal case is found for @xmath71 .
the results for the random model are consistently inferior to those of our algorithm .
the trend of better healing with increasing @xmath15 remains , but now even for unrestricted @xmath72 , the value of the repaired largest cluster can not reach the optimal value , @xmath62 . ( a ) top row : comparison of the result for the self - healing algorithm ( filled symbols ) with adding the same number of links starting from random nodes ( empty symbols ) . in both cases ,
bottom row : comparison of the self - healing algorithm ( continuous lines ) with the random case ( dashed lines ) , for different @xmath15 values . from bottom to top : @xmath15=2 , 3 , 4 , and @xmath73 .
( b ) comparison of the largest cluster size after an intentional attack . ] until now , we studied the healing process on networks where the nodes were randomly selected for removal .
another important removal technique simulates intentional attack on the network , where nodes are removed in descending order of their degree , i.e. the highest connected nodes are removed first .
this attack requires a much smaller percentage of removed nodes to destroy the network , since all the hubs that glue the network together are removed and the network disintegrates into small pieces .
the self - healing algorithm is efficient in this case , too ( fig .
[ fig8]b ) . the critical value of @xmath0 for a scale - free network with degree exponent @xmath74=2.5 moves from @xmath75 to @xmath55 .
this is a huge difference that allows the network to operate even after losing the most connected nodes in its backbone .
similarly , in the case of the airport connection network , the removal of the hubs leads to a disconnected network when 20% of the nodes are removed , but the healed network manages to preserve the largest cluster even if more than 60% of the nodes are removed .
in this paper we make explicit a fundamental feature of damage in complex networks and use it to design a very simple self - healing algorithm to restore connectivity - based function .
we show that even just this simple example is enough to restore large - scale connectivity in a severely damaged network , while keeping the reconnection cost low . in this particular scenario of structure enabling function , a node decides based on its own link losses to add a new link to one of its second neighbors ( provided at least one survives ) without considering any other parameter .
this approach has the added benefit that it is strictly local , and every node can make the decision autonomously .
the cost is also lower with this strategy compared to selecting random nodes for the same number of links .
the critical insight that working to restore function may not rely on restoring the initial structure allowed us to design a threshold , @xmath14 .
this threshold acts as a natural filter to direct the addition of the links towards the areas where function has been most affected by the damage in this connectivity - based scenario .
in fact , this strategy represents a worse - case scenario , because the nodes make new connections only after the removal of all nodes , or equivalently if the attack takes place faster than the nodes can react . in an alternate version
, the algorithm could be applied in a dynamic fashion so that a node continuously monitors its neighbors and decides to add new links based on current information , i.e. as soon as it notices that the number of its neighbors falls below the threshold . under this scenario ,
a node could establish more than one new links during the process and it would be easier to preserve long - range connectivity , simply because of the addition of a larger number of links .
what we showed above is that this extension is not necessary since a similar result can be achieved with much fewer new connections .
a few interesting parallels can be drawn between the self - healing algorithm and achlioptas processes @xcite . in the latter ,
we start from an empty network of @xmath20 nodes which grows by adding new links according to the process rules .
these rules can either favor or discourage the emergence of a largest cluster . in our case ,
the starting point is an already fractioned network , and the goal is to merge all clusters into one giant network with as few links as possible .
a key difference is that we only use local decisions , while in typical achlioptas processes the information of the involved cluster sizes is required .
we thank the dept . of homeland security for funds in support of this research through the ccicada center at rutgers , and
nsf easm grant no
. 1049088 .
we analyzed four different networks , based on the following datasets : 1 .
email network : email messages sent at the computer sciences department of london s global university http://lisgi1.engr.ccny.cuny.edu/~makse/social/emailcontacts.dat.gz .
cond - mat co - authorship : the network of co - authorship in preprints submitted to the cond - mat section of arxiv.org @xcite .
irvine social network : the dataset was downloaded from http://toreopsahl.com/datasets/#online_social_network and has been analyzed in @xcite . it includes online messages sent among students at the university of california , irvine , through a facebook - like social network .
usa airport network : a link indicates that two networks are connected by a direct flight .
this dataset refers to flights in 1997 and can be downloaded at http://vlado.fmf.uni-lj.si/pub/networks/data/mix/usair97.net .
r. cohen , k. erez , d. ben - avraham , and s. havlin , _ phys .
lett . _ * 85 * , 4626 ( 2000 ) .
r. albert , h. jeong , and a. l. barabasi , _ nature _ * 406 * , 378 ( 2000 ) .
s. boccaletti , v. latora , y. moreno , m. chavez , and d. hwang , _ phys .
rep . _ * 424 * , 175 ( 2006 ) .
d. helbing , _ nature _ * 497 * , 51 ( 2013 ) .
d. j. watts , _ proc . natl .
sci . _ * 99 * , 5766 ( 2002 ) .
r. cohen , k. erez , d. ben - avraham , and s. havlin , _ phys .
_ * 86 * , 3682 ( 2001 ) .
d. s. callaway , m. e. j. newman , s. h. strogatz , and d. j. watts , _ phys .
lett . _ * 85 * , 5468 ( 2000 ) .
a. a. moreira , j. s. andrade jr .
, h. j. herrmann , and j. o. indekeu , _ phys .
lett . _ * 102 * , 018701 ( 2009 ) . c. moore , g. ghoshal , and m. e. j. newman , _ phys .
e _ * 74 * , 036121 ( 2006 ) . n. farid and k. christensen , _ new j. phys . _
* 8 * , 212 ( 2006 ) .
j. saldana , _ phys .
e _ * 75 * , 027102 ( 2007 ) . c. m. schneider , l. de archangelis , and h. j. herrmann , _ europhys .
lett . _ * 95 * , 16005 ( 2011 ) .
h. bauke , c. moore , j. b. rouquier , and d. sherrington , _ eur .
b _ * 83 * , 519 ( 2011 ) . c. m. schneider , a. a. moreira , j. s. andrade jr .
, s. havlin , and h. j. herrmann , _ proc .
_ * 108 * , 3838 ( 2011 )
. m. faloutsos , p. faloutsos , and c. faloutsos , in ( acm , 1999 ) , p. 251 .
j. onnela , j. saramaki , j. hyvonen , g. szabo , d. lazer , k. kaski , j. kertesz , and a. barabasi , _ proc .
sci . _ * 104 * , 7332 ( 2007 ) .
s. h. strogatz , _ nature _ * 410 * , 268 ( 2001 ) .
r. v. sole , m. rosas - casals , b. corominas - murtra , and s. valverde , _ phys .
e _ * 77 * , 026102 ( 2008 ) .
b. monechi , v. d. servedio , and v. loreto , _ plos one _ * 10 * , e0125546 ( 2015 ) . c. ducruet , c. rozenblat , and f. zaidi , _ j. transp
. geogr . _ * 18 * , 508 ( 2010 ) .
e. lopez , r. parshani , r. cohen , s. carmi , and s. havlin , _ phys .
_ * 99 * , 188701 ( 2007 ) .
w. quattrociocchi , g. caldarelli , and a. scala , _ plos one _ * 9 * , e87986 ( 2014 )
. y. shang , _ phys .
e _ * 91 * , 042804 ( 2015 ) . m. stippinger and j. kertesz , _ physica a _ * 416 * , 481 ( 2014 ) .
a. majdandzic , b. podobnik , s. v. buldyrev , d. y. kenett , s. havlin , and h. e. stanley , _ nature physics _ * 10 * , 34 ( 2014 ) .
a. majdandzic , l. a. braunstein , c. curme , i. vodenska , s. levy - carciente , h. e. stanley , and s. havlin , arxiv preprint arxiv:1502.00244 ( 2015 ) .
j. wang , _ physica a _ * 392 * , 2257 ( 2013 ) . c. liu , d. li , e. zio , and r. kang , _ plos one _ * 9 * , e112363 ( 2014 )
. m. e. j. newman , _ networks : an introduction _ ( oxford university press , new york , 2010 ) .
j. m. kleinberg , _ nature _ * 406 * , 845 ( 2000 ) .
h. d. rozenfeld , c. song , and h. a. makse , _ phys .
lett . _ * 104 * , 025701 ( 2010 ) .
m. e. newman , _ proc .
_ * 103 * , 8577 ( 2006 ) .
a. clauset , m. e. j. newman , and c. moore , _ phys .
* 70 * , 066111 ( 2004 ) .
d. achlioptas , r. m. dsouza , and j. spencer , _ science _ * 323 * , 1453 ( 2009 ) .
m. e. j. newman , _ proc .
* 98 * , 404 ( 2001 ) .
t. opsahl and p. panzarasa , _ soc .
networks _ * 31 * , 155 ( 2009 ) . | the process of destroying a complex network through node removal has been the subject of extensive interest and research .
node loss typically leaves the network disintegrated into many small and isolated clusters .
here we show that these clusters typically remain close to each other and we suggest a simple algorithm that is able to reverse the inflicted damage by restoring the network s functionality .
after damage , each node decides independently whether to create a new link depending on the fraction of neighbors it has lost . in addition to relying only on local information , where nodes do not need knowledge of the global network status , we impose the additional constraint that new links should be as short as possible ( i.e. that the new edge completes a shortest possible new cycle ) .
we demonstrate that this self - healing method operates very efficiently , both in model and real networks .
for example , after removing the most connected airports in usa , the self - healing algorithm re - joined almost 90% of the surviving airports . |
frustrated magnets have been a focal point of the research on magnetism for the past two decades . in this article , the influence of geometrical frustration on the magnetic properties of the family of rare - earth strontium oxides , @xmath0 , ( where @xmath1gd , dy , ho , er , tm , and yb ) is discussed . given the nature of this special issue of _ low temperature physics _ on antiferromagnetism an extensive general introduction to magnetically frustrated systems
is omitted and the reader is instead referred to a collection of reviews available on the subject @xcite .
we start with a description of the crystal structure and general properties of @xmath0 and other closely related compounds and then present the recently obtained experimental results on their low - temperature magnetic properties by our group and others .
particular attention is paid to the zero - field ground state of @xmath2 , @xmath3 , and @xmath4 ( section [ section_zero_field ] ) , as well as the field - induced behaviour of these compounds ( section [ section_field ] ) . the penultimate section briefly reviews the other @xmath0 compounds and discusses the importance of crystal field effects .
the concluding section compares different members of the family and includes a brief summary . positions of the magnetic rare - earth @xmath5 ions within the @xmath0 compounds , with the two crystallographically inequivalent @xmath5 sites shown in different colours .
the left - hand panel emphasises the honeycomb arrangement of the @xmath5 ions when viewed along the @xmath6 axis , while the right - hand panel demonstrates the formation of zigzag ladders running along the @xmath6 axis which link the honeycomb layers and give rise to geometric frustration .
the blue box represents a crystal unit cell of the @xmath7 space group . ] the members of the @xmath0 family crystallise in the form of calcium ferrite @xcite , with the space group @xmath7 ; the crystal structure of these materials ( see fig .
[ fig1_structure ] ) can be viewed as a network of linked hexagons and triangles @xcite .
the most important feature of the linked hexagon ( or _ honeycomb _ " ) lattice is that it has the lowest coordination number , 3 , in two dimensions .
this feature attracts a lot of theoretical attention to the lattice , but being bipartite , the honeycomb lattice is not frustrated if only the nearest - neighbour interactions are considered .
the frustration in a honeycomb lattice can be induced by further neighbour interaction and numerous models of frustrated honeycomb lattices which include heisenberg , @xmath8 or ising @xmath9 interactions have been extensively studied theoretically , particularly for the @xmath10 quantum case . in the @xmath0 family , however , the cause of frustration is different ; it arises from the triangular ( or zigzag " ) ladders running along the @xmath6 axis which link the honeycomb layers . in this respect
@xmath0 compounds are similar to recently reported @xmath11 @xcite and perhaps to @xmath12 compounds @xcite . the term zigzag " has also been used to describe the spin - chain structure of another honeycomb lattice compound @xmath13 @xcite , but in a different context - to describe the arrangement of magnetic moments formed there .
the orthorhombic unit cell of the @xmath0 compounds contains 4 sr atoms on a single site , 8 @xmath14 atoms ( split equally between two sites ) and 16 oxygen atoms occupying 4 sites ; all sites are of the @xmath15 type with the coordinates @xmath16 @xcite
. the @xmath17 and @xmath18 axes of the unit cell are typically quite long , about 10 and 12 respectively , while the @xmath6 axis is the shortest , at around 3.4 on average . the magnetic @xmath14 atoms are surrounded by the distorted oxygen octahedra , forming the chains running along the @xmath6-axis .
the shortest @xmath19 separation is along the chains ; there is a slightly larger separation between the chains formed by the @xmath14 atoms occupying the same sites ( which are shown in fig .
[ fig1_structure ] in the same colour ) , while the distance between @xmath14 atoms from different sites ( red and green in fig .
[ fig1_structure ] ) is much greater @xcite .
such a crystal structure predetermines the quasi one - dimensional magnetic properties of the @xmath0 compounds , as for rare - earth ions in insulating materials direct exchange is often the most important mechanism for magnetic coupling .
it is rather useful to note the equivalence of the well - studied linear chain model with nearest and next - nearest interactions @xcite and the ladders of rare - earth ions described here if these were `` stretched '' along the @xmath6 axis .
an important observation to make prior to the description of their properties is that both polycrystalline and single crystal samples of the @xmath0 compounds have been used for investigations .
the progress achieved to date in the understanding of their complex behaviour is , however , largely due to the availability of high quality single crystals .
crystals of magnetic @xmath0 oxides and their non - magnetic analogues ( with @xmath1 lu or y ) have been synthesised by the floating zone technique by our group and others @xcite .
examples of the single crystals grown @xcite are shown in fig .
[ fig2_xtals ] ; the size of the crystals available is certainly sufficient for neutron scattering experiments , including inelastic studies . in comparison , the magnetic properties of structurally similar @xmath20 ( ref . ) , @xmath21 ( ref . ) and @xmath22 ( ref . )
compounds have not yet been probed to a significant degree , as only polycrystalline samples are available . as grown boules of @xmath3 ( left panel ) and @xmath4 ( right panel ) single crystals , using growth speeds of 6 to 8 mm / h .
figure is from ref . @xcite.,title="fig : " ] as grown boules of @xmath3 ( left panel ) and @xmath4 ( right panel ) single crystals ,
using growth speeds of 6 to 8 mm / h .
figure is from ref .
@xcite.,title="fig : " ]
magnetic structure of @xmath2 as determined from rietveld refinements of the neutron - diffraction pattern at @xmath23 k. the same structure is shown twice to emphasise different arrangements of the magnetic moments along the @xmath6 axis and with respect to the hexagons in the plane .
two different er sites and their surroundings are shown in different colours .
only one of the sites carries a significant magnetic moment .
figure is from ref .
[ ] . ]
@xmath2 is found to order magnetically at @xmath24 k with a @xmath25 antiferromagnetic ( afm ) structure ( depicted in fig .
[ fig3_seo_str ] ) consisting of ferromagnetic chains running along the @xmath6 axis , with adjacent chains arranged antiferromagnetically @xcite .
the refinement of the powder neutron diffraction ( pnd ) data suggested that the moments point along the @xmath6 direction and that only one of the two er@xmath26 sites possesses a sizeable magnetic moment .
it was not possible to determine which particular site contributed to the ordering , as the magnetic moments may be swapped between the two sites without changing the calculated pnd pattern significantly @xcite .
the situation with the low - temperature magnetic structure of @xmath2 became much clearer after the publication of single crystal polarised neutron diffraction results @xcite , which are summarised in fig .
[ fig4_seo_diff ] .
the presence of a magnetic component with long - range order ( lro ) below 0.75 k was confirmed by the observation of sharp resolution - limited bragg peaks at integer @xmath27 positions ( see fig . [
fig4_seo_diff]a ) .
these peaks are replaced by broad and much weaker diffuse scattering features above @xmath28 ( see fig . [
fig4_seo_diff]b ) .
surprisingly , another distinct magnetic component corresponding to a short - range incommensurate structure was also detected .
this component manifests itself by the presence of a strong diffuse signal , forming the undulated planes of scattering , which are seen as _ rods _ " in a particular scattering plane . fig .
[ fig4_seo_diff]c , for example , clearly shows the two rods are at positions @xmath29 and @xmath30 , where @xmath31 is dependent upon @xmath32 .
a monte carlo simulation @xcite showed that a simple model based on a ladder of triangles in which the nearest - neighbour interactions are approximately five times stronger than the next - nearest - neighbour interactions satisfactorily mimics the observed diffuse scattering patterns . from the width of the diffuse _ rods _ " at the base temperature , the estimates for correlation length along the @xmath6 axis vary from 130 to 70 depending on which
_ rod _ " is considered @xcite , but in any event the afm correlations are rather long and include more than 20 magnetic ions .
the interpretation of these data is that apart from the @xmath25 lro component shown in fig .
[ fig3_seo_str ] ) the magnetic structure of @xmath2 consists of highly correlated afm chains running along the @xmath6 axis , but the correlations between the chains are rather weak .
on warming from the base temperature of a dilution cryostat to much higher temperatures , the partially ordered component does not undergo a pronounced phase transition unlike the @xmath25 component .
instead it gradually loses the intensity , but it could be easily seen at 0.8 k ( see fig .
[ fig4_seo_diff]d ) and in fact much higher temperatures ( not shown ) . from the polarisation analysis @xcite , the magnetic moments in the long - range commensurate and short - range incommensurate structures are found to be predominantly pointing along the @xmath6 and @xmath17 axes , respectively . at a first glance
, a refinement of the low - temperature magnetic structure of @xmath3 looks very similar to @xmath2 .
the pnd data @xcite returned a collinear afm @xmath25 component , very similar to the one shown in fig .
[ fig3_seo_str ] , below the ordering temperature of 0.68 k , with only a half of the ho@xmath26 ions carrying a significant moment .
the presence of another magnetic component was also observed as a pronounced scattering around the @xmath33 positions .
further single crystal diffraction data @xcite , however , revealed a more complicated picture .
the observed broad diffraction peaks show that the @xmath25 component ( corresponding to a collinear antiferromagnetically coupled structure ) is of short - range order type .
the planes of diffuse scattering corresponding to another kind of magnetic order appear to be nearly perfectly commensurate , _
i.e. _ the parameter @xmath31 is almost zero for them , although the variations of the intensity have been seen in both the @xmath34 and @xmath35 planes in reciprocal space .
this observation suggests that the second type of short - range order present in @xmath3 is principally one - dimensional in nature , that is the magnetic structure is essentially a collection of afm coupled chains running along the @xmath6 axis with the intrachain correlations remaining rather weak down to lowest temperatures .
similarly to what have been observed in @xmath2 , a magnetic component with the propagation vector @xmath25 exists below a well - defined transition temperature , while the one - dimensional scattering is observed at much higher temperatures .
correlation lengths associated with the broad peaks are about 150 in the @xmath36 plane and about 190 along the @xmath6 axis , while the correlation length associated with the diffuse scattering planes is 230 along the @xmath6 axis at the lowest temperature . from the polarisation analysis @xcite ,
the magnetic moments in the @xmath25 and quasi one - dimensional structures are found to be pointing along the @xmath6 and @xmath18 axes , respectively .
temperature dependence of the specific heat divided by temperature of @xmath4 in zero field .
the inset shows the temperature dependence of the entropy , @xmath37 ( solid line ) , calculated as the area under the @xmath38 curve which has been extended linearly down to @xmath39 k. the dashed line indicates the position of 2r@xmath40 , which corresponds to the magnetic contribution for a system with an effective @xmath10 .
figure is from ref .
in contrast to the other members of the @xmath0 family investigated so far , @xmath4 does not show any sign of magnetic phase transition down to the lowest available temperatures . in zero field , heat capacity @xmath41 measurements indicate that this compound appears to be magnetically disordered down to at least 0.39 k ( see fig .
[ fig5_sdo_cvst ] ) .
the @xmath38 curve shows a very broad maximum at 0.77 k and a nearly linear temperature dependence below this peak .
there are no sharp features in the heat capacity curve which can be attributed to a phase transition to a magnetically ordered state .
pnd data for @xmath4 show no signs of any long - range magnetic order down to 20 mk , as the scattering pattern in zero field is dominated by broad diffuse scattering peaks @xcite .
the magnetic entropy recovered in @xmath4 between zero temperature and @xmath42 k ( see inset in fig .
[ fig5_sdo_cvst ] ) amounts to @xmath43 , which suggests that at the lowest temperature the system is essentially a doublet with the magnetic moments restricted to point only along the easy axis ( ising ) direction .
the higher - temperature magnetisation curves for the polycrystalline samples of these compounds have been reported by karunadasa _ et al . _
@xcite and revealed non - linear behaviour of magnetisation in field with pronounced maxima in the derivatives , @xmath44 as a function of applied field .
further single crystal magnetisation @xcite and heat capacity @xcite measurements , however , revealed highly anisotropic behaviour , which is partially masked in the polycrystalline samples .
we therefore summarise in this section the results obtained on single crystal samples .
magnetisation versus field curves @xmath45 and their field derivatives @xmath44 obtained for @xmath3 for a field applied along the principal symmetry axes are shown in fig .
[ fig6_sho_mvsh ] . for @xmath46 ( which is a hard magnetisation direction ) , @xmath45 remains rather small in any field .
for other two directions of an applied field , a significant portion of the total magnetic moment is recovered , although no complete saturation of magnetisation is observed , as the @xmath44 values remain nonzero even in a field of 70 koe @xcite .
this implies that the spins of the ho@xmath26 ions are not fully aligned at this field . for @xmath47
the magnetisation process is characterised by a double phase transition ( seeing most clearly as two maxima in the @xmath44 curves in the bottom - left panel of fig .
[ fig6_sho_mvsh ] ) indicative of the appearance of magnetisation plateau .
although the plateau is not well - pronounced , i.e. the derivative @xmath44 remains positive and relatively large between the maxima , one has to remember that the temperature for these measurements was relatively high , 0.5 k. for @xmath48 k ( see right - hand panels in fig .
[ fig6_sho_mvsh ] ) , the plateau in the magnetisation disappears .
therefore , it is likely that at the temperatures approaching 0 k the plateau will be much more obvious .
the value of the magnetisation on the plateau is about a third of the value observed in higher field .
this fact allowes to conjecture @xcite that the magnetic structure on the plateau is of collinear _ up - up - down _ type , where on each triangle the two moments are pointing along the applied field and the third moment is antiparallel to them .
for @xmath49 the magnetisation process in @xmath3 is characterised by a single , relatively sharp phase transition , above which the magnetisation remains practically constant . for @xmath48
k ( see right - hand panels in fig . [ fig6_sho_mvsh ] ) , the plateau in the magnetisation disappears , the maximum in @xmath44 for @xmath49 broadens and shifts to slightly higher fields , while the @xmath45 curve for @xmath46 remains unchanged . in @xmath2 and @xmath4
the field dependence of the magnetisation looks rather similar to what have been observed in @xmath3 , but the directions of applied field along which the plateaux and sharp single phase transitions appear ( as well as the actual values of critical fields ) are different . in all three compounds a single and relatively sharp increase in magnetisation is seen for @xmath50 , a direction along which the magnetic moments are pointing in the @xmath25 structures of @xmath2 and @xmath3 . in @xmath2 the application of field along the @xmath17 axis results in a magnetisation plateau , while the @xmath18 axis seems to be a hard magnetisation direction . in @xmath4
a magnetisation plateau appears for @xmath51 , while the @xmath17 axis seems to be a hard magnetisation direction . magnetic @xmath52 phase diagram of @xmath4 for @xmath53 $ ] obtained from the heat capacity measurements .
colour represents the heat capacity divided by temperature in the units of j/(mol k@xmath54 ) .
figure is from ref .
] the fact that in all three compounds the observed plateaux in magnetisation appear at approximately a third of the magnetisation saturation values suggests that for the field applied along either the @xmath17 or @xmath18 axes the contribution from the @xmath25 structures remains rather weak , that is , the magnetic moments in these structures remain pointing along the @xmath6 axis . further insight into the field - induced properties of @xmath4 can be gained from the heat capacity measurements @xcite .
[ fig7_sdo_phd ] ( where the value of the heat capacity divided by temperature is represented by the colour scale shown on the right of the figure ) shows a magnetic phase diagram of @xmath4 obtained by combining the heat capacity field - scans for @xmath51 . at the lowest experimentally available temperature of 0.39 k , a sharp double peak at about 20 koe is the main feature in the @xmath55 curve .
the peaks indicate multiple magnetic field - induced transitions in @xmath4 for this direction of an applied field , but from the bulk - property measurements alone it is impossible to determine whether any of the field - induced phases are long - range in nature . therefore further neutron diffraction experiments are required to answer this question .
remarkably for @xmath50 the application of a magnetic field does not result in any features in the @xmath55 curves sharp enough to be indicative of a phase transition @xcite which emphasises once again the highly anisotropic nature of the magnetisation process in the @xmath0 compounds .
( a ) magnetic powder pattern of @xmath56 collected on d7 at 30 mk . the magnetic structure where the arrows represent the yb@xmath26 ions spins ( yb1 blue , yb2 red ) ( b ) along the zigzag chains and ( c ) projected onto the ab plane .
the sr@xmath57 and o@xmath58 ions are represented by yellow and red circles , respectively .
figure is from ref .
[ ] . ] magnetic phase diagram of @xmath56 with magnetic field along ( a ) @xmath17 , ( b ) @xmath18 , and ( c ) @xmath6 axes .
the colours indicate the heat capacity in units of @xmath59/k .
the circles indicate the critical fields extracted from magnetocaloric effect measurements and the the triangles the critical fields extracted from magnetisation measurements .
black solid lines show second - order phase transitions .
dash - dot black lines indicate the transition from the afm phase to a less ordered one .
dashed black lines show metamagnetic crossovers . for the phase diagram along the @xmath6 axis , above 4.5 t
there are just three heat capacity scans at 6 , 9 , and 12 t , the colours between them result from the interpolation of the data . below 4.5 t ,
the data were collected every 0.2 t. the temperature axis is in logarithmic scale .
figure is from ref . [ . ] ] apart from @xmath2 , @xmath3 and @xmath4 reviewed above , the only other family - member for which the low - temperature properties have been reported is @xmath56 .
heat capacity measurements @xcite revealed a magnetic phase transition to lro at @xmath60 k. neutron diffraction measurements @xcite ( see fig .
[ fig8_syo_diff]a ) showed that the structure is a noncollinear @xmath25 antiferromagnet in which the magnetic moments of two inequivalent yb@xmath26 ions lie in the @xmath36 plane , but have different moment sizes and directions .
both moments are reduced from the fully ordered moment of yb@xmath26 ( see figs . [ fig8_syo_diff]b and [ fig8_syo_diff]c ) .
similarly to what has been observed for other @xmath0 compounds , the application of a relatively strong field , 140 koe , along any direction does not result in a recovery of a full moment expected for the yb@xmath26 ions @xcite .
very interesting and highly anisotropic magnetic phase diagrams ( see fig . [ fig9_syo_phd ] ) have been reconstructed for @xmath56 from the magnetocaloric and the magnetisation measurements @xcite
. a large number of transitions and crossovers were found which has been taken as an indication of the presence of metamagnetic phases due to spin - flip and spin - flop processes as well as possible competition between exchange interactions and magnetic anisotropy , however , the exact nature of the field - induced phases in @xmath56 remains presently unknown .
we have recently started investigations of the magnetic properties of @xmath61 and from the heat capacity measurements have found @xcite that the compound undergoes two magnetic transitions at 2.72 and 0.47 k. the initial characterisation of @xmath61 by karunadasa _
et al . _
@xcite missed the higher temperature transition .
since gd@xmath26 is expected to be nearly isotropic , it is quite surprising that the transition temperature of @xmath61 is much higher than that of the other members of the family .
the higher ordering temperature of @xmath61 is , however , consistent with the properties of structurally similar @xmath20 family , in which @xmath62 orders at 2.6 k , while the rest of the compounds do not order down to at least 1.7 k@xcite . in the absence of neutron diffraction data
the magnetic structure of @xmath61 remains unknown at present .
the only other established fact about low - temperature properties of @xmath61 is that it undergoes a field - induced transformation at 20.5 koe ( for @xmath63 k ) in a field applied along the @xmath6 axis @xcite .
it would certainly be interesting to expand the @xmath0 family and to test the magnetic properties of tm , tb , sm , nd containing compounds provided that single crystal samples can be prepared . from the findings presented above for the @xmath0 compounds , which vary greatly from one @xmath14 ion to another , it is rather obvious that the low - lying crystalline electric field ( cef ) levels must play an important role in the formation of the highly anisotropic magnetic properties . at present
the cef schemes remain unknown and the task of establishing them may not be trivial : there are 8 @xmath14 ions on two district crystallographic sites in the unit cell .
the symmetry is rather low , therefore the number of cef levels is expected to be large .
also , the positions of the levels at lower temperature can potentially be influenced by the development of short - range magnetic order .
inelastic neutron scattering ( ins ) data for @xmath4 and @xmath3 have been collected back in 2005 by kenzelman _
et al . _
the more recent ins results reported for @xmath3 by ghosh _
et al . _
@xcite are largely in agreement with the previous data .
we have also collected further ins data for @xmath2@xcite , but to date neither group have reported any cef schemes . moreover , there are further indications @xcite that the problem could prove to be difficult to solve .
an additional motivation for preparing this review was to alert the frustrated magnetism community to the presence of such a challenging , but potentially very important problem .
[ cols="^,>,^,^,^,<",options="header " , ] we conclude this review by listing in table [ table1 ] the most important magnetic parameters of the @xmath0 compounds , such as weiss temperature @xmath64 , effective moment @xmath65 in bohr magnetons @xmath66 , magnetic ordering temperature @xmath28 as well as indicating the nature of the zero field ground state and the presence of critical fields @xmath67 for different directions of an applied field
. important pieces of information missing from table [ table1 ] include the values of the various exchange interactions and details on the magnetic anisotropy in the @xmath0 compounds .
this information which is typically obtained from inelastic neutron scattering experiments is so far unavailable . only after establishing the absolute values ( including signs ) and relative strengths of the relevant exchange interactions , as well as details of the magnetic anisotropy ,
could one classify the @xmath0 compounds as a collection of weakly interacting chains of magnetic moments , or as a network of ladders consisting of triangles . apart from neutron scattering , further monte carlo simulations
, both direct and reverse , as well as density - functional theory band - structure calculations may play an important role in determining the magnetic interactions . in @xmath2 the lro @xmath25 afm phase ( see fig .
[ fig3_seo_str ] ) in which the magnetic moments point along the @xmath6 axis appears below 0.75 k while in @xmath3 a very similar phase appearing below 0.68 k remains short - range ordered down at least 50 mk .
apart from this phase , a sro quasi one - dimensional afm component is found in both compounds , but the direction along which the spins are pointing is different - it is parallel to the @xmath17 axis in @xmath2 and parallel to the @xmath18 axis in @xmath3 . in @xmath2 the diffuse scattering signal corresponding to
the quasi 1d component appears to be much more structured compared to the @xmath3 , which could be indicative of the importance of further neighbour exchange interactions . in @xmath56
the magnetic moments are confined to the @xmath36 plane ( see figs . [ fig8_syo_diff]b and [ fig8_syo_diff]c ) , with the two different yb@xmath26 sites having very different moment sizes and directions @xcite .
no long - range magnetic order has been found in @xmath4 down to 20 mk . despite having the weakest magnetic interactions ( as demonstrated by the lowest weiss temperature ) @xmath61 orders at the highest temperature of 2.72 k and undergoes another transition at 0.47 k. this observation suggests an immense importance of the magnetic anisotropy in establishing the ground state of the @xmath0 compounds and the potential competition between the exchange interactions and the single - ion effects .
for all the @xmath0 compounds the application of an external magnetic field results in the appearance of complex and highly anisotropic phase diagrams revealing multiple phase transitions , magnetisation plateaux and cross - over regions .
the magnetic structure of the field - induced phases remains presently unknown .
we hope that this review will stimulate further research on the magnetic properties of the @xmath0 and related honeycomb lattice compounds .
the author is very grateful to g. balakrishnan , m.r .
lees , d.mck .
paul , n.r .
wilson , t.j .
hayes , o. young , t.h .
cheffings , p. manuel , d.d .
khalyavin , d.t .
adroja , f. demmel , b.d .
rainford , a.r .
wildes , b. ouladdiaf , l.c .
chapon , s. de brion , e. suard , c. ritter , p.p .
deen , r.j .
mason , a.k.r .
briffa , m.w .
long , j. mercer , m.l .
plumer , a. desilets - benoit , a.d .
bianchi , and d.l .
quintero - castro for the collaborations on the projects involving the magnetic properties of the @xmath0 compounds .
99 a.p .
ramirez , annu .
* 24 * , 453 ( 1994 ) ; m.f .
collins and o.a .
petrenko , can .
j. phys .
* 75 * , 605 ( 1997 ) ; s.t . bramwell and m.j.p .
gingras , science * 294 * , 1495 ( 2001 ) ; h.t .
diep ( ed . ) , _ frustrated spin systems _ , world scientific , singapore , ( 2005 ) ; j.s .
gardner , m.j.p .
gingras and j.e .
greedan , rev .
* 82 * , 53 ( 2010 ) ; c. lacroix , p. mendels and f. mila ( ed . ) , _ introduction to frustrated magnetism _ , springer series in solid - state sciences * 164 * , ( 2011 ) .
decker and j.s .
kasper , acta cryst . * 10 * , 332 ( 1957 ) .
l. m. lopato and a. e. kushchevskii , ukr . khim .
zh . * 39 * , 7 ( 1973 ) . h. karunadasa , q. huang , b.g .
ueland , j.w .
lynn , p. schiffer , k.a .
regan , and r.j .
cava , phys .
b * 71 * , 144414 ( 2005 ) .
f. damay , c. martin , v. hardy , a. maignan , g. andre , k. knight , s.r .
giblin , and l.c .
chapon , phys .
b * 81 * , 214405 ( 2010 ) ) ; f. damay , c. martin , v. hardy , a. maignan , c. stock , and s. petit , phys . rev .
b * 84 * , 020402 ( 2011 ) .
s. das , a. niaz , y. mudryk , v.k .
pecharsky , and d.c .
johnston , phys .
b * 81 * , 104432 ( 2010 ) . f. ye , s. chi , h. cao , b.c .
chakoumakos , j.a .
fernandez - baca , r. custelcean , t.f .
qi , o.b . korneta , and g. cao , phys .
b * 85 * , 180403 ( 2012 ) .
j. chaloupka , g. jackeli , and g. khaliullin , phys .
lett . * 110 * , 097204 ( 2013 ) a. yoshimori , j. phys .
japan * 14 * , 807 ( 1959 ) .
g. balakrishnan , t.j .
hayes , o.a .
petrenko and d. mck .
paul , j. phys .
matter * 21 * , 012202 ( 2009 ) .
s. ghosh , h.d .
zhou , l. balicas , s. hill , j.s .
gardner , y. qiu , and c.r .
wiebe , j. phys .
matter * 23 * , 164203 ( 2011 ) .
quintero - castro , b. lake , m. reehuis , a. niazi , h. ryll , a.t.m.n .
islam , t. fennell , s.a.j .
kimber , b. klemke , j. ollivier , v. garcia sakai , p.p .
deen , and h. mutka , phys .
b * 86 * , 064203 ( 2012 ) .
y. doi , w. nakamori , and y. hinatsu , j. phys .
matter * 18 * , 333 ( 2006 ) .
k. hirose , y .
doi , and y. hinatsu , j. solid state chem .
* 182 * , 1624 ( 2009 ) .
o. ofer , j. sugiyama , j.h .
brewer , e.j .
ansaldo , m. mnsson , k.h .
chow , k. kamazawa , y. doi , and y. hinatsu , phys .
b * 84 * , 054428 ( 2011 ) .
y. misawa , y. doi , and y. hinatsu , j. cer .
japan , * 117 * , 85 ( 2009 ) o.a .
petrenko , g. balakrishnan , n.r .
wilson , s. de brion , e. suard and l.c .
chapon , phys . rev .
b * 78 * , 184410 ( 2008 ) . t.j .
hayes , g. balakrishnan , p.p .
deen , p. manuel , l.c .
chapon and o.a .
petrenko , phys .
b * 84 * , 174435 ( 2011 ) . o. young , l.c .
chapon and o.a .
petrenko , j. phys .
series * 391 * , 012081 ( 2012 ) .
o. young , a.r .
wildes , p. manuel , b. ouladdiaf , d.d .
khalyavin , g. balakrishnan and o.a .
petrenko , phys .
b * 88 * , 024411 ( 2013 )
stewart , p.p .
deen , k.h .
andersen , h. schober , j .- f .
barthemy , j.m .
hillier , a.p .
murani , t. hayes , and b. lindenau , j. appl . crystallogr .
* 42 * , 69 ( 2009 ) .
cheffings , m.r .
lees , g. balakrishnan and o.a .
petrenko , j. phys .
matter * 25 * , 256001 ( 2013 ) .
hayes , o. young , g. balakrishnan , and o.a .
petrenko , j. phys .
japan * 81 * , 024708 ( 2012 ) .
petrenko , l.c .
chapon , and t.j .
hayes _ ill exp .
_ 5 - 31 - 1695
. o. young , g. balakrishnan and o.a .
petrenko , unpublished ( 2013 ) .
m. kenzelmann and b. rosendahl hansen , isis experimental report 510364 .
petrenko , t.j .
hayes and f. demmel : isis experimental report 920327 .
a. desilets - benoit , thesis , universite de montreal ( 2010 ) . | recent progress in the understanding of the complex magnetic properties of the family of rare - earth strontium oxides , @xmath0 , is reviewed .
these compounds consisting of hexagons and triangles are affected by geometrical frustration and therefore exhibit its characteristic features , such as a significant reduction of magnetic ordering temperatures and complex phase diagrams in an applied field .
some of the observed features appear to be rather remarkable even in the context of the unusual behaviour associated in geometrically frustrated magnetic systems .
of particular interest is the coexistence at the lowest temperature of different magnetic structures ( exhibiting either long or short - range order ) characterised by different propagation vectors in materials without significant chemical or structural disorder . |
in microrheology , the mechanical properties of a substance are studied by monitoring the trajectories of single particles @xcite .
thereby , one distinguishes between two different modes : passive microrheology , in which the equilibrium properties of the material are probed by studying the motion of a single particle @xcite , and active microrheology , in which a single particle is subjected to an external drive . in the latter case ,
the response to strong external perturbations is of particular interest since it allows one to study the properties of the system in an out - of - equilibrium situation @xcite .
recently , this method has been applied to study the properties of glass - forming systems in both , experiments @xcite and simulations @xcite . for these systems ,
several remarkable effects have been observed like a nonlinear thinning of supercooled liquids @xcite , local melting of glassy samples @xcite and intermittent superdiffusivity in the supercooled regime @xcite . because of this , there is recently a field of vivid research to provide a theoretical description of the out - of - equilibrium properties @xcite , mostly in the framework of mode - coupling theory @xcite . for diluted colloidal solutions ,
a theoretical description in terms of the two - particle correlations between the probe and the host liquid has been developed by brady et al.@xcite .
a standard experiment in the context of active microrheology is the pulling of a single probe particle through a supercooled liquid with a constant external force @xmath0 @xcite , see fig .
[ fig : rheology ] ( a ) of a schematic visualization .
thereby , a characteristic quantity is the mobility @xmath1 of the probe , which is defined as the ratio of the stationary velocity @xmath2 of the particle along the force direction and the strength of the external field @xmath0 .
the mobility is determined by the resistance the particle experiences during the motion through the supercooled liquid and is therefore a direct measure for the mechanical properties of the system @xcite . as it can be seen in fig .
[ fig : rheology](b ) , the mobility is constant for small forces , which is in accordance with the einstein relation @xcite @xmath3 where @xmath4 denotes the equilibrium diffusion constant , and @xmath5 the thermal energy of the system . for large forces , the mobility becomes itself force - dependent , thereby indicating a thinning of the local environment .
for the physical description of non - equilibrium states , one often finds the concept of _ effective temperatures _ in literature @xcite .
thereby , a new temperature @xmath6 is defined which includes , on the one hand , the temperature imposed by the thermostat and , on the other hand , the control parameter of non - equilibrium , e.g. the maximum strain of a shear cycle or the waiting time in an aging system .
the exact definition can result from evaluating the fluctuation - dissipation theorem @xcite , studying the population of energies @xcite , applying weakly interacting degrees of freedom that act as a thermometer @xcite or simply via scaling relations of different dynamic quantities @xcite .
for the case of force - driven active microrheology the latter relations have been observed for the inverse mobility @xmath7 and for the perpendicular diffusion coefficient of the tracer particle @xmath8 , however , with quite different values of @xmath6 . since one would expect a `` physical '' temperature to be independent from the studied quantity , the insights these scaling temperatures are somewhat limited .
a novel way to describe microrheology in supercooled liquids is the application of the potential energy landscape ( pel ) approach .
the basic idea of this approach is to relate the dynamics of the system to its thermodynamic state , characterized by its overall potential energy @xcite .
in particular it turns out that the dynamics of supercooled liquids can be related to escape events out of mesoscopic funnels in the pel , formed by a set of local minima , which is called metabasins ( mb ) @xcite .
although the escape out of a single mb is a complicated process , which may contain several transitions between local minima , it can be described as a single activated process over a energy barrier that depends on the depth of the funnel @xcite .
this is in striking accordance to the trap model @xcite that has been successfully applied to understand the connection between thermodynamics and dynamics in supercooled liquids @xcite .
on a more coarse grained level , the sequence of escapes from the mbs is largely uncorrelated and can be described in terms of a continuous time random walk ( ctrw ) scheme @xcite .
recently , the pel approach has been used to describe the dynamics of a driven tracer particle in terms of a ctrw through its underlying pel @xcite .
thereby it has been found that the pel approach offers a non - trivial decomposition of the linear and nonlinear response of @xmath9 into two independent quantities . in particular
, one finds that the spatial part of the mobility behaves essentially constant , whereas the average waiting time @xmath10 in between two transitions or , equivalently , the average rate @xmath11 of a transition strongly varies ; thereby , @xmath12 denotes a number average , whereas @xmath13 denotes the average taken with respect to the boltzmann distribution of energies , see eq .
( [ eq : defavgamma ] ) further below .
this very different behavior means that the temporal part of the mobility is the main source of nonlinearity and , thus , that the understanding of the nonlinearity of the relevant timescale is identical to understanding the nonlinearity of the mobility itself . in this article
we offer a quantitative microscopic description of the thinning behavior of force - dependent mobility @xmath9 . for this purpose
we have conducted computer simulations of a typical glass - forming model liquid which allows us to study the real space dynamics simultaneous to the dynamics of the system in its underlying pel ( for details , see methods ) .
we first analyze the effect of the external force on the energy distribution , which can be rationalized in terms of an effective temperature @xmath6 . for this effective temperature , a universal scaling with a single parameter
is observed which can be applied for nearly all forces and temperatures studied in this work .
second , we study the force dependence of the local escape rates and provide a description in terms of an extended trap model .
thereby we find that the force dependence of this rates is governed by the same effective temperature as it was found for the thermodynamic quantity .
third , both quantities can be combined to obtain a quantitative description of the average escape rate @xmath11 in terms of @xmath6 . as a consequence
the dynamics of the system behaves during the application of an external force similar to the dynamics of a quiescent system at a higher ( bath ) temperature , resulting in a force - temperature superposition .
this result can be directly applied to the real - space mobility . in the last step ,
we apply the concept , within some approximations though , to the diffusive properties of the system .
thereby we observe a similar scaling with the same effective temperature as for the mobility , however , the quality of the scaling depends on the studied quantity , due to the different validity of the applied approximation .
we critically discuss that a slightly different choice of the scaling factor of the effective temperature can enhance the scaling properties of the diffusion coefficient .
however , this can also lead to a loss of the physical meaning of the so defined effective temperature .
we have conducted molecular dynamics simulations ( md ) of a binary ( 80:20 ) mixture lennard - jones particles where the potential parameters of kob and andersen @xcite have been used . since the analysis of the underlying pel requires small system sizes @xcite the potential cutoff radius has been reduced to @xmath14 ( in dimensionless lennard - jones units ) which allows us to perform simulations of systems with @xmath15 particles .
this system size is known to display only very small finite size effects concerning its diffusive behavior in equilibrium @xcite .
as is can be seen in fig .
[ fig : fsize ] , also the mobility studied in this work is only very weakly dependent on the size of the system and the small system only displays deviations about @xmath16 . at the beginning of the simulation one of the majority particles of the system
is randomly chosen and pulled by a constant force along a certain direction , thereby reaching a steady - state regime @xcite .
constant temperature conditions have been achieved by coupling the bath particles ( all particles except of the tracer ) to a nos - hoover thermostat @xcite .
we show in the supplemental material @xcite that , in the range of forces we are studying in this work , the driving of the tracer causes no unphysical fluctuations of the kinetic energy of the bath particles .
we study a ( bath ) temperature range of @xmath17 which is slightly above the mode - coupling temperature @xmath18 of the system @xcite . in this regime neither incomplete structural relaxation nor aging has been observed .
the key step of our analysis is the tracking of the path the system has taken through its pel . for this purpose
we compute the current inherent structure of the system at certain points in time by minimizing the potential energy of the system via conjugate gradient method .
the minimization is done with a copy of the system so that the md trajectories are not influenced . for the computation of the potential energy ,
only the lennard - jones interactions are taken into account .
this choice is necessary to compare the states of the driven and to those of the undriven system , since only in this case they all belong to the very same pel . at the end of the simulations , all inherent structures , between which forward - backwards transitions are observed ,
are combined to metabasins , whereas the structure with the lowest potential energy , which is typically the most frequently visited one , defines the energy . the exact protocol to determine the metabasins involves some minor subtleties which are described in @xcite .
in the pel description , the overall escape rate from an arbitrary mb depends on two formally independent quantities : the distribution @xmath19 of mb energies @xmath20 populated by the system , which is a purely thermodynamic quantity and the rates @xmath21 to escape from a mb of energy @xmath20 , which describe the local kinetics .
the escape rates @xmath21 are defined as the inverse average waiting time @xmath22 before leaving a mb of energy @xmath20 .
the brackets thereby denote the average over different realizations . the total average escape rate or , equivalently , the inverse average waiting time is given by @xcite @xmath23 where the index @xmath24 of @xmath25 shall emphasize that the average has been taken with respect to the energy distribution @xmath19 .
applying an external perturbation may alter the thermodynamics , the local kinetics or both in a different manner , giving rise to the observed nonlinear behavior of @xmath11 .
thus , the understanding of the force dependences of @xmath26 and @xmath27 immediately offer the understanding of the nonlinear mobility in general . *
( a ) * * ( b ) * * ( c ) * * ( d ) * the thermodynamic state of the system is characterized by the boltzmann distribution of energies @xmath19 .
numerically , @xmath19 can simply determined by computing the energy of a mb at arbitrary chosen points in time . in fig .
[ fig : energy](a ) , @xmath19 is shown for a quiescent system at different temperatures .
all curves display a gaussian shape , which is shifted to larger values as the temperature is increased .
in contrast to the average value , the width as well as the height of the distribution is not dependent on the temperature . in the quiescent system for @xmath28 , at which the anharmonic contributions of the local minimum are fully negligible @xcite , the energy distribution results from the density of states ( dos ) @xmath29 via @xmath30 where the states are weighted by the thermal energy of the system . with this equation
, one can determine the underlying dos by rescaling the distributions shown in fig .
[ fig : energy](a ) . as it is shown in @xcite ,
the underlying dos is a gaussian with the same width as @xmath19 .
the dynamics of the system can still be described for @xmath31 in terms of the original ( @xmath32 ) potential energy minima since the minima including the external force are , as we show in the supplemental material @xcite , basically the same in the range of forces displayed in .
it will be of key importance how the additional force changes the kinetic and thermodynamic properties of the system . in fig .
[ fig : energy](b ) @xmath27 is shown for different force strengths .
one observes that the gaussian shape of the distribution is recovered also in the non - equilibrium case , but the distribution is shifted to higher energies .
the width of the distribution is not altered by the application of the force .
this is remarkable , since it displays the same characteristics which have already been observed for @xmath19 in the quiescent system at different temperatures , see fig . [
fig : energy](a ) .
in particular , microrheological forcing and heating are indistinguishable in terms of the energy distribution .
this striking similarity between forcing and heating motivates the definition of an effective temperature which characterizes the shift of @xmath27 .
since , as mentioned above , @xmath29 is by construction independent from the force , we can rewrite eq .
( [ eq : pepst ] ) and insert the @xmath27 for the energy distribution : @xmath33 thereby it is implicitly assumed that the energy distribution is a boltzmann distribution , however , controlled by the effective temperature @xmath6 rather than the bath temperature @xmath34 .
the corresponding data are shown in fig .
[ fig : energy](c ) for different forces .
it can be observed that plotting the data according to eq .
( [ eq : defteff ] ) offers well defined straight lines whose slopes can be interpreted as the inverse effective temperatures .
this observation supports the idea that the energy distribution can be regarded at least formally as boltzmann distributed .
the definition for @xmath6 used above is similar to that in @xcite where it has been applied to characterize the non - equilibrium thermodynamics of a aging system and a system under an oscillatory shear strain .
however , in these references only the average value of the energy distribution has been used to determine @xmath6 .
while in this particular case both definitions would lead essentially to the same values , it has been found in another context @xcite , that the average value alone might be misleading , since it can result from very different distributions . using
the procedure depicted above would uncover such deviations by not showing straight lines in fig .
[ fig : energy](c ) . for later purposes ,
we rewrite eq .
( [ eq : pepst ] ) for @xmath31 according to the effective temperature defined in eq .
( [ eq : defteff ] ) : @xmath35 how is this effective temperature dependent on the external force ? since
the shift of the energy distribution must be symmetric concerning the direction of the applied force , it reasonable to assume that it is an even function in @xmath0 .
furthermore , in the limit of @xmath36 , @xmath6 must recover the bath temperature @xmath34 .
because of this , a natural guess for the force dependence of @xmath6 is @xmath37 where @xmath38 is a force and temperature independent parameter .
a similar guess has been justified by a mean - field theory of a brownian particle in a viscoelastic medium @xcite . in this reference , the prefactor has , however , been identified with the mobility of the brownian particle , which in the present case is strongly dependent on force and temperature .
thus , the relevance of the results from @xcite for microrheology of supercooled liquids is unclear .
equation [ eq : tefff ] is tested in fig .
[ fig : energy](d ) for the microrheologically driven systems at different temperatures .
it turns out that eq .
( [ eq : tefff ] ) allows to superimpose the thermodynamic data points for a very broad range of forces onto a master curve . only at high temperatures and large forces ,
small deviations occur . by fitting the master curve with a linear function ,
on can determine the value of @xmath38 for this system as @xmath39 .
an estimate for the value of @xmath38 based on properties of the underlying pel will be given at the end of the next subsection . instead of eq .
( [ eq : tefff ] ) , one could also think about a scaling like @xmath40 with @xmath41 denoting the inverse effective temperature and @xmath42 denoting the inverse bath temperature , respectively .
( [ eq : befff ] ) would fulfill the symmetry conditions discussed above as well and in the limit of @xmath36 , both eq .
( [ eq : tefff ] ) and eq .
( [ eq : befff ] ) are identical . however , eq . ( [ eq : befff ] )
offers a much poorer scaling for large forces , whereas eq .
( [ eq : tefff ] ) , as already discussed , allows a nearly perfect scaling for all forces and temperatures ( see fig .
[ fig : energy](d ) ) . * ( a ) * * ( b ) * * ( c ) * * ( d ) * the other central quantity is the average escape rate @xmath21 to leave a mb of energy @xmath20 . for a quiescent system at different temperatures , these local escape rates
are shown in fig .
[ fig : kinetics](a ) .
one finds that the temperature - dependence of the apparent escape rates is dependent on the energy of the mb .
this result is already known from previous simulations @xcite and can be understood by the fact that the escape from singe mb , although they occur via a multi - minima path , can be efficiently described as a single activated process : @xmath43 thereby both , the apparent activation energy @xmath44 , the system has to overcome , and the prefactor @xmath45 , which is of entropic nature and accounts for the number of possible paths to leave the minimum , are dependent on the energy of the particular mb . by fitting the escape rates of particular energies at different temperatures via eq .
( [ eq : get ] ) , one is able to determine @xmath44 and @xmath45 , see fig .
[ fig : kinetics](b ) . the resulting curve for @xmath44 is shown in fig . [ fig : kinetics](c ) .
one observes two distinct regimes : @xmath46 with the critical mb energy of the system @xmath47 and a slope @xmath48 .
since the dynamics of the system are merely determined by the slow escape rates , corresponding to large energy barriers , the following analysis will focus on the low - energy part of eq .
( [ eq : eapp ] ) . for the case
@xmath49 , the activation energy is in agreement with the trap model @xcite , where the mbs serve as traps : the system has to overcome a barrier equal to the depth of the trap plus a small additional barrier @xmath50 .
what is the meaning of the factor @xmath51 at low energies ?
although our simulations are conducted with a very small system , only a small set of particles is moving at the same time @xcite . to estimate the implications for @xmath44
one may consider that the simulated system consists of @xmath52 non - interacting subsystems , each one relaxing independently from each other .
the escape rates one measures is therefore the rate one observes for the overall system , i.e. sum of the rates of each individual subsystem @xmath53 : @xmath54 .
if one now assumes that the energy of the system is distributed equally among the @xmath52 subsystems , one immediately obtains @xmath55 , thus , the actual barrier , which has to be overcome , is reduced by a factor @xmath56 @xcite .
these ideas give rise to the notion of an `` extended trap model '' @xcite .
considering that the activation energy is proportional to the trap depth ( eq . ( [ eq : eapp ] ) ) , one can rewrite eq .
( [ eq : get ] ) as @xmath57 where the right hand side has been identified with the ratio of the energy distribution and the density of states , see eq .
( [ eq : pepst ] ) .
this connection of the escape dynamics on the one side and the thermodynamics on the other side is a fundamental property of the extended trap model .
the escape rates of a microrheologically driven system are shown in fig .
[ fig : kinetics](d ) for a broad range of forces .
one observes that the escape rates at low mb energies are strongly enhanced as compared to the quiescent case whereas at high mb energies the rates only slightly grow . by visual inspection on
finds that this behavior is rather similar to what one observes in fig .
[ fig : kinetics](a ) for a quiescent system at different temperatures .
this coincidence suggests that not only the energy distribution but also the escape kinetics are driven by an effective temperature .
this can further be justified in terms of the extended trap model : in eq .
( [ eq : getrap ] ) , a fundamental connection between the thermodynamics and kinetics is established which _ a priori _ holds for the quiescent system only . for the driven system ,
the right hand side of this equation can be replaced by eq .
( [ eq : pepsf ] ) , thus , the control parameter of the thermodynamics is the effective temperature @xmath6 .
( [ eq : getrap ] ) still holds for this system , it follows that also the kinetics are determined by @xmath6 defined in eq .
( [ eq : tefff ] ) , i.e. @xmath58 for this relation it is implicitly assumed that the factor @xmath51 of the extended trap model is not altered by the external force , thus , that the number of subsystems is the same as in the quiescent system .
this assumption can be justified by the observation that the average number of particles , involved in a mb transition , is basically @xmath0-independent ( see supplemental material @xcite for more details ) .
thus , the value of @xmath51 defined in eq .
( [ eq : eapp ] ) , describes the @xmath31 case as well .
equation ( [ eq : gef ] ) can be tested by inserting equations ( [ eq : tefff ] ) , ( [ eq : get ] ) and ( [ eq : eapp ] ) for different forces , as it is shown in fig .
[ fig : kinetics](d )
. one finds a very good agreement between the theoretical prediction and the numerical data .
our results show that both thermodynamics and kinetics are controlled by the same effective temperature @xmath6 which is dependent on a single parameter @xmath38 .
its microscopic origin can be qualitatively explored in terms of a one - dimensional cosine potential which is tilted along the force direction . as it is demonstrated in the supplemental material @xcite ,
one expects for this toy model that the escape rate out of a single minimum is for small forces given by @xmath59 with @xmath60 where @xmath61 denotes the barrier and @xmath62 the distance between two minima . for the binary lennard - jones mixture ,
a reasonable guess for the energy scale is the width @xmath63 of the dos or , equivalently , of the energy distribution @xmath19 .
the distance between to mb is given by the squared diffusive length @xmath64 of a single particle which can be determined from the mean - squared displacement @xmath65 of the particle after a number of mb transitions @xmath66 .
since both , @xmath63 and @xmath67 are independent from temperature @xcite , also the value of @xmath68 does not dependent on @xmath34 . with @xmath69 and @xmath70 , one can estimate @xmath71 which agrees with the actually measured value of @xmath39 up to a factor @xmath72 .
such deviations can be rationalized by considering that barriers between two mb do not lie exactly in the middle of two mb but are shifted closer to the starting minimum . in any event , already this simple model offers a reasonable estimate for the parameter @xmath38 , which is purely based on properties of the quiescent energy landscape . * ( a ) * * ( b ) * * ( c ) * * ( d ) * in the next step , one can revisit eq .
( [ eq : defavgamma ] ) to study the force - dependence of the average escape rate @xmath11 with provides the time scale of the mobility @xmath9 .
since both of its constituents are characterized exclusively by the effective temperature , also @xmath11 is a function of @xmath6 only .
this result is shown in fig .
[ fig : ftsmu](a ) where the average escape rate of a broad range of forces and temperature is shown as a function of their effective temperatures .
one finds that all data points can be collapsed onto a single master curve .
this property may be denoted as _ force - temperature superposition_. a subtlety emerges due to the fact that eq .
( [ eq : pepst ] ) has been used to compute the energy distribution for eq .
( [ eq : defavgamma ] ) , since this equation requires anharmonic effects to be fully negligible ( @xmath28 ) .
this difference is directly visible in fig .
[ fig : ftsmu](a )
. interestingly , one finds that the prediction via eqs .
( [ eq : defavgamma ] ) and ( [ eq : pepst ] ) offers a better description of the superimposed data points .
the observation that in case of microrheology the harmonic approximation holds for large effective temperatures suggest that the fluctuations inside a minimum are still controlled by the low bath temperature given by the thermostat .
this is is in agreement with the result for a sheared system in @xcite where it has been found that the short - time dynamics are determined by the bath temperature whereas the dynamics on long timescales are governed by a higher effective temperature . in fig .
[ fig : ftsmu](b ) the average escape rates are shown as a function of the force for different temperatures , together with the theoretical prediction based on the numerical integration of eq .
( [ eq : defavgamma ] ) .
one finds a quantitative agreement for all forces and temperatures .
we would like to emphasize that the data for a temperature @xmath73 ( full green circles in fig .
[ fig : ftsmu ] ) has been included which is below the mode - coupling temperature of the system ( @xmath74 @xcite ) .
for this temperature , the quiescent state is inaccessible due to very long relaxation times , thus , it can be regarded as below the ( computer ) glass transition . since eq .
( [ eq : defavgamma ] ) can still be evaluated , the present theory nevertheless allows a qualitative prediction of its highly non - equilibrium state .
the insights obtained above can eventually be used to predict the force - dependent mobility of the tracer particle in the supercooled liquid . for this purpose
one has to recall , that the mobility in terms of the pel approach given by @xmath75 where @xmath76 denotes the spatial part which depends on force and temperature .
however , as it has been shown in @xcite and the supplemental material @xcite , the spatial part displays only a weak dependence on the external force and is essentially given by @xmath77 where @xmath67 again denotes the diffusive length scale of a single particle in the quiescent system .
the one - dimensional diffusion coefficient @xmath4 of a single particle in quiescent system is given by @xcite @xmath78 thus , one can combine eqs .
( [ eq : muf1])-([eq : dt ] ) to obtain for the mobility : @xmath79 here , the force - dependent mobility is brought into relation with the temperature dependence of the equilibrium diffusion coefficient and can therefore be tested with real space quantities only .
the mobility @xmath9 as a function of the effective temperature is shown in fig .
[ fig : ftsmu](c ) .
there one can observe that again all data points fall onto a single master curve which is in a good agreement with the theoretical predictions via eq .
( [ eq : muf2 ] ) .
the deviations at rather low effective temperatures occur due to the fact , that the force - dependence of the spatial part in eq .
( [ eq : muf1 ] ) has not been considered in eq .
( [ eq : muf2 ] ) . at high effective temperatures ,
the theoretical prediction differs from the superimposed data point due to the breakdown of the harmonic approximation , similar to what already has been observed in fig .
[ fig : ftsmu](a ) for the average escape rate .
we would like to mention that the structure eq .
( [ eq : muf2 ] ) may seem to be a little bit odd , since two different temperatures explicitly enter : the effective temperature which determines the diffusion coefficient and the bath temperature in the denominator . however , this behavior is a direct result of the pel approach , since in eq .
( [ eq : xf ] ) , @xmath34 can not be replaced by @xmath6 . as a consequence , the application of @xmath80 for
the rescaling of the data results in a much poorer force - temperature superposition in the nonlinear regime , see fig .
[ fig : ftsmu](d ) . in summary , eq .
( [ eq : muf2 ] ) offers very reasonable description of the force - dependent mobility , based on real space properties of the quiescent system and a single parameter @xmath38 for a broad range of forces and temperatures . * ( a ) * * ( b ) * * ( c ) * * ( c ) * finally , one may ask whether the effective temperature also governs the diffusive properties of the system . for this purpose ,
we have determined the nonlinear diffusion coefficient @xmath8 of the tracer particle orthogonal to the force direction as well as @xmath81 of the type a bath particles in the same direction by fitting the long - time limit of the mean - squared displacements .
the orthogonal direction has been chosen since the tracer particle displays a strong anomalous behavior in direction of the force , including intermittent superdiffusivity as well as a transition to a strongly enhanced long - time diffusion .
this behavior has been intensively discussed in @xcite , where a quantitative theoretical description of the anomalous diffusion in terms of the pel approach is provided .
the fact that the diffusion coefficient of the bath particles displays a nonlinear behavior has of course to be treated with some care , since it is only valid for the rather small system studied in this work , where the bath particles and the tracer belong to the same elementary system .
nevertheless , the analysis of this quantity may be instructive for the question of the universality of the effective temperature in the elementary system . in analogy to the mobility in iii.c .
, the nonlinear diffusion coefficients can be separated into its spatial and temporal compounds @xcite : @xmath82 where @xmath83 or @xmath84 characterizes the different species .
this equation is basically the same as eq .
( [ eq : dt ] ) , however , both the diffusive length scales @xmath67 and the average hopping rate @xmath25 have been replaced by their force dependent counter parts .
the only difference between for @xmath8 and @xmath81 is the relevant diffusive length scale , either @xmath85 or @xmath86 , which can be determined from the mean - squared displacements of the respective particle types after a certain number of mb transitions . for the evaluation purely in real space , eq .
( [ eq : df ] ) is less helpful since the pel parameters are inaccessible .
similarly to what has been done for the mobility in the previous section , one one can therefore make the assumption that the diffusive length scales are basically independent of the external force ( and temperature , since @xmath87 ) .
if this assumption holds , eq .
( [ eq : df ] ) simplifies to @xmath88 for both diffusion coefficients .
the test of this assumption is shown in fig .
[ fig : ftsd](a),(b ) . for the diffusion coefficient of
the tracer particle one finds that the theoretical prediction is only in a qualitative agreement with the rescaled data points .
please notice , however , that in the analyzed temperature range the diffusion coefficient varies over three orders of magnitude , which is reasonably reflected by the theory .
for @xmath89 , a much better agreement can be observed : the data points collapse on a master curve which agrees up to the harmonic breakdown at @xmath90 with eq .
( [ eq : df2 ] ) .
the reason for the different qualities of the prediction is the validity of the assumption that the spatial part is not dependent on the applied force .
as one can see in fig .
[ fig : ftsd](c ) , the spatial part of the bath particles ( lower branch ) varies only very little as compared to the quiescent length scale .
in contrast , the length scale of the tracer particle grows half an order of magnitude ( upper branch ) . in the first case ,
the applied assumption is much better fulfilled , resulting in a better scaling in fig .
[ fig : ftsd](b ) . for the tracer particle diffusion ,
please consider , that the spatial part of the tracer diffusion varies _ at most _ half an order of magnitude .
in the same regime , the temporal part , which can be very well described by our model , extends over one to two orders of magnitude ( see fig .
[ fig : ftsmu](b ) ) .
this explains why the presented approach still allows a reasonable description of the data in fig .
[ fig : ftsd](a ) .
our results shown above are in some disagreement with the work presented in @xcite where it has been shown that the mobility and the orthogonal diffusion of the tracer particle obey a force - temperature superposition , however , with different effective temperatures .
adopted to our notation , these two temperatures differ in the value of @xmath38 which turns out to be a factor of @xmath91 larger for the orthogonal diffusion than for the nonlinear mobility .
indeed , also for our data a larger value of @xmath38 ( here increased by @xmath92 for a different system ) allows a nearly perfect scaling of the numerical data on a master curve , see fig .
[ fig : ftsd](d ) .
however , the effective temperature resulting from such a rescaling is somewhat artificial , since it has lost its connection to the thermodynamic and , in this application , kinetic properties it was originally derived from .
effective temperatures purely derived from scaling laws should therefore be treated with some caution , because its importance for the description of non - equilibrium thermodynamics is a priori unclear .
in this work , we have derived a microscopic approach to describe the thinning behavior of nonlinear active microrheology in supercooled liquids . for this purpose
we have applied the pel approach to describe the thermodynamics and dynamics of the driven system which offers valuable insight into the relationship of each other , even far away from equilibrium .
our results could show that the thermodynamic and dynamic properties of the system far from equilibrium are governed by a single effective temperature that can qualitatively connected to the properties to the quiescent energy landscape . since one can transfer this behavior , within some approximations however , to real space quantities like the force - dependent mobility and the orthogonal diffusion coefficient
, it may encourage tests of this results in experiments as well as in other theoretical models of supercooled liquids .
this work was financially supported by dfg research unit 1394 `` nonlinear response to probe vitrification '' and nrw graduate school of chemistry .
we furthermore acknowledge helpful discussions with e. bouchbinder , d. lacks , n. h. siboni and l. smith .
49ifxundefined [ 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 http://stacks.iop.org/0034-4885/68/i=3/a=r04 [ * * , ( ) ] link:\doibase 10.1039/b706004c [ * * , ( ) ] link:\doibase 10.1146/annurev - fluid-121108 - 145608 [ * * , ( ) ] http://stacks.iop.org/0953-8984/26/i=24/a=243101 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.74.1250 [ * * , ( ) ] http://stacks.iop.org/0295-5075/67/i=3/a=477 [ * * , ( ) ] link:\doibase 10.1021/jp8079028 [ * * , ( ) ] http://stacks.iop.org/0295-5075/93/i=5/a=58007 [ * * , ( ) ] link:\doibase 10.1103/physreve.74.011403 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.248302 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.028303 [ * * , ( ) ] link:\doibase 10.1103/physreve.87.012302 [ * * , ( ) ] link:\doibase 10.1063/1.4772627 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.110.067801 [ * * , ( ) ] link:\doibase 10.1103/physreve.91.032313 [ * * , ( ) ] link:\doibase 10.1140/epjst / e2013 - 02060 - 5 [ * * , ( ) ] link:\doibase 10.1103/physreve.87.032304 [ * * , ( ) ] link:\doibase 10.1063/1.1960607 [ * * , ( ) ] \doibase http://dx.doi.org/10.1122/1.2085174 [ * * , ( ) ] link:\doibase 10.1017/s0022112006009608 [ * * , ( ) ] link:\doibase 10.1017/s0022112010001606 [ * * , ( ) ] link:\doibase 10.1063/1.4770335 [ * * , ( ) ] link:\doibase 10.1103/physreve.55.3898 [ * * , ( ) ] http://stacks.iop.org/0295-5075/49/i=5/a=590 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.86.107 [ * * , ( ) ] link:\doibase 10.1063/1.1460862 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.89.095703 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.93.255501 [ * * , ( ) ] http://stacks.iop.org/0295-5075/79/i=2/a=26001 [ * * , ( ) ] link:\doibase 10.1063/1.1672587 [ * * , ( ) ] link:\doibase 10.1126/science.267.5206.1935 [ * * , ( ) ] link:\doibase 10.1103/physreve.67.031506 [ * * , ( ) ] http://stacks.iop.org/0305-4470/29/i=14/a=012 [ * * , ( ) ] link:\doibase 10.1103/physreve.72.021503 [ * * , ( ) ] link:\doibase 10.1103/physreve.67.030501 [ * * , ( ) ] link:\doibase 10.1103/physreve.78.011504 [ * * , ( ) ] link:\doibase 10.1103/physreve.51.4626 [ * * , ( ) ] link:\doibase 10.1103/physreve.60.6507 [ * * , ( ) ] http://stacks.iop.org/0953-8984/15/i=11/a=309 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.117801 [ * * , ( ) ] link:\doibase 10.1080/00268978400101201 [ * * , ( ) ] link:\doibase 10.1103/physreva.31.1695 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.463940 [ * * , ( ) ] @noop http://stacks.iop.org/0953-8984/20/i=37/a=373101 [ * * , ( ) ] http://stacks.iop.org/0953-8984/12/i=29/a=325 [ * * , ( ) ] link:\doibase 10.1103/physreve.83.031505 [ * * , ( ) ] link:\doibase 10.1021/jp204459b [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1644538 [ * * , ( ) ] | in active microrheology the mechanical properties of a material are tested by adding probe particles which are pulled by an external force . in case of supercooled liquids ,
strong forcing leads to a thinning of the host material which becomes more pronounced as the system approaches the glass transition . in this work we provide a quantitative theoretical description of this thinning behavior based on the properties of the potential energy landscape ( pel ) of a model glass - former .
a key role plays the trap - like nature of the pel .
we find that the mechanical properties in the strongly driven system behave the same as in a quiescent system at an enhanced temperature , giving rise to a well - characterized effective temperature .
furthermore , this effective temperature turns out to be independent of the chosen observable and individually shows up in the thermodynamic and dynamic properties of the system .
based on this underlying theoretical understanding , we can estimate its dependence on temperature and force by the pel - properties of the quiescent system .
we furthermore critically discuss the relevance of effective temperatures obtained by scaling relations for the description of out - of - equilibrium situations . *
( a ) * * ( b ) * |
in comparison with a central region of very small rapidities , @xmath2 , the forward rapidity region allows to study processes corresponding to much higher initial energies accessible at mid rapidities .
if a particle with mass @xmath3 and transverse momentum @xmath1 is produced in a hard reaction then the corresponding values of bjorken variable in the beam and the target are @xmath4 .
thus , at forward rapidities the target @xmath5 is @xmath6- times smaller than at mid rapidities .
this allows to study already at rhic coherence phenomena ( shadowing , color glass condensate ( cgc ) ) , which are expected to suppress particle yields .
forward rapidity physics , manifested itself as a strong nuclear suppression , has been already investigated in variety of processes at different energies : in production of different species of particles in @xmath7 collisions @xcite , in charge pion @xcite and charmonium production @xcite at sps , in the drell - yan process and charmonium production at fermilab @xcite and later on at larger rhic energies by measurements of high-@xmath1 particles in @xmath8 collisions @xcite .
althought forward rapidity region at rhic allows to investigate small-@xmath9 coherence phenomena , one should be carreful with interpretation of observed suppression .
such a suppression is arisen globally for any reaction studied so far at any energy .
namely , all fixed target experiments have too low energy for the onset of coherence effects since @xmath5 is not small .
the rise of suppression with @xmath10 ( with feynman @xmath0 ) shows the same pattern as observed at rhic .
this universality of suppression favors also another mechanism which should be common for all reactions studied at any energy .
such a mechanism was proposed in @xcite and allows to describe a strong suppression via energy conservation effects in initial state parton rescatterings .
it can be also interpreted alternatively as a parton effective energy loss proportional to initial energy leading so to @xmath0 scaling of nuclear effects .
the projectile hadron can be decomposed over different fock states .
a nucleus has a higher resolution than a proton due to multiple interactions and so can resolve higher fock components containing more constituents .
corresponding parton distributions fall off steeper at @xmath11 where any hard reaction can be treated as a large rapidity gap ( lrg ) process where no particle is produced within rapidity interval @xmath12 .
the suppression factor as a survival probability for lrg was estimated in @xcite , @xmath13 .
each of multiple interactions of projectile partons produces an extra @xmath14 and the weight factors are given by the agk cutting rules @xcite .
then the effective parton distribution correlates with the nuclear target @xcite , f^(a)_q / n(x , q^2,b ) = cf_q / n(x , q^2 ) exp , [ 10 ] where @xmath15 is the nuclear thickness function defined at nuclear impact parameter @xmath16 , @xmath17 mb @xcite and the normalization factor @xmath18 is fixed by the gottfried sum rule . in this paper
we study a suppression of the drell - yan ( dy ) process on a nucleus with respect to a nucleon target and the rise of this suppression with @xmath10 ( @xmath19 , @xmath0 ) in various kinematic regions .
first we compare our predictions with data from the fixed target e772 experiment at fnal @xcite .
then similar nuclear effects are predicted also for the rhic forward region expecting the same suppression pattern as seen at fnal .
finally we perform for the first time predictions in the kinematic range corresponding to a new e906 experiment planned at fnal where no coherence effects are expected .
the dy process in the target rest frame can be treated as radiation of a heavy photon / dilepton by a projectile quark .
the transverse momentum @xmath1 distribution of photon bremsstrahlung in quark - nucleon interactions , @xmath20 , reads @xcite : & & = _ in , f d^2 r_1d^2 r_2 e^i_t(_1 - _ 2 ) _ ^ * q^*(,_1 ) _ ^ * q(,_2 ) ( , r_1,r_2 ) [ 20 ] where @xmath21 , @xmath22 and the light - cone ( lc ) wave functions of the projectile @xmath23 fluctuation @xmath24 are presented in @xcite .
feynman variable is given as @xmath25 and in the target rest frame @xmath26 .
for the dipole cross section @xmath27 in eq .
( [ 20 ] ) we used gbw @xcite and kst @xcite parametrizations .
the hadron cross section is given convolving the parton cross section , eq .
( [ 20 ] ) , with the corresponding parton distribution functions ( pdfs ) @xmath28 and @xmath29 @xcite , = _
x_1 ^ 1 _ q z_q^2 \ { f_q(,q^2 ) + f_|q(,q^2 ) } , [ 30 ] where @xmath30 is the fractional quark charge , pdfs @xmath31 and @xmath32 are used with the lowest order ( lo ) parametrization from @xcite at the scale @xmath33 and the factor @xmath34 accounts for decay of the photon into a dilepton .
the rest frame of the nucleus is very convenient for study of coherence effects .
the dynamics of the dy process is controlled by the coherence length , l_c = = , [ 40 ] where @xmath35 and @xmath36 is the energy and mass of the projectile quark and the center of mass energy squared @xmath37 .
the fraction of the proton momentum @xmath38 carried by the quark is related to @xmath19 as @xmath39 .
+ the coherence length is related to the longitudinal momentum transfer , @xmath40 , which controls the interference between amplitudes of the hard reaction occurring on different nucleons .
the condition for the onset of shadowing in a hard reaction is sufficiently * long coherence length * ( lcl ) in comparison with the nuclear radius , @xmath41 , here the special advantage of the color dipole approach allows to incorporate nuclear shadowing effects via a simple eikonalization of @xmath42 @xcite , i.e. replacing @xmath42 in eq .
( [ 20 ] ) by @xmath43 : _ |qq^a = 2 d^2 b\{1 - ^a } . [ 50 ] the corresponding predictions for nuclear broadening in dy reaction based on the theory @xcite for lcl limit were presented in @xcite .
+ in the * short coherence length * ( scl ) regime the coherence length is shorter than the mean internucleon spacing , @xmath44fm . in this limit
there is no shadowing due to very short duration of the @xmath45 fluctuation .
the corresponding theory for description of the quark transverse momentum broadening can be found in @xcite . in this regime the transverse momentum distribution for an incident proton can be obtained integrating over @xmath46 similarly as in eq .
( [ 30 ] ) : = _
x_1 ^ 1 _ q z_q^2 \{f_q(,q^2 ) + f_|q(,q^2 ) } ^qa(,p_t ) , [ 250 ] where @xmath47 represents the cross section for an incident quark to produce a photon on a nucleus @xmath48 with transverse momentum @xmath1 .
this cross section can be expressed convolving the probability function @xmath49 with the cross section @xmath50 ( see eq .
( [ 20 ] ) ) , ^qa(,p_t ) = d^2 k_t w^qa(_t , x_q ) ^qn ( , _ t ) , [ 240 ] where @xmath51 .
probability distribution in eq .
( [ 240 ] ) that a quark will acquire transverse momentum @xmath52 on the nucleus , @xmath49 , is obtained by the averaging procedure over the nuclear density @xmath53 : w^qa(_t , x_q ) = d^2 b dz _ a(b , z ) w_a^q(_t , x_q,,z ) , [ 210 ] where @xmath54 means now the partial probability distribution that a valence quark arriving at the position @xmath55 in the nucleus @xmath48 will have acquired transverse momentum @xmath52 .
it can be written in term of the quark density matrix , @xmath56 , w_a^q(_t , x_q,,z ) = d^2 r_1 d^2 r_2e^i_t(_1 - _ 2 ) _
q(_1,_2 ) e^- _ |qq ( x_q,_1 - _ 2 ) t_a ( + , z ) , [ 180 ] where @xmath57 with @xmath58@xmath59 representing the mean - square charge radius of the proton . @xmath60 in eq .
( [ 180 ] ) is the partial nuclear thickness function , @xmath61 .
nuclear effects in @xmath7 collisions are usually investigated via the so called nuclear modification factor , defined as @xmath62 , where the numerator is calculated in scl and lcl regimes as described above .
corrections for the finite coherence length was realized by linear interpolation using nuclear longitudinal formfactor @xcite ( for more sophisticate green function method see @xcite ) .
note that at rhic energy and at forward rapidities the eikonal formula for lcl regime , eqs .
( [ 30 ] ) and ( [ 50 ] ) , is not exact since higher fock components containing gluons lead to additional corrections , called gluon shadowing ( gs ) .
the corresponding suppression factor @xmath63 was derived in @xcite and included in calculations replacing in eq .
( [ 50 ] ) @xmath64 by @xmath65 .
gs leads to reduction of the cronin effect @xcite at medium - high @xmath1 and to additional suppression ( see fig .
[ rhic ] ) . in the fixed target fnal energy range , for elimination of the coherence effects one can study production of dileptons at large @xmath3 ( see eq .
( [ 40 ] ) ) as has been realized by the e772 collaboration @xcite .
another possibility is to study the dy process at large @xmath66 , when also @xmath67 , and @xmath68 in this limit ( see eq .
( [ 40 ] ) )
. differential cross section of dileptons in @xmath69 collisions at @xmath70 and @xmath71gev vs. e866 data @xcite . ]
we start with the dy process in @xmath69 collisions . besides calculations based on eq .
( [ 30 ] ) using grv98 pdfs @xcite ( see the solid line in fig . [ e866-pp ] ) we present by the dashed and dotted line also predictions using proton structure functions from @xcite and cteq6l parametrization of pdfs from @xcite , respectively .
[ e866-pp ] shows a reasonable agreement of the model with data from the e866/nusea collaboration @xcite .
this encourages us to apply the color dipole approach to nuclear targets as well . of drell - yan cross sections on w and d vs. e772 data for @xmath72gev ( left ) and @xmath73gev ( right ) .
predictions correspond to scl ( dotted curves ) , lcl ( dashed curves ) regimes and their interpolation ( solid curves ) .
thick and thin curves are calculated with and without corrections ( [ 10 ] ) for energy conservation , respectively . ]
the e772 collaboration @xcite found a significant suppression of dy pairs at large @xmath19 ( see fig .
[ e772 ] ) .
large invariant masses of the photon allows to minimize shadowing effects ( see a small differences between dotted and solid lines in fig .
[ e772 ] ) . if effects of energy conservation , eq . ( [ 10 ] ) , are not included one can not describe a strong suppression at large @xmath19 . in the opposite case a reasonable agreement of our model with data is achieved .
one can approach the kinematic limit increasing @xmath1 .
therefore we present also predictions for @xmath1 dependence of the nuclear modification factor @xmath74 at rhic energy and at several fixed values of @xmath0 .
similarly as in @xcite instead of usual cronin enhancement , a suppression is found ( see fig . [ rhic ] ) .
the onset of isotopic effects ( ie ) in @xmath8 collisions at large @xmath1 gives the values @xmath75 depending on @xmath0 . in @xmath76 collisions the corresponding ratio @xmath77 from above and no nuclear effects are assumed at large @xmath1 expecting so qcd factorization .
however , we predict a strong onset of effective energy loss effects , eq . (
[ 10 ] ) , at large @xmath0 ( see fig . [ rhic ] ) quantifying itself as a large deviation of suppression from the above values @xmath78 .
the predicted huge rise of suppression with @xmath0 in fig .
[ rhic ] reflects much smaller survival probability @xmath79 at larger @xmath0 and can be tested in the future by the new data from rhic .
note that effects of gs depicted in fig .
[ rhic ] by the thick lines lead to additional suppression which rises with @xmath0 . at @xmath80gev for several fixed values of @xmath0 without effects of effective energy loss .
( right ) the same as ( left ) but with effects of effective energy loss , eq .
( [ 10 ] ) , ( thin lines ) .
thick lines additionally include gs effects . ] of drell - yan cross sections on w and d for @xmath72gev ( left ) and @xmath73gev ( right ) realized for the kinematic range of the planned e906 experiment at fermilab .
solid and dashed curves are calculated with and without effects of effective energy loss , eq .
( [ 10 ] ) , respectively . ]
finally we present in fig .
[ e906 ] for the first time predictions for @xmath19 dependence of the nucleus - to - nucleon ratio in the kinematic range corresponding to a new e906 experiment planned at fermilab .
we shoud not expect any shadowing effects since initial energy is small , @xmath81gev and a strong nuclear suppression at large @xmath19 is caused predominantly by the energy conservation constraints , eq .
( [ 10 ] ) .
we demonstrate that besides an onset of coherence a nuclear suppression at forward rapidities ( large @xmath19 , @xmath0 ) can be induced also by energy conservation effects in multiple parton rescatterings interpreted alternatively as a parton effective energy loss proportional to initial energy .
universality of this treatment is in its applicability to any reaction studied at any energy also in the kinematic regions where coherence phenomena ( shadowing , cgc ) can not be manifested .
first we apply this approach to the dy process and explain well a significant suppression at large @xmath19 in accordance with the e772 data .
the fnal energy range and large invariant masses of the photon allow to minimize the effects of coherence , what does not leave much room for other mechanisms , such as cgc .
then we predict a significant suppression also for @xmath8 collisions at rhic in the forward region ( see fig .
[ rhic ] ) . at small @xmath1
we show an importance of gs effects and their rise with @xmath0 .
finally we present for the first time predictions for strong nuclear effects expected in a new e906 experiment planned at fnal .
much smaller beam energy than in e772 experiment allows to exclude safely interpretations based on coherence phenomena .
this work was supported in part by the slovak funding agency , grant 2/0092/10 and by grants vz mmt 6840770039 and lc 07048 ( ministry of education of the czech republic ) .
9 barton d s , et al .
1983 _ phys .
d * 27 * 2580 ; geist w m 1991 _ nucl .
phys . _ a * 525 * 149c ; beretvas a , et al .
1986 _ ibid . _ * 34 * 53 ; binkley m , et al .
1976 _ phys .
lett . _ * 37 * 571 ; bailey r , et al .
1984 _ z. phys .
_ c * 22 * 125 ; skubic p , et al . 1978 _ phys .
_ d * 18 * 3115 | data from e772 and e866 experiments on the drell - yan process exhibit a significant nuclear suppression at large feynman @xmath0 .
we show that a corresponding kinematic region does not allow to interpret this as a manifestation of coherence or a color glass condensate .
we demonstrate , however , that this suppression can be treated alternatively as an effective energy loss proportional to initial energy . to eliminate suppression coming from the coherence , we perform predictions for nuclear effects also at large dilepton masses .
our calculations are in a good agreement with available data . since the kinematic limit can be also approached in transverse momenta @xmath1 , we present in the rhic energy range corresponding predictions for expected large-@xmath1 suppression as well .
since a new experiment e906 planned at fnal will provide us with more precise data soon , we present also predictions for expected large-@xmath0 nuclear suppression in this kinematic region . |
in @xcite , we proved the riemmanian penrose inequality with charge for multiple black holes . [ crpi ]
let @xmath0 be a strongly asymptotically flat initial data set for the einstein - maxwell equations with outermost minimal surface boundary of area @xmath1 , with adm mass @xmath2 , and total charge @xmath3 , satisfying the charged dominant energy condition and the maxwell constraints without charged matter . then @xmath4 with equality if and only if the data set arises as the canonical slice of the reissner - nordstrm spacetime .
here @xmath5 is a three dimensional manifold , @xmath6 a riemannian metric on @xmath5 , @xmath7 and @xmath8 vector fields on @xmath5 , and @xmath9 is called the _ area radius _ of the outermost minimal surface . when @xmath10 , the positive mass theorem with charge @xcite , @xmath11 , which was accomplished with the following result .
[ thm - imcf ] let @xmath0 be a strongly asymptotically flat initial data set for the einstein - maxwell equations with outermost minimal surface boundary of area @xmath1 , with adm mass @xmath2 , and total charge @xmath3 , satisfying the charged dominant energy condition and the maxwell constraints without charged matter .
if @xmath12 , then @xmath13 with equality if and only if the data set arises as the canonical slice of the reissner - nordstrm spacetime . we proved this theorem using a conformal flow method adapted from @xcite . in @xcite ,
a counterexample to based on the majumdar - papapetrou solutions was constructed when @xmath14 , showing that the hypothesis @xmath15 is necessary in theorem [ thm - imcf ] . in this paper
we examine whether the hypothesis @xmath16 , i.e. the absence of charges outside the horizon , is necessary .
we will prove the following two theorems ; a positive and a negative result .
[ crpi - charges ] let @xmath0 be a strongly asymptotically flat initial data set for the einstein - maxwell equations with outermost minimal surface boundary of area @xmath1 , with adm mass @xmath2 , and total charge @xmath3 , satisfying the charged dominant energy condition . if @xmath17 and @xmath18 are compactly supported , then holds .
[ counterexample ] there is a spherically symmetric counterexample to with @xmath19 , satisfying all the conditions of theorem [ crpi - charges ] except that @xmath17 is not compactly supported .
the existence of a spherically symmetric counterexample was conjectured in @xcite .
note that in theorem [ crpi - charges ] the rigidity statement has been omitted .
in fact , we show that a counterexample to rigidity exists .
[ norigidity ] there is a non - trivial arbitrarily small spherically symmetric perturbation of the reissner - nordstrmcanonical slice @xmath20 , which coincides with reissner - nordstrm outside of an annulus , satisfies all the hypothesis of theorem [ crpi - charges ] in addition to @xmath15 , and saturates inequality : @xmath21 we also examine the question of whether the conformal flow technique developed in @xcite can be extended to other matter models .
it is expected that for a riemannian penrose inequality to hold , the matter model should admit a unique stable static black hole solution . at the moment , we are only aware of two such models , the einstein - abelian - yang - mills ( eaym ) , and the einstein - maxwell - dilaton ( emd ) .
we prove a riemannian penrose inequality for eaym black holes , and conjecture that a riemannian penrose inequality for emd black holes also holds .
[ abelian - ym ] let @xmath22 be a strongly asymptotically flat initial data set for the eaym equations with an outermost minimal surface boundary of area @xmath23 , with adm mass @xmath2 , and total charge @xmath3 , satisfying the eaym dominant energy condition .
if @xmath24 and @xmath25 are compactly supported , then holds . if there are no charges outside the horizon then equality holds if and only if the data set arises as the canonical slice of an eaym reissner - nordstrm spacetime .
[ dilaton ] let @xmath26 be a strongly asymptotically flat initial data set for the emd equations with an outermost minimal surface boundary of area @xmath1 , adm mass @xmath2 , and total charge @xmath3 , satisfying the emd dominant energy condition .
if the charge densities @xmath27 and @xmath18 are compactly supported , then @xmath28 if @xmath29 and @xmath8 are divergence free , then equality holds if and only if the data set arises as the canonical slice of the gibbons emd black hole .
the plan of the paper is as follows .
in the next section , we briefly present some of the background , heuristics , and prior work on the riemannian penrose inequality without and with charge . in that section
we also discuss the dominant energy conditions for eaym data sets and for emd data sets , as well as the emd black hole . in section [ charges ] , we prove theorem [ crpi - charges ] , theorem [ counterexample ] and proposition [ norigidity ] . in section [ matter ] , we prove theorem [ abelian - ym ] and discuss conjecture [ dilaton ] .
the penrose inequality was originally proposed by penrose as a test of the cosmic censorship conjecture , or more generally a test of the standard picture of gravitational collapse @xcite .
consider a strongly asymptotically flat cauchy surface in a spacetime satisfying the dominant energy condition , with adm mass @xmath30 , and containing an event horizon of area @xmath31 , which undergoes gravitational collapse and settles to a kerr - newman black hole . since the adm mass @xmath2 of the final state is no greater than @xmath30 , and since the end state area radius @xmath9 is no less than @xmath32 by hawking s area theorem @xcite , and since for the final state @xmath33 in order to avoid naked singularities , it must have been the case that @xmath34 also at the beginning of the evolution .
the event horizon is indiscernible in the original slice without knowing the full evolution .
however , one may replace the event horizon by the outermost minimal area enclosure of the apparent horizon , the boundary of the region admitting trapped surfaces , and reach the same conclusion .
a counterexample to the penrose inequality would therefore have suggested cauchy data which leads under the einstein evolution to naked singularities , while a proof of the inequality could be viewed as evidence in support of cosmic censorship .
it is usually assumed that the end state of gravitational collapse is a single kerr - newman black hole . however , a more appropriate assumption for the end state is a finite number of mutually distant kerr - newman black holes moving apart with asymptotically constant velocity .
this should be the result , if for instance , two distant black holes were initially moving away from each other sufficiently fast .
we will now describe the heuristic penrose argument for the charged penrose inequality in this setting .
it appears that this has not been previously considered in the literature .
let @xmath35 , @xmath36 , @xmath37 , @xmath38 denote the adm masses , horizon areas , total charges , and angular momenta of the end state black holes .
then the total ( adm ) mass , horizon area radius , and charge of the end state is @xmath39 , @xmath40 , @xmath41 .
the area radius of the kerr - newman black hole @xcite is given by @xmath42 it follows that @xmath43 let @xmath44 , @xmath45 , @xmath46 denote the adm mass , horizon area radius , and total charge of an initial state . under the assumption that no charged matter is present ,
the total charge is conserved @xmath47 . moreover , by the hawking area theorem @xmath48 , and since gravitational waves may only carry away positive energy @xmath49 .
let @xmath50 , @xmath51 be real numbers , then @xmath52 this follows directly from the triangle inequality for @xmath53 points in @xmath54 .
now let @xmath55 and @xmath56 , then we have @xmath57 or equivalently @xmath58 thus we conclude : @xmath59 the penrose inequality further simplifies in the time - symmetric case , where the apparent horizon coincides with the outermost minimal surface .
moreover , the dominant energy condition reduces now to non - negative scalar curvature of the cauchy hypersurface , leading to the riemannian version of the inequality .
[ rpi ] let @xmath60 be a three dimensional strongly asymptotically flat riemannian manifold of non - negative scalar curvature , with an outermost minimal surface boundary @xmath61 of area @xmath62 , then @xmath63 with equality if and only if the manifold is a canonical slice of the schwarzschild spacetime . the first published proof of this theorem was given by huisken - ilmanen in @xcite using inverse mean curvature flow , but required the assumption that @xmath61 be connected .
another proof by bray in @xcite used a conformal flow and applied more generally to a non - connected boundary @xmath61 . nonetheless , one advantage of the first proof was that it could be immediately generalized to obtain the einstein - maxwell case @xcite , i.e. to prove theorem [ thm - imcf ] in the case of a connected @xmath61 .
it is important to point out that implies both and a lower bound : @xmath64 in fact this lower bound follows , when @xmath61 is connected , from the stability of the outermost horizon @xcite .
indeed , the absence of charges outside the horizon , the cauchy - schwartz inequality , the charged dominant energy condition , the stability of the outermost horizon , and the gauss - bonnet theorem , imply @xmath65 \\
\leq \int_{\sigma } \left ( |e|^2 + |b|^2 \right ) da \leq \int_\sigma \frac{1}{2}r\,da
\leq \int_\sigma k = 4\pi n , \end{gathered}\ ] ] where @xmath66 is the scalar curvature , and @xmath67 , @xmath68 , and @xmath53 are respectively the number of components , the gauss curvature , and the unit normal of @xmath61 . in particular one obtains the area - charge inequality @xmath12 , or more generally @xmath69 this inequality was obtained in @xcite . if @xmath70 , the lower bound follows immediately : @xmath71 we note that the inequality implies that if @xmath72 , then there must be at least @xmath67 black holes , demonstrating that inequalities between these geometric quantities have topological consequences .
all these results are summarized in figure [ graphical ] .
the gauge group for the eaym matter model is @xmath73 .
the connection , and the field strength @xmath74 , have values in the abelian lie algebra @xmath75 . as usual
, the electric and magnetic fields are given by @xmath76 , @xmath77 , where @xmath78 is the future directed timelike unit normal to the initial data slice , and @xmath79 is the hodge dual .
the eaym equations of motions are @xmath80 where barred quantities refer to spacetime objects , and greek indices run @xmath81 . from the expression for the energy - momentum - stress tensor
, we obtain the energy density of the slice after contributions from the yang - mills fields have been removed : @xmath82 where @xmath83 is the second fundamental form of the slice . in the time - symmetric case , @xmath84 , and the dominant energy condition
is @xmath85 we note that the dominant energy condition is usually stronger , namely @xmath86 , while here we only use the weaker condition @xmath87 .
the total charges are defined by @xmath88 the emd action has the lagrangian @xmath89 there are other emd models where the coupling @xmath90 is replaced by @xmath91 , in which @xmath92 is a coupling constant . while it is possible to conjecture a riemannian penrose inequality also for @xmath93
, we leave this for future work . from the lagrangian ,
one obtains the emd equations of motion @xmath94 as above @xmath95 in the time - symmetric case , @xmath84 and @xmath96 , thus from @xmath97 we get the emd dominant energy condition @xmath98 the charges contained in a surface @xmath99 are given by @xmath100 we note that if @xmath101 these charges depend only on the homology class of @xmath99 .
the total charges are obtained by taking @xmath102 .
a static spherically symmetric emd black hole was discovered by gibbons @xcite : @xmath103 where @xmath104 is the round metric on the unit 2-sphere .
it was proven to be the unique static solution in @xcite .
we note that the scalar curvature of this emd black hole is given by @xmath105 and hence is an equality in this case . for more details on this solution , see @xcite .
the proof follows directly from @xcite , after noting that the only place which requires @xmath7 and @xmath8 to be divergence free is when one applies the inverse mean curvature flow at the end of the proof . according to the exhaustion
result the flowing surfaces @xmath106 eventually become connected and enclose any large coordinate sphere .
thus , if @xmath7 and @xmath8 are divergence free outside of a large coordinate sphere @xmath107 , then we may apply the inverse mean curvature flow argument once @xmath106 encloses @xmath107 .
let the spherically symmetric metric be given by @xmath108 it is easy to check that @xmath109 is the hawking mass of the coordinate sphere @xmath110 .
furthermore , a horizon occurs at @xmath111 if @xmath112 .
if @xmath109 is increasing there is only one horizon , hence @xmath113 is the outermost horizon .
we will use the charged hawking mass @xcite @xmath114 where @xmath3 is the total charge . note that on the horizon , we have @xmath115 and at infinity @xmath116 tends to the adm mass @xmath117 . clearly , if @xmath118 for @xmath119 , then will be violated
. a violation of with @xmath120 implies a violation of .
thus , in order to construct a counterexample it suffices to find two functions @xmath109 and @xmath121 , where @xmath122 , satisfying the following conditions : a. the charged dominant energy condition @xmath123 b. total charge is @xmath3 : @xmath124 as @xmath125 . c. the condition @xmath126 .
d. asymptotic flatness : @xmath127 as @xmath125 .
e. the inequality @xmath128 choose @xmath129 .
we begin by solving @xmath130 so that @xmath131 and is satisfied for @xmath119 .
one finds @xmath132 clearly asymptotic flatness is satisfied , and in fact @xmath133 furthermore @xmath134 , and consequently @xmath135 , for @xmath119 . finally observe that @xmath136 it follows that we can find a function @xmath121 satisfying @xmath137 , the charged dominant energy condition @xmath138 for @xmath119 , and @xmath139 as @xmath125 .
let @xmath140 denote the canonical slice of the reissner - nordstrm spacetime , and let @xmath141 denote the anisotropic radial coordinate for this data .
consider the annulus @xmath142 where @xmath143 , and fix @xmath144 to be nonnegative and not identically zero . since the scalar curvature @xmath145 is nonconstant in @xmath142 , the formal @xmath146-adjoint of the linearized scalar curvature operator has trivial kernel .
it then follows from @xcite that there exists a smooth contravariant 2-tensor @xmath147 , such that the scalar curvature of the metric @xmath148 is given by @xmath149 for @xmath150 sufficiently small .
it is also clear from the proof in @xcite that if @xmath151 is chosen to be spherically symmetric , then @xmath152 is also spherically symmetric .
define @xmath153 , and observe that @xmath154 since @xmath155 , the implicit function theorem may be used to find @xmath156 such that @xmath157 .
therefore @xmath158 so that the initial data set @xmath159 satisfies the charged dominant energy condition .
moreover , it is clear that this initial data set satisfies all desired hypotheses , except possibly the outermost condition for the minimal surface boundary .
however , it can easily be seen that the minimal boundary is outermost by choosing @xmath160 sufficiently small .
namely , by choosing @xmath160 sufficiently small , the mean curvature of each coordinate sphere @xmath107 , @xmath161 , remains positive .
thus , the exterior region is foliated by surfaces of positive mean curvature , showing that no other minimal surface enclosing the boundary can exist ( by the maximum principle for minimal surfaces ) .
apply a rotation @xmath162 so that all the transformed charges @xmath163 and @xmath164 vanish except possibly @xmath165 , and @xmath166 .
if @xmath167 then the positive mass theorem with charge applies , and we get as explained in the introduction . otherwise , starting with this new data , one may apply the proof in @xcite , conformally deforming all the electromagnetic fields in the same manner @xmath168 , @xmath169 .
both the charged conformal flow and the inverse mean curvature flow arguments now proceed as in theorem [ crpi - charges ] , and we obtain . if there are no charges and equality holds , we obtain as in @xcite that the transformed data is a canonical slice of reissner - nordstrm . after rotating back , we find that the original data is a canonical slice of an eaym reissner - nordstrm black hole . an initial data set for the emd equations consists of @xmath170 , where @xmath171 is the initial time derivative of @xmath172 . in the time - symmetric case @xmath83 and
@xmath171 vanish .
we assume the data is strongly asymptotically flat , satisfies the emd dominant energy condition , and that charges are absent , @xmath173 outside the horizon , or more generally outside a compact set .
our first observation is that a penrose heuristic argument , as in section [ penrose ] , is still valid .
start from data with paramaters @xmath30 , @xmath32 , @xmath174 , and assume the data settles to an emd black hole as in with parameters @xmath2 , @xmath9 , and @xmath3 .
then , the charge is conserved @xmath175 , the hawking area theorem applies giving @xmath176 , and as before the masses satisfy @xmath177 .
since for the emd black hole we have @xmath178 it follows that @xmath179 and hence holds for the initial data .
we now point out a number of significant differences between this setting and our work in @xcite .
first , there are analogs of the majumdar - papapetrou multiple neck solutions @xcite , but they do not pose an obstruction to the lower bound . indeed , the right hand side of is always monotone in @xmath9 unlike the right hand side of , hence we do not expect an auxiliary inequality , such as the area / charge inequality in theorem [ thm - imcf ] , to be necessary to prove .
these extremal solutions have , in a conformally related `` string '' metric , asymptotically cylindrical infinitely long necks just as in majumdar - papapetrou , but in the spacetime metric these have cross sectional area tending to zero at a finite distance
. since we have @xmath180 in this case , with @xmath181 , there are emd black holes with nearby parameters that have a single connected component horizon , and charge to area radius ratio arbitrarily small .
thus the phenomenon described in section [ section - rpi ] whereby excess charge leads to high multiplicity of horizon components is absent in the emd model .
the mathematical reason is that in the emd case , to replace the integral over @xmath99 with the integral over @xmath182 in inequality , would require @xmath183 and @xmath184 to be divergence free rather than , as is assumed , @xmath29 and @xmath8 .
notwithstanding these observations , the many similarities we now outline , lead us to surmise that the same approach used in @xcite should apply in the emd setting .
first , a positive mass theorem with charge holds @xcite .
[ dilaton - pmt ] let @xmath170 be a strongly asymptotically flat initial data set for the emd equations with an apparent horizon boundary , satisfying the emd dominant energy condition , with adm mass @xmath2 and total charge @xmath3 .
if charge densities @xmath27 and @xmath18 vanish , then @xmath185 observe that the initial data is not assumed to be time - symmetric . on the other hand ,
the rigidity statement has not yet been proved .
we suspect that in analogy with the em case , equality holds if and only if the data can be embedded as a cauchy slice in one of the ghs static extremal emd solutions from @xcite , and if the data is time - symmetric equality holds if and only if that slice is the cannonical slice .
that said , contrary to @xcite , theorem [ dilaton - pmt ] likely plays no role in the proof of .
instead , we believe that an emd charged hawking mass @xmath186 exists as in @xcite satisfying the following properties : ( i ) @xmath186 is monotonically non - decreasing under the huisken - ilmanen weak inverse mean curvature flow , and it is constant only for the central spheres in the spherically symmetric emd black hole ; ( ii ) @xmath186 converges to @xmath187 on the outermost horizon , and to the adm mass at infinity . this would prove conjecture [ dilaton ] including the rigidity statement , provided the horizon is connected and there are no charges outside the horizon .
finally , we conjecture that there exists an emd conformal flow satisfying : ( i ) the emd dominant energy condition is preserved under the flow ; ( ii ) the absence of charges is preserved under the flow ; ( iii ) the adm mass is monotonically non - increasing under the flow and constant only for the spherically symmetric emd black holes ; ( iv ) the area of the boundary is constant under the flow ; ( v ) the boundary eventually encloses any compact set on the original manifold .
this would prove the full conjecture . for ( iii ) , a critical ingredient , used in both @xcite and @xcite , namely a conformal factor used to prove uniqueness in a doubling argument , is already available @xcite . | in this paper we investigate the extension of the charged riemannian penrose inequality to the case where charges are present outside the horizon .
we prove a positive result when the charge densities are compactly supported , and present a counterexample when the charges extend to infinity .
we also discuss additional extensions to other matter models . |
this paper is mostly an expository survey , centered on the topology of complements of hyperplane arrangements , their milnor fibrations , and their boundary structures .
the presentation is loosely based on a set of notes for a mini - course given at the conference
arrangements in pyrnes " , held in pau , france in june 2012 . although we expanded the scope of those notes , and provided many more details and explanations , we made every effort to maintain the original spirit of the lectures , which was to give a brisk , self - contained introduction to the subject , and provide motivation for further study .
an arrangement is a finite collection of hyperplanes in a finite - dimensional , complex vector space .
there are various ways to understand the topology of such an object . in this paper , we describe several topological spaces associated to a hyperplane arrangement , all connected to each other by means of inclusions , bundle maps , or covering projections . associated to these spaces , there is a plethora of topological invariants of an algebraic nature : fundamental group and lower central series , betti numbers and torsion coefficients , cohomology ring and massey products , characteristic and resonance varieties , and so on .
one of the main goals of the subject is to decide whether a given invariant is combinatorially determined , and , if so , to express it explicitly in terms of the intersection lattice of the arrangement .
as the title indicates , the focus of the paper is on the milnor fibration of the complement of a hyperplane arrangement .
we use cohomology with coefficients in rank @xmath0 local systems on the complement to compute the homology groups of the milnor fiber , with coefficients in a field not dividing the number of hyperplanes , and determine the characteristic polynomial of the monodromy operator acting on these homology groups . in the process , we also show how to compute various homological invariants of the boundary of the complement and the boundary of the milnor fiber . given an arrangement @xmath1 in @xmath2 , the most direct approach is to study the complement , @xmath3 .
the fundamental group of this space can be computed algorithmically , using the braid monodromy associated to a generic plane section .
the cohomology ring of the complement was computed by brieskorn in @xcite .
his result shows that @xmath4 is a formal space ; in particular , all rational massey products vanish . in @xcite ,
orlik and solomon gave a simple combinatorial description of the algebra @xmath5 : it is the quotient of the exterior algebra on classes dual to the meridians , modulo a certain ideal determined by the intersection lattice , @xmath6 .
a comprehensive treatment of this topic can be found in @xcite .
starting in the mid - to - late 1990s , the subject underwent a considerable shift of emphasis towards the study of the cohomology jump loci of the complement .
these loci come in two basic flavors : the characteristic varieties , which are the jump loci for cohomology with coefficients in complex , rank one local systems , and the resonance varieties , which are the jump loci for the homology of the cochain complexes arising from multiplication classes in @xmath7 .
since the complement of the arrangement is a smooth , quasi - projective variety , its characteristic varieties are finite unions of torsion - translates of algebraic subtori of the character group @xmath8 .
since the complement is a formal space , the resonance varieties of @xmath4 coincide with the tangent cone at the origin to the corresponding characteristic varieties , and thus are finite unions of rationally defined linear subspaces of @xmath7 . for the degree-@xmath0 resonance varieties ,
these subspaces were described combinatorially by libgober and yuzvinsky @xcite , and later by falk and yuzvinsky @xcite , solely in terms of multinets on sub - arrangements of @xmath1 . in general , though , the degree-@xmath0 characteristic varieties of an arrangement may contain components which do not pass through the origin , and it is still an open problem whether such components are combinatorially determined .
the characteristic and resonance varieties of an arrangement complement may also be defined for ( algebraically closed ) fields of positive characteristic .
the nature of these varieties is less predictable , though .
for instance , as noted in @xcite , both the tangent cone formula and the linearity of the resonance components fail in this setting .
furthermore , as shown in @xcite , non - vanishing massey triple products may appear in positive characteristic .
a more refined topological invariant of an arrangement @xmath1 is the milnor fibration of its complement . for each hyperplane @xmath9
, choose a linear form @xmath10 whose kernel is @xmath11 .
the polynomial @xmath12 , then , is homogeneous of degree @xmath13 . as shown by milnor @xcite in a more general context ,
the restriction @xmath14 is a smooth fibration . the typical fiber , @xmath15 ,
is called the milnor fiber of the arrangement .
we give a topological description of this fibration , building on previous joint work with cohen and denham @xcite .
it turns out that @xmath16 is a regular , cyclic @xmath17-fold cover of the projectivized complement , @xmath18 .
the classifying homomorphism for this cover , @xmath19 , takes each meridian generator to @xmath0 .
embedding @xmath20 into @xmath21 by sending @xmath0 to a primitive @xmath17-th root of unity , we may view @xmath22 as a character on @xmath23 .
the relative position of this character with respect to the characteristic varieties of @xmath24 determines the betti numbers of the milnor fiber @xmath16 , as well as the characteristic polynomial of the algebraic monodromy .
more generally , given multiplicities @xmath25 for each hyperplane @xmath26 , we consider the milnor fibration of the multi - arrangement @xmath27 , defined by the homogeneous polynomial @xmath28 .
the milnor fiber of this polynomial is now a regular , @xmath29-fold cyclic cover of the complement , where @xmath29 is the sum of the multiplicities .
not too surprisingly , the betti numbers and other topological invariants of the milnor fiber @xmath30 vary with the choice of @xmath31 .
although the complement of an arrangement and its milnor fiber share some common features ( for instance , they are both smooth , quasi - projective varieties ) , there are some striking differences between the two . for one , the homology groups of the milnor fiber need not be torsion - free , as recent examples from @xcite show . for another , the milnor fiber may have non - vanishing massey products , and thus be non - formal , as examples from @xcite show . finally ,
as noted in @xcite , the cohomology jump loci of the milnor fiber may differ from those of the complement .
all these novel qualitative features are related to the nature of multinets supported by the arrangement in question , or one of its sub - arrangements .
both the projectivized complement and the milnor fiber of an arrangement are non - compact manifolds ( without boundary ) .
replacing these spaces by their compact versions allows us to study the behavior of the complement and the milnor fibration as we approach the boundary , thereby revealing otherwise hidden phenomena . cutting off a regular neighborhood of the arrangement @xmath1 yields a compact manifold with boundary , @xmath32 , onto which @xmath24 deform - retracts .
the boundary manifold of the arrangement , then , is the smooth , compact , orientable @xmath33-dimensional manifold @xmath34 .
the homology groups are torsion - free , and the ring @xmath35 is functorially determined by @xmath36 , via a doubling " construction . especially interesting
is the case @xmath37 , when the manifold @xmath34 is a graph - manifold , in the sense of waldhausen . following the approach from @xcite , we describe the fundamental group , the cohomology ring , and the cohomology jump loci of @xmath34 in terms of the underlying graph structure , which in turn can be read off the intersection lattice of @xmath1 . yet
significant differences with the complement exist .
for one , the boundary manifold is never formal , unless @xmath1 is a pencil or a near - pencil .
for another , the resonance varieties of @xmath34 may have non - linear components .
intersecting now the milnor fiber with a ball in @xmath2 centered at the origin yields a compact manifold with boundary , @xmath38 , onto which @xmath16 deform - retracts .
the boundary of the milnor fiber , then , is the smooth , compact , orientable @xmath33-dimensional manifold @xmath39 .
we observe here that @xmath39 is a regular , cyclic @xmath17-fold cover of @xmath34 , where @xmath13 , and identify a classifying homomorphism for this cover .
when @xmath37 , the boundary of @xmath38 is a @xmath40-dimensional graph - manifold . using work of nmethi and szilrd @xcite , we give a formula for the characteristic polynomial of the monodromy operator acting on @xmath41 , purely in terms of the mbius function of @xmath42 .
we also show that the manifold @xmath39 is never formal , unless @xmath1 is a pencil or a near - pencil , and point out that @xmath43 typically has non - trivial @xmath17-torsion .
the paper is divided roughly into three parts , following the approach outlined so far in this introduction . the first part deals with the complement of a hyperplane arrangement @xmath1 . in
[ sect : ma1 ] , we discuss the combinatorics of @xmath1 , as it relates to the cohomology ring and resonance varieties of the complement , with special emphasis on multinets . in [ sect : ma2 ] , we study the topology of @xmath44 , as reflected in the fundamental group and characteristic varieties , with special emphasis on orbifold fibrations and translated subtori in those varieties .
the second part covers the milnor fibration defined by a multi - arrangement . in [ sect : mf arr ] we set up in all detail the regular covers into which the corresponding milnor fiber fits . in [ sect : mf bis ] , we study the homology of the milnor fiber and the monodromy action , and discuss the formality properties and the cohomology jump loci of the milnor fiber .
the third part deals with the boundary structures associated to an arrangement . in
[ sect : boundary ] we discuss the boundary manifold of an arrangement , while in [ sect : bdry mf ] we discuss the boundary of the milnor fiber . sprinkled throughout the text
there are about two dozen open problems .
most of these problems have been raised before ; some are well - known , but some appear here for the first time .
we collect the necessary background material in three appendices at the end .
appendix [ sect : cvs ] serves as a quick introduction to the cohomology jump loci of a space , and some of their properties .
appendix [ sect : covers ] deals with the homology groups and jump loci of finite , regular abelian covers of a space .
finally , appendix [ sect : formal ] discusses formality properties of spaces , especially as they pertain to finite covers and jump loci .
i wish to thank the organizers of the arrangements in pyrnes conference for giving me the opportunity to lecture in such a wonderful setting , and for their support and hospitality .
i also wish to thank the university of sydney for its support and hospitality while most of this paper was written up .
in this section we describe the cohomology ring and the resonance varieties of the complement of a complex hyperplane arrangement .
an _ arrangement of hyperplanes _ is a finite set @xmath1 of codimension-@xmath0 linear subspaces in a finite - dimensional , complex vector space @xmath2 . throughout
, we will assume that the arrangement is central , that is , all the hyperplanes pass through the origin .
the combinatorics of the arrangement is encoded in its _ intersection lattice _ , @xmath6 ; this is the poset of all intersections of @xmath1 , ordered by reverse inclusion , and ranked by codimension .
the arrangement is said to be essential if the intersection of all flats in @xmath6 is @xmath45 . for each hyperplane @xmath9 ,
let @xmath46 be a linear form with kernel @xmath11 .
the product @xmath47 then , is a defining polynomial for the arrangement , unique up to a ( non - zero ) constant factor .
notice that @xmath48 is a homogeneous polynomial of degree equal to @xmath49 , the cardinality of the set @xmath1 . on occasion
, we will allow multiplicities on the hyperplanes .
multi - arrangement _ is a pair @xmath27 , where @xmath1 is a hyperplane arrangement , and @xmath50 is a function with @xmath51 for each @xmath52 .
we may also assign multiplicities to subspaces @xmath53 , by letting @xmath54 .
a defining polynomial for the multi - arrangement @xmath27 is the homogeneous polynomial @xmath55 the main topological invariant associated to an arrangement @xmath1 is its complement , @xmath56 .
this is a smooth , quasi - projective variety , with the homotopy type of a connected , finite cw - complex of dimension @xmath57 .
[ ex : boolean ] the boolean arrangement @xmath58 consists of the coordinate hyperplanes @xmath59 in @xmath60 .
the intersection lattice is the boolean lattice of subsets of @xmath61 , ordered by reverse inclusion , while the complement is the complex algebraic torus @xmath62 .
[ ex : config ] the reflection arrangement of type @xmath63 , also known as the braid arrangement , consists of the diagonal hyperplanes @xmath64 in @xmath60 .
the intersection lattice is the lattice of partitions of @xmath65=\set{1,\dots , n}$ ] , ordered by refinement .
the complement is the configuration space @xmath66 of @xmath17 ordered points in @xmath67 . in the early 1960s , fox and
neuwirth showed that @xmath68 , the pure braid group on @xmath17 strings , while neuwirth and fadell showed that @xmath66 is aspherical .
the group @xmath21 acts freely on @xmath69 via @xmath70 .
the orbit space is the complex projective space of dimension @xmath71 , while the orbit map , @xmath72 , @xmath73 $ ] , is the hopf fibration .
the set @xmath74 is an arrangement of codimension @xmath0 projective subspaces in @xmath75 ; its complement , @xmath76 , coincides with the quotient @xmath77 .
the hopf map restricts to a bundle map , @xmath78 , with fiber @xmath79 . fixing a hyperplane @xmath9
, we see that @xmath80 is also the restriction of the bundle map @xmath81 .
clearly , this latter bundle is trivial ; hence , we have a diffeomorphism @xmath82 [ ex : pencil ] let @xmath83 be the arrangement of @xmath84 lines in @xmath85 defined by the polynomial @xmath86 .
then @xmath87 consists of @xmath84 points in @xmath88 .
thus , @xmath89 , and so @xmath90 is homotopy equivalent to @xmath91 .
the ( integral ) cohomology ring of a hyperplane arrangement complement was computed by brieskorn in @xcite , building on pioneering work of arnold on the cohomology ring of the pure braid group . in @xcite , orlik and
solomon gave a simple description of this ring , solely in terms of the intersection lattice of the arrangement .
let @xmath1 be an arrangement , with complement @xmath92 .
fix a linear order on @xmath1 , and let @xmath93 be the exterior algebra over @xmath94 with generators @xmath95 in degree @xmath0 .
next , define a differential @xmath96 of degree @xmath97 , starting from @xmath98 and @xmath99 , and extending @xmath100 to a linear operator on @xmath93 , using the graded leibniz rule @xmath101 for homogeneous elements @xmath102 and @xmath103 .
finally , let @xmath104 be the ideal of @xmath93 generated by @xmath105 , where @xmath106 runs through all sub - arrangements of @xmath1 which are not in general position , and @xmath107 .
then @xmath5 is isomorphic , as a graded @xmath94-algebra , to the quotient ring @xmath108 . under this isomorphism ,
the basis @xmath109 of @xmath110 is dual to the basis of @xmath111 given by the meridians @xmath112 around the hyperplanes , oriented compatibly with the complex orientations on @xmath2 and the hyperplanes .
furthermore , all the homology groups of @xmath4 are torsion - free ; the generating function for their ranks is given by @xmath113 where @xmath114 is the mbius function of the intersection lattice , given inductively by @xmath115 and @xmath116 .
for details on all this , we refer to @xcite .
recall that a finite cw - complex is formal if its rational cohomology ring is quasi - isomorphic to its sullivan s algebra of polynomial differential forms ( more details can be found in appendix [ sect : formal ] ) .
it follows from brieskorn s work that the complement of a hyperplane arrangement @xmath1 is formal , in a very strong sense .
indeed , for each hyperplane @xmath9 , the @xmath0-form @xmath117 on @xmath2 restricts to a @xmath0-form on @xmath44 .
let @xmath118 be the subalgebra of the de rham algebra @xmath119 generated over @xmath120 by these @xmath0-forms . then ,
as shown in @xcite , the correspondence @xmath121 $ ] induces an isomorphism @xmath122 , and this readily implies the formality of @xmath44 .
similar considerations apply to the homology and cohomology of the projectivized complement , @xmath76 .
in particular , if we let @xmath13 be the number of hyperplanes , then @xmath123 , with canonical basis the set @xmath124 , and @xmath125 we will denote by @xmath126 the image of @xmath127 in @xmath128 .
let @xmath129 be an algebraically closed field .
the above - mentioned isomorphisms allow us to identify @xmath130 with the affine space @xmath131 , and @xmath132 with the affine space @xmath133 let @xmath134 be the resonance varieties of the arrangement complement .
from the general theory reviewed in
[ subsec : rv ] , we know that each of these sets is a homogeneous subvariety of @xmath131 , and that @xmath135 , for all @xmath136 .
the diffeomorphism , together with the product formula yields an identification @xmath137 thus , we may view the resonance varieties of @xmath1 as lying in the affine space @xmath138 .
if @xmath139 is a proper sub - arrangement , the inclusion @xmath140 induces an epimorphism on fundamental groups . by proposition [ prop : epi cv ] ,
the induced homomorphism in cohomology restricts to an embedding @xmath141 .
the irreducible components of @xmath142 that lie in the image of such an embedding are called _ non - essential _ ; the remaining components are called _
essential_. the description of the orlik solomon algebra given in [ subsec : os alg ] makes it clear that the resonance varieties @xmath134 depend only on the intersection lattice , @xmath6 , and on the characteristic of the field @xmath129 .
a basic problem in the subject is to find concrete formulas making this dependence explicit .
we will briefly discuss the positive characteristic case in [ subsec : res p ] , but for now we will concentrate on the case when @xmath143 and @xmath144 .
the complex resonance varieties @xmath145 were first defined and studied by falk in @xcite .
soon after , cohen
suciu @xcite , libgober @xcite , and libgober
yuzvinsky @xcite showed that the varieties @xmath146 consist of linear subspaces of the vector space @xmath147 , and analyzed the nature of these subspaces .
let us summarize those results .
[ thm : res arr ] for a hyperplane arrangement @xmath1 , the following hold . 1 . [ r1 ]
each irreducible component of @xmath148 is either @xmath45 , or a linear subspace of @xmath149 of dimension at least @xmath150 .
[ r2 ] two distinct components of @xmath148 meet only at @xmath151 .
[ r3 ] @xmath152 is either @xmath45 , or the union of all components of @xmath148 of dimension greater than @xmath153 .
an elegant method , due to falk and yuzvinsky @xcite , describes explicitly the linear subspaces comprising @xmath148 . before proceeding , we need to introduce the relevant combinatorial notion , following the treatment from @xcite .
[ def : multinet ] a _ multinet _
@xmath154 on an arrangement @xmath1 consists of the following data : an integer @xmath155 , and a partition of @xmath1 into @xmath156 subsets , say , @xmath157 .
an assignment of multiplicities on the hyperplanes , @xmath158 .
a subset @xmath159 , called the base locus . moreover , the following conditions must be satisfied : 1 . [
m1 ] there is an integer @xmath160 such that @xmath161 , for all @xmath162 $ ] .
2 . [ m2 ] for any two hyperplanes @xmath11 and @xmath163 in different classes , the flat @xmath164 belongs to @xmath165 .
[ m3 ] for each @xmath166 , the sum @xmath167 is independent of @xmath168 .
[ m4 ] for each @xmath162 $ ] , the space @xmath169 is connected .
we say that a multinet as above has @xmath156 classes and weight @xmath160 , and refer to it as a @xmath170-multinet . without essential loss of generality
, we may assume that @xmath171 .
if all the multiplicities are equal to @xmath0 , the multinet is said to be _
reduced_. if , furthermore , every flat in @xmath165 is contained in precisely one hyperplane from each class , the multinet is called a _ @xmath170-net_. for any multinet on @xmath1 , the base locus @xmath165 is determined by the partition @xmath172 . indeed , for each @xmath173 , we have that @xmath174 the next lemma ( also from @xcite ) is an easy consequence of this observation , and definition [ def : multinet ] . for completeness
, we include a proof .
[ lem : fysums ] for any @xmath170-multinet , the following identities hold : 1 . @xmath175 .
2 . @xmath176 , for any @xmath9 .
3 . @xmath177 .
clearly , @xmath178 to verify the second identity , let @xmath162 $ ] be such that @xmath179 ; then , for any @xmath180 , we have @xmath181 finally , for any @xmath173 , @xmath182 and this completes the proof . [ ex : local component ] let @xmath183 , and assume that the sub - arrangement @xmath184 has size at least @xmath40 .
we may then form a net on @xmath185 by assigning to each hyperplane the multiplicity @xmath0 , putting one hyperplane in each class , and setting @xmath186 .
( -0.5,3 )
( 2.5,-3 ) ; ( 0.5,3 ) ( -2.5,-3 ) ; ( -3,-2 ) ( 3,-2 ) ; ( 3,-2.68 ) ( -2,0.68 ) ; ( -3,-2.68 ) ( 2,0.68 ) ; ( 0,-3.1 ) ( 0,3.1 ) ; at ( -2,-2 ) @xmath187 ; at ( 2,-2 ) @xmath187 ; at ( 0,2 ) @xmath187 ; at ( 0,-0.7 ) @xmath187 ; at ( -2.5,0.4 ) @xmath188 ; at ( 2.5,0.4 ) @xmath189 ; at ( 0,-3.5 ) @xmath190 ; at ( -3.7,-2 ) @xmath191 ; at ( -2.5,-3.3 ) @xmath192 ; at ( 2.5,-3.3 ) @xmath193 ; [ ex : braid arr ]
let @xmath1 be a generic @xmath40-section of the reflection arrangement of type @xmath194 , defined by the polynomial @xmath195 .
figure [ fig : braid ] shows a plane section of @xmath1 , with the hyperplanes labeled by their defining linear forms . ordering the hyperplanes as the factors of @xmath48
, the flats in @xmath196 may be labeled as @xmath197 , @xmath198 , @xmath199 , and @xmath200 .
the @xmath201-net on @xmath1 depicted in the picture corresponds to the partition @xmath202 ; the base locus @xmath165 consists of the triple points indicated by dark circles .
[ ex : b3 ] let @xmath1 be the reflection arrangement of type @xmath203 , defined by the polynomial @xmath204 . figure [ fig : b3 arr ] shows a plane section of @xmath1 , together with a @xmath205-multinet on it . ( 0,0 ) circle ( 3.1 ) ; at ( -2.4,0.25 ) @xmath206 ; at ( 0,-2.5 ) @xmath207 ; at ( 3.35,0.5 ) @xmath208 ; ( 0,0 ) circle ( 2.9 ) ; ( -1,-2.1 ) ( -1,2.5 ) ; ( 0,-2.2 ) ( 0,2.5 ) ; ( 1,-2.1 ) ( 1,2.5 ) ; ( -2.5,-1 ) ( 2.5,-1 ) ; ( -2.5,0 ) ( 2.5,0 ) ; ( -2.5,1 ) ( 2.5,1 ) ; ( -2,-2 ) ( 2,2 ) ; ( -2,2 ) ( 2,-2 ) ; at ( -2.1,-0.75 ) @xmath191 ; at ( -2.1,1.2 ) @xmath190 ; at ( -1,-2.35 ) @xmath192 ; at ( 1,-2.35 ) @xmath188 ; at ( -2,-1.4 ) @xmath189 ; at ( 2,-1.4 ) @xmath193 ; the next theorem , which combines results of pereira
yuzvinsky @xcite and yuzvinsky @xcite , summarizes what is known about the existence of non - trivial multinets on arrangements .
[ thm : pyy ] let @xmath209 be a multinet on @xmath1 , with multiplicity vector @xmath31 and base locus @xmath165 . 1
. [ multi1 ] if @xmath210 , then @xmath211 or @xmath212 .
[ multi2 ] if there is a hyperplane @xmath9 such that @xmath213 , then @xmath211 .
although infinite families of multinets with @xmath211 are known , only one multinet with @xmath214 is known to exist : the @xmath215-net on the hessian arrangement , which we will discuss in example [ ex : hesse ] .
in fact , it is a conjecture of yuzvinsky @xcite that the only @xmath216-multinet is the hessian @xmath215-net .
we now return to the ( degree @xmath0 ) complex resonance varieties of a hyperplane arrangement @xmath1 .
recall that @xmath217 is a subvariety of the affine space @xmath218 , consisting of linear subspaces meeting only at the origin .
let us describe these subspaces in a concrete way
. given a multinet @xmath219 on @xmath1 , with parts @xmath220 and multiplicity vector @xmath31 , set @xmath221 for each @xmath222 , and put @xmath223 by construction , @xmath224 is a linear subspace of @xmath149 . as shown in (
* theorem 2.4 ) , this subspace lies inside @xmath148 and has dimension @xmath225 .
now suppose there is a sub - arrangement @xmath139 which supports a multinet @xmath219 with @xmath156 parts . by the above
, the linear space @xmath226 lies inside @xmath227 . by the discussion from [ subsec : res var ] , the inclusion @xmath228 induces an embedding @xmath229 .
thus , @xmath226 is a linear subspace of @xmath148 , of dimension @xmath225 .
conversely , it is shown in ( * ? ? ?
* theorem 2.5 ) that all ( non - zero ) irreducible components of @xmath148 arise in this fashion . summarizing the above discussion
, we have the following description of the first resonance variety of an arrangement .
[ thm : res fy ] the positive - dimensional , irreducible components of @xmath148 are in one - to - one correspondence with the multinets on sub - arrangements of @xmath1 , and so @xmath230 using now theorem [ thm : res arr ] , part [ r3 ] , we find that @xmath231 the simplest components of @xmath148 arise in the following fashion .
let @xmath232 be a rank @xmath150 flat lying at the intersection of at least @xmath40 hyperplanes .
recall from example [ ex : local component ] that @xmath232 determines a net on the sub - arrangement @xmath233 consisting of those hyperplanes in @xmath1 that contain @xmath232 .
the corresponding component of @xmath148 , @xmath234 has dimension @xmath235 , and is called a _
local _ component . if @xmath236 , then all components of @xmath148 are local .
for @xmath237 , though , the resonance variety @xmath148 may have non - local components .
it follows from theorems [ thm : pyy ] and [ thm : res fy ] that any such component must have dimension either @xmath150 or @xmath40 .
[ ex : braid ] let @xmath1 be the braid arrangement from example [ ex : braid arr ] .
the variety @xmath238 has @xmath212 local components , corresponding to the flats @xmath239 , and one non - local component , corresponding to the net @xmath219 indicated in figure [ fig : braid ] : @xmath240 since all these components are @xmath150-dimensional , @xmath241 .
[ ex : b3 bis ] let @xmath1 be the @xmath203-arrangement from example [ ex : b3 ] . ordering the hyperplanes as the factors of the defining polynomial , the multinet @xmath219 indicated in figure [ fig : b3 arr ] has associated partition @xmath242 .
the variety @xmath243 has @xmath244 local components , corresponding to @xmath212 triple points and @xmath40 quadruple points , @xmath245 components corresponding to braid sub - arrangements , and one essential , @xmath150-dimensional component , corresponding to the above multinet , @xmath246 ( -2,-0.5 ) rectangle ( 6,6.5 ) ; ( 0,4 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = blue ] ( v1 ) ; ( 2,4 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = red!90!white ] ( v2 ) ; ( 4,4 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkgreen!80!white ] ( v3 ) ; ( 0,2 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = blue ] ( v4 ) ; ( 2,2 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = red!90!white ] ( v5 ) ; ( 4,2 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkgreen!80!white ] ( v6 ) ; ( 0,0 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = blue ] ( v7 ) ; ( 2,0 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = red!90!white ] ( v8 ) ; ( 4,0 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkgreen!80!white ] ( v9 ) ; ( v1 ) ( v2 ) ( v3 ) ; ( v4 ) ( v5 ) ( v6 ) ; ( v7 ) ( v8 ) ( v9 ) ; ( v1 ) ( v4 )
( v7 ) ; ( v2 ) ( v5 ) ( v8 ) ; ( v3 ) ( v6 ) ( v9 ) ; ( v2 ) ( v4 ) ( v8 ) ( v6 ) ( v2 ) ; ( v1 ) ( v5 ) ( v9 ) ; ( v3 ) ( v5 ) ( v7 ) ; ( v7 ) .. controls ( -5,6 ) and ( -1,6.8 ) .. ( v2 ) ; ( v4 ) .. controls ( -4,7 ) and ( 0,7.8 ) .. ( v3 ) ; ( v1 ) .. controls ( 5,7.8 ) and ( 8,7 ) .. ( v6 ) ; ( v2 ) .. controls ( 6,6.8 ) and ( 9,6 ) .. ( v9 ) ; [ ex : ceva3 ] let @xmath1 be the ceva(@xmath40 ) arrangement , also known as the monomial arrangement of type @xmath247 , associated to the complex reflection group @xmath248 , and defined by the polynomial @xmath249 .
figure [ fig : ceva3 ] shows the corresponding matroid , denoted @xmath250 in @xcite .
the resonance variety @xmath243 has @xmath251 local components , corresponding to the triple points , and @xmath212 essential components corresponding to the @xmath252-nets defined by the partitions @xmath253 ,
@xmath254 , @xmath255 , and @xmath256 .
the first of these nets is depicted in figure [ fig : ceva3 ] .
the next example describes the only arrangement for which non - local components of dimension @xmath40 are known to occur .
[ ex : hesse ] let @xmath1 be the hessian arrangement in @xmath257 , also known as the monomial arrangement of type @xmath258 associated to the complex reflexion group of type @xmath259 , and defined by the polynomial @xmath260 .
the projective configuration consists of @xmath251 projective lines meeting at @xmath261 quadruple points , and @xmath251 double points ; each line contains @xmath40 quadruple points and @xmath150 double points .
figure [ fig : hessian ] shows the corresponding matroid , which is obtained from the projective plane over @xmath262 ( also known as the @xmath263 matroid , cf .
@xcite ) by deleting a point .
the resonance variety @xmath148 has @xmath261 local components , and a single , @xmath40-dimensional , essential component , corresponding to the @xmath215-net depicted in figure [ fig : hessian ] .
( -1.5,-3.5 ) rectangle ( 6,6.5 ) ; ( 0,4 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = blue ] ( v1 ) ; ( 2,4 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = red!90!white ] ( v2 ) ; ( 4,4 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkgreen!80!white ] ( v3 ) ; ( 0,2 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = red!90!white ] ( v4 ) ; ( 2,2 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkgreen!80!white ] ( v5 ) ; ( 4,2 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = blue ] ( v6 ) ; ( 0,0 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkgreen!80!white ] ( v7 ) ; ( 2,0 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = blue ] ( v8 ) ; ( 4,0 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = red!90!white ] ( v9 ) ; ( v1 ) ( v2 ) ( v3 ) ; ( v4 ) ( v5 ) ( v6 ) ; ( v7 )
( v9 ) ; ( v1 )
( v4 ) ( v7 ) ; ( v2 ) ( v5 ) ( v8 ) ; ( v3 ) ( v6 ) ( v9 ) ; ( v4 ) ( v8 ) ; ( v2 ) ( v6 ) ; ( v1 ) ( v5 ) ( v9 ) ; ( v7 ) .. controls ( -5,6 ) and ( -1,6.8 ) .. ( v2 ) ; ( v4 ) .. controls ( -4,7 ) and ( 0,7.8 ) .. ( v3 ) ; ( 2,-3 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkbrown!75!white ] ( v10 ) ; ( 6.12,-2.12 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkbrown!75!white ] ( v11 ) ; ( 7,2 ) node[draw , ultra thick , shape = circle , inner sep=1.3pt , outer sep=0.7pt , color = dkbrown!75!white ] ( v12 ) ; ( v8 ) .. controls ( 2.65,-0.6 ) and ( 3,-1.5 ) .. ( v11 ) ; ( v6 ) .. controls ( 4.6,1.35 ) and ( 5.5,1 ) .. ( v11 ) ; ( v7 ) .. controls ( 0.4,-2.2 ) and ( 1.2,-2.6 ) .. ( v10 ) ; ( v9 ) .. controls ( 3.6,-2.2 ) and ( 2.8,-2.6 ) .. ( v10 ) ; ( v8 ) ( v10 ) ; ( v9 ) ( v11 ) ; ( v6 ) ( v12 ) ; ( v3 ) .. controls ( 5,3.8 ) and ( 6,3.6 ) .. ( v12 ) ; ( v9 ) .. controls ( 5,0.2 ) and ( 6,0.4 ) .. ( v12 ) ; the varieties @xmath264 over fields @xmath129 of characteristic @xmath265 were first defined and studied by matei and suciu in @xcite , and further investigated by falk in @xcite . as illustrated in the next few examples , all three properties from theorem [ thm : res arr ] , which hold over @xmath266 , fail in this setting .
[ ex : nonfano ] let @xmath1 be the realization of the non - fano plane , obtained from the @xmath267 arrangement from figure [ fig : b3 arr ] by deleting the planes @xmath268 and @xmath269 , and order the hyperplanes as the factors of @xmath270 .
the variety @xmath271 has @xmath272 local components corresponding to triple points , and @xmath40 components arising from braid sub - arrangements , but no essential components .
now let @xmath273 .
then , as noted in @xcite , the variety @xmath274 has a single non - local component , defined by the equations @xmath275 , while @xmath276 has a single , @xmath0-dimensional component , defined by the equations @xmath277 and @xmath278 .
thus , property [ r3 ] fails . it should be noted that the analogue of the tangent cone formula also fails in this case .
indeed , a computation from @xcite shows that @xmath279 ; thus , the tangent cone to this variety is @xmath45 , which is is strictly included in @xmath276 .
[ ex : db3 ] let @xmath1 be the arrangement obtained by deleting the plane @xmath280 from the @xmath267 arrangement , and let @xmath273 .
then , as shown in @xcite , the variety @xmath281 has two distinct components which intersect outside @xmath151 . thus , property [ r2 ] fails .
[ ex : hesse again ] let @xmath1 be the hessian arrangement from example [ ex : hesse ] , and let @xmath282 .
then , as shown in @xcite , the variety @xmath281 has a component which is isomorphic to an irreducible cubic hypersurface in @xmath283 .
thus , property [ r1 ] fails .
[ rem : p - massey ] in @xcite , matei uses the resonance varieties @xmath281 over a field @xmath129 of characteristic an odd prime @xmath284 to detect non - vanishing massey products in the cohomology ring @xmath285 of certain arrangements @xmath1 .
the simplest such example is the hessian arrangement from example [ ex : hesse ] , for which @xmath286 . these examples and the above remark
raise a natural question .
[ quest : p - formal ] given an arrangement @xmath1 , find the primes @xmath284 ( if any ) for which @xmath44 is not @xmath129-formal , over a field @xmath129 of characteristic @xmath284 .
in this section we describe the fundamental group and the characteristic varieties of the complement of a complex hyperplane arrangement . as usual , let @xmath1 be a hyperplane arrangement in @xmath2 .
the complement @xmath44 is a path - connected space .
thus , we may fix a basepoint @xmath287 , and consider the fundamental group @xmath288 . for each hyperplane
@xmath9 , pick a meridian curve about @xmath11 , oriented compatibly with the complex orientations on @xmath2 and @xmath11 , and let @xmath127 denote the based homotopy class of this curve , joined to the basepoint by a path in @xmath44 . by the van kampen theorem , then , the group @xmath288 is generated by the set @xmath124 .
an explicit presentation for this group can be obtained via the braid monodromy algorithm from @xcite .
let us briefly describe this algorithm .
let @xmath289 be a generic two - dimensional section of @xmath1 . by the lefschetz - type theorem of hamm and l ,
the inclusion @xmath290 between the respective complements induces an isomorphism on fundamental groups .
thus , for the purpose of computing fundamental groups , we may as well replace @xmath1 by @xmath289 .
so let @xmath291 be an arrangement of affine lines in @xmath85 .
let @xmath292 be the intersection points of the lines .
if @xmath293 is one such intersection point , write @xmath294 for the corresponding increasingly ordered tuple , and let @xmath295 be the braid in the pure braid group @xmath296 which performs a full twist on the strands corresponding to @xmath104 , leaving the other strands fixed .
there are then braids @xmath297 in the full braid group @xmath298 such that the arrangement group has presentation @xmath299 \rangle.\ ] ] clearly , this is a commutator - relators presentation .
furthermore , the presentation is minimal , in that the number of generators equals @xmath300 , and the number of relators equals @xmath301 . if @xmath1 is the complexification of a real arrangement , the conjugating braids @xmath302 may be obtained by a procedure that goes back to @xcite .
each vertex set @xmath303 gives rise to a partition @xmath65 = i'_{q}\cup i_{q } \cup i''_{q}$ ] into lower , middle , and upper indices .
let @xmath304 .
then @xmath302 is the subword of the full twist @xmath305}$ ] given by @xmath306 in the general case , the braids @xmath307 can be read off a braided wiring diagram " associated to the arrangement , see @xcite for further details and references .
[ rem : ryb ] in work from the early 1990s that appeared in @xcite , rybnikov constructed a pair of arrangements , @xmath308 and @xmath309 , both realizations of the same matroid , for which @xmath310 .
thus , the fundamental group of an arrangement complement is _ not _ determined by the intersection lattice , in general .
[ rem : lcs ] nevertheless , if @xmath311 is an arrangement group , the ranks of the associated graded lie algebra , @xmath312 , are combinatorially determined , due to the formality of @xmath44 . moreover ,
as shown in @xcite , the ranks of the chen lie algebra , @xmath313 , are also combinatorially determined .
as observed in @xcite , there are interesting connections between the chen ranks of the group @xmath314 and the dimensions of the components of the resonance variety @xmath148 .
for a recent survey of these topics , we refer to @xcite . returning to the general situation ,
let @xmath1 be an essential arrangement in @xmath2 .
fix an ordering of the hyperplanes , say , @xmath315 , and choose linear forms @xmath316 with @xmath317 . assembling these forms together
, we obtain an injective linear map @xmath318 now let @xmath58 be the boolean arrangement in @xmath319 , and identify @xmath320 with @xmath321 .
recall that @xmath322 if and only if @xmath323 ; thus , the map @xmath324 restricts to an inclusion @xmath325 . consequently , we may view the complement @xmath44 as a linear slice of the complex @xmath17-torus : @xmath326 clearly , the map @xmath327 is equivariant with respect to the diagonal action of @xmath21 on both source and target .
thus , @xmath324 descends to a map @xmath328 , which defines an embedding @xmath329 .
[ lem : h1ofa ] let @xmath315 be a hyperplane arrangement . then , 1 . [ iota1 ] the inclusion @xmath330 is a classifying map for the universal abelian cover @xmath331 .
[ iota2 ] the inclusion @xmath332 is a classifying map for the universal abelian cover @xmath333 .
we start with the first statement .
set @xmath334 .
since @xmath321 is a classifying space for the group @xmath335 , it is enough to to check that the induced homomorphism , @xmath336 , coincides with the abelianization map @xmath337 .
by naturality of the hurewicz homomorphism , we have a commuting diagram @xmath338 \ar@{->>}^{\operatorname{{ab}}}[d ] & \pi_1(m({{\mathcal{b}}}_n ) ) \ar@{=}[d ] \\
h_1(m({{\mathcal{a}}}),{\mathbb{z } } ) \ar^{\iota_{*}}[r ] & h_1(m({{\mathcal{b}}}_n),{\mathbb{z}})\ , . } \ ] ] plainly , the homomorphism @xmath339 takes each meridian @xmath127 in @xmath123 to the corresponding meridian in @xmath340 .
hence , in these meridian bases , the map @xmath341 is the identity map , and we are done . for the second statement ,
recall from that @xmath342 .
an argument as above shows that @xmath343 is the abelianization map , and this ends the proof .
let @xmath311 and @xmath344 be the fundamental groups of the complement and the projectivized complement , respectively .
let @xmath129 be an algebraically closed field .
the isomorphism @xmath345 allows us to identify the character group @xmath346 with the algebraic torus @xmath347 .
similarly , the isomorphism @xmath348 allows us to identify the character group @xmath349 with the algebraic torus @xmath350 let @xmath351 be the characteristic varieties of the arrangement complement . from the general theory reviewed in [ subsec : char var ] , we know that each set @xmath351 is a zariski closed subset of the algebraic torus @xmath347 .
the diffeomorphism , together with the product formula yields an identification @xmath352 .
in particular , we may view the characteristic varieties of @xmath1 as lying in the torus .
[ ex : circle ] let @xmath1 be the arrangement in @xmath67 consisting of the single point @xmath151 .
then @xmath353 is homotopy equivalent to a circle @xmath354 . under the identifications
@xmath355 and @xmath356 $ ] , the equivariant chain complex of the universal cover @xmath357 takes the form @xmath358 , where @xmath359 and @xmath360 . given a character @xmath361 , we tensor this chain complex with the local system @xmath362 , and obtain a new chain complex , @xmath363 .
this chain complex is exact , except for @xmath364 , when @xmath365 .
the upshot is that @xmath366 and @xmath367 , otherwise .
[ ex : cv boolean ] more generally , let @xmath58 be the boolean arrangement in @xmath319 , with complement @xmath368 . proceeding as in the previous example , we see that @xmath369 \emptyset & \text{otherwise}. \end{cases}\ ] ] [ ex : cv free ] let @xmath83 be a pencil of @xmath370 lines in @xmath85 as in example [ ex : pencil ] .
recall that @xmath371 , and thus @xmath372 . identifying @xmath373 with @xmath374 , we find that @xmath375 although an explicit formula for the characteristic varieties @xmath351 is far from known in general , a structural result is known in the case when @xmath266 . a theorem of arapura @xcite , as strengthened by dimca @xcite , libgober @xcite , artal - bartolo , cogolludo and matei @xcite , budur and wang @xcite , and others , states the following .
[ thm : arapura ] the characteristic varieties @xmath376 of an arrangement complement are finite unions of torsion - translates of algebraic subtori in @xmath377 .
we now describe in more detail the characteristic varieties of @xmath44 in degree @xmath144 .
since these varieties depend only on @xmath378 , we may as well assume @xmath1 is a ( central ) arrangement in @xmath257 .
a key point here is the naturality property enjoyed by these varieties .
suppose @xmath379 is a map between ( projectivized ) arrangement complements , and that the induced homomorphism , @xmath380 , is surjective .
then , by proposition [ prop : epi cv ] , the corresponding monomorphism between character groups restricts to an embedding @xmath381 .
in particular , if @xmath139 is a sub - arrangement , the inclusion @xmath382 induces an epimorphism on fundamental groups , and thus defines an embedding @xmath381 . building on arapura s work , dimca @xcite and artal bartolo
, cogolludo and matei @xcite showed that the varieties @xmath383 are unions of torsion - translated subtori inside the algebraic torus @xmath384 .
the isolated ( torsion ) points in these characteristic varieties are still poorly understood , but the positive - dimensional components in @xmath385 can be described in very concrete terms .
a ( genus @xmath151 ) _ orbifold fibration _ is a surjective holomorphic map @xmath386 , for some @xmath387 , that has connected generic fiber , and extends to a map between the respective compactifications , @xmath388 , which is also a surjective , holomorphic map with connected generic fibers .
the map @xmath389 is a locally trivial bundle map , away from a finite ( possibly empty ) set of points @xmath390 inside @xmath391 ; let @xmath392 ( @xmath393 ) denote the multiplicities of the respective fibers .
let @xmath394 be the induced homomorphism on fundamental groups .
since the generic fiber of @xmath389 is connected , this homomorphism is surjective .
moreover , @xmath395 lifts to a ( surjective ) homomorphism @xmath396 , where @xmath397 is the orbifold fundamental group of the base .
thus , @xmath398 induces an injective morphism , @xmath399 , between character groups .
note that @xmath400 where @xmath401 is the identity component and @xmath402 . by proposition [ prop : epi cv ] ,
the above morphism restricts to an inclusion @xmath403 .
a computation as in example [ ex : circle ] shows that @xmath404 \big ( \widehat{\pi}\setminus \widehat{\pi}^{\circ}\big ) \cup \{{{\mathbf 1}}\ } & \text{if $ r=1 $ and $ s\ge 1$}. \end{cases}\ ] ] note that each irreducible component of @xmath405 has dimension @xmath406 .
the next theorem says that _ all _ positive - dimensional components of @xmath385 arise by pullback along such orbifold fibrations , which we call _ large _ in the first case , and _ small _ in the second case .
[ thm : v1a ] the first characteristic variety of an arrangement @xmath1 is given by @xmath407 where the unions are over the equivalence classes of pencils @xmath408 of the types indicated , and @xmath409 is a finite set of torsion characters . as shown by falk and yuzvinsky in @xcite , the irreducible components of @xmath385 passing through the origin can be described in terms of multinets on the intersection lattice of @xmath1 .
indeed , let @xmath410 be a defining polynomial for @xmath1 . given a multinet @xmath219 on @xmath1 , with parts @xmath220 and multiplicity vector @xmath31 ,
write @xmath411 for @xmath412 , so that the defining polynomial for @xmath27 factors as @xmath413 .
next , define a rational map @xmath414 by @xmath415.\ ] ] by definition of multinets , the degrees of the polynomials @xmath416 are independent of @xmath168 ; hence , the map @xmath417 factors through a rational map @xmath418 . as shown in @xcite , there is a set @xmath419 , \dots , [ a_k{\!:\!}b_k ] } \subset { \mathbb{cp}}^1 $ ] such that each of the polynomials can be written as @xmath420 , and , furthermore , the image of @xmath421 misses @xmath422 .
identify @xmath423 .
the restriction @xmath424 then , is a large pencil which gives rise by pullback to a @xmath425-dimensional component of @xmath385 which passes through the origin of @xmath426 .
[ ex : cv braid ] let @xmath1 be the braid arrangement from examples [ ex : braid arr ] and [ ex : braid ] .
the characteristic variety @xmath385 has @xmath212 local components of dimension @xmath150 , corresponding to the @xmath212 triple points . additionally , the @xmath201-net @xmath27 depicted in figure [ fig : braid ] defines a rational map , @xmath427 , @xmath428 .
this map restricts to a pencil @xmath429 , which yields another @xmath150-dimensional component , @xmath430 .
in general , the characteristic variety @xmath385 also has irreducible components not passing through the origin .
following @xcite , we describe a combinatorial construction which , under certain assumptions , produces @xmath0-dimensional translated subtori in @xmath385 .
fix a hyperplane @xmath9 , and let @xmath431 be the _ deletion _ of @xmath1 with respect to @xmath11 .
a _ pointed multinet _ on @xmath1 is a multinet @xmath432 , together with a distinguished hyperplane @xmath9 for which @xmath213 , and @xmath433 for each flat @xmath434 such that @xmath435 .
[ prop : del ] suppose @xmath1 admits a pointed multinet , and @xmath289 is obtained from @xmath1 by deleting the distinguished hyperplane @xmath11
. then @xmath436 has a component which is a @xmath0-dimensional subtorus of @xmath437 , translated by a character of order @xmath438 . without loss of generality
, we may assume that @xmath439 .
consider the regular map given by , @xmath440 .
since @xmath10 does not divide @xmath441 , we may extend @xmath417 to a regular map @xmath442 . by construction , @xmath443 , and so @xmath444}$ ] .
furthermore , @xmath445 equals @xmath446,[a_3{\!:\!}b_3]}$ ] .
passing to the projective complement , and taking the corestriction of @xmath447 to its image yields an orbifold fibration , @xmath448 . by hypothesis ,
the degrees of the restrictions of @xmath449 and @xmath441 to the hyperplane @xmath11 are both divisible by @xmath438 .
thus , the fiber of @xmath447 over @xmath450 $ ] has multiplicity @xmath438 , and so @xmath447 is a small pencil .
the desired conclusion now follows from theorem [ thm : v1a ] .
[ ex : deleted b3 ] let @xmath1 be the @xmath203 arrangement from examples [ ex : b3 ] and [ ex : b3 bis ] , and let @xmath289 be the arrangement obtained by deleting the hyperplane @xmath280 , as in example [ ex : db3 ] . as noted in @xcite , the characteristic variety @xmath451 has @xmath244 local components , corresponding to @xmath272 triple points and one quadruple point , @xmath452 components corresponding to braid sub - arrangements , and a component of the form @xmath453 , where @xmath454 and @xmath455 . to explain where this translated torus comes from ,
let @xmath31 be the multinet on @xmath1 depicted in figure [ fig : b3 arr ] , and let @xmath456 be the corresponding factorization of the defining polynomial of @xmath27 .
we then obtain a ( large ) pencil @xmath457 , [ 0{\!:\!}1 ] , [ 1{\!:\!}1 ] \}$ ] , given by @xmath458\mapsto [ z^2(x^2-y^2):y^2(x^2-z^2)]$ ] .
extending @xmath417 to the complement of @xmath289 by allowing @xmath280 yields an orbifold fibration , @xmath459 , [ 1{\!:\!}1 ] \}$ ] .
note that @xmath460)=[0{\!:\!}y^2x^2]$ ] , and so the fiber over @xmath450 $ ] has multiplicity @xmath150 ; thus , @xmath417 is a small pencil .
the orbifold fundamental group of the base of the pencil is @xmath461 , and the pullback of @xmath462 along @xmath417 yields the translated torus @xmath463 .
[ prob : trans ] do all positive - dimensional translated tori in the first characteristic variety of an arrangement arise in the manner described in proposition [ prop : del ] ? finally , there are also arrangements @xmath1 with isolated torsion points in the characteristic variety @xmath385 .
here is an example , also from @xcite .
[ ex : grunbaum ] let @xmath1 be the arrangement in @xmath257 defined by the polynomial @xmath464
. then @xmath385 has @xmath465 local components , corresponding to @xmath244 triple and @xmath40 quadruple points , @xmath466 components corresponding to braid sub - arrangements , @xmath0 component corresponding to a @xmath267 sub - arrangement , @xmath40 translated components corresponding to deleted @xmath267 sub - arrangements , and @xmath150 isolated points of order @xmath272 .
[ prob : tors v1 ] find a concrete description of the @xmath151-dimensional components of the first characteristic variety @xmath385 of an arrangement @xmath1 .
are all such components determined by the intersection lattice @xmath6 ?
answering in the affirmative problems [ prob : trans ] and [ prob : tors v1 ] would lead to a solution ( at least for @xmath467 ) of the following well - known problem , which is central to the theory of hyperplane arrangements .
[ prob : cv comb ] given a hyperplane arrangement @xmath1 , are the characteristic varieties @xmath376 determined by the intersection lattice @xmath6 ?
in this section , we introduce our main object of study : the milnor fibration attached to a multi - arrangement . in the process
, we describe several covering spaces related to the milnor fiber .
let @xmath1 be an arrangement of hyperplanes in @xmath2 .
recall we associated to each multiplicity vector @xmath468 a homogeneous polynomial @xmath469 of degree @xmath470 .
note that @xmath471 is a proper power if and only if @xmath472 , where @xmath473 . as before , let @xmath44 be the complement of the arrangement .
the polynomial map @xmath474 restricts to a map @xmath475 . as shown by j. milnor @xcite in a much more general context , @xmath471 is the projection map of a smooth , locally trivial bundle , known as the _ ( global ) milnor fibration _ of the multi - arrangement @xmath27 .
the typical fiber of this fibration , @xmath476 is called the _ milnor fiber _ of the multi - arrangement , while the milnor fibration itself is written as @xmath477 & m({{\mathcal{a } } } ) \ar^(.55){{{q_m({{\mathcal{a}}})}}}[r ] & { \mathbb{c}}^*. } \ ] ] clearly , the milnor fiber is a smooth manifold of dimension @xmath478 .
in fact , @xmath479 is a stein domain of complex dimension @xmath71 , and thus has the homotopy type of a finite cw - complex of dimension @xmath71 . in the case
when all the multiplicities @xmath438 are equal to @xmath0 , the polynomial @xmath480 is the usual defining polynomial for the arrangement , and has degree @xmath13 .
moreover , @xmath481 is the usual milnor fiber of @xmath1 .
[ rem : mdep ] although the polynomials @xmath471 depend on the choice of multiplicities , they all have the same zero set ; thus , they all share the same complement , namely , @xmath44 . on the other hand , the milnor fibers @xmath479 do depend on the various choices of @xmath31 , and not just on @xmath1 .
the next example illustrates this point .
[ ex : point ] let @xmath1 be the arrangement in @xmath67 consisting of the single subspace @xmath482 , and assign a multiplicity @xmath483 to that point
. then @xmath484 , and so @xmath479 is the set of @xmath31-roots of unity . of course , the milnor fibers in the previous example are connected only when @xmath485 . for an arbitrary arrangement @xmath1 ,
the number of connected components of @xmath479 equals @xmath486 .
@xmath487 \clip ( 0,0 ) circle ( 3 ) ; \foreach \a in { 0 , 60, ...
,359 } \draw(0 , 0 ) -- ( \a:8 ) ; \draw[style = thick , color = red , scale=1 , domain=1:2.5,smooth , variable=\y ] plot ( { 0.5*sqrt ( \y^3 - 1)/sqrt(\y ) } , { \y } ) ; \draw[style = thick , color = red , scale=1 , domain=1:2.5,smooth , variable=\y ] plot ( { -0.5*sqrt ( \y^3 - 1)/sqrt(\y ) } , { \y } ) ; \draw[style = thick , color = red , scale=1.6 , domain=-1.25:-0.08,smooth , variable=\y ] plot ( { 0.5*sqrt ( 1-\y^3)/sqrt(-\y ) } , { \y } ) ; \draw[style = thick , color = red , scale=1.6 , domain=-1.25:-0.08,smooth , variable=\y ] plot ( { -0.5*sqrt ( 1-\y^3)/sqrt(-\y ) } , { \y } ) ; \node at ( 0.7,1.5 ) [ label = right:${{\mathcal{a}}}$ ] { } ; \node at ( 1,-1 ) [ label = right:$f({{\mathcal{a}}})$ ] { } ; \end{tikzpicture } \hspace*{0.7 in } \begin{tikzpicture}[baseline=(current bounding box.center),scale=0.7 ] \draw ( -1,0 ) to[bend left ] ( 1,0 ) ; \draw ( -1.2,.1 ) to[bend right ] ( 1.2,.1 ) ; \draw[rotate=0 ] ( 0,0 ) ellipse ( 100pt and 50pt ) ; \node[circle , draw , inner sep=1pt , fill = black ] at ( 2.2,0 ) { } ; \node[circle , draw , inner sep=1pt , fill = black ] at ( -1.1,0.9 ) { } ; \node[circle , draw , inner sep=1pt , fill = black ] at ( -1.1,-0.9 ) { } ; \draw[->,>=stealth , shorten > = 1pt , dashed ] ( 1.9 , 0.4 ) to[bend right ] ( -0.6,1 ) ; \node at ( 0.75,1.2 ) [ label = right:$h$ ] { } ; \node at ( 1,-2 ) [ label = right:$f({{\mathcal{a}}})$ ] { } ; \end{tikzpicture}\ ] ] consider the milnor fibration @xmath488 defined by the polynomial @xmath489 . for
each @xmath490 $ ] , let @xmath491 be the fiber over the point @xmath492 , so that @xmath493 . for each @xmath494 , the path @xmath495 \to { \mathbb{c}}^*$ ] , @xmath496 lifts to a path @xmath497 \to m({{\mathcal{a}}})$ ] , @xmath498 , satisfying @xmath499 . notice that @xmath500 ; thus , if @xmath501 , then @xmath502 . by definition ,
the _ monodromy _ of the milnor fibration is the diffeomorphism @xmath503 given by @xmath504 . in view of the above discussion
, this diffeomorphism can be written as @xmath505 clearly , the map @xmath506 has order @xmath29 . furthermore , note that the complement @xmath44 is homotopy equivalent to the mapping torus of @xmath506 : @xmath507/(z,0 ) \sim ( h(z),1).\ ] ] [ ex : mf boolean ] let @xmath58 be the boolean arrangement in @xmath60 , and identify the complement @xmath320 with the algebraic torus @xmath321 . given a multiplicity vector @xmath31 , the map @xmath508 , @xmath509 is a morphism of algebraic groups .
hence , the milnor fiber @xmath510 is an algebraic subgroup , realized as the disjoint union of @xmath486 copies of @xmath511 .
the monodromy automorphism , @xmath512 , permutes those copies in a circular fashion . if @xmath513 , the algebraic subgroup @xmath514 is , in fact , an algebraic torus isomorphic to @xmath511 .
moreover , the monodromy automorphism is isotopic to the identity , through the isotopy @xmath515 .
thus , the bundle @xmath516 is trivial in this case .
the previous example is , in some sense , a classifying object for milnor fibrations of arrangements . to make this statement more precise ,
let @xmath315 be an ( ordered ) essential arrangement of hyperplanes in @xmath2 , with defining linear forms @xmath517 .
recall we defined in
[ subsec : abelcov ] a linear map @xmath518 by @xmath519 . restricting this map to the complement yields an embedding of @xmath44 into @xmath368 .
since @xmath1 and @xmath58 have the same number of hyperplanes , a multiplicity vector for one of them yields a multiplicity vector for the other .
the next lemma shows that the corresponding milnor fibrations are compatible .
[ lem : iota mab ] for each @xmath520 , the map @xmath325 is compatible with the milnor fibrations @xmath521 and @xmath522 .
write @xmath523 and @xmath524 .
clearly , then , the following equality holds : @xmath525 thus , the map @xmath324 restricts to an inclusion @xmath526 , which fits into the commuting diagram @xmath527 \ar^{\iota_m({{\mathcal{a}}})}[d ] & m({{\mathcal{a } } } ) \ar^(.55){{{q_m({{\mathcal{a}}})}}}[r ] \ar^{\iota({{\mathcal{a}}})}[d ] & { \mathbb{c}}^ * \ar@{=}[d ] \\
f_m({{\mathcal{b}}}_n ) \ar[r ] & m({{\mathcal{b}}}_n ) \ar^(.55){q_m({{\mathcal{b}}}_n)}[r ] & { \mathbb{c}}^*. } \ ] ] in other words , @xmath324 is a bundle map , as claimed . as a consequence of this lemma
, we may expresses the milnor fiber of @xmath27 as the intersection of two familiar objects .
[ cor : fam bnm ] the milnor fiber @xmath479 is obtained by intersecting the complement @xmath44 , viewed as a subvariety of the algebraic torus @xmath368 via the inclusion @xmath324 , with @xmath528 , viewed as an algebraic subgroup of @xmath321 : @xmath529 using the above comparison between the two milnor fibrations , we identify now the homomorphism induced on fundamental groups by the projection map @xmath531 . as usual , denote by @xmath532 the standard generating set for @xmath378 .
[ prop : qm pi1 ] the induced homomorphism @xmath533 is given by @xmath534 . from , we know that @xmath535 . by lemma [ lem : h1ofa ] , the homomorphism @xmath536 may be identified with the abelianization map , @xmath537 . on the other hand ,
the homomorphism induced in @xmath530 by the algebraic morphism @xmath508 , @xmath538 may be identified with the linear map @xmath539 , @xmath540 . the conclusion follows . using this proposition
, we can derive in a novel way a well - known lower - bound on the first betti number of the milnor fiber of a ( multi- ) arrangement .
[ corollary : qpi ] suppose @xmath513 .
then @xmath541 , where @xmath13 . from the assumption , we know that the milnor fibers @xmath479 and @xmath514 are connected . applying the @xmath530 functor to diagram , and using the homotopy long exact sequences associated to the respective milnor fibrations , as well as lemma [ lem : h1ofa ] and proposition [ prop : qm pi1 ] , we obtain the following commuting diagram with exact rows : @xmath542&\pi_1({{f_m({{\mathcal{a}}})}})\ar[rr ] \ar@{->>}^{\iota_m({{\mathcal{a}}})_{\sharp}}[dd ] \ar@{ .. >>}_(.4){}@/_28pt/[ddd ] & & \pi_1(m({{\mathcal{a } } } ) ) \ar@{->>}^{\iota({{\mathcal{a}}})_{\sharp}}[dd ] \ar^(.62){{{q_m({{\mathcal{a}}})}}_{\sharp}}[rr ] \ar@{ .. >>}_(.38){\operatorname{{ab}}}@/_28pt/[ddd ] & & { \mathbb{z}}\ar[r ] \ar@{=}[dd ] & 1 \,\phantom{. } \\ \\
1\ar[r]&\pi_1(f({{\mathcal{b}}}_n))\ar[rr ] \ar@{=}[d ] & & \pi_1(m({{\mathcal{b}}}_n ) ) \ar^(.62){q_m({{\mathcal{b}}}_n)_{\sharp}}[rr ] \ar@{=}[d ] & & { \mathbb{z}}\ar[r ] & 1 \ , .
\\ & { \mathbb{z}}^{n-1 } & & { \mathbb{z}}^n \ar@{->>}_{m}[urr ] } \ ] ] this diagram yields an epimorphism @xmath543 , indicated by a dotted arrow in .
this completes the proof .
we will come back to this topic in [ sect : mf bis ] , where we will use a different approach to provide improved estimates on the betti numbers of the milnor fiber .
a standard construction turns any continuous map into a fibration , at least up to homotopy , see e.g. @xcite . in the case of an inclusion @xmath544 ,
the construction is particularly simple : let @xmath545 be the space of paths in @xmath93 starting in @xmath16 .
then the inclusion @xmath546 , @xmath547 is a homotopy equivalence , while the map @xmath548 , @xmath549 is a fibration .
now suppose @xmath550 is a fibration , and all three spaces are path - connected cw - complexes .
pick a basepoint @xmath551 , and let @xmath552 be the homotopy fiber of the inclusion @xmath544 .
let @xmath553 be the space of loops in @xmath554 , based at @xmath555 .
then @xmath553 has the homotopy type of a cw - complex , and we have a map @xmath556 , which sends a loop at @xmath557 to its lift at @xmath558 , traversed in the opposite direction . comparing the long exact sequences of homotopy groups of the two fibrations , @xmath559 and @xmath560 , we conclude that @xmath389 is a homotopy equivalence . in our situation , consider the milnor fibration associated to @xmath27 .
the homotopy fiber of the inclusion @xmath561 , then , is homotopic to @xmath562 .
[ prop : mf cyclic ] the homotopy fibration defined by the inclusion @xmath563 is equivalent to the regular , infinite cyclic cover classified by the homomorphism @xmath564 , @xmath534 .
the exponential map @xmath565 , @xmath566 is a regular @xmath94-cover . pulling back this cover along the map @xmath567
, we obtain the diagram @xmath568 \ar[d ] & { \mathbb{z}}\ar[d ] \\ & f \ar^{\simeq}[r ] \ar^{\simeq}[d ] & e \ar^{q}[r ] \ar^{p}[d ] \ar@{ .. >>}_{\phi}[dl ] & { \mathbb{c}}\ar^{\exp}[d ] \\ { \mathbb{z}}\ar[r ] & { { f_m({{\mathcal{a}}})}}\ar^{j}[r ] & m({{\mathcal{a } } } ) \ar^(.55){q}[r ] & { \mathbb{c}}^ * } \ ] ] where @xmath16 is the homotopy fiber of @xmath569 , the homotopy equivalence @xmath570 is the restriction of @xmath284 to ( homotopy ) fibers , and @xmath571 is the homotopy fiber of @xmath572 . by construction ,
the pullback cover , @xmath573 , is a regular @xmath94-cover , classified by the homomorphism @xmath574 .
furthermore , the homotopy equivalence @xmath575 defined by diagram satisfies @xmath576 .
thus , the cover @xmath573 may be identified with the homotopy fibration associated to the inclusion @xmath561 .
the desired conclusion now follows from proposition [ prop : qm pi1 ] .
[ rem : df ] in general , an infinite cyclic cover of a finite cw - complex need not have finite betti numbers , let alone have the homotopy type of a finite cw - complex .
for instance , if @xmath577 is a wedge of two circles , and @xmath578 is an arbitrary @xmath94-cover , then @xmath579 . in our situation , though , the geometry of the situation conspires to insure that all the @xmath94-covers corresponding to milnor fibrations of arrangements have finite cell decompositions . for
more on the geometric and homological finiteness properties of ( infinite ) abelian covers , we refer to @xcite
. as before , fix a multiplicity vector @xmath31 on @xmath1 with @xmath513 , and set @xmath470 .
the monodromy automorphism @xmath580 , given by @xmath581 , generates a cyclic group @xmath582 .
as noted in @xcite , this group acts freely on @xmath479 , and the quotient space may be identified with the projective complement , @xmath76 .
we thus have a regular , @xmath29-fold cyclic cover , @xmath583 .
as shown in @xcite , this cover can be described in purely group - theoretic terms .
we give here a self - contained , full - detail proof of this basic result .
the proof uses a somewhat different approach , which draws in part from @xcite . to start with , recall from [ subsec : ma ] that the hopf fibration @xmath584 restricts to a ( trivializable ) bundle map , @xmath78 , with fiber @xmath21 .
[ thm : mf ] the map @xmath78 restricts to a regular , @xmath582-cover @xmath585 .
furthermore , this cover is classified by the homomorphism @xmath586 for simplicity , we will drop the arrangement @xmath1 from the notation . by homogeneity of the polynomial @xmath587
, we have that @xmath588 , for every @xmath589 and @xmath590 .
thus , the restriction of @xmath587 to a fiber of @xmath591 may be identified with the covering projection @xmath592 , @xmath593 .
now , if both @xmath594 and @xmath595 belong to @xmath596 , then @xmath597 , and so @xmath598 .
thus , the restriction @xmath599 is the orbit map of the free action of the geometric monodromy on @xmath596 .
hence , @xmath600 is a regular @xmath582-cover , as claimed . to determine the classifying homomorphism @xmath601 for this cover
, we will make some auxiliary constructions .
we refer to diagram for the various spaces and maps introduced along the way .
@xmath602 \ar^{\psi}[d ] \ar^{p_1}[dr ] & { \mathbb{c}}^ * \ar[d ] \ar^{q}[dr ] \\ f_m\ar[r ] \ar_(.5){\pi_m}[dr ] & m \ar^{\pi}[d ] \ar^{q_m}[r ] & { \mathbb{c}}^*\ , .\\ & u}\ ] ] let @xmath603 , and denote by @xmath604 and @xmath605 the restrictions of the coordinate projections .
clearly , @xmath606 thus , @xmath607 is a regular @xmath582-cover , obtained as the pullback of @xmath569 along @xmath587 .
of course , the cover @xmath569 is classified by the canonical projection @xmath608 .
hence , by proposition [ prop : qm pi1 ] , the cover @xmath607 is classified by the homomorphism @xmath609 we now proceed to the other side of diagram . given an element @xmath610 , note that @xmath611 .
hence , we may define a map @xmath612 by @xmath613 .
plainly , @xmath614 therefore , the pullback of @xmath600 along @xmath591 coincides with the cover @xmath607 .
now , by lemma [ lem : h1ofa ] , the homomorphism @xmath615 is given by @xmath616 .
thus , the cover @xmath607 is classified by the homomorphism @xmath617 comparing the answers obtained in and , we find that @xmath618 , and we are done .
we continue our discussion of the milnor fibration with a study of its homology groups and monodromy operators . using the interpretation of the milnor fiber of a multi - arrangement as a finite cyclic cover of the projectivized complement
, we may compute the homology groups of the milnor fiber and the characteristic polynomial of the algebraic monodromy in each degree , at least over a field of characteristic not dividing the order of the cover .
[ thm : hmf ] let @xmath27 be a multi - arrangement . set @xmath470 , and let @xmath619 be the homomorphism given by @xmath620
. if @xmath129 is a coefficient field of characteristic not dividing @xmath29 , then 1 .
the homology groups of the milnor fiber @xmath479 are given by @xmath621 2 .
the characteristic polynomial of the algebraic monodromy of the milnor fibration , @xmath622 , is given by @xmath623 by theorem [ thm : mf ] , the restriction of the hopf fibration to the milnor fiber , @xmath624 , is a regular , @xmath582-cover classified by the homomorphism @xmath625 .
the desired conclusions follow from theorem [ thm : betti cover ] .
[ rem:1evalue ] the @xmath0-eigenspace of the linear transformation @xmath626 acting on the @xmath129-vector space @xmath627 is isomorphic to @xmath628 , which has dimension @xmath629 , independent of @xmath129 . in particular , @xmath626 is the identity on @xmath627 if and only if @xmath630 . the other eigenvalues of @xmath626 , and
the dimensions of their eigenspaces , depend in principle on both the multiplicity vector @xmath31 and the characteristic of @xmath129 .
[ rem : primitive ] in the case when all multiplicities @xmath438 are equal to @xmath0 , we can say more : in the product , only non - primitive roots of @xmath0 appear , provided @xmath266 and @xmath631 , see ( * ? ? ?
* corollary 2.2 ) . for more refined versions of this result
, we refer to ( * ? ? ?
* proposition 2.1 ) and ( * ? ? ?
* theorem 3.13 ) .
[ ex : mf pencil ] let @xmath83 be the pencil of @xmath84 lines through the origin of @xmath85 from examples [ ex : pencil ] and [ ex : cv free ] . we know that the milnor fiber @xmath632 is an @xmath29-fold cover of @xmath633 , where @xmath29 is the sum of the multiplicities .
we also know that @xmath632 is a riemann surface with @xmath29 punctures .
a standard euler characteristic argument now shows that the genus of this surface is @xmath634 ; thus , @xmath635 for all fields @xmath129 . alternatively , we know from that @xmath636 and @xmath637 .
thus , we may recover the calculation from by means of formula , at least in the case when @xmath638 . in view of theorem [ thm : v1a ] , the above result yields a rather explicit formula for the first betti number of the milnor fiber @xmath639 , and for the characteristic polynomial @xmath640 of its algebraic monodromy , @xmath641 .
let @xmath642 be the @xmath406-th cyclotomic polynomial , and let @xmath643 be its degree .
[ cor : b1 mf ] let @xmath1 be an arrangement of @xmath17 hyperplanes , and let @xmath639 be its milnor fiber . then , @xmath644 where @xmath645 is the diagonal " character , given by @xmath646 , for all @xmath9 , and @xmath647 .
in particular , @xmath648 [ rem : trivial ] in formula , only the essential components of @xmath385 may contribute to the sum .
indeed , if a component lies on a ( proper ) coordinate subtorus @xmath649 , then the diagonal subtorus , @xmath650 , intersects @xmath651 only at the origin . in particular , local components , or , more generally , components arising from multinets supported on proper sub - arrangements of @xmath1 , do _ not _ produce jumps in the first betti number of @xmath639
on the other hand , if @xmath27 is a multi - arrangement with some @xmath213 , then even the non - essential components in the characteristic varieties of @xmath76 may produce jumps in the betti numbers of @xmath652 , according to formula .
[ rem : cdo ] in ( * ? ? ?
* theorem 13 ) , cohen , dimca , and orlik give some nice combinatorial upper bounds on the exponents of the cyclotomic polynomials appearing in formula . for more on this , we refer to @xcite .
we now illustrate with a few examples the range of applicability of theorem [ thm : hmf ] ( see @xcite for more computations of this sort . )
[ ex : mf braid arr ] let @xmath1 be the braid arrangement from examples [ ex : braid arr ] , [ ex : braid ] , and [ ex : cv braid ] . we know that @xmath385 has a single essential component , namely , the subtorus @xmath430
clearly , @xmath653 , yet @xmath654 ; hence , @xmath655 .
[ ex : cv mf b3 ] let @xmath1 be the reflection arrangement of type @xmath267 from examples [ ex : b3 ] and [ ex : b3 bis ] .
we know that @xmath385 has an essential component , corresponding to the @xmath205-multinet depicted in figure [ fig : b3 arr ] .
it is readily verified that this is the only such component , and that the diagonal subtorus intersects this component only at the origin .
hence , @xmath656 .
[ ex : ceva bis ] let @xmath1 be the ceva(@xmath40 ) arrangement from example [ ex : ceva3 ] .
the @xmath252-net depicted in figure [ fig : ceva3 ] defines a rational map , @xmath418 , @xmath657 , which restricts to a pencil @xmath658 .
let @xmath659 be the essential @xmath150-dimensional component of @xmath385 obtained by pullback along this pencil .
it is readily verified that the subgroup generated by the diagonal character @xmath22 intersects @xmath660 in two points , both lying on @xmath659 , and both of order @xmath40 .
hence , @xmath661 .
as the next example shows , the first betti number of the milnor fiber of an arrangement in @xmath257 does not only depend on the number and type of multiple points of @xmath662 , but also on their relative position .
[ ex : pappus ] consider the arrangements @xmath663 and @xmath664 defined by the polynomials @xmath665 the arrangement @xmath663 is a realization of the pappus configuration @xmath666 , while @xmath664 is a realization of the configuration @xmath667 .
each projective configurations has @xmath261 double points and @xmath261 triple points ; additionally , the first configuration supports a @xmath252-net , but the second one supports no essential multinet . applying formula , we find that @xmath668 and @xmath669 , as in @xcite . such examples , and many others from the literature ( see for instance @xcite for some recent progress in this direction ) raise the following well - known question , dating from the 1980s .
[ quest : mf comb ] given an arrangement @xmath1 , are the betti numbers of the milnor fiber @xmath639 and the characteristic polynomial of the algebraic monodromy determined by the intersection lattice @xmath6 ? a long - standing question , raised by randell and dimca nmethi among others
, asks whether the milnor fiber of an arrangement can have non - trivial torsion in homology .
examples from @xcite first showed that @xmath670 may have torsion , for suitable multi - arrangements @xmath27 . in @xcite ,
these examples were recast in a more general framework , leading to examples of high - dimensional hyperplane arrangements @xmath1 for which @xmath671 has torsion , for some @xmath672 .
let us sketch here this circle of ideas .
[ thm : milnor multi ] suppose @xmath1 admits a pointed multinet , with distinguished hyperplane @xmath11 and multiplicity @xmath31 .
let @xmath284 be a prime dividing @xmath438 .
there is then a choice of multiplicities @xmath673 on the deletion @xmath674 such that @xmath675 has non - zero @xmath284-torsion .
( sketch ) by proposition [ prop : del ] , the variety @xmath436 has a component of the form @xmath453 , where @xmath659 is a @xmath0-dimensional subtorus , and @xmath676 is a torsion character , of order divisible by @xmath284
. using this fact , and some further machinery , it is shown in ( * ? ? ?
* theorem 6.1 ) , that , for all sufficiently large integers @xmath677 not divisible by @xmath284 , there exists a regular , @xmath406-fold cyclic cover @xmath678 such that @xmath679 has @xmath284-torsion . on the other hand , it is also shown in ( * ? ? ?
* proposition 6.7 ) that any finite cyclic cover of an arrangement complement is dominated by a milnor fiber corresponding to a suitable choice of multiplicities .
hence , there exists a choice of multiplicities @xmath673 for which the cover @xmath680 factors through @xmath545 , and @xmath284 does not divide @xmath681 .
a standard transfer argument then shows that @xmath675 also has @xmath284-torsion .
[ ex : mf deleted b3 ] let @xmath289 be the deletion of the @xmath267 arrangement from example [ ex : deleted b3 ] . proceeding as above , let @xmath682 and consider the @xmath40-fold cover @xmath678 classified by the vector @xmath683 . choose @xmath684 as a positive integer lift of @xmath685 .
then @xmath686 , and @xmath687 factors through @xmath545 ; thus , there is @xmath150-torsion in the first homology of @xmath687 .
explicit calculation shows that , in fact , @xmath688 and @xmath689 .
this leads to a couple of natural questions ( see also @xcite ) .
[ prob : tors mf ] is there a hyperplane arrangement @xmath1 ( without multiplicities ) such that @xmath690 has non - trivial torsion ? [ prob : tors comb ] given a hyperplane arrangement @xmath1 , is the torsion in @xmath691 determined by the intersection lattice @xmath6 ?
given a multi - arrangement @xmath27 , we define in @xcite a new hyperplane arrangement , @xmath692 , called the _ polarization _ of the multi - arrangement .
the rank of @xmath106 equals @xmath693 , while the number of hyperplanes in @xmath106 equals @xmath694 .
a crucial point of this construction is the connection between the milnor fiber of the ( simple ) arrangement @xmath106 and the milnor fiber of the multi - arrangement @xmath27 : the pullback of the cover @xmath695 along the canonical inclusion @xmath696 is equivalent to the cover @xmath697 . using this fact , together with theorem [ thm : milnor multi ] ,
the following result is proved in @xcite .
[ thm : polar tors ] suppose @xmath1 admits a pointed multinet , with distinguished hyperplane @xmath11 and multiplicity @xmath31 .
let @xmath284 be a prime dividing @xmath438 .
there is then a choice of multiplicities @xmath673 on the deletion @xmath674 such that @xmath698 has @xmath284-torsion , where @xmath699 and @xmath700 .
[ ex : polar delb3 ] let @xmath289 be the deleted @xmath267 arrangement from examples [ ex : deleted b3 ] and [ ex : mf deleted b3 ]
. then the choice of multiplicities @xmath701 produces an arrangement @xmath702 of @xmath703 hyperplanes in @xmath704 , such that @xmath705 has @xmath150-torsion of rank @xmath706 .
very little is known about the homology with coefficients in rank @xmath0 local systems of the milnor fiber of a ( multi- ) arrangement @xmath27 . since @xmath479 is a smooth , quasi - projective variety , deep theorems of arapura @xcite and budur
wang @xcite guarantee that the characteristic varieties @xmath707 are unions of torsion - translated subtori .
let us analyze in more detail these varieties , following the approach of dimca and papadima from @xcite . as in [ subsec
: mf cover ] , let @xmath708 be the restriction of the hopf fibration to the milnor fiber , and let @xmath709 be the monodromy of the milnor fibration . since @xmath591 is a finite , regular cover , proposition [ prop : jumpcov ] guarantees that @xmath710 and @xmath711 , for all @xmath712 and @xmath136 . if @xmath626 acts as the identity on @xmath713 , more can be said . indeed , corollary
[ cor : trivnom ] implies that @xmath714 is an isomorphism , for all @xmath136 .
furthermore , by theorem [ thm : tcone ] , the map @xmath715 restricts to an isomorphism between the components through the identity of @xmath716 and @xmath717 .
let us restrict now our attention to the usual milnor fiber , @xmath639 , and its degree @xmath0 characteristic varieties , @xmath718 .
in general , the inclusion @xmath719 is strict .
for instance , suppose @xmath1 admits a non - trivial , reduced multinet @xmath154 , and let @xmath720 be the corresponding component of @xmath385 .
it is then shown in @xcite that @xmath721 has a component passing through the identity and containing @xmath722 as a proper subset .
[ ex : cv mf braid ] let @xmath1 be the braid arrangement from figure [ fig : braid ] . recall from examples [ ex : cv braid ] and [ ex : mf braid arr ] that @xmath385 has four local components , @xmath723 , corresponding to the triple points , and an essential component @xmath659 , corresponding to a @xmath201-net . by the above discussion
, the characteristic variety @xmath724 has @xmath150-dimensional components @xmath725 , as well as a component @xmath726 strictly containing @xmath727 . as noted in @xcite
, @xmath726 is a @xmath212-dimensional algebraic subtorus .
direct computation shows that @xmath721 has no other irreducible components .
[ ex : hesse mf cv ] let @xmath1 be the hessian arrangement from example [ ex : hesse ] . we know that @xmath385 has @xmath465 components passing through the origin , all @xmath40-dimensional : @xmath261 of those are local components , while the last one , @xmath659 , corresponds to the @xmath215-net from figure [ fig : hessian ] . as noted in @xcite , the component @xmath726 of @xmath721 containing @xmath727 has dimension @xmath261 . in particular
, this shows that @xmath728 .
[ quest : mf cv ] find a more precise description of the characteristic varieties @xmath729 , and , more generally , @xmath730 .
the following question was raised in @xcite , in a more general context : is the milnor fiber of a hyperplane arrangement always formal ?
of course , if @xmath731 , then @xmath639 has the homotopy type of a wedge of circles , and so it is formal . in general , though , the answer is no , as illustrated by the following example from @xcite .
[ ex : zuber ] let @xmath1 be the ceva(@xmath40 ) arrangement from example [ ex : ceva3 ] , and let @xmath659 be the @xmath150-dimensional , essential component of @xmath385 described in example [ ex : ceva bis ] .
the pullback @xmath727 is a @xmath212-dimensional subtorus of @xmath732 , of the form @xmath733 , for some linear subspace @xmath734 . using the mixed hodge structure on the cohomology of the milnor fiber
, zuber @xcite shows that @xmath735 can not possibly be a component of the resonance variety @xmath736 .
hence , the tangent cone formula from theorem [ thm : tcone ] is violated , and thus , the space @xmath639 can not be @xmath0-formal .
this example raises several questions , all interrelated .
[ quest : mf tc ] find a concrete description of the resonance varieties @xmath737 , and , more generally , @xmath738 . for which arrangements @xmath1 does the tangent cone formula @xmath739 hold ?
[ quest : mf nonf ] give a purely topological explanation for the non - formality of the milnor fibers of arrangements such as the ceva(@xmath40 ) arrangement .
such an explanation could involve either an explicit computation of the resonance varieties of milnor fibers ( as in problem [ quest : mf tc ] ) , or the study of triple massey products in the cohomology of the milnor fiber , or perhaps some completely new method .
[ quest : mf formal ] given a multi - arrangement @xmath27 , determine whether the milnor fiber @xmath652 is formal or not .
does this formality property depend only on the underlying arrangement @xmath1 , or also on the multiplicity vector @xmath31 ?
the boundary manifold of a hyperplane arrangement @xmath1 in @xmath2 is the boundary of a regular neighborhood in @xmath75 of the union of the projective hyperplanes comprising @xmath662 . in this section ,
we survey a number of known results ( primarily from @xcite ) regarding the cohomology ring , fundamental group , jump loci , and formality properties of boundary manifolds of arrangements .
let @xmath1 be a ( central ) arrangement of hyperplanes in @xmath2 ( @xmath740 ) with union @xmath741 and complement @xmath742 .
likewise , let @xmath743 be the corresponding arrangement in @xmath75 , with union @xmath744 and complement @xmath18 .
a regular neighborhood @xmath745 of the algebraic hypersurface @xmath746 may be constructed as follows .
let @xmath747 be the smooth function defined by @xmath748 ) = { \left| q(z ) \right|}^2 / { |\!|z|\!|}^{2n}$ ] , where @xmath749 is a defining polynomial for the arrangement , and @xmath13 .
then , for sufficiently small @xmath750 , the preimage @xmath751)$ ] is a closed , regular neighborhood of @xmath726 .
( since @xmath749 is homogeneous , we may simply take @xmath752 . )
alternatively , one may triangulate @xmath75 with @xmath726 as a subcomplex , and take @xmath745 to be the closed star of @xmath726 in the second barycentric subdivision . as shown by durfee @xcite
, these constructions yield isotopic neighborhoods , independent of the choices made . evidently , @xmath745 is a compact , orientable , smooth manifold with boundary , of dimension @xmath478 ; moreover , @xmath745 deform - retracts onto @xmath726 .
the _ exterior _ of the projectivized arrangement , denoted @xmath32 , is the complement in @xmath75 of the open regular neighborhood @xmath753 .
clearly , @xmath32 is a compact , connected , orientable , smooth @xmath478-manifold with boundary .
moreover , @xmath24 deform - retracts onto @xmath32 , and thus @xmath754 .
[ def : bdry mfd ] the _ boundary manifold _ of a hyperplane arrangement @xmath1 in @xmath2 is the common boundary @xmath755 of the exterior @xmath32 and the regular neighborhood @xmath745 defined above .
note that @xmath34 is a compact , orientable , smooth manifold ( without boundary ) , of dimension @xmath756 .
the inclusion map @xmath757 is a @xmath758-equivalence , see ( * ? ? ?
* proposition 2.31 ) ; in particular , @xmath759 for @xmath760 . if @xmath761 , then @xmath662 consists of @xmath17 distinct points in @xmath88 ; thus , the boundary manifold @xmath34 consists of @xmath17 disjoint small circles around those points , and there is not much else to say .
consequently , we will assume from now on that @xmath762 , in which case @xmath34 is connected .
let us illustrate these definitions with a couple of examples , extracted from @xcite .
[ ex : bdry pencil ] let @xmath1 be a pencil of @xmath17 hyperplanes in @xmath2 , defined by the polynomial @xmath763 . if @xmath764 , then @xmath765 and @xmath766 .
otherwise , @xmath24 may be realized as the complement of @xmath767 parallel hyperplanes in @xmath768 . in this case ,
@xmath769 and thus @xmath34 is diffeomorphic to the connected sum @xmath770 .
[ ex : boundary near - pencil ] let @xmath1 be a near - pencil of @xmath17 planes in @xmath771 , defined by the polynomial @xmath772 . in this case ,
@xmath773 , where @xmath774 denotes the orientable surface of genus @xmath775 ( see also example [ ex : bd mf near - pencil ] ) .
as shown in @xcite , the long exact sequence of the pair @xmath776 breaks into split , short exact sequences , @xmath777 & h_{q+1}({{\overline{u}}},{{\partial{{{\overline{u}}}}}},{\mathbb{z}})\ar[r ] & h_q({{\partial{{{\overline{u}}}}}},{\mathbb{z } } ) \ar^{j_*}[r ] & h_q({{\overline{u}}},{\mathbb{z } } ) \ar[r ] & 0}.\ ] ] by lefschetz duality , @xmath778 .
since @xmath754 , and since the homology groups of @xmath24 are torsion - free , we obtain a direct sum decomposition @xmath779 hence , @xmath780 .
the cohomology groups of @xmath34 admit a decomposition similar to the one from . to describe the cup - product structure in @xmath35 , we first need to review some notions .
let @xmath781 be a graded , finite - dimensional algebra over @xmath94 .
assume that @xmath781 is graded - commutative , of finite type ( i.e. , each graded piece @xmath782 is a finitely generated @xmath94-module ) , and connected ( i.e. , @xmath783 ) .
then , the @xmath94-dual @xmath784 is an @xmath781-bimodule , with left and right multiplication given by @xmath785 and @xmath786 , respectively .
moreover , this bimodule structure is compatible with the gradings : if @xmath787 and @xmath788 , then both @xmath789 and @xmath790 belong to @xmath791 .
[ theo : coho double ] let @xmath1 be an arrangement in @xmath2 , let @xmath792 be the cohomology ring of the projectivized complement , and let @xmath793 be the cohomology ring of the boundary manifold .
then @xmath794 , with multiplication given by @xmath795 , and grading @xmath796 .
consequently , the cohomology ring of the boundary manifold depends only on the intersection lattice of the arrangement . in the case
when @xmath37 , the structure of this ring can be described in more concrete terms , as follows .
write @xmath797 , and fix ordered bases , @xmath798 for @xmath110 and @xmath799 for @xmath800
. the multiplication map , @xmath801 , is then given by @xmath802 , for some integer coefficients satisfying @xmath803 .
now write @xmath804 , and pick dual bases @xmath805 for @xmath806 and @xmath807 for @xmath808 .
the multiplication map @xmath809 restricts to @xmath810 on @xmath811 , vanishes on @xmath812 , while on @xmath813 , it is given by @xmath814 . finally , @xmath815 , where @xmath816 is the generator of @xmath817 dual to @xmath818 .
the resonance varieties of the boundary manifold of an arrangement @xmath1 in @xmath2 were studied in detail in @xcite . if @xmath819 , then @xmath820 , and the resonance varieties of @xmath34 can be expressed solely in terms of the resonance varieties of @xmath24 : @xmath821 \bigcup_{i+j = s } \bigl({\mathcal{r}}^{d-1}_{i}(u ) \cap { \mathcal{r}}^{d}_{j}(u)\bigr ) & \text{if $ q = d-1 $ or $ q = d$ , } \\[3pt ] { \mathcal{r}}^{2d - k-1}_{s}(u ) & \text{if $ q \ge d+1$. } \end{cases}\ ] ] now suppose @xmath1 is an arrangement of @xmath17 planes in @xmath257 , and let us consider the resonance varieties @xmath822 of its boundary manifold . for depth
@xmath823 , these varieties admit a particularly simple description : @xmath824 the higher - depth resonance varieties , though , can be much more complicated , as the following example from ( * ? ? ?
* corollary 6.11 ) illustrates .
[ ex : res lines ] let @xmath1 be an arrangement of @xmath84 planes in @xmath257 , in general position .
if @xmath825 and @xmath826 , then the resonance variety @xmath822 is the zero locus of the pfaffians ( of appropriate size ) of a generic @xmath827 skew - symmetric matrix . in particular
, @xmath822 is a singular , irreducible variety . as noted in
[ subsec : bdry nbhd ] , if @xmath1 is an arrangement in @xmath2 , with @xmath828 , then @xmath829 .
thus , from the point of view of the fundamental group of the boundary manifold , the most interesting dimension to study is @xmath830 .
so assume for the rest of this section that @xmath1 is an arrangement of planes in @xmath257 .
in this case , the boundary manifold admits another interpretation , that arises in the work of jiang and yau @xcite , hironaka @xcite , and westlund @xcite .
let @xmath831 be the projectivized line arrangement in @xmath832 .
suppose @xmath662 has @xmath406 non - transverse intersection points , i.e. , points @xmath833 , of multiplicity @xmath834 .
blowing up @xmath832 at each of these points , we obtain an arrangement @xmath835 in the rational surface @xmath836 , consisting of the proper transforms of the lines of @xmath662 , together with the exceptional lines arising from the blow - ups .
this construction realizes the boundary manifold @xmath34 as a graph manifold , in the sense of waldhausen .
the underlying graph @xmath837 has vertex set @xmath838 , with vertices in one - to - one correspondence with the lines of @xmath839 : the vertex corresponding to @xmath840 is labeled @xmath841 , while the vertex corresponding to @xmath842 is labeled @xmath843 .
the edge set @xmath844 consists of an edge @xmath845 from @xmath841 to @xmath846 if @xmath847 and the corresponding lines @xmath840 and @xmath848 are transverse , and an edge @xmath849 from @xmath843 to @xmath841 if @xmath850 .
each vertex gets assigned a weight , equal to the self - intersection number of the corresponding line in the blow - up : @xmath841 has weight @xmath851 , and @xmath843 has weight @xmath852 . applying a method due to hirzebruch @xcite to the graph manifold structure described above
, westlund @xcite obtained a presentation for the fundamental group of the boundary manifold of an arrangement in @xmath257 , as follows .
let @xmath853 be the weighted graph associated to @xmath662 , and choose an orientation on this graph .
pick a maximal tree @xmath854 , and list the remaining edges as @xmath855 , where @xmath153 is the number of linearly independent cycles in @xmath853 .
with these choices , the fundamental group of @xmath856 has presentation @xmath857 y_1,\dots , y_s \end{array } \bigg| \begin{array}{ll}[x_i , x_j^{u_{ij } } ] , & ( i , j)\in{\mathsf{e}}\\ [ 2pt ] \prod_{j=1}^{n+r } x_j^{u_{ij } } , & 1\le i\le n \end{array } \right\rangle,\ ] ] where @xmath858 here @xmath859=aba^{-1}b^{-1}$ ] , @xmath860 is the identity element , and @xmath861 for @xmath862 .
note that if @xmath863 and @xmath864 , then @xmath865 .
[ prop : comm ] the presentation may be simplified to a commutator - relators presentation for @xmath866 , with generators @xmath867 and @xmath868 as above .
instead of reproducing the precise set of relations here , let us first make a remark , and then illustrate with a simple example .
[ rem : bdry map ] recall from that the inclusion map @xmath757 induces a split surjection in first homology , leading to the direct sum decomposition @xmath869 .
the group @xmath870 is freely generated by homology classes @xmath871 corresponding to the meridians around the first @xmath767 lines . on the other hand ,
the group @xmath872 is isomorphic to @xmath873 , and thus admits a basis @xmath868 corresponding to disks in the exterior whose boundaries are the chosen cycles in the graph . with respect to these generating sets ,
then , the abelianization map @xmath874 takes @xmath875 to @xmath875 and @xmath876 to @xmath876 , while the induced homomorphism @xmath877 takes @xmath875 to @xmath875 and @xmath876 to @xmath151 .
( -0.5,3 ) ( 2.5,-3 ) ; ( 0.5,3 ) ( -2.5,-3 ) ; ( -3,-2 ) ( 5,-2 ) ; ( -2.2,1.3 ) ( 5,-2.6 ) ; at ( -3.5,-2 ) @xmath878 ; at ( -2.6,1.5 ) @xmath879 ; at ( -0.75,3.4 ) @xmath880 ; at ( 0.8,3.4 ) @xmath881 ; ( 0,0 ) ( 2,3.4641 ) ( 4,0 ) cycle ; ( 0,0 ) ( 2,1.1547 ) ; ( 2,3.4641 ) ( 2,1.1547 ) ; ( 4,0 ) ( 2,1.1547 ) ; ( 0,0 ) circle ( 3.2pt ) ; ( 2,3.4641 ) circle ( 3.2pt ) ; ( 4,0 ) circle ( 3.2pt ) ; ( 2,1.1547 ) circle ( 3.2pt ) ; at ( -0.3,-0.3 ) @xmath882 ; at ( 2,3.8 ) @xmath883 ; at ( 4.3,-0.3 ) @xmath884 ; at ( 2.4,1.3 ) @xmath885 ; at ( 2,-0.3 ) @xmath886 ; at ( 0.7,1.8 ) @xmath887 ; at ( 3.3,1.8 ) @xmath888 ; [ ex : general position ] let @xmath1 be an arrangement of @xmath212 planes in @xmath257 in general position . in this case , @xmath837 is the complete graph @xmath889 . using the maximal tree indicated by dashed edges in figure [ fig:4genlines ]
, we obtain the following presentation for the fundamental group of the boundary manifold of @xmath1 : @xmath890 x_1,x_2,x_3 \\[2.5pt ] y_{1},y_{2},y_{3 } \end{array } \bigg| \begin{array}{ll}x_1x_2x_3=x_1x_{2}^{y_{1}^ { } } x_{3}^{y_{2}^ { } } = x_2 x_1^{y_{3}^{-1 } } x_{3}^{y_{1}^ { } } = x_3 x_1^{y_{2}^{-1 } } x_{2}^{y_{1}^{-1 } } \\ [ 3pt ] [ x_1^{},x_2^{y_{3 } } ] = [ x_1^{},x_3^{y_{2 } } ] = [ x_2^{},x_3^{y_{1}}]=1 \end{array } \right\rangle.\ ] ] the next result expresses the alexander polynomial of the boundary manifold in terms of the underlying graph structure .
[ thm : alex poly arr ] let @xmath1 be an essential arrangement of planes in @xmath257 , and let @xmath853 be the associated graph .
then the alexander polynomial of the boundary manifold @xmath34 is given by @xmath891 where @xmath892 denotes the degree of the vertex @xmath893 , and @xmath894 .
note that @xmath895 . in view of proposition
[ prop : v1 3 m ] , we obtain the following decomposition of the first characteristic variety of @xmath34 into irreducible components : @xmath896 in particular , every component of @xmath897 is a codimension @xmath0 algebraic subtorus of the character torus of @xmath34 .
as we saw in examples [ ex : bdry pencil ] and [ ex : boundary near - pencil ] , for simple arrangements such as pencils or near - pencils , the boundary manifold is built out of spheres by successive product and connected sum operations , and thus is formal .
on the other hand , consider the following , equally simple , example .
[ ex : tc lines ] let @xmath1 be an arrangement of @xmath452 planes in @xmath257 , in general position . from the discussion in example [ ex : res lines ]
, it follows that @xmath898 is isomorphic to @xmath899 , an irreducible quadric with an isolated singular point at @xmath151 . by theorem [ thm : tcone ] , then , the manifold @xmath34 is not @xmath0-formal .
alternatively , formula implies that @xmath900 , for all @xmath136 .
thus , @xmath901 , showing again that @xmath34 is not @xmath0-formal .
the general situation was elucidated in ( * ? ? ?
* theorem 9.7 ) .
[ thm : bdry ] let @xmath1 be an arrangement of planes in @xmath257 , and let @xmath34 be the corresponding boundary manifold .
the following are equivalent : 1 .
[ f1 ] the manifold @xmath34 is formal .
[ f2 ] the group @xmath866 is @xmath0-formal .
[ f3 ] @xmath902 .
[ f4 ] @xmath1 is either a pencil or a near - pencil .
[ f5 ] @xmath34 is either @xmath903 or @xmath904 , where @xmath905 .
implications @xmath906 @xmath906 are true in general ( see appendix [ sect : formal ] ) , implication @xmath906 was discussed in examples [ ex : bdry pencil ] and [ ex : boundary near - pencil ] , while @xmath906 is obvious . to prove @xmath906 , suppose @xmath1 is neither a pencil , nor a near - pencil .
then , by formula , @xmath907 .
on the other hand , formula shows that @xmath897 is a union of codimension @xmath0 subtori in @xmath908 .
thus , @xmath909 , and we are done .
in this section , we study the boundary of the closed milnor fiber of a hyperplane arrangement , and the way it relates to the boundary manifold of the arrangement . as usual
, let @xmath1 be a ( central ) arrangement of hyperplanes in @xmath2 ( @xmath740 ) , with defining polynomial @xmath410 .
let @xmath910 be the union of the hyperplanes in @xmath1 , and let @xmath92 be its complement .
now let @xmath31 be a ( primitive ) multiplicity vector for @xmath1 , and let @xmath911 . as mentioned in
[ sect : mf arr ] , the polynomial map @xmath912 restricts to a smooth fibration , @xmath913 , with fiber @xmath914 .
intersecting the global milnor fiber with a ball in @xmath2 of large enough radius , we obtain a compact , smooth , orientable @xmath478-dimensional manifold with boundary , @xmath915 which we call the _ closed milnor fiber _ of the multi - arrangement .
clearly , @xmath652 deform - retracts onto @xmath916 ; hence , @xmath917 .
the _ boundary of the milnor fiber _ of @xmath27 is the compact , smooth , orientable , @xmath33-dimensional manifold @xmath918 as usual , we will abbreviate @xmath919 and @xmath920 when all multiplicities @xmath438 are equal to @xmath0 , and will drop @xmath1 from the notation when the arrangement is understood . as noted in ( * ? ? ? * proposition 2.4 ) , the pair @xmath921 is @xmath758-connected . in particular , if @xmath922 , the boundary of the milnor fiber is connected , and the inclusion - induced homomorphism @xmath923 is surjective .
returning to the general situation , let @xmath924 be the global milnor fiber of a multi - arrangement .
recall that the monodromy of the fibration @xmath925 is the diffeomorphism @xmath926 given by @xmath581 , where @xmath470 .
clearly , the map @xmath506 restricts to a diffeomorphism @xmath927 .
let @xmath928 be the mapping torus of this diffeomorphism ; we then have a smooth fibration @xmath929 & y_m({{\mathcal{a } } } ) \ar[r ] & s^1}.\ ] ] now let @xmath930 be the link of the singularity , and let @xmath931 be a closed , regular neighborhood of @xmath932 inside the sphere . then , as shown by milnor in @xcite , there is a smooth fibration @xmath933 & s^{2d+1}\setminus { { \stackrel{\circ}{\nu}}}(k ) \ar^(.62){q_m/{\left| q_m \right|}}[rr ] & & s^1}.\ ] ] furthermore , the local milnor fibration is equivalent , through a fiber - preserving diffeomorphism , to the fibration .
( 0,0 ) circle ( 3 ) ; ( -3,0 ) ( 3,0 ) ; ( 0,-3 ) ( 0,3 ) ; ( 0,0 ) circle ( 2.4 ) ; plot ( , 1/ ) ; plot ( , 1/ ) ; ( 0,0 ) circle(2.4 ) ( 1.8,0 ) circle ( 2.4 ) ; ( 0,0 ) circle ( 2.4 ) ( 1.8,0 ) circle(2.4 ) ; ( 0,0)circle(2.4 ) ; ( -63:2.4 ) arc ( -63:-73:2.4 ) ; ( 1.8,0)circle(2.4 ) ; ( 1.8,0)+(107:2.4 ) arc ( 107:117:2.4 ) ; ( 63:2.4 ) arc ( 63:73:2.4 ) ; [ ex : seifert ] let @xmath1 be the arrangement in @xmath85 defined by the polynomial @xmath934 .
then @xmath919 is a seifert surface for the @xmath17-component hopf link .
this is a compact , connected , orientable surface of genus @xmath935 , with @xmath17 open disks removed ; its boundary , @xmath920 , consists of @xmath17 disjoint circles . for @xmath936 ,
the surface is the twisted cylinder depicted in figure [ fig : mf local ] , and the monodromy is a dehn twist about the core of the cylinder .
( although this diffeomorphism is isotopic to the identity , no such isotopy can be the identity on the two boundary circles . )
there are several ways in which the closed milnor fiber and its boundary are related to the boundary manifold of an arrangement .
the next proposition ( whose proof is adapted from the discussion in @xcite ) details one such relationship , in the special case when the defining polynomial splits off a linear factor not involving the other variables .
first recall a basic notion : the _ double _ of a manifold @xmath726 with non - empty boundary @xmath937 is the ( closed ) manifold @xmath938)= w\cup_{\partial w } w$ ] .
[ prop : double mf ] let @xmath1 be a hyperplane arrangement in @xmath2 , defined by a polynomial of the form @xmath939 .
let @xmath940 be the arrangement in @xmath941 defined by @xmath942 , and let @xmath943 be the corresponding milnor fibration .
then , the boundary manifold of @xmath1 is diffeomorphic to the double of @xmath944 ; moreover , @xmath856 fibers over the circle , with fiber the double of @xmath945 .
let @xmath946 be the subvariety of @xmath941 defined by @xmath942 .
its projective closure , @xmath947 , is the zero - set of @xmath749 ; moreover , @xmath948 .
forming the union of a regular neighborhood @xmath949 with a tubular neighborhood of the hyperplane at infinity , , and rounding off corners , we obtain a regular neighborhood @xmath950 .
clearly , @xmath951 is diffeomorphic to @xmath952 .
hence , @xmath953 by the discussion from [ subsec : alt mf ] , each of the two sides in the above decomposition is diffeomorphic to @xmath944 , and the gluing is done along their common boundary , @xmath954 .
thus , @xmath34 is the double of @xmath944 .
the last assertion follows at once .
[ ex : boolean 2 ] let @xmath58 be the boolean arrangement in @xmath60 .
from example [ ex : mf boolean ] , we know that @xmath955 . hence , @xmath956)^{n-1}\cong t^{n-1}\times d^{n-1}$ ] , and so @xmath957 ( compare with ( * ? ? ?
* , exercise 1.15 ) ) .
now note that @xmath958 .
we also know from example [ ex : mf boolean ] that the bundle @xmath959 is trivial . applying proposition [ prop : double mf ] , then , shows that @xmath960 .
[ ex : bd mf near - pencil ] let @xmath1 be a near - pencil of @xmath17 planes in @xmath771 , defined by the polynomial @xmath961 , where @xmath962 .
then @xmath944 admits a fibration over the circle ( different from the milnor fibration ) , whose fiber is @xmath963 with @xmath767 open disks removed , and whose monodromy is a dehn twist about the boundary of @xmath963 .
hence , @xmath773 . as noted in (
* example 19.10.7 ) , we also have that @xmath964 .
we now describe a construction that realizes the boundary of the milnor fiber associated to a hyperplane arrangement @xmath1 as a regular , finite cyclic cover of the boundary manifold of the arrangement @xmath1 . as usual ,
set @xmath13 .
recall that the hopf fibration @xmath584 restricts to a ( trivializable ) bundle map , @xmath965 , with fiber @xmath21 , which in turn restricts to a regular @xmath20-cover , @xmath966 .
[ lem : bd cover ] the map @xmath584 restricts to regular , cyclic @xmath17-fold covers , @xmath967 and @xmath968 .
let @xmath969 .
recall that @xmath970 \in { \mathbb{cp}}^{2d } \mid { \left| q(z ) \right|}^2 / { |\!|z|\!|}^{2n}\ge 1\}.\notag\end{aligned}\ ] ] thus , the map @xmath591 restricts to a map of pairs , @xmath971 .
hence , the cover @xmath966 restricts to covers @xmath967 and @xmath968 , and we are done .
note that the inclusion @xmath972 is a fiber - preserving homotopy equivalence . summarizing
, we have a commuting ladder @xmath973 \ar[d ] & { \mathbb{z}}_n \ar@{=}[r ] \ar[d ] & { \mathbb{z}}_n \ar[r ] \ar[d ] & { \mathbb{c}}^ * \ar@{=}[r ] \ar[d ] & { \mathbb{c}}^ * \ar[d ] \\ { { \partial{{{\overline{f}}}}}}\ar^{\pi}[d ] \ar[r ] & { { \overline{f}}}\ar^{\pi}[d ] \ar^{\simeq}[r ] & f \ar^{\pi}[d ] \ar[r ] & m \ar[r ] \ar^{\pi}[d ] & { \mathbb{c}}^{d+1 } \setminus\{0\ } \ar^{\pi}[d ] \\ { { \partial{{{\overline{u}}}}}}\ar[r ] & { { \overline{u}}}\ar^{\simeq}[r ] & u \ar@{=}[r ] & u \ar[r ] & { \mathbb{cp}}^{d } } \ ] ] where the horizontal arrows are inclusions , and the maps denoted by @xmath591 are principal bundles with fiber either @xmath21 or @xmath20 .
assume now that @xmath37 , and let @xmath974 be the associated arrangement of lines in @xmath832 .
by proposition [ prop : comm ] , the group @xmath866 has generators @xmath867 corresponding to the meridians around the first @xmath767 lines , and generators @xmath868 corresponding to the cycles in the associated graph @xmath853 .
[ prop : bdf cover ] the @xmath20-cover @xmath968 is classified by the homomorphism @xmath975 given by @xmath976 and @xmath977 . by lemma [ lem : bd cover ] , the cover @xmath968 is the pullback along the inclusion @xmath757 of the cover @xmath967 . by theorem [ thm : mf ] ,
this latter cover is classified by the homomorphism @xmath978 , @xmath979 . by remark [ rem : bdry map ] , the homomorphism @xmath980 is given by @xmath981 and @xmath977 .
the desired conclusion follows .
[ ex : bmf pencil ] let @xmath1 be a pencil of @xmath84 planes in @xmath257 .
from example [ ex : bdry pencil ] , we know that @xmath982 . since @xmath983 is a cover with @xmath84 sheets , we find that @xmath984 .
hence , by standard @xmath40-manifold topology , @xmath985 .
( see also ( * ? ? ?
* example 19.10.6 ) . ) a detailed study of the boundary of the milnor fiber of a non - isolated surface singularity was done by nmethi and szilrd in @xcite .
when applied to arrangements in @xmath257 , their work yields the following result .
[ thm : charpoly ] let @xmath1 be an arrangement of @xmath17 planes in @xmath257 .
the characteristic polynomial of the algebraic monodromy acting on @xmath41 is given by @xmath986 in particular , the first betti number of @xmath39 is determined solely by the mbius function of @xmath6 :
@xmath987 this shows that the first betti number of the boundary of the milnor fiber is a much less subtle invariant than @xmath988 , in that @xmath989 depends only on the number and type of multiple points of @xmath662 , but not on their relative position .
the next example illustrates this phenomenon .
[ ex : pappus bis ] let @xmath663 and @xmath664 be the two arrangements from example [ ex : pappus ] .
recall that the characteristic polynomials of the monodromy operators acting on the first homology of the milnor fibers of these arrangements are @xmath990 and @xmath669 , respectively .
on the other hand , the monodromy operators acting on the first homology of the boundaries of the respective milnor fibers share the same characteristic polynomial , namely , @xmath991 . nevertheless , there are pairs of arrangements for which the milnor fibers have the same first betti number , but the boundaries of the milnor fibers have different first betti numbers .
let us illustrate this claim with a simple example .
[ ex:4planes ] let @xmath663 be an arrangement of @xmath212 planes in general position and let @xmath664 be a near - pencil of @xmath212 planes .
let @xmath992 and @xmath993 be the respective closed milnor fibers .
the monodromy operators acting on @xmath994 and @xmath995 have characteristic polynomial @xmath996 , yet the monodromy operators acting on @xmath997 and @xmath998 have characteristic polynomials equal to @xmath999 and @xmath1000 , respectively .
next , we determine which milnor fibers of plane arrangements have formal boundaries .
[ prop : bdf formal ] let @xmath1 be an arrangement of planes in @xmath257 , and let @xmath39 be the boundary of its milnor fiber .
the following are equivalent : 1 .
[ bf1 ] the manifold @xmath39 is formal .
[ bf2 ] @xmath1 is either a pencil or a near - pencil .
3 . [ bf3
] @xmath39 is either @xmath1001 or @xmath904 , where @xmath905 .
implication @xmath906 was discussed in examples [ ex : bmf pencil ] and [ ex : bd mf near - pencil ] , while implication @xmath906 is obvious . to prove @xmath906 , suppose @xmath1 is neither a pencil , nor a near - pencil . by theorem [ thm : bdry
] , then , the boundary manifold @xmath34 is not formal . on the other hand , we know from lemma [ lem :
bd cover ] that @xmath39 is a regular , cyclic @xmath17-fold cover of @xmath34 .
hence , by lemma [ lem : kfc ] , the manifold @xmath39 is not formal . in @xcite , nmethi and a. szilrd list a number of open problems regarding the boundary of the milnor fiber of a non - isolated surface singularity .
problem 24.4.19 on the list reads as follows .
[ quest : h1bdf ] find a nice formula for the torsion of @xmath43 .
the next example shows that torsion can indeed occur in the first homology of @xmath39 ( compare with ( * ? ? ?
* example 19.10.9 ) ) .
[ ex : gen 4 ] let @xmath1 be a general position arrangement of @xmath212 planes in @xmath257 .
as noted above , the characteristic polynomial of the algebraic monodromy acting on @xmath41 is given by @xmath1002 .
direct computation shows that , in fact , @xmath1003 . for a generic arrangement of @xmath17 planes in @xmath257
, we expect that @xmath1004 for an arbitrary arrangement @xmath1 , it would be interesting to see whether all the torsion in @xmath1005 consists of @xmath20-summands , where @xmath13 .
the next problem summarizes problems 24.2.1 and 24.2.2 from @xcite .
[ quest : coho bdf ] determine the cohomology ring and the resonance varieties of the boundary of the milnor fiber of an arrangement .
let @xmath1 be an arrangement . from lemma [ lem :
bd cover ] , we know that the hopf fibration restricts to a a regular , finite cover , @xmath968
. proposition [ prop : jumpcov ] , then , insures that @xmath1006 and @xmath1007 , for all @xmath712 and @xmath136
. if @xmath1 is a near - pencil of planes in @xmath257 , then all these inclusions are , in fact , equalities . in general , though , these inclusions are strict . for instance , if @xmath1 is a pencil of @xmath84 planes in @xmath257 , and @xmath1008 , then @xmath1009 , whereas @xmath1010 .
let @xmath232 be a connected cw - complex . without loss of generality
, we may assume @xmath232 has a single @xmath151-cell .
let @xmath1011 be the fundamental group of @xmath232 , based at this @xmath151-cell .
we will assume throughout that @xmath232 has finitely many @xmath0-cells .
of course , these cells must be attached at the unique @xmath151-cell @xmath1012 . by cellular approximation ,
the based homotopy classes of the @xmath0-cells generate the fundamental group ; hence , @xmath314 is a finitely generated group .
now let @xmath129 be an algebraically closed field , and let @xmath1013 be the affine algebraic group of @xmath129-valued , multiplicative characters on @xmath314 .
the _ characteristic varieties _ of @xmath232 over @xmath129 are the jumping loci for homology with coefficients in @xmath129-valued , rank-@xmath0 local systems on @xmath232 : @xmath1014 as long as @xmath232 has finite @xmath569-skeleton , these loci are zariski closed subsets of @xmath1015 ; moreover , we have a descending filtration @xmath1016 as shown in @xcite , the depth-@xmath0 characteristic varieties @xmath1017 behave well under direct products .
more precisely , suppose @xmath1018 and @xmath1019 are two connected , finite - type cw - complexes , with fundamental groups @xmath1020 and @xmath1021 .
identify the character group of @xmath1022 with @xmath1023 .
then , for all @xmath712 , @xmath1024 let @xmath1025 be the cohomology algebra of @xmath232 . if @xmath1026 has @xmath150-torsion , assume that @xmath1027 . then , for each @xmath1028 , we have @xmath1029 , by graded - commutativity of the cup product .
thus , left - multiplication by @xmath102 defines a cochain complex , @xmath1030 & a^1\ar^{a}[r ] & a^2 \ar[r ] & \cdots},\ ] ] the jump loci for the cohomology of this complex define a natural filtration of the affine space @xmath1031 .
the _ resonance varieties _ of @xmath232 are the sets @xmath1032 if @xmath232 has finite @xmath569-skeleton , the sets @xmath1033 form a descending filtration of @xmath110 by homogeneous , zariski closed subsets of @xmath110 .
note that , if @xmath1034 , then @xmath1035 , for all @xmath1036 . in degree @xmath151
, we have @xmath1037 , and @xmath1038 , for @xmath1039 .
the resonance varieties respect field extensions : if @xmath1040 , then @xmath1041 . furthermore ,
as noted in @xcite , the depth-@xmath0 resonance varieties @xmath1042 behave well under direct products : @xmath1043 given a finitely generated group @xmath314 , we may define its degree @xmath0 characteristic and resonance varieties as those of a presentation @xmath150-complex for the group .
it is readily checked that this definition does not depend on a choice of presentation for @xmath314 .
if the group admits a finite presentation , @xmath1044 , we may compute the sets @xmath1045 and @xmath1046 directly from the presentation , by means of the fox calculus .
the algorithm goes as follows .
let @xmath1047 be the free group with generators @xmath1048 , and let @xmath1049 be the augmentation map , given by @xmath1050 . for each @xmath1051
, there is a linear operator @xmath1052 , known as the @xmath572-th fox derivative , uniquely defined by the following rules : @xmath1053 , @xmath1054 , and @xmath1055 .
now let @xmath1056 be the ring morphism obtained by composing the abelianization map @xmath1057 with the presentation homomorphism @xmath1058 , and extending linearly to group rings .
we then define the alexander matrix of the given presentation as the @xmath31 by @xmath17 matrix @xmath1059 with entries @xmath1060 in the group algebra @xmath1061 .
note that @xmath1061 is the coordinate ring of the algebraic group @xmath1062 .
the variety @xmath1045 , then , is the zero locus of the codimension @xmath153 minors of @xmath1063 , at least away from the trivial character , see @xcite .
a somewhat similar interpretation of the resonance varieties @xmath1046 was given in @xcite , at least in the case when @xmath314 admits a commutator - relators presentation .
more precisely , if all the relators @xmath1064 belong to @xmath1065 $ ] , we may define the linearized alexander matrix , @xmath1066 , as the @xmath31 by @xmath17 matrix with entries @xmath1067 in the polynomial ring @xmath1068 $ ] .
of course , @xmath422 may be viewed as the coordinate ring of the affine space @xmath1069 . the variety @xmath1046 , then , is the zero locus of the codimension @xmath153 minors of @xmath1070 .
for instance , if @xmath1071 , then @xmath1072 , for all @xmath1073 .
every group homomorphism @xmath1074 induces a morphism between character groups , @xmath1075 , given by @xmath1076 .
likewise , @xmath1077 induces a homomorphism in cohomology , @xmath1078 .
clearly , if @xmath1077 is surjective , then both @xmath1079 and @xmath1080 are injective .
the next proposition describes a nice functoriality property enjoyed by the characteristic and resonance varieties of groups .
the proposition extends results from @xcite ; the first part of the proof is modeled on the proof of ( * ? ? ?
* lemma 2.13 ) , while the second part of the proof is modeled on the proof of ( * ? ? ? * lemma 5.1 ) .
[ prop : epi cv ] let @xmath314 be a finitely generated group , and let @xmath1081 be a surjective homomorphism .
then , for each ground field @xmath129 , and each @xmath136 , the following hold . 1 . [ epi1 ]
the induced morphism between character groups , @xmath1082 , restricts to an embedding @xmath1083 .
[ epi2 ] the induced morphism between cohomology groups , @xmath1084 , restricts to an embedding @xmath1085 . for the first part ,
let @xmath1086 be a character .
the @xmath452-term exact sequence associated to the extension @xmath1087 and the @xmath1088-module @xmath1089 ends in @xmath1090 , where @xmath1091 is the module of coinvariants under the action of @xmath932 .
clearly , @xmath1092 is isomorphic as a @xmath1093-module to @xmath1094 .
hence , @xmath1095 is bounded below by @xmath1096 .
thus , if @xmath1097 , then @xmath1098 , and we are done . for the second part ,
consider the commuting diagram @xmath1099^(.6){\mu_q } \ar@{^{(}->}[d]^{\varphi^1\wedge \varphi^1 } & h^2(q,{\bbbk } ) \ar[d]^{\varphi^2 } \\
h^1(g,{\bbbk } ) \wedge h^1(g,{\bbbk } ) \ar[r]^(.6){\mu_g } & h^2(g,{\bbbk } ) } , \ ] ] where @xmath1100 and @xmath1101 are the respective cup - product maps .
let @xmath102 be a nonzero element in @xmath1102 . by definition
, @xmath102 belongs to @xmath1103 if there exist linearly independent elements @xmath1104 such that @xmath1105 in @xmath1106 , yet @xmath1107 .
clearly , @xmath1108 are linearly independent in @xmath1109 and @xmath1110 , by injectivity of @xmath1080 .
moreover , @xmath1111 , by commutativity of the diagram .
hence , @xmath1112 belongs to @xmath1046 . as before
, let @xmath1044 be a finitely presented group . for simplicity
, we will assume @xmath1113 is torsion - free ( otherwise , we need to mod out its torsion subgroup ) , and will fix the coefficient field @xmath266 .
the _ alexander polynomial _ of the group @xmath314 , denoted @xmath1114 , is the greatest common divisor of the minors of size @xmath767 of the alexander matrix @xmath1115 .
as such , it is an element in the ring of laurent polynomials @xmath1116 , well - defined up to units .
the alexander polynomial defines a hypersurface , @xmath1117 , in the complex algebraic torus @xmath1015 . as shown in (
* corollary 3.2 ) , this hypersurface can be recovered from the characteristic variety @xmath1118 .
more precisely , either @xmath1119 , or @xmath1120 where @xmath1121 denotes the union of all codimension - one irreducible components of @xmath1122 .
if @xmath314 is a @xmath40-manifold group , more can be said .
[ prop : v1 3 m ] let @xmath4 be a compact , connected , orientable @xmath40-manifold without boundary .
let @xmath1123 , and suppose @xmath1113 is torsion - free .
then @xmath1124 in other words , at least away from the origin , the characteristic variety @xmath1125 is the hypersurface defined by the alexander polynomial @xmath1126 .
as before , let @xmath232 be a connected cw - complex with finite @xmath0-skeleton , and basepoint at the unique @xmath151-cell , @xmath1012 . consider a covering map , @xmath1127 , with connected total space @xmath545
the cell structure on @xmath232 lifts to a cell structure on @xmath545 , in such a way that @xmath284 is a cellular map .
fix a basepoint @xmath1128 ; the induced homomorphism , @xmath1129 , is injective .
the assignment @xmath1130 , then , establishes a one - to - one to correspondence between basepoint - preserving equivalence classes of connected covers of @xmath232 and subgroups of @xmath1131 . in this context , a special role is played by the _
regular _ covers of our space @xmath232 , that is , those covers @xmath1132 for which @xmath1133 is a normal subgroup of @xmath1011 .
for such a cover , let @xmath1134 be the quotient group , and let @xmath1135 be the canonical projection .
we say that @xmath781 is the group of deck transformations , and @xmath685 is a classifying homomorphism for the cover @xmath284 .
conversely , if @xmath1136 is an epimorphism to a ( necessarily finitely generated ) group @xmath781 , there is a regular cover @xmath1137 , whose classifying homomorphism is @xmath685
. now let @xmath1138 be a map , and let @xmath1139 be the cover obtained by pulling back the cover @xmath1137 along @xmath389 .
then @xmath1139 is a regular @xmath781-cover , classified by the homomorphism @xmath1140 .
the next theorem records a formula for the homology groups @xmath1141 , in the case when the group of deck - transformations @xmath781 is finite , and @xmath129 is an algebraically closed field of characteristic not dividing the order of @xmath781 . in the case
@xmath144 , this formula is well - known , and due to libgober @xcite , sakuma @xcite and e. hironaka @xcite for @xmath266 , and to matei
suciu @xcite for other fields @xmath129 .
we provide a self - contained proof , following ( * ? ? ?
* theorem 2.5 ) .
let @xmath1142 be an epimorphism to a finite abelian group @xmath781 , and let @xmath1143 be the induced morphism between character groups .
then @xmath1144 is injective , and its image is ( non - canonically ) isomorphic to @xmath781 .
furthermore , if @xmath1145 is a character belonging to @xmath1146 , there is a unique character @xmath1147 such that @xmath1148 .
[ thm : betti cover ] let @xmath1137 be the regular cover defined by an epimorphism @xmath685 from @xmath1011 to a finite abelian group @xmath781 .
let @xmath129 be an algebraically closed field of characteristic not dividing the order of @xmath781 .
then , for each @xmath712 , @xmath1149 the epimorphism @xmath685 puts a left @xmath1150$]-module structure on the group algebra @xmath1151 $ ] . by shapiro s lemma ,
@xmath1152 is isomorphic , as a right @xmath1151$]-module , to @xmath1153)$ ] . by our assumption on @xmath129 ,
the algebra @xmath1151 $ ] is completely reducible , with one - dimensional , irreducible representations parametrized by @xmath1154 . as a left @xmath1150$]-module
, each such representation is isomorphic to the image of @xmath1155 .
thus , @xmath1156)\cong\bigoplus_{\rho\in\operatorname{im}(\widehat{\chi } ) } h_q(x,{\bbbk}_\rho).\ ] ] by the remark above , @xmath1157 is isomorphic , as a @xmath1151$]-module , to @xmath1158 where @xmath1159 . by definition , @xmath1160 if and only if @xmath1161 .
putting things together , we obtain an isomorphism of @xmath1151$]-modules , @xmath1162 taking dimensions on both sides completes the proof .
we now specialize to the case when @xmath1137 is a regular cover , classified by an epimorphism @xmath1163 to a finite cyclic group @xmath781 .
choose a generator @xmath1164 of @xmath781 .
let @xmath1165 be the corresponding monodromy automorphism , and let @xmath1166 be the induced map in homology .
as in @xcite , then , we obtain the following application of theorem [ thm : betti cover ] .
[ thm : alg mono ] assume @xmath1167 .
then , the characteristic polynomial of the algebraic monodromy , @xmath1168 , is given by @xmath1169 from the hypothesis , the @xmath1151$]-module @xmath1170)$ ] is completely reducible .
therefore , the automorphism @xmath626 is diagonalizable .
furthermore , the eigenvalues of @xmath626 , counted with multiplicity , are indexed by the irreducible @xmath1151$]-modules appearing in decomposition , and we are done . in degree @xmath144 , and for @xmath266 , the polynomial @xmath1171 can be related to the alexander polynomial of the group @xmath1172 , as follows : @xmath1173 where @xmath1174 is an integer depending only on @xmath314 . the next proposition and its corollary
were proved in @xcite in the case when @xmath129 has characteristic @xmath151 . for completeness
, we include a proof , valid in a slightly more general context .
[ prop : jumpcov ] let @xmath781 be a finite group , and let @xmath1127 be a regular @xmath781-cover .
suppose that @xmath1167 , and @xmath1027 if @xmath1026 has @xmath150-torsion
. then @xmath1175 and @xmath1176 , for all @xmath712 and @xmath136 . for the first claim ,
let @xmath1177 be a @xmath129-valued , rank-@xmath0 local system on @xmath232 , and consider the hochschild - serre spectral sequence of the cover , @xmath1178 .
since @xmath781 is finite and @xmath1179 , we have that @xmath1180 , for all @xmath1181 and @xmath1182 ; thus , the spectral sequence collapses to an isomorphism , @xmath1183 . by duality
, we obtain an injection , @xmath1184 , and we are done . for the second claim , a standard transfer argument ,
using again the hypothesis on @xmath1185 , allows us to identify the induced algebra map in cohomology , @xmath1186 , with the inclusion @xmath1187 . for a class @xmath1188 ,
the monodromy action of @xmath781 on @xmath1189 gives rise to an action on the chain complex @xmath1190 , with fixed subcomplex @xmath1191 .
we thus obtain an inclusion @xmath1192 , and we are done .
the proof of the second part of the proposition has an immediate corollary .
[ cor : trivnom ] with notation as above , suppose @xmath781 acts trivially on @xmath1193 .
then @xmath1194 is an isomorphism , for all @xmath136 .
the assumption of the corollary is really necessary .
for instance , if @xmath232 is a wedge of @xmath1071 circles , then @xmath545 is a wedge of @xmath1195 circles ; thus , if @xmath1196 , then the map @xmath1197 is a proper inclusion , sending @xmath131 into @xmath1198 .
let @xmath232 be connected cw - complex with finite @xmath0-skeleton .
to such a space , sullivan associated the commutative differential graded algebra ( cdga ) of polynomial differential forms on @xmath232 with coefficients in @xmath1199 , denoted @xmath1200 .
let @xmath1201 be the rational cohomology algebra of @xmath232 , endowed with the zero differential .
the space @xmath232 is said to be _ formal _ if there is a zig - zag of cdga morphisms connecting @xmath1200 to @xmath1201 , with each such morphism inducing an isomorphism in cohomology .
the space @xmath232 is merely _ @xmath156-formal _ ( for some @xmath1202 ) if each of these morphisms induces an isomorphism in degrees up to @xmath156 , and a monomorphism in degree @xmath1203 .
if @xmath232 is a @xmath156-formal and @xmath1204 , then @xmath232 is formal .
if @xmath232 is a smooth manifold , sullivan s algebra may be replaced by the de rham algebra @xmath1205 of smooth , differential forms on @xmath232 .
examples of formal spaces include suspensions , rational cohomology tori , surfaces , compact connected lie groups , as well as their classifying spaces . on the other hand , the only nilmanifolds which are formal are tori .
formality is preserved under wedges and products of spaces , and connected sums of manifolds .
the @xmath0-minimality property of a space @xmath232 depends only on its fundamental group , @xmath1011 . alternatively ,
a finitely generated group @xmath314 is @xmath0-formal if and only if its malcev lie algebra ( defined as the lie algebra of primitive elements in the @xmath104-adic completion of the group - algebra @xmath1206 $ ] ) admits a quadratic presentation .
examples of @xmath0-formal groups include free groups and free abelian groups of finite rank , surface groups , and groups with first betti number equal to @xmath151 or @xmath0 .
the @xmath0-formality property is preserved under free products and direct products .
a classical obstruction to formality is provided by the higher - order massey products
. in particular , if @xmath232 is @xmath0-formal , and @xmath1207 are elements in @xmath1208 such that @xmath1209 , then the massey triple product @xmath1210 must vanish , modulo indeterminacy .
suppose @xmath1212 is a map such that @xmath1213 is an isomorphism up to degree @xmath156 , and a monomorphism in degree @xmath1203 .
then @xmath232 is @xmath156-formal if and only if @xmath545 is @xmath156-formal .
let @xmath853 be the group of deck - transformations , so that @xmath1214 .
as shown in ( * ? ? ?
* remark 3.30(2 ) ) , if @xmath545 is formal , then @xmath232 is also formal .
the same argument applies to @xmath156-formality .
a particular case is worth mentioning .
suppose @xmath314 is a finitely generated group , and @xmath1215 is a finite - index , normal subgroup . if @xmath11 is @xmath0-formal , then @xmath314 is also @xmath0-formal . as the next example shows
, the converse does not hold . [
ex : non1f ] as is well - known , the heisenberg group @xmath1216 $ central } \rangle$ ] admits non - trivial triple massey products in @xmath1217 ; thus , @xmath11 is not @xmath0-formal . on the other hand ,
the semi - direct product @xmath1218 defined by the involution @xmath1219 , @xmath1220 has @xmath1221 , and so @xmath314 is @xmath0-formal .
[ lem : kformal covers ] let @xmath1222 be a finite , regular cover , and suppose the group of deck - transformations acts trivially on @xmath1223 , for all @xmath1224
. then @xmath545 is @xmath156-formal if and only if @xmath232 is @xmath156-formal .
let @xmath1011 .
the homomorphism @xmath1225 , @xmath1226 induces a homomorphism @xmath1227 . given a subvariety of @xmath1228 ,
define its _ exponential tangent cone _ at the origin to be the set @xmath1229 it turns out that @xmath1230 is a finite union of rationally defined linear subspaces ( see @xcite , and also @xcite for more details ) .
moreover , @xmath1230 is included in @xmath1231 , the usual tangent cone to @xmath726 at the origin .
now fix an integer @xmath1036 , and consider the varieties @xmath1232 and @xmath1233 . by work of libgober @xcite , the tangent cone to @xmath1234 at @xmath0
is included in @xmath1235 .
thus , we get a chain of inclusions , @xmath1236 each of which is a proper inclusion , in general . the main connection between the formality property of a space and its cohomology jump loci is provided by the following theorem . as a consequence
, the irreducible components of @xmath1235 are all rationally defined subspaces , while the components of @xmath1238 passing through the origin are all algebraic subtori of the form @xmath733 , with @xmath735 running through the irreducible components of @xmath1235 .
e. artal bartolo , j. cogolludo , d. matei , _ characteristic varieties of quasi - projective manifolds and orbifolds _ , geom . topol .
* 17 * ( 2013 ) , no . 1 , 273309 .
http://www.ams.org/mathscinet-getitem?mr=3035328[mr3035328 ]
e. brieskorn , _ sur les groupes de tresses ( daprs v. i. arnold ) _ , sminaire bourbaki , 24me anne ( 1971/1972 ) , exp .
401 , pp . 2144 , lecture notes in math .
317 , springer , berlin , 1973 .
( 54 # 10660 ) ] n. budur , a. dimca , m. saito , _ first milnor cohomology of hyperplane arrangements _ , in : _ topology of algebraic varieties and singularities _
, 279292 , contemp .
, vol . 538 , amer .
providence , ri , 2011 .
http://www.ams.org/mathscinet-getitem?mr=2777825[mr2777825 ( 2012f:32040 ) ] d. cohen , a. dimca , p. orlik , _ nonresonance conditions for arrangements _ , annales institut fourier ( grenoble ) * 53 * ( 2003 ) , no . 6 , 18831896 .
http://www.ams.org/mathscinet-getitem?mr=2038782[mr2038782 ( 2005a:32027 ) ] d. cohen , a. suciu , _ the braid monodromy of plane algebraic curves and hyperplane arrangements _ ,
helvetici * 72 * ( 1997 ) , no . 2 , 285315 .
http://www.ams.org/mathscinet-getitem?mr=1470093[mr1470093 ( 98f:52012 ) ] d. cohen , a. suciu , _ the boundary manifold of a complex line arrangement _ , geometry & topology monographs * 13 * ( 2008 ) , 105146 .
http://www.ams.org/mathscinet-getitem?mr=2508203[mr2508203 ( 2010c:32051 ) ] g. denham , _ homological aspects of hyperplane arrangements _ , in : _
arrangements , local systems and singularities _
, 3958 , progress in math . , vol . 283 , birkhuser , basel , 2010 .
http://www.ams.org/mathscinet-getitem?mr=3025859[mr3025859 ] g. denham , a. suciu , _ multinets , parallel connections , and milnor fibrations of arrangements _ , proc . london math . soc .
( first published online december 11 , 2013 ) , ` doi:10.1112/plms / pdt058 ` , available at ` arxiv:1209.3414v2 ` .
a. dimca , s. papadima , _ finite galois covers , cohomology jump loci , formality properties , and multinets _ , ann . sc . norm .
* 10 * ( 2011 ) , no . 2 , 253268 .
http://www.ams.org/mathscinet-getitem?mr=2856148[mr2856148 ( 2012i:32033 ) ] a. dimca , s. papadima , a. suciu , _ alexander polynomials : essential variables and multiplicities _ , int .
res . notices * 2008 * , no . 3 , art .
i d rnm119 , 36 pp .
http://www.ams.org/mathscinet-getitem?mr=2416998[mr2416998 ( 2009i:32036 ) ] a. dimca , s. papadima , a. suciu , _ topology and geometry of cohomology jump loci _ , duke math .
journal * 148 * ( 2009 ) , no . 3 , 405457 . http://www.ams.org/mathscinet-getitem?mr=2527322[mr2527322 ( 2011b:14047 ) ] m. falk , _ resonance varieties over fields of positive characteristic _ , int .
research notices * 2007 * ( 2007 ) , no . 3 , article i d rnm009 , 25 pages .
http://www.ams.org/mathscinet-getitem?mr=2337033[mr2337033 ( 2008m:14107 ) ] m. falk , s. yuzvinsky , _ multinets , resonance varieties , and pencils of plane curves _
, compositio math .
* 143 * ( 2007 ) , no . 4 , 10691088 .
( 2009e:52043 ) ] y. flix , j. oprea , d. tanr , _ algebraic models in geometry _
, oxford grad .
texts in math .
17 , oxford univ .
press , oxford , 2008 .
http://www.ams.org/mathscinet-getitem?mr=2403898[mr2403898 ( 2009a:55006 ) ] e. hironaka , _ abelian coverings of the complex projective plane branched along configurations of real lines _ , memoirs a.m.s .
, vol . 502 , amer .
soc . , providence , ri , 1993 .
http://www.ams.org/mathscinet-getitem?mr=1164128[mr1164128 ( 94b:14020 ) ] e. hironaka , _ alexander stratifications of character varieties _ , annales de linstitut fourier ( grenoble ) * 47 * ( 1997 ) , no . 2 , 555583 .
http://www.ams.org/mathscinet-getitem?mr=1450425[mr1450425 ( 98e:14020 ) ] f. hirzebruch , _ the topology of normal singularities of an algebraic surface ( after d. mumford ) _ , sminaire bourbaki , vol . 8 , exp .
250 , 129137 , soc .
france , paris , 1995 .
http://www.ams.org/mathscinet-getitem?mr=1611536[mr1611536 ]
t. jiang , s.s .- t .
yau , _ topological invariance of intersection lattices of arrangements in @xmath832 _ , bull .
* 29 * ( 1993 ) , no . 1 , 8893 .
http://www.ams.org/mathscinet-getitem?mr=1197426[mr1197426 ( 94b:52022 ) ] t. jiang , s.s .- t .
yau , _ intersection lattices and topological structures of complements of arrangements in @xmath832 _ , ann . scuola norm .
( 4 ) * 26 * ( 1998 ) , no . 2 , 357381 ; http://www.ams.org/mathscinet-getitem?mr=1631597[mr1631597 ( 99g:52020 ) ]
a. libgober , _
characteristic varieties of algebraic curves _ , in : applications of algebraic geometry to coding theory , physics and computation ( eilat , 2001 ) , pp . 215254 , nato sci .
36 , kluwer acad .
dordrecht , 2001 .
http://www.ams.org/mathscinet-getitem?mr=1866902[mr1866902 ( 2003e:14008 ) ] a. libgober , _ first order deformations for rank one local systems with a non - vanishing cohomology _ , topology appl .
* 118 * ( 2002 ) , no . 1 - 2 , 159168 .
http://www.ams.org/mathscinet-getitem?mr=1877722[mr1877722 ( 2002m:52025 ) ] a. libgober , _
eigenvalues for the monodromy of the milnor fibers of arrangements _ , trends in singularities , pp .
141150 , trends math . ,
birkhuser , basel , 2002 . http://www.ams.org/mathscinet-getitem?mr=1900784[mr1900784 ( 2003f:32035 ) ] a. libgober , _ on combinatorial invariance of the cohomology of milnor fiber of arrangements and catalan equation over function field _ , in : arrangements of hyperplanes ( sapporo 2009 ) , 175187 , adv .
pure math .
62 , math .
japan , tokyo , 2012 .
http://www.ams.org/mathscinet-getitem?mr=2933797[mr2933797 ]
a. libgober , s. yuzvinsky , _ cohomology of orlik solomon algebras and local systems _ , compositio math .
* 21 * ( 2000 ) , no . 3 , 337361 .
http://www.ams.org/mathscinet-getitem?mr=1761630[mr1761630 ( 2001j:52032 ) ] a. mcinic , s. papadima , _ on the monodromy action on milnor fibers of graphic arrangements _ , topology appl .
* 156 * ( 2009 ) , no . 4 , 761774 .
http://www.ams.org/mathscinet-getitem?mr=2492960[mr2492960 ( 2010i:32028 ) ]
d. matei , _ massey products of complex hypersurface complements _ , in : _ singularity theory and its applications _ , 205219 ,
studies in pure math .
43 , math .
japan , tokyo , 2007 . http://www.ams.org/mathscinet-getitem?mr=2325139[mr2325139 ( 2008h:32041 ) ] d. matei , a. suciu , _ cohomology rings and nilpotent quotients of real and complex arrangements _ , in : arrangements
tokyo 1998 , pp . 185215 , adv .
pure math .
27 , math .
japan , tokyo , 2000 . http://www.ams.org/mathscinet-getitem?mr=1796900[mr1796900 ( 2002b:32045 ) ] d. matei , a. suciu , _ hall invariants , homology of subgroups , and characteristic varieties _ , internat . math .
notices * 2002 * ( 2002 ) , no . 9 , 465503 .
( 2003d:20055 ) ] j. milnor , _ singular points of complex hypersurfaces _ , annals of math .
studies , vol .
61 , princeton univ .
press , princeton , nj , 1968 .
( 39 # 969 ) ] a. nmethi , a. szilrd , _ milnor fiber boundary of a non - isolated surface singularity _ , lecture notes in math , vol .
2037 , springer - verlag , berlin heidelberg , 2012 .
http://www.ams.org/mathscinet-getitem?mr=3024944[mr3024944 ]
s. papadima , a. suciu , _ geometric and algebraic aspects of @xmath0-formality _ , bull .
roumanie * 52 * ( 2009 ) , no . 3 , 355375 .
http://www.ams.org/mathscinet-getitem?mr=2554494[mr2554494 ( 2010k:55018 ) ] s. papadima , a. suciu , _ bieri neumann strebel renz invariants and homology jumping loci _ ,
london math .
* 100 * ( 2010 ) , no . 3 , 795834 .
( 2011i:55006 ) ] g. rybnikov , _ on the fundamental group of the complement of a complex hyperplane arrangement _
* 45 * ( 2011 ) , no . 2 , 137148 http://www.ams.org/mathscinet-getitem?mr=2848779[mr2848779 ( 2012i:14067 ) ] a. suciu ,
_ fundamental groups of line arrangements : enumerative aspects _ , in : advances in algebraic geometry motivated by physics ( lowell , ma , 2000 ) , pp . 4379 , contemp .
, vol 276 , amer . math . soc . ,
providence , ri , 2001 .
http://www.ams.org/mathscinet-getitem?mr=1837109[mr1837109 ( 2002k:14029 ) ] a. suciu ,
_ translated tori in the characteristic varieties of complex hyperplane arrangements _ , topology appl .
* 118 * ( 2002 ) , no . 1 - 2 , 209223 .
http://www.ams.org/mathscinet-getitem?mr=1877726[mr1877726 ( 2002j:32027 ) ] a. suciu , _ fundamental groups , alexander invariants , and cohomology jumping loci _ , in : topology of algebraic varieties and singularities , 179223 , contemp . math .
538 , amer .
soc . , providence , ri , 2011 .
http://www.ams.org/mathscinet-getitem?mr=2777821[mr2777821 ( 2012b:20092 ) ] a. suciu , _ geometric and homological finiteness in free abelian covers _
, in : _ configuration spaces : geometry , combinatorics and topology ( centro de giorgi , 2010 ) _ , 461501 , publications of the scuola normale superiore , vol
. 14 , edizioni della normale , pisa , 2012 , available at ` arxiv:1112.0948v2 ` .
a. suciu , _
characteristic varieties and betti numbers of free abelian covers _ , intern .
notices ( first published online november 16 , 2012 ) , ` doi:10.1093/imrn / rns246 ` , available at ` arxiv:1111.5803v3 ` .
s. yuzvinsky , _ a new bound on the number of special fibers in a pencil of curves _ ,
* 137 * ( 2009 ) , no . 5 , 16411648 .
http://www.ams.org/mathscinet-getitem?mr=2470822[mr2470822 ( 2009k:14015 ) ]
s. yuzvinsky , _ resonance varieties of arrangement complements _ , in : _ arrangements of hyperplanes ( sapporo 2009 ) _ , 553570 , advanced studies pure math . , vol .
62 , kinokuniya , tokyo , 2012 . | there are several topological spaces associated to a complex hyperplane arrangement : the complement and its boundary manifold , as well as the milnor fiber and its own boundary .
all these spaces are related in various ways , primarily by a set of interlocking fibrations .
we use cohomology with coefficients in rank @xmath0 local systems on the complement of the arrangement to gain information on the homology of the other three spaces , and on the monodromy operators of the various fibrations .
tant donn un arrangement dhyperplans , il y a plusieurs espaces topologiques quon puisse lui associer : le complmentaire et sa varit de bord , ainsi que la fibre de milnor et son bord .
tous ces espaces sont relis , en premier lieu par des fibrations .
on utilise la cohomologie avec coefficients dans les systmes locaux de rang @xmath0 sur le complmentaire dun arrangement dhyperplans pour tudier lhomologie des trois autres espaces , et les oprateurs de monodromie des fibrations associes . |
sensitivity of diffractive lepto and photoproduction to the gluon density in the proton gives an excellent tool to test these structure functions .
intensive experimental study of diffractive processes were performed in desy @xcite .
theoretical analysis shows that the cross sections of diffractive hadron production are expressed in terms of skewed parton distributions ( spd ) @xcite .
the diffractive charm @xmath0 production and @xmath2 production are determined by the gluon spd @xmath3 because the charm component in the proton is small . for
light hadron production effects of the quark spd should be important for not small @xmath4 . in the polarized case
, the spin - dependent gluon distributions can be investigated . in future , there will be an excellent possibility of studying spin effects with transversely polarized target at hermes and compass . in this report
, we consider double spin asymmetries for longitudinally polarized leptons and transversely polarized protons in diffractive vector @xmath0 production at high energies ( see @xcite ) which is expressed in terms of the polarized cross sections @xmath5 at small @xmath4 the diffractive contribution to the asymmetry @xmath1 determined by the pomeron exchange should be important . in the qcd-
based models the pomeron is usually regarded as a two - gluon state . within the two - gluon exchange model the two - gluon coupling with the proton can be written as follows : @xmath6 here the structure proportional to @xmath7 determines the spin - non - flip contribution .
the term @xmath8 leads to the transverse spin - flip at the vertex .
it has been found within the model approaches @xcite that the ratio @xmath9 and has a weak energy dependence ( weak @xmath4 dependence ) .
the weak energy dependence of spin asymmetries in exclusive reactions is not in contradiction with the experiment @xcite .
the diffractive @xmath0 production in the lepton - proton reaction is determined by the photon - two - gluon fusion .
the spin - average and spin - dependent cross section can be written in the form @xmath10 here @xmath11 is a normalization coefficient , the functions @xmath12 are determined by a sum of all graphs integrated over the gluon momenta @xmath13 and @xmath14 .
we calculate here the imaginary parts of the photon - two - gluon fusion amplitudes which are expressed directly in terms of the functions @xmath15 and @xmath16 from ( [ ver ] ) .
the function @xmath17 has the form @xmath18 here @xmath19 and @xmath20 with @xmath21 .
the gluon structures @xmath22(@xmath23 ) are connected with the @xmath24(@xmath25 ) spd ( see ( [ bqq],[kqq ] ) ) .
thus , the functions @xmath15 and @xmath16 are the nonintegrated gluon distributions . the hard part @xmath26 in ( [ np ] ) can be calculated perturbatively when @xmath27 is not small , about @xmath28 or larger .
the spin - dependent contribution @xmath29 has two terms proportional to the scalar products ( @xmath30 ) and ( @xmath31 ) @xcite .
we consider here only the term proportional to @xmath32 which can be written as follows : @xmath33.\ ] ] the other term @xmath34 has been discussed in @xcite .
the asymmetry is determined by the ratio @xmath35 . at small @xmath4 the gluon structure functions
have large imaginary part . in this case
the asymmetry can be approximated as @xmath36 in numerical calculations we use a simple parameterization of the spd as a product of the form factor and the ordinary gluon distribution @xcite . in our estimations
we use the value @xmath37 .
we analyze the case when the @xmath38 asymmetry has a maximal value ( the momentum @xmath39 is parallel to the target polarization @xmath40 ) . the predicted @xmath1 asymmetry in diffractive light @xmath0 production at @xmath41
is shown in fig .
this asymmetry is not small for @xmath42 .
the @xmath38 asymmetry has a strong mass dependence . for heavy quark production
this asymmetry becomes negative , fig .
it is interesting to have predictions for a light quark production at the slac and hermes low energy range @xmath43 . here
the `` window '' for perturbative calculation is quite small .
really , it can be found in this case that the maximum value of the transverse momentum is limited by @xmath44 because we have the restriction @xmath45 from ( [ sigma ] ) . in
target experiments , it is usually difficult to detect the final hadron and determine the momentum transfer . in this case , we estimate the asymmetry integrated over momentum transfer @xmath46 @xmath47 ; @xmath48 .
the predicted integrated asymmetry is about 3% .
note that we have calculated here only the gluon contribution to the asymmetry . at hermes energies
the contribution of the quark spd to the @xmath1 asymmetry should be important .
to conclude , we would like to emphasize that the diffraction contribution to the @xmath1 asymmetry is found to be proportional to the ratio of @xmath49 structure functions .
the predicted coefficient @xmath50 in ( [ cltqq ] ) is not small , about 0.3 - 0.5 .
this shows the possibility of studying the transverse distribution @xmath51 in future experiments with a transversely polarized target ( hermes , compass and future erhic facilities )
. these results could be applicable to reactions with heavy quarks . for processes with light hadron production
, our predictions can be used in the small @xmath4 region ( @xmath52 e.g. ) where the contribution of the quark spd is expected to be small .
the recoil particle detector is needed to distinguish the diffractive events .
really , in the case when the recoil detector is absent , the diffractive events are detected together with nondiffractive ones .
the measured asymmetry in this case looks like @xmath53 here @xmath54 and @xmath55 .
the ratio @xmath56 integrated over @xmath4 has been found at hera to be about 0.200.30 @xcite . in this case , the diffractive contribution to asymmetry will be smaller by the factor 3 - 5 .
99 h1 collaboration , s. aid et al .
b472 * , 3 ( 1996 ) ; + zeus collab . , j. breitweg et al . , z. phys , * c75 * , 215 ( 1997 ) .
zeus collab . , j. breitweg et al . , eur .
j. * c5 * , 41 ( 1998 ) ; + h1 collab .
, c. adloff et al . , eur .
j. * c6 * 421 ( 1999 ) .
radyushkin , phys.rev , d * 56 * , 5524 ( 1997 ) .
x. ji , phys.rev .
d * 55 * 7114 ( 1997 ) .
goloskokov , hep - ph/0112268 to appear in euro .
goloskokov , s.p .
kuleshov , o.v .
selyugin , z. phys . *
c50 * , 455 ( 1991 ) .
goloskokov , p. kroll , phys .
d * 60 * , 014019 ( 1999 ) .
h1 collaboration , aid s. et al , _ z. phys . _ * c69 * , 27 ( 1995 ) . | we consider double spin asymmetries for longitudinally polarized leptons and transversely polarized protons in diffractive @xmath0 production which is connected with @xmath1 asymmetry .
the predicted asymmetry is large and can be used to obtain the information on the polarized skewed gluon distributions in the proton . |
time - of - flight measurements are a standard tool for many experimentalists . since the particles or atoms involved are usually fast , their center - of - mass motion is typically treated classically , yielding a simple description of the time - of - flight measurement . but as the diffraction and interference experiments using temporal ( instead of spatial ) slits of szriftgiser et .
al have shown @xcite , such a description of the center - of - mass motion by means of classical physics is not always sufficient : the advance of cooling techniques has made it possible to create ultracold gases in a trap and produce very slow atoms , e.g. , by opening the trap .
whenever such ultracold atoms are involved , the spatial extent and the spreading of the wave function can show noticeable effects .
even the seemingly simple question of the time spent by a particle in a given region of space does not possess a simple and definite answer .
related to this `` dwell - time '' problem are the problems of `` passage time , '' concentrating on those particles that actually cross the region of interest and are not reflected , and `` tunneling time '' in the case of a barrier that classically can not be traversed inside the region of interest .
these problems have on one hand been treated axiomatically aiming at ideal quantities relying only on the system of interest , see , e.g. , refs .
other approaches may be called `` operational '' in the sense that a sort of `` measurement device '' is introduced to which the system of interest is coupled ; see , e.g. , refs . @xcite . for a critical review on different approaches to tunneling times
see , e.g. , refs .
@xcite . a distinction that can be drawn between the various time - related quantities recalled above concerns whether they are pre- or post- decoherence . since the present work models the measurement apparatus , it can be thought of as post - decoherence ; in fact it is one of our objectives to include decoherence - inducing processes , thereby eliminating some of the black magic of quantum measurement . on the other hand
, one can take , for example , a path integral approach to tunneling time @xcite in which the paths are sorted by ( variously defined ) times spent in the barrier region .
for these paths , amplitudes are retained , so it is very much a pre - decoherence calculation . as a result interpretive issues arise ( discussed for example by @xcite ) which do not enter in the present work . in the present paper
we investigate a particular measurement scheme for passage times , mimicking the way a time - of - flight experiment is typically performed : employing a particular spin - boson detector model for measuring quantum arrival times investigated in ref .
@xcite , our measurement scheme involves two measurements of arrival time by means of this detector , one upon entering the region of interest and one when exiting ; see fig .
[ passage_setup ] .
this scheme can be expected to distort the particle s wave function less than a scheme based on the continuous coupling of the particle to a measurement device .
indeed , it will be seen that a passage - time measurement by means of `` slow '' detectors yields a rather broad passage - time distribution . but
detectors responding quickly to the presence of the particle also yield a broad distribution .
this will be shown to be a quantum effect involving the heisenberg uncertainty relation .
however , there is an intermediate range for the detector parameters yielding optimal results . in the best case
the precision of the measurement can be estimated to behave like @xmath0 , improving the results obtained from clock models @xcite . in sec .
[ model ] we review the detector model and its application to quantum arrival - times . in sec .
[ reset - section ] the particle s wave function immediately after the detection is investigated , and in sec .
[ passage - section ] analytical formulas for the application of the detector model to passage times are derived .
numerical examples are investigated in sec .
[ numerical ] . for simplicity , we concentrate there on passage times for free particles ; the extension to passage times in the presence of barriers , and thus to tunneling times , should proceed on similar lines .
the precision of the measurement scheme is estimated in sec .
[ estimate ] .
we briefly review the detector model introduced in refs . @xcite and the arrival - time distribution obtained from this model in ref .
the detector consists of a three - dimensional array of spins with ferromagnetic interaction in a metastable state .
the spins are weakly coupled to the environment , modeled as a bath of bosons .
the effect of the particle to be observed is to strongly enhance the spin - bath coupling when the particle s wave function overlaps with that of a spin .
thus when the particle is close to a spin , this spin flips with strongly increased probability due to the enhanced spin - bath coupling . by means of the ferromagnetic interaction , this in turn triggers the subsequent spontaneous flipping of the neighboring spins , and finally , by a domino effect , of all spins , even in the absence of the particle .
in this way , the single spin flip induced by the presence of the particle is amplified to a macroscopic event , and either the change in the detector state or in the bath state can be measured .
the hamiltonian for this model is @xmath2 where we use the following definitions : @xmath3 is the free hamiltonian of the particle ; @xmath4 with @xmath5 denoting the excited state of the @xmath6 spin and @xmath7 is the free hamiltonian of the detector ; @xmath8 is the energy difference between ground state and excited state of the @xmath6 spin , and @xmath9 is the coupling energy between the spins @xmath10 and @xmath11 .
further , @xmath12 where @xmath13 is the annihilation operator for a boson with wave vector @xmath14 , is the free hamiltonian of the environment , modeled as a bath of bosons , and @xmath15 with @xmath16 and the coupling constants @xmath17 and the phases @xmath18 depending on the particular realization of the detector and the bath , describes the permanent spin - bath coupling .
the spin - bath coupling is strongly enhanced in the particle s presence due to @xmath19 with @xmath20 thus , the enhanced coupling of the @xmath6 spin to the bath is proportional to a sensitivity function @xmath21 which vanishes outside the region @xmath22 where the @xmath6 spin is located .
an example would be the characteristic function which is @xmath23 on @xmath22 and zero outside .
it will be assumed in the sequel that the markov property [ see eq .
( [ markov ] ) ] holds in an appropriate continuum limit .
initially , the system is prepared in the state @xmath24 where @xmath25 is the ground state of the bath ( no bosons present ) , and @xmath26 denotes the spatial wave function of the particle . for the @xmath8 only slightly above the energetic threshold set by the ferromagnetic spin - spin coupling , and @xmath17 sufficiently small , the probability of a spontaneous spin flip ( false positive ) is very small @xcite .
but when the particle is close to the @xmath6 spin , the excited state @xmath5 decays much more quickly , due to the enhanced coupling , `` @xmath27 , '' of the spin to the bath .
then , the ferromagnetic force experienced by its neighbors is strongly reduced , and these spins can therefore flip rather quickly , even in the absence of the particle , by means of the @xmath17 .
the first spin flip will be amplified to a macroscopic event by the previously mentioned domino effect @xcite . in ref .
@xcite we investigated , by means of the quantum jump approach @xcite , the application of this detector model to arrival - time measurements .
the bath modes were indexed by wave vectors @xmath28 and the coupling constants were assumed to be of the form @xmath29 with @xmath30 , @xmath31 , and @xmath32 we introduce `` modified resonance frequencies '' @xmath33 which incorporate the ferromagnetic spin - spin coupling .
the correlation function @xmath34 is defined by @xmath35 and similarly for @xmath36 , @xmath37 , and @xmath38 .
it is assumed that in the continuum limit , @xmath39 , the correlation functions satisfy the markov property , i.e. , @xmath40 for some small correlation time @xmath41 , and likewise for the other correlation functions .
this is the case , e.g. , if @xmath42 is independent of @xmath43 , as in quantum optics .
let @xmath44 denote the time evolution of the spatial wave function under the condition that no boson is detected at least until time @xmath45 ( `` conditional time evolution '' ) .
then @xmath46 is the probability that no detection occurred until @xmath45 , which is decreasing in time , and @xmath47 is the probability density for the first detection to occur at time @xmath45 . in ref .
@xcite it was shown that @xmath44 obeys a `` conditional schrdinger equation '' of the form @xmath48 with a hamiltonian containing a complex potential , @xmath49\ , , \label{three - d - schroedinger}\ ] ] where @xmath50 and @xmath51 are defined as follows .
@xmath51 is a position - dependent detector decay rate , @xmath52\nonumber\\ & & \phantom{\sum_j } \quad \times \int \frac{d \omega_\mathbf{e}}{(2 \pi)^2 } \left ( \left| \gamma^{(j ) } \left ( \tilde{\omega}_0^{(j ) } , \mathbf{e } \right ) \right|^2 \chi^{(j)}({\mathbf{x}})^2 \right .
\nonumber \\ & & \phantom{\sum_j \quad \times \int \frac{d \omega_\mathbf{e}}{(2 \pi)^2 } } \qquad \left .
+ \left| \gamma^{(j)}_\mathrm{spon } \left ( \tilde{\omega}_0^{(j ) } , \mathbf{e } \right ) \right|^2 \right ) \
, , \label{athree } \nonumber \\\end{aligned}\ ] ] where the @xmath53 integral is taken over the unit sphere and where the contributions from @xmath54 and @xmath55 have been neglected , due to ( [ gg ] ) .
the terms have the familiar form of the einstein coefficients in quantum optics , where there would also be a sum over polarizations .
the real part of the potential , @xmath56 , is given by @xmath57 in quantum optics , this would correspond to a line - shift .
since the @xmath58 term leads to a constant it just gives an overall phase factor and can be omitted .
note that the hamiltonian on the r.h.s . of eq.([three - d - schroedinger
] ) is not norm conserving due to the imaginary contribution @xmath59 , in accordance with eq .
( [ nodet ] ) . from eqs .
( [ three - d - schroedinger ] ) and ( [ density ] ) one easily finds @xmath60 which is an average of the position - dependent decay rate @xmath61 of the detector , weighted with the probability density for the particle to be at position @xmath62 and as yet undetected .
it was shown in ref .
@xcite , using an example with a one - dimensional , simplified detector model , that @xmath63 essentially agrees with the probability density obtained from the discrete model with 40 bath modes by means of standard unitary quantum mechanics , up to the time of recurrences due to the discrete nature of the bath .
for the intended application of the detector model we will need more than the time evolution up to the first detection .
in particular , we require knowledge of the `` reset state '' @xcite , the particle state immediately after the first boson detection . let the complete system ( bath , detector , particle ) be described at a particular time by a density matrix of the form @xmath64 ( with @xmath65 the particle density matrix ) , i.e. , no boson and all spins up .
if a boson is found in a broadband boson measurement a time @xmath66 later , the density matrix for the corresponding subensemble is obtained by sandwiching the above expression with @xmath67 by the von neumann - lders projection rule @xcite , and the trace gives the probability .
the subsystem consisting solely of the particle is then described after the detection of a boson by a partial trace , @xmath68 which defines the operation @xmath69 . to calculate this
we go to the interaction picture w.r.t . @xmath70 and use standard second - order perturbation theory for @xmath71 .
the zeroth order does not contribute to eq .
( [ partial ] ) , and neither does the first order since @xmath72 acts once on @xmath73 . in second order only the term with @xmath74 on the left and right survives , and in second order one obtains after some calculation @xmath75 \right ] \ , \varrho_\mathrm{p } \overline{\left[\gamma_{\pmb{\ell}}^{(j ) } + g_{\pmb{\ell}}^{(j ) } \chi^{(j ) } [ \hat{\mathbf{x}}(t_2 ) ] \right]}e^{{\frac{i}{\hbar}}h_\mathrm{part}\delta t } , \nonumber \\
\label{reset_full}\end{aligned}\ ] ] where @xmath76 is the free time development of @xmath77 in the heisenberg picture . to obtain the second equality ,
the rectangular integration area @xmath78 $ ] has been split into two triangles , @xmath79,~t_2\in [ 0,t_1]$ ] and @xmath80,~t_1 \in [ 0 , t_2]$ ] . the phases @xmath81 in @xmath74 have canceled since we are taking the trace over detector states . as an example we consider a simplified model with only one spatial dimension and only one spin .
this simplification is reasonable if the radius of the region @xmath82 is smaller than the distance between spins .
( our assumption of locality of the interaction though is a bit stronger than this however , since below , for calculational convenience , we will extend the region @xmath82 to a half - line , i.e. , @xmath83 . ) the vectors @xmath84 and @xmath14 are replaced by @xmath85 and @xmath86 , and there is of course no summation over the spin index @xmath10 ; we will temporarily drop the superscript @xmath87 .
the detector hamiltonian @xmath88 [ eq . ( [ hdet ] ) ] simplifies to @xmath89 and the modified resonance frequency @xmath90 [ eq.([modified ] ) ] is replaced by the resonance frequency of the single spin , @xmath91 .
also , we will neglect @xmath92 in view of the assumption @xmath93 .
the time development in this model for a wave packet incident from the left with detector and bath initially in state @xmath94 has been investigated , among other questions , in ref .
@xcite .
we assume a maximal boson frequency @xmath95 and @xmath96 boson modes , @xmath97 we further take @xmath98 where @xmath99 denotes heaviside s step function . as the particle we consider a cesium atom , and the initial state at @xmath100
is assumed to be a pure state , @xmath101 , with @xmath102 with these simplifications the first line of eq .
( [ reset_full ] ) yields @xmath103 \right| \psi \right\rangle \nonumber \\ & & \phantom{\sum_\ell \left| g_\ell \right|^2 \int\limits_0^{\delta t } dt_1 dt_2 \ , e^{i ( \omega_\ell - \omega_0 ) ( t_1 - t_2 ) } } \quad \quad \times \left\langle \psi \left| \theta \left [ \hat{x } ( t_2 ) \right ] e^{i h_\mathrm{part } \delta t / \hbar } \right| x \right\rangle \nonumber \\ & = & \sum_\ell \left| g_\ell \int\limits_0^{\delta t } dt \ ,
e^{i ( \omega_\ell - \omega_0 ) t } \left\langle x \left| e^{-i h_\mathrm{part } ( \delta t - t ) / \hbar } \theta \left ( \hat{x } \right ) e^{-i h_\mathrm{part } t / \hbar } \right| \psi \right\rangle \right|^2 .
\label{reset_example}\end{aligned}\ ] ] the contribution @xmath104 has an intuitive explanation : the initial state evolves freely until @xmath45 and is then projected onto the detector region . in an intuitive picture , the time @xmath45 may be viewed as time of occurrence of a boson . since a boson
can only be created when the spin couples to the bath , the particle has to be inside the detector at @xmath45 , hence the projection .
then the state continues to evolve freely until @xmath105 .
the integration is understood as sum over all possible `` paths '' satisfying the above picture , i.e. , as sum over all times @xmath45 .
the second line of eq .
( [ reset_example ] ) can be evaluated , e.g. , by inserting @xmath106 with momentum eigenfunctions @xmath107 on the left of @xmath108 , and @xmath109 on its right , and noting that @xmath110 a numerical illustration of @xmath111 [ note the division by @xmath112 as compared to eq .
( [ reset_full ] ) or eq.([reset_example ] ) ] for @xmath113 bosons modes is given in fig.[reset_comp ] ( dots ) .
we return to the general expression in eq .
( [ reset_full ] ) . for simplicity we will assume in the following @xmath114 in view of eq .
( [ gamma_small ] ) and the remarks after eq.([initial ] ) we neglect the @xmath17 terms . we introduce the collective correlation function @xmath115 since we assume that the coupling constants are such that the markov property holds in the continuum limit , i.e. , @xmath116 for some small correlation time @xmath117 , in the double integral of eq .
( [ reset_full ] ) only times with @xmath118 contribute , and if @xmath117 is small enough one can write @xmath119 \approx \chi [ \hat{\mathbf{x}}(t_2 ) ] .
\label{approx}\ ] ] then , with a change of variable , @xmath120 , the integration over @xmath121 can be extended to @xmath122 if @xmath123 .
specializing the definition eq .
( [ athree ] ) to the case at hand , we put @xmath124 and note that @xmath125 is given by the right side of eq .
( [ athree ] ) without the terms @xmath126 and @xmath127 .
one then obtains @xmath128 hence , to first order in @xmath112 , one finally has @xmath129 for the state immediately after a detection .
@xmath69 is called the reset operation .
if @xmath65 is a pure state , then the reset state is also a pure state . in particular ,
if @xmath130 , then the reset state is given by the wave function @xmath131 we note that this result is physically very reasonable since it means that right after a detection of the particle by a detector located in a specific region the particle is localized in that region .
we also note that [ see eq .
( [ w ] ) ] @xmath132 which is the probability density for the first detection .
we apply eq .
( [ state ] ) to the continuum limit of the example of subsection [ example ] . in that one - dimensional and single - spin case one has with eqs .
( [ ex ] ) and ( [ kappa ] ) @xmath133 thus one obtains in the continuum limit @xmath134 - 1\right),\end{aligned}\ ] ] and @xmath41 is of the order of @xmath135 . the resulting spatial probability density @xmath136 is plotted in fig .
[ reset_comp ] for the same initial state and same parameters as in the discrete case .
the plots are in very good agreement .
we note that in the present model with @xmath18 independent of @xmath62 there is no explicit recoil on the particle from the created boson .
this is in line with the original idea of a minimally invasive measurement .
the absence of an explicit recoil distinguishes the present detector model from other models which are based on the direct interaction with the particle s internal degrees of freedom .
an example for such a model would be the detection by means of laser induced fluorescence . in that case , the reset state after the detection of the first fluorescence photon explicitely incorporates a recoil due to the momentum of the emitted photon @xcite .
it appears reasonable that in the present model no such recoil on the particle occurs : after all , the boson is emitted not by the particle but by the spin lattice .
hence , the recoil should be experienced by this lattice , rather than by the particle , similar to what occurs in the mssbauer effect .
of course , the projection of the wave packet onto the detector region by means of the reset operation also changes the momentum distribution of the wave packet .
after detection of a boson , the further interaction of the particle with the detector depends on the particular choice of parameters of the detector model .
the internal dynamics of the detector after the first spin flip has been investigated in detail in refs.@xcite . based on these results , several choices are possible such that the amplification of the first spin flip will not significantly change the spatial wave function after the first spin flip , @xmath137 .
this means that effectively only one reset operation , associated with the very first spin flip , has to be performed on the spatial wave function .
such `` minimally invasive '' detector models will be of interest if one is interested in actual quantum mechanical limitations of a passage - time measurement .
as a simple example , consider a ring of identical spins with nearest - neighbor interaction , @xmath138 with @xmath139 kronecker s @xmath140 , @xmath141 as before , and @xmath142 identified with @xmath143 .
the rate for the neighbors of the first - flipped spin to flip into _ their _ ground state is denoted by @xmath144 and is given by the right side of eq .
( [ athree ] ) , but with @xmath90 replaced by @xmath91 since the ferromagnetic forces on these neighboring spins cancel .
choosing parameters as , e.g. , @xmath145 and @xmath146 independent of @xmath43 as well as @xmath147 , one has . by a kind of domino effect the whole ring flips into the ground state ; the mean time needed for this given by @xmath148 @xcite .
if this time is very short , as one can achieve by making @xmath91 large ( while @xmath90 remains small to prevent spontaneous spin flips before the first particle - induced spin flip ) , the reset operations associated with these subsequent spin flips will not significantly change the particle s state since the wave function has been projected onto the detector by the first reset operation already . another possibility to prevent the spatial wave function from being changed by the amplification process would be to couple only one spin to the particle , by choosing @xmath149 in eq .
( [ general_coupling ] ) .
( in fact , this is the _ de facto _ setup of the detector actually investigated in refs .
@xcite : effectively only one spin couples to the particle , and subsequently the other spins flip spontaneously , i.e. , without particle - enhanced spin - bath coupling . )
we now consider two detectors separated by some distance .
as indicated in the preceding section , we assume that the amplification of the first spin flip to a macroscopic event is very fast and does not change the spatial wave function .
thus , we take the probability density @xmath63 [ eq . ( [ w ] ) ] for the first spin flip to be the `` measured arrival - time distribution '' and @xmath137 [ eq . ( [ state ] ) ] as the particle s state after the detection .
then , the joint probability density for the first detector detecting the particle at @xmath150 _ and _ the second one detecting it at @xmath151 is given by @xmath152 where the superscripts indicate the detector under consideration and the second argument is the initial state for which the respective probability density is calculated . since @xmath153 is bilinear in the wave function
[ see eq .
( [ w ] ) ] , this simplifies to @xmath154 by eq .
( [ reset_norm ] ) . the desired measured passage - time distribution is then given by integration over the entry time @xmath150 : @xmath155 in order to investigate basic features of a quantum passage - time distribution obtained from the detector model in the continuum limit and in the above measurement scheme , we consider as a simple example a cesium atom in one dimension .
the initial wave packet is prepared in the remote past far away from the detector such that the free wave packet ( with no detectors present ) would be described at @xmath100 by a gaussian minimal uncertainty packet around @xmath156 with @xmath157 and average velocity @xmath158 .
each of the two identical detectors is described in the continuum limit by an absorbing potential @xmath159 extending from @xmath160 to @xmath161 or @xmath162 to @xmath163 , resp .
, where we consider three different examples @xmath164 .
these parameters are chosen in such a way that transmission and reflection without detection , which are typical for imaginary potentials @xcite and have been extensively studied in the framework of quantum arrival times @xcite , play no significant role .
consequently , all distributions shown in the following are normalized to 1 to a good approximation .
the passage - time distributions , calculated as described in section [ general ] , are shown in fig .
[ passage_fig ] .
it is seen that small and large values of @xmath125 give rise to rather broad distributions , while the intermediate value yields a narrower one .
the reason for the broad distribution arising for small @xmath125 can be understood by looking at the arrival - time distribution measured by the first detector ( see fig .
[ arrival_fig ] ) : it is already this distribution which is rather broad for small @xmath125 . physically , small @xmath125 means that the detector is responding only slowly to the presence of the particle ; the undetected amplitude @xmath44 decays only slowly , yielding a broad arrival - time distribution at each of the two detectors and consequently a broad passage - time distribution . in other words ,
the poor quality of the passage - time measurement arises from the poor quality of the individual arrival - time measurements .
the measurement of the arrival time , however , can be improved by making @xmath125 large , that is , by making the detector responding faster to the presence of the particle . in fig .
[ arrival_fig ] it is seen , e.g. , that for @xmath165 one comes much closer to kijowski s arrival - time distribution ; this in turn is known to have minimum standard deviation among all those distributions fulfilling certain axioms transferred to quantum mechanics from classical arrival - time distributions @xcite , and thus provides a sort of `` ideal distribution . '' while the broad passage - time distribution for small @xmath125 can be understood simply as due to the poor quality of the individual measurements , the broad distribution for large @xmath125 is more interesting .
it can be understood looking at the reset state immediately after the detection by the first detector .
of course , this reset state depends on the instant of time when the detection took place ; as an example , we consider detection times @xmath166 if @xmath167 and @xmath168 if @xmath169 , which are close to the maximum of the respective probability distribution @xmath63 ( see fig . [ arrival_fig ] ) .
the reset states and the free packet at the respective times are shown in fig .
[ reset_x ] .
it is seen that the fast detection has a strong impact on the wave function .
large @xmath125 means that @xmath44 , in particular that part of @xmath44 which overlaps with the detector , decays very fast .
but since the reset state ( [ state ] ) immediately after the detection is essentially the projection of @xmath44 onto the detector region , the fast decay of this overlap yields a reset state which is very narrow in position space , located at the very beginning of the detector .
thus , by the heisenberg uncertainty relation , it is very broad in momentum space as can be seen in fig .
[ reset_k ] .
it is intuitively clear that such a broad momentum distribution immediately after the measurement of the entry time by the first detector yields a broad passage - time distribution .
so the broad passage - time distribution in case of large @xmath125 is due to the strong distortion of the wave function by the first measurement .
since the broad momentum distribution of the typical reset states in this case is enforced by the heisenberg uncertainty relation , this is a pure quantum effect .
in refs . @xcite a measurement scheme for passage times was investigated , based on the ( semi- ) continuous coupling of a particle to a clock .
it was argued in these references that the precision of the measurement behaves like @xmath1 where @xmath170 denotes the kinetic energy of the particle .
one now may wonder whether or not the @xmath1 behavior is a fundamental quantum limit for measuring passage times .
we argue that this is not the case since the present measurement scheme by means of two spatially separated detectors yields , for optimal parameter choices , passage - time densities with widths behaving like @xmath0 .
thus , for low energies , we have an example which breaks the @xmath1 limitation of the clock model . in this subsection
we give an estimate for the width of the passage - time density obtained from the present measurement scheme .
we assume that the detectors can be described by the continuum limit and that transmission and reflection without detection are negligible .
this assumption is justified in the examples of the preceding section , which employed rectangular sensitivity functions .
it can also be justified in general if one drops the restriction to rectangular sensitivity functions @xmath171 @xcite .
considering particles with mean velocity @xmath172 , the detector is assumed to be constructed in such a way that the first detection occurs with high probability in a spatial interval of length @xmath173 .
the length @xmath173 is related to @xmath125 of eq .
( [ a ] ) , an average detection rate , @xmath173 being of the order of @xmath174 .
the length @xmath173 imposes an upper limit on the width of the reset state in position space , @xmath175 by the heisenberg uncertainty relation , this immediately yields a lower bound for the width @xmath176 of the reset state in momentum space , @xmath177 note that this is only a very rough estimate , without taking into account details of the actual wave packet .
if the incident wave packet is very narrow in position space , then @xmath178 also may be much smaller than @xmath173 , and consequently @xmath179 may be much larger than @xmath180 .
also , the reset state may be far from being a gaussian , and then already the first inequality in eq .
( [ dp ] ) may underestimate the width of @xmath179 seriously .
let @xmath181 denote the width of the measured passage - time distribution .
there are several contributions to this width .
first , a particle with velocity @xmath172 takes the time @xmath182 to travel the distance @xmath183 between the two detectors , and therefore the width of the reset state in position space contributes @xmath184 to @xmath181 .
second , the width in momentum space contributes according to @xmath185 third , the width of the delay , @xmath186 , for the first detection due to the spin - boson interaction is roughly @xmath187 an estimate for the width of the passage - time distribution is given by the sum of these contributions , where @xmath186 has to be counted twice since it arises in both detectors : @xmath188 from this estimate it is again seen that both small as well as large values of @xmath125 , i.e. , both slow as well as fast detectors , lead to rather broad passage - time distributions ( due to @xmath189 and @xmath190 , respectively ) .
having established the general estimate for @xmath181 in eq.([delta_tau ] ) , we now turn to the task of finding optimal parameters , minimizing @xmath181 .
we are interested in measuring the passage time through a spatial interval of length @xmath183 , the distance between the starting points of the two detectors , which we regard as fixed .
first , we consider given detectors , i.e. , a given detection rate , @xmath125 .
again , particles with mean velocity @xmath172 would be detected within an interval with length @xmath173 given approximately by @xmath191 .
this means that the velocities must not be too large since in order to avoid undetected transmission @xmath173 must not exceed the actual length of the detector ( and in particular @xmath173 must not exceed @xmath183 ) .
quantum effects , however , are expected to play a role for slow particles while fast particles may be treated classically , hence this is not a serious drawback .
we assume that the reset state is a gaussian wave packet , or at least close to a gaussian , so that in the second line of eq.([delta_tau ] ) the approximate equality holds , @xmath192 we note that @xmath193 is a purely detector - related quantity , while the remaining contribution to @xmath181 is determined by the shape of the reset state only .
we will first optimize this latter quantity . for given particles with given mean velocity , i.e. , @xmath194 and @xmath172 fixed
, this is minimal for @xmath195 substituting this into eq .
( [ delta_tau_gauss ] ) and writing @xmath196 for the kinetic energy of the incident particles yields @xmath197 the subscript `` @xmath198 '' indicates that only the reset state was optimized while the detection rate @xmath125 was considered as a given quantity . aiming at optimizing also the detection rate @xmath125 for minimal @xmath199 , one would like to choose @xmath125 as large as possible in order to reduce the @xmath193 contribution to @xmath181 .
however , one has to take into account that the width of the reset state is bounded by eq.([reset_x_bound ] ) .
thus , given @xmath183 , @xmath194 , and @xmath172 , the decay rate must be at most of the order of @xmath200 with @xmath201 as in eq .
( [ delta_x_opt ] ) .
in fact , we may choose the incoming state such that it forms a gaussian minimal uncertainty packet at the starting point of the first detector with width in position space @xmath202 and choose further @xmath203 we note that , by this parameter choice , yet another requirement of a good measurement scheme is fulfilled : the detection by the first detector will not change the wave function too strongly .
the reset operation after this detection is essentially a projection onto the detector region , and the detection is slow enough that at typical detection times most of the wave packet overlaps with the detector , hence the projection does not change the wave function too much .
thus , the reset state will be close to a gaussian wave packet with width @xmath204 . substituting eq.([a_opt ] ) into eq .
( [ delta_tau_opt_reset ] ) finally yields @xmath205 we stress that , independent of the detection rate @xmath125 , the optimal energy dependence of @xmath181 is limited to @xmath0 already by the dependence of @xmath181 on the width of the reset state in position space , @xmath206 [ eq .
( [ delta_tau_x ] ) ] , and on its width in momentum space , @xmath207 [ see eq.([delta_tau_p ] ) ] .
for the example of a cesium atom with @xmath208 and a distance of @xmath209 between the detectors ( with rectangular sensitivity function ) of the preceding section , optimal values would be according to eqs .
( [ delta_x_opt ] ) and ( [ a_opt ] ) @xmath210 considering the wave packet with @xmath211 actually investigated in the examples , the optimal decay rate according to eq .
( [ a_opt ] ) would be @xmath212 .
this is consistent with the observation that the example with @xmath125 closest to @xmath213 yields the narrowest distribution .
we have investigated the continuum limit of a fully quantum mechanical spin - boson model for the detection of a moving particle and its application to passage - time measurements .
the continuum limit has been derived under the condition that the spin - boson interaction satisfies the markov property , an assumption that was explicitly verified in ref .
analytical expressions for the state immediately after the first detection have been obtained , and for an example with a simplified detector model and 15 boson modes it was shown numerically that the continuum limit is in very good agreement with the discrete model .
further , analytical expressions for the passage - time distribution have been obtained , and numerical examples for passage - time measurements have been discussed .
detectors with a very slow response yield broad distributions , due to the poor quality of the individual measurements , and so do very fast detectors , due to the strong distortion of the particle s wave function by the measurement .
intermediate detectors yield narrower distributions .
the optimal precision of the present measurement scheme has been estimated to behave like @xmath0 , where @xmath170 is the kinetic energy of the incident particle . for slow particles
this is better than , and in contrast to , a scheme based on quantum clocks which yields an @xmath1 behavior .
this work was supported in part by nsf grant phy 0555313 . | the passage - time distribution for a spread - out quantum particle to traverse a specific region is calculated using a detailed quantum model for the detector involved .
that model , developed and investigated in earlier works , is based on the detected particle s enhancement of the coupling between a collection of spins ( in a metastable state ) and their environment .
we treat the continuum limit of the model , under the assumption of the markov property , and calculate the particle state immediately after the first detection .
an explicit example with 15 boson modes shows excellent agreement between the discrete model and the continuum limit .
analytical expressions for the passage - time distribution as well as numerical examples are presented .
the precision of the measurement scheme is estimated and its optimization discussed . for slow particles ,
the precision goes like @xmath0 , which improves previous @xmath1 estimates , obtained with a quantum clock model . |
in the mixed state of type - ii superconductors the magnetic field is concentrated in an array of flexible flux bundles that , much like ordinary matter , can form crystalline , liquid and glassy phases.@xcite in clean systems the vortex solid melts into a flux liquid via a first order phase transition.@xcite if the barriers to vortex line crossing are high , a rapidly cooled vortex liquid can bypass the crystal phase and get trapped in a metastable polymer - like glass phase.@xcite the diversity of vortex structures is further increased by pinning from material disorder , which leads to a variety of novel glasses .
disorder - driven glass transitions are continuous , with diverging correlation lengths and universal critical behavior.@xcite of particular interest is the dynamics of the vortex array in the various phases and in the proximity of a phase transition . in the liquid phase the vortex array flows yielding a linear resistivity . in the presence of large scale spatial inhomogeneities , the liquid flow can be highly nonlocal due to interactions and entanglement.@xcite the correlation length controlling the nonlocality of the flow grows with the liquid shear viscosity , which becomes large as the liquid freezes . at a continuous liquid - glass transition this correlation length diverges with a universal critical exponent . in the solid phase the vortex array moves as a single elastic object under uniform drive ,
provided the shear stresses are not too large . in the presence of strong spatial inhomogeneities
, plastic flow occurs for large drives ( or even for vanishingly small drives in a glassy solid ) and the response is always nonlinear.@xcite the dynamical correlation length can be identified with the separation between free dislocations and diverges at a continuous melting transition . probing spatial velocity correlations can therefore give information on vortex dynamics within a given phase , as well as on the nature of the phase transitions connecting the various phases .
as for ordinary matter , the shear rigidity of the vortex array can be probed by forcing the vortices to flow in confined geometries .
@xcite this type of experiments was pioneered by kes and collaborators to study the shear rigidity of the two - dimensional vortex liquid near freezing in thin films.@xcite more recently , patterned irradiation of cuprate superconductors with heavy ions has made it possible to create samples with controlled distributions of damage tracks.@xcite we recently showed that an analysis of such experiments that combines an inhomogeneous scaling theory with the hydrodynamics of viscous flux liquids can be used to infer the critical behavior near a continuous glass transition , as well as to distinguish between continuous transitions , such as that to a bose glass , and nonequilibrium transition to a polymer - like glass driven by interaction and entanglement.@xcite .
, in the liquid annulus of an irradiated corbino disk .
the inner and outer radii are @xmath0 and @xmath1 , and @xmath2 .
the dashed line is the @xmath3 field profile in an uncorrelated liquid , with @xmath4 .
inset : a sketch of the disk the bose glass contacts are not shown.,title="fig:",width=220 ] [ corbino1 ] large scale spatial inhomogeneities can also be introduced in the flow , even in the absence of pinning , by applying a driving force with controlled spatial gradients , as done recently by the argonne group using the corbino disk geometry.@xcite in this paper we illustrate the analysis of spatially inhomogeneous vortex motion in both the liquid and the solid using the corbino disk as a prototype of a novel class of experiments exploiting the effect of geometry to study the dynamics of vortex matter .
in the corbino disk , with magnetic field along the disk axis ( @xmath5 direction ) , a uniform radial current density of magnitude @xmath6 is introduced in the sample by injecting current at the center and removing it at the outer circumference of the disk ( inset of fig .
1 ) . the current drives the vortices to move in circles about the axis . in the flux liquid , the dynamics on scales larger than the intervortex spacing , @xmath7 , is described by hydrodynamic equations for the flow velocity @xmath8 , which determines the local field from flux motion , @xmath9 , with @xmath10 . for simple geometries like the corbino disk , where the current is spatially homogeneous in the @xmath5 direction ,
hydrodynamics reduces to a single equation , @xcite @xmath11 where @xmath12 is the friction , @xmath13 is the viscosity controlling the viscous drag from interactions and entanglement , and the term on the right hand side is the lorentz force density driving flux motion .
it is instructive to rewrite eq .
( [ hydro ] ) as an equation for the local field,@xcite @xmath14 with @xmath15 the viscous correlation length and @xmath16 the flux flow resistivity .
if the viscous force is negligible , eq .
( [ viscouse ] ) is simply ohm s law and the radial field is @xmath17 . to probe the viscous drag ,
it is necessary to force large scale spatial inhomogeneities in the flow .
this may be achieved by suitable pinning boundaries . as an example , we imagine selectively irradiating a cylindrical central region and an outer annular region of the disk to obtain the structure sketched in the inset of fig .
the vortices in the heavily irradiated central and outer regions ( shaded ) are in the bose glass phase , while vortices in the unirradiated ( white ) annular region are in the flux liquid phase . a radial current drives tangential flow in the resistive flux liquid annulus , which is impeded by the `` bose - glass contacts '' at the boundaries . the field profile obtained by solving eq .
( [ viscouse ] ) with no - slip boundary conditions @xcite is spatially inhomogeneous on length @xmath18 , as shown in fig . 1 .
one can probe this profile and extract @xmath18 by placing a string of radial contacts at @xmath19 , for @xmath20 , and measuring the voltage @xmath21 across each successive pair ( inset of fig .
if the viscosity is small ( @xmath22 ) , the voltage decreases logarithmically as one moves from the inner to the outer contacts , as in a freely flowing uncorrelated liquid , where @xmath23 .
when @xmath18 grows , the onset of rigidity in the liquid becomes apparent ( fig .
an elastic vortex solid would rotate as a rigid object under the radial drive , with @xmath24 and @xmath25 , for @xmath26 . indeed for @xmath27 , @xmath21 is no longer monotonic with @xmath28 and it exhibits a solid - like growth with @xmath28 within a boundary layer of width @xmath18 . across pairs of contacts
@xmath29 , with @xmath30 , for @xmath31 , @xmath32 , and @xmath33 the contact spacing .
the symbols refer to @xmath34 ( triangles ) , @xmath35 ( squares ) and @xmath36 ( circles ) .
solid lines are guides to the eye .
inset : top view of the corbino disk with bose glass contacts , with @xmath37.,title="fig:",width=240,height=211 ] [ corbino ] the resistance per unit thickness of the disk , @xmath38 , is @xmath39 with @xmath40 a function determined by the geometry that is obtained from the solution of the hydrodynamic equation . if @xmath41 , viscous effect are not important and @xmath42 .
conversely , when the viscous length exceeds the sample size ( @xmath43 ) , then @xmath44 , for @xmath26 .
when the effects of geometry are negligible , resistivity measurements directly probe the friction , @xmath45 , while measurements in channels narrow compared to @xmath18 probe the flux liquid viscosity , @xmath46 .
experiments in artificial pinning structures can therefore be used to infer both @xmath45 and @xmath46 .
the behavior of these two parameters near a phase transition can in fact be used to identify the transition , as summarized in table 1 .
.behavior of @xmath45 and @xmath46 near various transitions of vortex matter . here
@xmath47 , with @xmath48 the relevant transition temperature in each case . [ cols="<,<,<",options="header " , ] + of particular interest is the case of continuous glass transitions , which are characterized by universal critical behavior.@xcite experiments of the type just described are especially powerful in this case as they can be used to map out the critical behavior.@xcite one can probe the bose glass transition by lightly irradiating the liquid annular region , and then lowering the temperature at constant field from @xmath49 to @xmath50 .
most physical properties near the transition can be described via a scaling theory in terms of diverging correlation lengths perpendicular and parallel to the field direction , @xmath51 and @xmath52 , with @xmath53 , and a diverging correlation time , @xmath54.@xcite scaling can then be used to relate physical quantities to these diverging length and time scales . in particular ,
the friction coefficient @xmath45 that determines the bulk flux flow resistivity @xmath55 is predicted to diverge as @xmath56 as @xmath57 .
@xcite as shown recently by marchetti and nelson , when flux flow in confined geometries is analyzed by combining hydrodynamics with the the bose glass scaling theory
generalized to the spatially inhomogeneous case
the bose glass correlation length @xmath58 is naturally identified with the viscous length @xmath18.@xcite it then follows that the liquid shear viscosity diverges at @xmath59 as @xmath60 .
furthermore , the scaling of the finite - geometry resistivity displayed in eq . [ rhor ] is a general property of the vortex liquid near a continuous glass transition .
hydrodynamics yields the _ precise _ form of the scaling function , which depends on the experimental geometry and can be found in @xcite for the corbino disk .
as shown in recent experiments by the argonne group , the corbino disk geometry can also be used to study the onset of plastic flow in a driven solid.@xcite in this case we consider an unirradiated disk , where the vortex array has a clear melting transition .
below melting the vortex solid moves as a rigid body , with @xmath24 , and the voltage grows as @xmath61 .
the @xmath3 dependence of the driving force yields , however , large elastic deformations of the medium , described by the solution of @xmath62 with free boundary conditions ( @xmath63 is the shear modulus of the lattice , assumed incompressible ) .
elastic deformations yield a finite shear stress , @xmath64 $ ] , that can unbind dislocations from bound pairs .
assuming for simplicity that vortices are straight along the field direction , the energy of a pair of dislocations of opposite burgers vectors , separated by a distance @xmath65 , is @xmath66 - 2e_ct$ ] , with @xmath67 the angle between @xmath68 and @xmath69 and @xmath70 the core energy per unit length of an edge dislocation .
an applied stress pulls the two dislocations in opposite directions.@xcite ignoring climb , and assuming the spatial variations of the stress field are on scales large compared to @xmath71 , the interaction energy is now @xmath72 , with @xmath73 the radial location of the center of mass of the pair .
the applied stress lowers the barrier that confines the bound pair . in the simplest model ,
unbinding occurs where the location @xmath74 of the barrier , defined by @xmath75_{x = x_b}=0 $ ] , becomes comparable to @xmath7 , or @xmath76 . by solving for @xmath73 ,
one obtains the critical radius @xmath77 where shear - induced dislocation unbinding occurs , yielding slippage of neighboring planes of the vortex lattice , @xmath78 , for @xmath26 , with @xmath79 a maximum current above which the entire vortex solid shear melts .
the melting radius @xmath80 increases with current , indicating that , since the stresses are largest near the axis of the disk , `` shear - induced melting '' occurs first in circular layers close to the axis .
this behavior is qualitatively consistent with the observations by the argonne group.@xcite the simple model described here suggests that at high fields the current scale @xmath81 is independent of field .
a more detailed calculation incorporating field and temperature dependence will be described elsewhere.@xcite | the fabrication of artificial pinning structures allows a new generation of experiments which can probe the properties of vortex arrays by forcing them to flow in confined geometries .
we discuss the theoretical analysis of such experiments in both flux liquids and flux solids , focusing on the corbino disk geometry . in the liquid , these experiments can probe the critical behavior near a continuous liquid - glass transition . in the solid , they probe
directly the onset of plasticity . |
the lennard - jones potential energy problem is to construct the most stable form of @xmath0 atoms of some material with the minimal energy structure .
its form in mathematics is very simple : @xmath8 where @xmath2 , @xmath9 is the coordinates of atom @xmath4 , @xmath6 , and @xmath7 .
however , it interests many researchers in the field of biology , physics , chemistry , mathematical optimization , computer science , and materials science .
the reason lies in the nonconvexity of the objective function and the huge number of local minima , which is growing exponentially with @xmath0 .
@xcite tells us the number of distinct local minima of @xmath0-atoms lennard - jones problem is about @xmath10 .
for example , when @xmath116 , 7 , 8 , 9 , 10 , 11 , 12 , 13 the numbers of distinct local minima of the problem are 2 , 4 , 8 , 18 , 57 , 145 , 366 , 988 respectively ( @xcite ) .
from the point of view of numerical optimization methods , this problem is an excellent test problem for local or global optimization methods .
however , the lennard - jones problem is very difficult and challenging to many optimization methods even when @xmath0 is not very large . by now , the best objective function vlues for @xmath12 @xcite are well known . at first
, around 1972 hoare and pal @xcite gave the best values for atoms 5@xmath1316 , 18@xmath1321 , 25 , 26 , 29 , 55 .
then , in 1987 , northby ( @xcite ) made a landmark@xmath14yielding most of the lowest values in the range @xmath15 ; and in 1999 @xcite presented the best results on @xmath16 with the exception of @xmath11192 , 201 @xcite , @xmath04=200 , 300 @xcite , @xmath11185 @xcite , @xmath11186 , 187 @xcite . at present , about thirty percent of northby s results have been improved @xcite .
the improvements are described as follows . in 1994
xue ( @xcite ) got the best values for @xmath1165,66,134 .
doye et al ( @xcite ) in 1995 gave the best values on @xmath0=38,75@xmath1377,102@xmath13104 ; and leary in 1997 ( @xcite ) presented best values for @xmath1169,78,88,107,113,115 .
wolf ( @xcite ) got the best values on @xmath1182 , 84 , 86 , 88 , 92 , 93 , 94 , 95 , 96 , 99 , 100 in 1998 . for the left so - called magic numbers sequence : 17,23,24,72,88 , freeman , farges , wille , coleman , deaven ( @xcite ) got the best results , respectively . the outline of this paper is as follows . in section 2 , we review the methods in the most important references above - mentioned .
section 3 contains a description of our techniques and methods to tackle problem ( [ ljproblem ] ) ; and results of numerical experiments for the problem are given in this section .
the structure of our optimal solutions is illustrated , and can be compared with those of others , at the end of this section .
section 4 concludes the paper .
some good summaries on the methods solving the lennard - jones problem in fact can be found in @xcite . in this section
we review the methods in the most important references @xcite .
hoare and pal s work ( @xcite ) may be the early most successful results on lennard - jones problem .
the idea is using build - up technique to construct the initial solutions which are expected to represent low energy states , and using those initial solutions as starting points for a local search method to relax to the optimal solution ( @xcite ) .
the starting seed is the regular unit tetrahedron with atoms at the vertices , the obvious global optimal solution for @xmath17 . beginning with this tetrahedron
, hoare and pal added one atom at a time to construct a sequence of polytetrahedral structures and at last got good results up to @xmath18 .
for example , for @xmath19 its globally optimal trigonal bi - pyramid ( bi - tetrahedron ) structure is gotten by adding an atom at the tetrahedral capping position over a triangular face ; following the bi - tetrahedron structure , the optimal structure of @xmath20 is tri - tetrahedron ( another known optimal structure for @xmath20 is octahedron ( using tetrahedral capping over triangular faces and half - octahedral capping over square faces ) , which is not a polytetrahedron ) ; for @xmath21 its best structure constructed is the pentagonal bi - pyramid , a structure with a five - fold axis of symmetry .
many computer science data structure procedures such as greedy forward growth operator and reverse greedy operator can make the build - up technique work well .
@xcite describes the application of methods of studying noncrystalline clusters to the study of spherical " face centred cubic ( fcc ) microcrystallites . @xcite
describes the chief geometrical features of the clustering of small numbers of interacting particles .
the data structure of northby ( @xcite ) in finding the good starting solution is the lattice based structure .
the lattice structures consist of an icosahedral core and particular combinations of surface lattice points .
@xcite first constructed a class of icosahedral packings by adding successively larger icosahedral shells in layers around a core central atom ; this icosahedral lattice can be described as 20 slightly flattened tetrahedrally shaped fcc units with 12 vertices on a sphere centered at the core atom .
atoms within each triangular face are placed in staggered rows in a two dimensional hexagonal close - packed arrangement .
each atom in the interior of a face in a given shell is a tetrahedral capping position relative to three atoms in the underlying shell .
northby relaxed the structure of @xcite to get his ic and fc multilayer icosahedral lattice structures .
the ic lattice can be referred to the fortran code in @xcite ; it consists of all those sites which will comprise the outer shell of the next complete mackay ( @xcite ) icosahedron .
fc lattice is a slight modification of ic lattice in that its outer shell maintains icosahedral symmetry and consists of points at the icosahedral vertices and the stacking fault positions of the outer ic shell .
basing on the ic and fc lattices , northy gave his algorithm first finding a set of lattice local minimizers and then relaxing those lattice minimizers by performing continuous minimization starting with those lattice minimizers .
the algorithm was summarized as algorithm 1 and algorithm 2 of @xcite .
the great majority of the best known solutions of northy are icosahedral in character .
the hybridization of global search and local search methods , usually , is more effective to solve the large scale problem than the global search method or local search method working alone .
catching those two ideas , @xcite combined a genetic algorithm with a stochastic search procedure on icosahedrally derived lattices .
the structures of the optimal solutions gotten in @xcite are either icosahedral or decahedral in character .
the best results of @xcite for @xmath22 were gotten by using a genetic algorithm alone .
deaven et al ( @xcite ) also using the genetic algorithm got the optimal value known for the magic number @xmath23 .
the successful works to improve northby s results ( @xcite ) were mainly done by xue ( @xcite ) , leary ( @xcite ) , and doye et al ( @xcite ) .
xue ( @xcite ) introduced a modified version of the northby algorithm .
he showed that in some cases the relaxation of the outer shell lattice local minimizer with a worse potential function value may lead to a local minimizer with a better value . in northby
s algorithm the lattice search part is a discrete optimization local search procedure , which makes a lattice move to its neighboring lattice with @xmath24 time complexity . in @xcite xue
introduced a simple storage data structure to reduce the time complexity to @xmath25 per move ; and then used a two - level simulated annealing algorithm within the supercomputer cm-5 to be able to solve fastly the lennard - jones problem with sizes as large as 100,000 atoms . in @xcite by employing avl trees ( @xcite ) data structure xue furthermore reduced the time complexity to @xmath26 if nn ( nearest neighbor ) potential function is used .
he ( @xcite ) relaxed _ every _ lattice local minimizer found instead of relaxing only those lattice local minimizers with best known potential function value by a powerful truncated newton local search method , and at last got the best results known for @xmath27 .
leary ( @xcite ) gave a successful big bang algorithm in 1997 for getting the best values known of @xmath28 . in @xcite the fcc lattice structure is discussed and its connections are made with the macrocluster problem .
it is also concluded in @xcite that almost all known exceptions to global optimality of the well - known northby multilayer icosahedral conformations for microclusters are shown to be minor variants of that geometry .
the big bang algorithm contains 3 steps : step 1 is an initial solution generating procedure which randomly generates each coordinate of the initial solution with the independently normal distribution ; step 2 is to generate the new neighborhood solution by discrete - typed fixed step steepest descent method , which is repeated until no further progress is made ; step 3 is to relax the best solution gotten in step 2 by a continuous optimization method@xmath14conjugate gradient method .
doye et al ( @xcite ) investigated the structures of clusters by mapping the structure of the global minimum as a function of both cluster size and the range of the pair potential which is appropriate to the clusters of diatomic molecule , @xmath29 molecule , and the ones between them both .
for the larger clusters the structure of the global minimum changes from icosahedral to decahedral to fcc as the range is decreased ( @xcite ) . in @xcite doye
et al predicted the growth sequences for small decahedral and fcc clusters by maximisation of the number of nn contacts .
lastly , in this section , we give a brief review of methods on the magic numbers @xmath30 and on the exceptions to @xcite .
freeman et al ( @xcite ) presented the best value for @xmath31 when the thermodynamic properties of argon clusters were studied by a combination of classical and quantum monte carlo methods .
the polyicosahedral growth of farges et al ( @xcite ) starts from a 13-atom primitive icosahedron containing a central atom and 12 surface atoms . on each one of the five tetrahedral sites , surrounding a particular vertex ,
a new atom is added and finally a sixth atom is placed on top to create a pentagonal cap . in this way a 19-atom structure being made of double interpenetrating icosahedra , which is a 13-atom icosahedra sharing 9 atoms , is obtained .
i.e. , for three pentagonal bipyramids each one shares an apex with its nearest neighbour . in this way a 23-atom model consisting of three interpenetrating icosahedra is gotten for the best value known .
wille ( @xcite ) used the simulated annealing method yielding low - lying energy states whose distribution depends on the cooling rate to find the best solution known for @xmath32 .
coleman et al ( @xcite ) proposed a build - up process to construct the optimal solution structures .
the hoc ( half octahedral cap ) structure of the optimal solution for @xmath33 is found by a prototype algorithm designed using the anisotropic effective energy simualted annealing method at each build - up stage ( @xcite ) .
wales and doye ( @xcite ) in 1997 gave the lowest values known for @xmath34 .
their method is so - called basin - hopping method , in which first the transformed function @xmath35 was defined and performed starting from @xmath36 by the pr conjugate gradient method ( see , for example , @xcite ) , and then the energy landscape for the function @xmath37 was explored using a canonical monte carlo simulation .
leary ( @xcite ) has developed techniques for moving along sequences of local minima with decreasing energies to arrive at good candidates for global optima and got the best value known on @xmath38 .
in this section we first briefly introduce the discrete gradient method ; then present a hybrid discrete gradient and simulated annealing method .
and then we give some techniques , with the discrete gradient method and the hybrid method , to tackle the lennard - jones problem .
discrete gradient method is a derivative - free local search method for nonsmooth optimization ( @xcite ) .
results of numerical experiments presented in @xcite show that this method can jump over stationary points , which are not local minima , so we can reduce the number of stationary points , which we meet .
details of this method can be found , for example , in @xcite .
simulated annealing method is a global search method .
it has received a great deal of attention in last several years .
first , this method was applied to combinatorial optimization , i.e. when the objective function is defined in a discrete domain ( see @xcite ) . later
this method was applied to solve continuous global optimization problems ( see @xcite ) .
convergence of simulated annealing method is studied , for example , in @xcite .
numerical experiments show that algorithms based on a combination of the global and local search techniques are more effective than the global search methods working alone ( see , for example , @xcite ) . in these methods
a local search algorithm is used to find a stationary point and a global search algorithm is used to escape from this stationary point and to find a new starting point for a local search algorithm . in this subsection , we develop a new hybrid simulated annealing and discrete gradient method . during the whole process of the method ,
both simulated annealing and discrete gradient methods need the the objective function values only .
the simulated annealing method for solving continuous global optimization problems was studied by many authors ( see , for example , @xcite ) .
_ neighborhood solution search _ procedure is one of the important steps in this method .
noticing that the neighborhood solution search for simulated annealing method should be at least based on two basic ideas : ( a ) neighbor means nearby " , ( b ) simulated annealing method is a stochastic method so that the neighborhood solution should be randomly taken , we may simply give a neighborhood solution search procedure for simulated annealing algorithm : when the feasible region of the optimization problem is the unit simplex @xmath39 , the neighborhood solution search procedure should be done a little modification : the hybrid method starts from an initial point , first executes discrete gradient method to find local minimum , then carries on simulated annealing method in order to escape from this local minimum and to find a new starting point for the discrete gradient method .
then we again apply the discrete gradient method starting from the last point and so on until the sequence of the optimal objective function values gotten is convergent .
the pseudo - code of the hybrid method is listed as following : 0.3 cm algorithm 1 : hybrid simulated annealing and discrete gradient method + 0.4 cm the convergence of the proposed hybrid method directly follows from the convergence of the simulated annealing and the discrete gradient methods .
the simulated annealing method part consists of two procedures : inner procedure for the search of neighborhood solution and outer procedure for the cooling temperature @xmath40 .
the number of iterations of inner and outer loops are taken to be large enough for guaranteeing that the sufficient iterations will be carried on to escape from the current local minimum . in implementing the hybrid method ,
we use @xmath41 as the temperature annealing schedule and the initial temperature is taken large enough according to the rule in @xcite .
we restrict the number of iterations for the outer procedure by 100 and number of iterations for the inner procedure by 1000 .
the discrete gradient method part is terminated when the distance between the approximation to the subdifferential and origin is less than a given tolerance @xmath42 ( @xmath43 ) . in this subsection ,
the proposed discrete gradient method and hybrid discrete gradient and simulated annealing method , with the techniques given in the below , have been applied for minimization of well - known lennard - jones potential function in protein folding problem ( [ ljproblem ] ) .
the nonconvex lennard - jones potential function has a huge number of local minima .
therefore many global optimization techniques fail to minimize it .
the proposed hybrid method fails to solve this problem when number of atoms @xmath44 . in order to reduce the number of local minima we suggest to approximate the function @xmath45 by the following function : @xmath46 where @xmath47 is the piecewise linear approximation of the function @xmath48 in segment @xmath49 $ ]
, @xmath50 is the piecewise linear approximation of this function over segment @xmath51 $ ] , and finally @xmath52 is the piecewise linear approximation over @xmath53 $ ] and @xmath54 is large enough number . here @xmath55{2},~~r_1=1/\sqrt[3]{2/7}\ ] ] such a approximation of the function @xmath48 allows us to remove many local minima of the lennard - jones potential function and to get a good approximation to the global minimum of the objective function @xmath56 in problem ( [ ljproblem ] ) . in numerical experiments
we take @xmath57 and divide the segment @xmath58 $ ] into 100 segments , the segment @xmath51 $ ] into 100 segments and the @xmath59 $ ] into 50 segments which allows one to get good approximations for the function @xmath48 .
the replacement of the function @xmath48 by the function @xmath60 makes the objective function nonsmooth . on the other side such a replacement
significantly reduce the number of local minima .
since the discrete gradient method is a method of nonsmooth optimization the proposed hybrid method can be applied for solving this transformed problem .
when solving the lennard - jones problem , first we use the discrete gradient method with build - up technique to relax to an initial solution .
then we apply the hybrid discrete gradient and simulated annealing method , with the good approximation for the objective function , to get another initial solution . starting from this initial solution
we again apply the derivative - free discrete gradient method and at last get the global solution .
results of numerical experiments are presented in table [ table1 ] as the appendix .
results from table [ table1 ] show that our techniques and methods effectively solve protein folding problem when number of atoms is not greater than 310 .
.our numerical results for lennard - jones protein problem [ cols="^,^,^",options="header " , ] [ table1 ] the structure of the optimal solutions corresponding to the above optimal values is illustrated and can be compared with the ones of @xcite in the following figures : [ ouratoms ] [ ouratoms ] [ ouratoms ] [ ouratoms ] [ ouratoms ] [ ouratoms ] [ ouratoms ] [ ouratoms ] [ ouratoms ] [ ouratoms ]
in this paper , a brief review and a bibliography of important computational algorithms on minimizing the lj potential energy are introduced in sections 1 and 2 .
section 3 of this paper illuminates many beautiful graphs for the three dimensional structures of molecules with minimal lj potential .
t. coleman , d. shalloway and z. j. wu , a parallel build - up algorithm for global energy minimizations of molecular clusters using effective energy simulated annealing " , _ j. global optimization _ * 4 * ( 1994 ) 171 - 185 .
a. corana , m. marchesi , c. martini , and s. ridella , minimizing multimodal functions of continuous variables with the simulated annealing algorithm " , _ acm transactions on mathematical software _ * 13 * ( 1987 ) 262 - 280 .
d. l. freeman and j. d. doll , quantum monte carlo study of the thermodynamic properties of argon clusters : the homogeneous nucleation of argon in argon vapor and magic number `` distributions in argon vapor '' , _ j. chem . phys . _
* 82 * ( 1985 ) 462 - 471 .
hedar and m. fukushima , hybrid simulated annealing and direct search method for nonlinear global optimization " , _ technical report , department of applied mathematics and physics , kyoto university _ * 2001 - 013 * ( december 2001 ) .
hedar and m. fukushima , simplex coding genetic algorithm for the global optimization of nonlinear functions " , _ technical report , department of applied mathematics and physics , kyoto university _ * 2002 - 006 * ( march 2002 ) . | the lennard - jones ( lj ) potential energy problem is to construct the most stable form of @xmath0 atoms of a molecule with the minimal lj potential energy .
this problem has a simple mathematical form @xmath1 where @xmath2 , @xmath3 is the coordinates of atom @xmath4 in @xmath5 , @xmath6 , and @xmath7 ; however it is a challenging and difficult problem for many optimization methods when @xmath0 is larger . in this paper ,
a brief review and a bibliography of important computational algorithms on minimizing the lj potential energy are introduced in sections 1 and 2 .
section 3 of this paper illuminates many beautiful graphs ( gotten by the author nearly 10 years ago ) for the three dimensional structures of molecules with minimal lj potential . * a brief review on results and computational algorithms for minimizing the lennard - jones potential * * jiapu zhang * _ centre for informatics and applied optimization & + graduate school of information technology and mathematical sciences , + university of ballarat , victoria 3353 , australia _ 0.2 cm |
in their celebrated paper @xcite , otto and villani proved that , in a smooth riemannian setting , the log - sobolev inequality implies the talagrand s transport - entropy inequality @xmath0 .
later , bobkov , gentil and ledoux @xcite proposed an alternative proof of this result .
both approaches are based on semi - group arguments .
more recently , the first named author gave a new proof , based on large deviation theory , valid on metric spaces @xcite . in this paper , on the one hand ,
we give yet another proof of otto and villani s theorem .
this proof does not use any semi - group argument nor large deviation and requires very few structure on the space .
we are thus able to recover and extend the result of @xcite in a general metric space framework . on the other hand , we recently introduced @xcite a new functional inequality , called inf - convolution log - sobolev inequality .
we proved that , in a euclidean framework , it is equivalent to the talagrand transport - entropy inequality @xmath0 , leading to a new characterization of such an inequality .
the present paper establishes the equivalence in a general metric space framework . as a byproduct , we prove that the inequality @xmath0 is stable under bounded perturbation ( holley - stroock perturbation lemma ) .
our strategy is very general and applies for a very large class of transport - entropy inequalities . in order to present our results ,
we need first to fix some notation .
we first introduce the notion of optimal transport cost .
then we give the definition of the transport - entropy inequality and of the @xmath1-log - sobolev inequality .
* general assumption . * in all this paper @xmath2 will always be a complete , separable metric space such that closed balls are compact
. + let @xmath3 be a continuous function .
given two probability measures @xmath4 and @xmath5 on @xmath6 , the _
optimal transport cost _ between @xmath4 and @xmath5 ( with respect to the cost function @xmath7 ) is defined by @xmath8 where the infimum runs over all the probability measures @xmath9 on @xmath10 with marginals @xmath4 and @xmath5 .
the notion of optimal transport cost is very old ( it goes back to monge @xcite ) .
it has been intensively studied and it is used in a wide class of problems running from geometry , pde theory , probability and statistics , see @xcite . here
we focus on the following transport - entropy inequality . in all the paper , the cost functions @xmath7 will be assumed to belong to the class of young functions .
( young functions ) a function @xmath11 is a _ young function _ if @xmath7 is an even convex , increasing function on @xmath12 such that @xmath13 and @xmath14 [ def : tci ] let @xmath7 be a young function ; a probability measure @xmath5 on @xmath6 is said to satisfy the _ transport - entropy inequality @xmath15 _ , for some @xmath16 if @xmath17 where @xmath18 is the relative entropy of @xmath4 with respect to @xmath5 and @xmath19 is the set of all probability measures on @xmath6 .
[ rem alpha en 0 ] it can be shown that if @xmath11 is an even convex function such that @xmath20 then the only probability measures that satisfy the transport inequality @xmath21 are dirac masses ( see e.g ( * ? ? ?
* proposition 2 ) ) .
this is the reason why , in our definition of young functions , we impose that @xmath22 .
popular young functions appearing in the literature , as cost functions in transport - entropy inequalities , are the functions @xmath23 defined by @xmath24 ( the case @xmath25 can be discarded according to the remark above ) .
when @xmath26 , we use the notation @xmath27 instead of @xmath28 transport - entropy inequalities imply concentration results as shown by marton @xcite , see also @xcite , @xcite , and @xcite for a full introduction to this notion .
the transport - entropy inequality related to the quadratic cost @xmath29 is the most studied in the literature . in this case , the transport - entropy inequality is often referred to as the talagrand transport - entropy inequality and is denoted by @xmath0 .
talagrand @xcite proved that , on @xmath30 ( where @xmath31 stands for the euclidean norm ) , the standard gaussian measure satisfies @xmath0 with the optimal constant @xmath32 .
the second inequality of interest for us is the log - sobolev inequality and more generally modified log - sobolev inequalities . to define these inequalities properly ,
we need to introduce additional notation .
recall that the fenchel - legendre transform @xmath33 of a young function @xmath7 is defined by @xmath34 a function @xmath35 is said to be _ locally lipschitz _ if for all @xmath36 , there exists a ball @xmath37 centered at point @xmath38 such that @xmath39 when @xmath40 is locally lipschitz , we define @xmath41_+}{d(y , x ) } & \mbox{if } x \text { is not an isolated point } \\ 0 & \mbox{otherwise } , \end{array } \right.\ ] ] and @xmath42_-}{d(y , x ) } & \mbox{if } x \text { is not an isolated point } \\ 0 & \mbox{otherwise } , \end{array } \right.\ ] ] where @xmath43_+=\max(a;0)$ ] and @xmath43_-=\max(-a;0)$ ] .
note that @xmath44 and @xmath45 are finite for all @xmath46 when @xmath40 is a smooth function on a smooth manifold , @xmath47 and @xmath48 equal the norm of the gradient of @xmath40 . finally , if @xmath5 is a probability measure on @xmath6 , recall that the entropy functional @xmath49 is defined by @xmath50 [ def : mlsi ] let @xmath7 be a young function ; a probability measure @xmath5 on @xmath6 is said to satisfy the _ modified log - sobolev inequality plus _ @xmath51 for some @xmath52 if @xmath53 for all locally lipschitz bounded function @xmath54 .
+ it verifies the _ modified log - sobolev inequality minus _
@xmath55 for some @xmath52 if @xmath56 for all locally lipschitz bounded function @xmath54 .
again , the quadratic cost @xmath29 plays a special role since in this case we recognize the usual log - sobolev inequality introduced by gross @xcite ( see also @xcite ) . in this case
, we will use the notation @xmath57 bobkov and ledoux @xcite introduced first the modified log - sobolev inequality with the function @xmath58 , in order to recover the celebrated result by talagrand @xcite on the concentration phenomenon for products of exponential measures . in particular these authors proved that , with this special choice of function , the modified log - sobolev inequality is actually equivalent to the poincar inequality .
after them , gentil , guillin and miclo @xcite established that the probability measure @xmath59 , @xmath60 and @xmath61 verifies the modified log - sobolev inequality associated to the function @xmath62 .
in a subsequent paper @xcite they generalized their results to a large class of measures with tails between exponential and gaussian ( see also @xcite and @xcite ) . finally , let us introduce the notion of inf - convolution log - sobolev inequality .
in a previous work @xcite , we proposed the following inequality @xmath63 where @xmath64 we called it inf - convolution log - sobolev inequality and we proved that it is equivalent - in a euclidean setting - to the transport - entropy inequality @xmath65 , for young functions @xmath7 such that @xmath66 is concave . also , we get an explicit comparison between the constants @xmath67 and @xmath68 , namely @xmath69 .
our proof relies in part on the hamilton - jacobi semi - group approach developed by bobkov , gentil and ledoux @xcite .
inequality is actually a family of inequalities , with a constant having a specific form ( _ i.e. _ @xmath70 ) in the right hand side . in this paper , in order to broaden this notion , we will call @xmath1-log - sobolev inequality rather than inf - convolution log - sobolev inequality the following inequality . [
def : qlsi ] let @xmath7 be a young function ; a probability measure @xmath5 on @xmath6 is said to satisfy the _ @xmath1-log - sobolev inequality _
@xmath71 for some @xmath72 if @xmath73 for all bounded locally lipschitz function @xmath74 where the inf - convolution operator @xmath75 is defined by @xmath76 when @xmath77 , we use the notation @xmath78 instead of @xmath79 the notation @xmath80 refers to the celebrated @xmath1-property introduced by maurey @xcite ( that uses the inf - convolution operator @xmath78 and that is also closely related to the transport - entropy inequality , see ( * ? ? ?
* section 8.1 ) ) .
of course implies @xmath81 , for any @xmath82 .
the other direction is not clear , _ a priori _ ( it would trivially be true if @xmath83 ) , even if the two inequalities have the same flavor .
thanks to theorem [ grs - thm - intro ] below , they appear to be equivalent , under mild assumptions on @xmath7 . in the next sections
, our objective will be to relate the log - sobolev inequalities @xmath85 and @xmath86 to the transport - entropy inequality @xmath87 .
this program works well if we suppose that @xmath7 verifies the classical doubling condition @xmath88 recall that a young function @xmath7 is said to satisfy the _
@xmath84-condition _ if there exists some positive constant @xmath89 ( that must be greater than or equal to @xmath90 ) such that @xmath91 the classical functions @xmath23 introduced in enjoy this condition .
the following observation will be very useful in the sequel .
[ lem r p ] if @xmath7 is a young function satisfying the @xmath84-condition , then @xmath92 where @xmath93 ( resp .
@xmath94 ) denotes the right ( resp .
left ) derivative of @xmath95 the proof of this lemma is in the appendix .
to understand these exponents @xmath96 and @xmath97 , observe that for the function @xmath98 , defined by , we have @xmath99 and @xmath100 . moreover , if @xmath101 are given numbers , and @xmath7 is a young function such that @xmath102 and @xmath103 , then it is not difficult to check that @xmath104 our first result states that the modified log - sobolev inequality ( plus or minus ) implies the transport - entropy inequality associated to the same @xmath7 ( otto - villani theorem ) .
[ ov - thm - intro ] let @xmath5 be a probability measure on @xmath6 and @xmath7 a young function satisfying the @xmath84-condition .
* if @xmath5 satisfies @xmath105 for some @xmath52 , then @xmath5 satisfies @xmath106 with @xmath107 * if @xmath5 satisfies @xmath108 for some @xmath52 , then @xmath5 satisfies @xmath109 with @xmath110 the numbers @xmath111 , @xmath112 are defined by . let us comment on this theorem .
first observe that @xmath113 and @xmath114 are of the same order since @xmath115 for the quadratic case @xmath29 , the constants reduce to @xmath116 .
this corresponds ( when @xmath6 is a smooth riemannian manifold ) to the usual otto - villani theorem @xcite ( see also @xcite ) .
let us mention that lott and villani @xcite generalized the result from riemannian manifolds to length spaces , for @xmath29 , with an adaptation of the hamilton - jacobi semigroup approach developed by bobkov , gentil and ledoux @xcite .
but their statement requires additional assumptions , such as a local poincar inequality , which are not needed in theorem [ ov - thm - intro ] .
also , in @xcite the authors prove that the modified log - sobolev inequality , in euclidean setting and with @xmath117 , with @xmath118 , implies the corresponding transport inequality @xmath21 , again using the hamilton - jacobi approach @xcite .
more recently , in @xcite , the first named author proved that @xmath119 implies @xmath120 in the quadratic case @xmath29 and on an arbitrary complete and separable metric space .
his proof can be easily extended to more general functions such as @xmath121 .
the scheme of proof is the following .
inequality @xmath0 is first shown to be equivalent to dimension free gaussian concentration . according to the well known herbst argument
, @xmath122 implies dimension free gaussian concentration , so it also implies @xmath123 finally , as shown by cattiaux and guillin @xcite , we mention that the talagrand transport - entropy inequality @xmath0 does not imply , in general , the log - sobolev inequality .
hence , there is no hope to get an equivalence in the above theorem .
however , the @xmath1-log - sobolev inequality appears to be equivalent to the transport - entropy inequality .
this is the main result of this paper .
[ grs - thm - intro ] let @xmath5 be a probability measure on @xmath6 and @xmath7 a young function satisfying the @xmath84-condition and let @xmath112 be defined by .
the following statements are equivalent 1 .
there exists @xmath67 such that @xmath5 satisfies @xmath15 .
2 . there exist @xmath124 , @xmath125 such that @xmath5 satisfies @xmath71 .
moreover , the constants are related in the following way @xmath126 where @xmath127 .
such a characterization appeared for the first time in @xcite , in a euclidean setting and with @xmath7 between linear and quadratic . here
our result is valid not only for a wider family of young functions @xmath7 but also on very general metric spaces . due to its functional form , it is easy to prove a perturbation lemma for the inequality @xmath80 .
this leads to the following general holley - stroock perturbation result for transport - entropy inequalities whose proof is given in section [ hs ] .
[ holley ] let @xmath5 be a probability measure on @xmath6 and @xmath7 a young function satisfying the @xmath84-condition and let @xmath112 be defined by .
assume that @xmath5 satisfies @xmath15 for some constant @xmath16 .
then , for any bounded function @xmath128 , the measure @xmath129 ( where @xmath130 is the normalization constant ) satisfies @xmath131 , with @xmath132 where @xmath133 , and @xmath134
. this theorem fully extends the previous perturbation result ( * ? ? ?
* corollary 1.8 ) obtained in a euclidean setting and for a young function @xmath7 such that @xmath66 is concave .
namely , for such an @xmath7 , the function @xmath135 is non - increasing ( * ? ? ?
* lemma 5.6 ) , and so @xmath136 .
the paper is divided into five sections and one appendix .
section [ preliminaries ] is dedicated to some preliminaries .
in particular we will give a characterization of transport - entropy inequalities ( close from bobkov and gtze one ) that might be of independent interest and that is one of the main ingredients in our proofs .
for the sake of completeness , we also recall how the transport - entropy inequality @xmath21 implies the @xmath1-log - sololev inequality ( @xmath137 of theorem [ grs - thm - intro ] ) , this argument had been first used in @xcite and then in @xcite . in section [ grs - section ] , we prove the other direction : the @xmath1-log - sololev inequality implies the transport - entropy inequality @xmath21 . in section [ ov - section ] , we give the proof of the generalized otto - villani result , theorem [ ov - thm - intro ] . the proof of the holley - stroock perturbation result is given in section [ hs ] .
finally , most of the technical results needed on young functions are proved in the appendix .
in this section , we first recall the proof of the first half of theorem [ grs - thm - intro ] , namely @xmath138 . in a second part , we give a useful `` dimension '' refinement of the characterization of transport - entropy inequalities by bobkov and gtze @xcite .
these characterization provides sufficient conditions for transport - entropy inequality to hold .
these conditions are the one obtained in the proofs of @xmath139 and @xmath140 . in ( * ? ? ?
* theorem 2.1 ) , we proved the following result which is the first half ( @xmath141 @xmath142 @xmath143 ) of theorem [ grs - thm - intro ] . for the sake of completeness , its short proof is recalled below .
[ grs ] let @xmath5 be a probability measure on @xmath6 and @xmath7 a young function .
if @xmath5 satisfies @xmath15 for some constant @xmath16 , then , for all @xmath82 , @xmath5 satisfies @xmath144 .
take @xmath35 a locally lipschitz function such that @xmath145 and consider the probability @xmath146 defined by @xmath147 jensen inequality implies that @xmath148 .
so , if @xmath9 is an optimal coupling between @xmath149 and @xmath150 , then it holds @xmath151 by definition of @xmath152 , @xmath153 since @xmath9 is optimal , it holds @xmath154 plugging the inequality @xmath155 in the inequality above with @xmath156 immediately gives @xmath144 . in this section , we show that bounds on the exponential moment of the tensorized inf - convolution or sup - convolution operator allows to recover the transport - entropy inequality ( see proposition [ p11 ] and corollary [ cor ] below ) .
these results are a key argument to recover transport - entropy inequality either from modified log - sobolev or from @xmath1-log - sobolev inequality .
it is known , since the work by bobkov and gtze @xcite ( see also @xcite ) , that transport - entropy inequalities have the following dual formulation .
[ bg ] let @xmath5 be a probability measure on a complete and separable metric space @xmath2 . then the following are equivalent : * the probability measure @xmath5 satisfies @xmath157 ; * for any bounded continuous function @xmath158 , it holds @xmath159 in the next proposition we show , using the law of large numbers , that the bound in point @xmath160 can be relaxed as soon as it holds in any dimension .
[ p11 ] let @xmath5 be a probability measure on a complete and separable metric space @xmath2 .
then the following are equivalent : * the probability measure @xmath5 satisfies @xmath157 ; * there exist three constants @xmath161 , @xmath162 , @xmath163 such that for any @xmath164 , for any bounded continuous function @xmath165 , it holds @xmath166 where @xmath167 note that the constants @xmath161 and @xmath162 do not play any role . on the other hand , notice that @xmath40 is only assumed to be non - negative .
observe that the transport - entropy inequality @xmath157 naturally tensorises ( see _ e.g. _ @xcite ) . applying bobkov and gtze result above
, we see that @xmath168 implies @xmath160 with @xmath169 and @xmath170 .
now let us prove that @xmath160 implies @xmath168 .
for that purpose , fix a bounded continuous function @xmath158 with mean @xmath171 under @xmath5 and , following @xcite ( see also @xcite ) , define @xmath172 on @xmath173 as @xmath174 , @xmath175 . then , @xmath176 where , as usual , @xmath177 .
it follows that @xmath178 now , according to the strong law of large numbers , @xmath179 in @xmath180 , where the @xmath181 s are i.i.d .
random variables with common law @xmath5 .
hence , @xmath182 when @xmath183 tends to infinity .
we conclude that @xmath184 since the latter is invariant by changing @xmath40 into @xmath185 for any constant @xmath186 , we can remove the assumption @xmath187 .
this ends the proof .
the next corollary will be used in the proofs of theorems [ ov - thm - intro ] and [ grs - thm - intro ] .
it gives a sufficient condition for the transport - entropy inequality @xmath21 to hold .
[ cor ] let @xmath5 be a probability measure on a complete and separable metric space @xmath2 .
define , for all @xmath188 , @xmath189 assume that there exist some constants @xmath190 , @xmath161 , @xmath191 and @xmath192 such that , for all integer @xmath164 and all bounded continuous functions @xmath165 , it holds @xmath193 then @xmath5 satisfies @xmath194 .
let @xmath195 and take a bounded continuous function @xmath196 .
in order to apply proposition [ p11 ] , we need to remove the spurious term @xmath197 .
observe on the one hand that for any @xmath198 , one has @xmath199 on the other hand , set @xmath200 .
it is bounded , non - negative and satisfies @xmath201 . moreover , since @xmath202 and @xmath203 are defined with the same cost function , it is elementary to verify that @xmath204 .
hence @xmath205 . therefore , since @xmath206 , by tchebychev s inequality and the assumption
, we have @xmath207 consequently , we get @xmath208 finally , proposition [ p11 ] provides that @xmath5 satisfies @xmath209 .
optimizing over @xmath210 leads to the expected result .
in this section we prove theorem [ ov - thm - intro ] . we have to distinguish between the modified log - sobolev inequalities plus and minus . as in @xcite , the proofs of theorems [ ov - thm - intro ] and [ grs - thm - intro ] use as a main ingredient the stability of log - sobolev type inequalities under tensor products .
let us recall this tensorisation property .
the entropy functional enjoys the following well known sub - additivity property ( see _ e.g. _ ( * ? ? ?
* chapter 1 ) ) : if @xmath211 , @xmath212 where , for all @xmath213 , the application @xmath214 is the @xmath215-th partial application defined by @xmath216 let us say that @xmath211 is _ separately locally lipschitz _ , if all the partial applications @xmath214 @xmath217 , @xmath213 are locally lipschitz on @xmath218 now , suppose that a probability @xmath5 on @xmath6 verifies @xmath105 for some @xmath219 then , using , we easily conclude that @xmath220 enjoys the following inequality : @xmath221 for all function @xmath188 separately locally lipschitz , where @xmath222 is defined by @xmath223_{+}}{d(y , x_{i})}.\ ] ] the same property holds for @xmath224 the first part of theorem [ ov - thm - intro ] , that we restate below , says that the modified log - sobolev inequality @xmath225 implies the transport - entropy inequality @xmath21 .
in fact we shall prove the following slightly stronger result . to any young function @xmath7
, we associate a function @xmath226 defined by @xmath227 where @xmath93 is the right derivative of @xmath7 .
note that @xmath226 is non - decreasing and may take infinite values .
[ ov - thm ] let @xmath5 be a probability measure on @xmath6 and @xmath7 a young function satisfying the @xmath84-condition .
if @xmath5 satisfies @xmath105 for some constant @xmath52 , then @xmath5 satisfies @xmath228 with @xmath229 .
the following lemma gives an estimation of @xmath226 .
[ lem estimation xi ] let @xmath7 be a young function satisfying the @xmath84-condition , and let @xmath111 , @xmath112 be the numbers defined by .
then , it holds @xmath230 with the convention @xmath231 if @xmath232 and @xmath233 otherwise .
the proof of this result is in the appendix .
+ using lemma [ lem estimation xi ] , we easily derive point ( i ) of theorem [ ov - thm - intro ] , with the explicit constant @xmath234 before turning to the proof of theorem [ ov - thm ] , let us say that the estimation is satisfactory at least for the small values of @xmath38 ( corresponding to the large values of @xmath235 ) , as shows the following exact calculation of @xmath226 when @xmath7 is the function @xmath23 defined by . [ exact calculation ] let @xmath236 and @xmath237 and let @xmath98 ; then @xmath100 and it holds @xmath238 moreover , for @xmath239 , it holds @xmath240 where @xmath241 and @xmath242 the proof of this lemma is in the appendix too .
our aim is to use herbst s argument together with proposition [ p11 ] .
let @xmath195 ; according to lemma [ locally lip ] below , for any bounded function @xmath188 , the function @xmath243 is separately locally lipschitz ( recall that the inf - convolution operator @xmath203 is defined by ) .
fix a non - negative bounded continuous function @xmath244 . applying to @xmath245 , @xmath246 , and using lemma [ lem : gradient+ ] below together with the fact that @xmath247 , one gets @xmath248 now , we proceed with the herbst argument .
set @xmath249 , @xmath246 .
since @xmath250 , the latter can be rewritten as @xmath251 set @xmath252 , @xmath246 , so that the previous differential inequality reduces to @xmath253 since @xmath254 , we get @xmath255 where we set @xmath256 ( thanks to lemma [ lem estimation xi ] above we are guaranteed that @xmath257 and that @xmath258 on @xmath259 ) .
since @xmath260 , we finally get @xmath261 which leads to the expected result , thanks to proposition [ p11 ] ( and after optimization over @xmath262 ) .
[ locally lip]let @xmath7 be a young function .
for any integer @xmath195 , any bounded function @xmath188 , the function @xmath243 is separately locally lipschitz on @xmath263 let @xmath264 ; then , for all @xmath213 and @xmath217 , it holds @xmath265 where @xmath266 is defined by the second infimum .
let us show that @xmath267 is locally lipschitz on @xmath218 observe that @xmath172 is bounded and define @xmath268 for all @xmath269 , and all @xmath270 such that @xmath271 , we have @xmath272 since @xmath273 , we conclude that @xmath274 let @xmath275 , and let @xmath276 be the closed ball of center @xmath277 and radius @xmath278 . if @xmath279 , then @xmath280 now , if @xmath281 , we see that for all @xmath282 , @xmath283 with @xmath284 the map @xmath285 is an infimum of @xmath286-lipschitz functions on @xmath276 , so it is @xmath286-lipschitz on @xmath287 this ends the proof .
[ lem : gradient+ ] let @xmath7 be a young function . for any integer @xmath183 , any @xmath288 and any bounded continuous function @xmath289 , @xmath290 where @xmath291 is any point such that @xmath292 fix @xmath183 , @xmath288 and a bounded function @xmath165 . for @xmath175 , @xmath293 and @xmath294 , we shall use the following notation @xmath295 let @xmath213 ; since @xmath40 is bounded continuous and closed balls in @xmath6 are assumed to be compact , it is not difficult to show that there exists @xmath296 such that @xmath297 for all @xmath298 and all @xmath217 , we have also @xmath299 since the maps @xmath300_{+}$ ] and @xmath7 are non - decreasing , it holds @xmath301_{+ } & \leq [ \alpha(d(z , y^x_{i } ) ) - \alpha ( d(x_i , y_{i } ) ) ] _ { + } \\&\leq [ \alpha(d(z , x_{i } ) + d(x_{i},y^x_{i } ) ) - \alpha ( d(x_i , y_{i } ) ) ] _ { + } .\end{aligned}\ ] ] therefore , @xmath302_+}{d(z , x_i)}\\ & = \alpha_+ ' ( d(x_i , y^x_i)).\end{aligned}\ ] ] hence , by the very definition of @xmath226 , @xmath303 in this section we prove the second part of theorem [ ov - thm - intro ] , that we restate ( in a slightly stronger form ) below , namely that the modified log - sobolev inequality minus @xmath304 implies the transport - entropy inequality @xmath21 .
let us define ( recall the defintion of @xmath226 given in ) @xmath305 note that , by lemma [ lem estimation xi ] , if @xmath7 satisfies the @xmath84-condition , then @xmath306 .
[ ov - thm- ] let @xmath5 be a probability measure on @xmath6 and @xmath7 a young function satisfying the @xmath84-condition .
if @xmath5 satisfies @xmath108 for some constant @xmath52 , then @xmath5 satisfies @xmath307 with @xmath308 . for more comprehension and to complete the proof of part ( ii ) of theorem [ ov - thm - intro ] , let us prove that @xmath309 . if @xmath310 then by lemma [ lem estimation xi ] , @xmath311 .
moreover , using that @xmath312 , @xmath313 , one has @xmath314 with @xmath315 when @xmath316 , since @xmath306 and using the fact that the function @xmath317 is non - increasing , we get @xmath318 the proof of theorem [ ov - thm- ] follows essentially the lines of the proof of theorem [ ov - thm ] . let @xmath195 ; thanks to the tensorisation property of @xmath108 , it holds @xmath319 for any @xmath320 separately locally lipschitz and bounded . take a non - negative bounded continuous function @xmath244 . recall that @xmath321 since @xmath322 , it follows from lemma [ locally lip ] that @xmath323 is separately locally lipschitz . applying to @xmath324 , @xmath246 ,
one gets @xmath325 observe that @xmath322 and that @xmath326 , for all @xmath327 .
so applying lemma [ lem : gradient+ ] , we see that for all @xmath213 , there is some @xmath291 such that @xmath328 so we get the following inequality @xmath329 as in the proof of theorem [ ov - thm ] , we proceed with the herbst argument .
set @xmath330 , @xmath331 .
since @xmath332 , the latter can be rewritten as @xmath333 set @xmath334 , @xmath335 , so that the previous differential inequality reduces to @xmath336 set @xmath337 ( which belongs to @xmath338 thanks to lemma [ lem estimation xi ] ) . since @xmath339 , solving the latter differential inequality , we easily get that for all @xmath331 @xmath340 applying corollary [ cor ] yields that @xmath341 holds for all @xmath331 . observing that the function @xmath342 is non - decreasing on @xmath343 , the proof is completed by optimizing in @xmath344 .
in this section , we prove the second part ( @xmath345 ) of theorem [ grs - thm - intro ] .
observe that @xmath346 is equivalent to @xmath347 .
hence , changing @xmath7 into @xmath348 , we can restate the first part of theorem [ grs - thm - intro ] as follows .
[ grs - thm ] let @xmath5 be a probability measure on @xmath6 and @xmath7 a young function satisfying the @xmath84-condition .
let @xmath112 be defined by .
if @xmath5 satisfies @xmath349 for some @xmath125 , then @xmath5 satisfies @xmath15 with @xmath350 where @xmath127 .
two proofs are given below .
the first one exactly follows the lines of the proof of @xmath351 , whereas the second one uses the equivalence between transport - entropy inequalities and dimension free concentration established in @xcite together with a change of metric argument . in each proof
, the first step is to tensorise the @xmath1-log - sobolev inequality .
let @xmath195 ; using the sub - additivity property of the entropy functional , we see that @xmath349 implies that @xmath352 where @xmath353 is the inf - convolution operator with respect to the @xmath215-th coordinate , namely @xmath354 ( using the notation introduced in section [ ov - section ] ) .
as in the proof of theorem [ ov - thm- ] , applying to @xmath355 , @xmath356 where @xmath172 belongs to some class of functions , we get @xmath357 as a main difference , the class of functions @xmath172 differs in each proof . in the first one , @xmath172 is any non - negative bounded separately locally lipschitz function , whereas in the second proof , @xmath172 is globally lipschitz in some sense . for both proofs ,
the next step is to bound efficiently the right - hand side of , in order to use some herbst argument .
this bound will be given by the following lemma .
[ petitlemmebis ] let @xmath7 be a young function satisfying the @xmath84-condition and let @xmath112 be defined by .
for any bounded continuous function @xmath320 , for any @xmath213 and @xmath358 , @xmath359 where @xmath291 is any point such that @xmath360 , and where @xmath361 we postpone the proof of lemma [ petitlemmebis ] to the end of the section .
the first proof of theorem [ grs - thm ] mimics the one of the implication @xmath351 . using ,
lemma [ petitlemmebis ] ensures that for every non - negative locally lipschitz bounded function @xmath172 , for every @xmath362 @xmath363 where @xmath364 . solving this differential inequality ,
exactly as in the proof of theorem [ ov - thm- ] ( we omit details ) , leads to @xmath365 with @xmath366 , @xmath367 the inequality @xmath15 then follows from corollary [ cor ] .
now , let us estimate the constant @xmath67 .
by convexity , one has for every @xmath368 $ ] , @xmath369 this inequality easily implies that for all @xmath362 , @xmath370 consequently , we obtain for all @xmath371 , @xmath372 optimizing in @xmath373 , we get @xmath374 with @xmath127 . the idea of this second proof is to prove the theorem in the particular case of the functions @xmath121 and then to treats the general case by a change of metric argument .
let us introduce some notation and definitions .
when @xmath376 , we will use the notation @xmath377 and @xmath378 instead of @xmath379 and @xmath380 .
let @xmath195 ; a function @xmath188 will said to be @xmath381-lipschitz @xmath382 , if @xmath383 we recall the following result from @xcite . [ gozlan p ] the probability @xmath5 verifies the transport - entropy inequality @xmath377 , for some @xmath16 if and only if it enjoys the following dimension free concentration property : for all @xmath195 and all @xmath384 such that @xmath385 for some @xmath386 , it holds @xmath387 so to show that a @xmath375 inequality holds it is enough to prove the right concentration inequality .
we will use the following result to estimate the right - hand side of .
[ omega_p ] let @xmath388 ; there exists a constant @xmath389 such that for all @xmath195 and all @xmath381-lipschitz function @xmath384 , and all @xmath213 , the function @xmath390 attains its maximum on the closed ball @xmath391 when @xmath2 is geodesic ( see below ) , then one can take @xmath392 recall that @xmath2 is geodesic , if for all @xmath393 , there is a path @xmath394}$ ] joining @xmath38 to @xmath395 and such that @xmath396 , for all @xmath397.$ ] this notion encompasses the case of riemannian manifolds .
the proof of the lemma is at the end of the section .
[ grs - p ] let @xmath398 ; if @xmath5 verifies the @xmath399 , then it verifies @xmath377 , with @xmath400 with @xmath401 let us compare the constants appearing in theorems [ grs - thm ] and [ grs - p ] for @xmath402 .
when @xmath402 , theorem [ grs - thm ] gives @xmath403 , and theorem [ grs - p ] gives @xmath404 .
a simple calculation shows that when @xmath402 , @xmath405 if @xmath406 , then @xmath407 and @xmath408 is smaller than @xmath409 .
but if @xmath410 ( which is the case , when @xmath2 is geodesic ) , then @xmath411 and @xmath409 is smaller than @xmath412 take a @xmath381 - lipschitz function @xmath413 to bound the right - hand side of , we use lemmas [ petitlemmebis ] and [ omega_p ] : @xmath414 so , letting @xmath415 provides @xmath416 equivalently , the function @xmath417 verifies @xmath418 since @xmath419 when @xmath420 , we conclude that @xmath421 where @xmath422 since @xmath172 is @xmath381 - lipschitz , it holds @xmath423 so , @xmath424 applying this inequality to @xmath425 with @xmath40 a @xmath381 - lipschitz function , we get @xmath426 so , optimizing over @xmath427 yields @xmath428 with @xmath429 using chebychev s argument we derive from that @xmath430 with @xmath431 applying theorem [ gozlan p ] , we conclude that @xmath5 verifies @xmath377 .
a change of metric technique , which is explained in the lemma below , enables us to reduce the study of the transport - entropy inequalities @xmath21 to the study of the inequalities @xmath375 , @xmath432 [ lem change metric ] let @xmath7 be a young function satisfying the @xmath84-condition and let @xmath112 be defined by .
the function @xmath433 is subadditive on @xmath434 : @xmath435 as a consequence , @xmath436 , @xmath393 is a distance on @xmath6 .
the proof of lemma [ lem change metric ] is at the end of the section .
let @xmath7 be a young function and @xmath5 a probability on @xmath6 . according to lemma [ lem change metric ] , the function @xmath437
, @xmath393 is a metric on @xmath218 furthermore , it is clear that @xmath5 verifies @xmath438 ( resp .
@xmath15 ) on @xmath2 if and only if @xmath5 verifies @xmath439 ( resp .
@xmath440 ) on @xmath441 .
we immediately deduce from theorem [ grs - p ] that if @xmath5 verifies @xmath438 , then it verifies @xmath442 , with @xmath443 , where @xmath444 ( since @xmath445 ) and @xmath446 defined in lemma [ petitlemmebis ] . note that the constant @xmath67 obtained using this approach
is in general bigger than the constant obtained in the first proof of theorem [ grs - thm ] .
fix @xmath447 , @xmath448 , and @xmath449 . then , @xmath450 let @xmath451 be such that @xmath452 , where @xmath453 by choosing @xmath454 in the expression below , it holds @xmath455 where , in the last line , we used the triangular inequality for the distance @xmath456 defined in lemma [ lem change metric ] . hence ,
optimizing yields @xmath457 - d_\alpha^{p_\alpha}(x_i , z ) \right\ } \\ & = \sup_{r > 0 } \left\ { t\left [ ( d_\alpha(x_i , y^x_i ) + r)^{p_\alpha } - d^{p_\alpha}_\alpha(x_i , y^x_i ) \right ] - r^{p_\alpha } \right\ } \\ & = t\varepsilon(t)d_\alpha^{p_\alpha}(x_i , y_i^x)= t\varepsilon(t)\alpha(d(x_i , y_i^x)).\end{aligned}\ ] ] taking the sum , we get the result .
let @xmath451 be a point where the function @xmath458 reaches its maximum .
then , for all @xmath459 , it holds @xmath460 choosing @xmath461 , we get @xmath462 now , assume that @xmath2 is geodesic . then the product space @xmath463 with @xmath464 is geodesic too . in the calculation above , take for @xmath465 a @xmath344-midpoint of @xmath38 and @xmath451 _ i.e _ choose @xmath459 such that @xmath466 and @xmath467 , with @xmath468 $ ] . then , letting @xmath469 , it holds @xmath470 and so @xmath471 letting @xmath472 gives the result .
let @xmath473 , @xmath474 .
then , by definition of @xmath97 @xmath475 so @xmath476 is non - increasing on @xmath477 thus , if @xmath478 , @xmath479
in this section we prove theorem [ holley ] . the proof follows the line of the original proof . using the following representation of the entropy @xmath480 with @xmath481
, we see that ( since @xmath482 ) @xmath483 since @xmath5 verifies @xmath15 , theorem [ grs ] implies that , for all @xmath82 , @xmath484 in other words , @xmath485 satisfies @xmath486 , for any @xmath82 .
now , applying theorem [ grs - thm - intro ] , we conclude that @xmath487 verifies @xmath131 , with @xmath488
in this appendix we prove the technical lemmas on young functions we used during the paper .
first , let us prove lemma [ lem r p ] , that we restate below using the convexity of @xmath7 , we see that @xmath490 .
this shows that @xmath491 on the other hand , the function @xmath7 is convex , so @xmath492 , for all @xmath474 .
since @xmath7 verifies the @xmath84-condition , there is some constant @xmath493 such that @xmath494 .
so we get @xmath495 , for all @xmath474 .
this proves that @xmath496 let us show that @xmath112 .
otherwise we would have @xmath497 ( since @xmath498 ) and so @xmath499 for all @xmath500 this would imply that @xmath7 is linear on @xmath501 .
this can not happen , since by assumption young functions are increasing and such that @xmath22 .
so @xmath502 now let us prove lemmas [ lem estimation xi ] and [ exact calculation ] whose statements is summarized below .
recall that the function @xmath226 is defined by @xmath503 + @xmath504 let @xmath7 be a young function satisfying the @xmath84-condition , and let @xmath111 , @xmath112 be the numbers defined by .
then , it holds @xmath505 with the convention @xmath506 if @xmath507 and @xmath233 otherwise .
+ @xmath504 let @xmath236 and @xmath237 and let @xmath98 ; then @xmath100 and it holds @xmath238 moreover , for @xmath239 , it holds @xmath508 where @xmath241 and @xmath242 defining @xmath509 , for all @xmath510 , we get @xmath511 from the convexity inequality @xmath512 , @xmath513 we deduce immediately that @xmath514 for all
@xmath515 thus @xmath516 so , all we have to show is that @xmath517 , for all @xmath510 .
define @xmath518 and @xmath519 for all @xmath520 as in the proof of lemma [ lem change metric ] , a simple calculation shows that @xmath476 is non - increasing and @xmath521 is non - decreasing . as a result , @xmath522.\end{aligned}\ ] ] taking the fenchel - legendre transform yields @xmath523.\end{aligned}\ ] ] equivalently , @xmath524,\\
\alpha^*(ux)&\leq x^{\frac{r_\alpha}{r_\alpha-1}}\alpha^*(u),\qquad \forall u\geq 0,\ \forall x\geq 1.\end{aligned}\ ] ] and since @xmath525 , we conclude that @xmath526 @xmath527 now , let us calculate @xmath528 , for @xmath236 , @xmath237 .
first observe that @xmath529 for all @xmath530 it will be more convenient to do the calculation with the function @xmath531 let us denote by @xmath532 , @xmath533 the conjugate exponents of @xmath534 and @xmath535 then the following identity holds : @xmath536 let us show that @xmath537 for @xmath538 the case @xmath539 is similar and left to the reader .
define @xmath540 we have to distinguish three cases : @xmath541 * * case 1 .
@xmath542**then @xmath543 + * case 2 .
* then @xmath545 if @xmath546 , then the function @xmath476 is non - increasing on @xmath547 $ ] , and so @xmath548 if @xmath549 , then the function @xmath476 is non - decreasing on @xmath547 $ ] and so @xmath550 * case 3 .
@xmath551 * then @xmath552 if @xmath546 , the function @xmath476 is non - increasing on @xmath553 and so @xmath554 if @xmath549 , the function @xmath476 is non - decreasing on @xmath555 and so @xmath556 observe , in particular , that @xmath476 never reaches its supremum at @xmath557 we conclude that @xmath558 and so @xmath559 since @xmath560 , the proof is complete . c. an , s. blachre , d. chafa , p. fougres , i. gentil , f. malrieu , c. roberto , and g. scheffer . ,
volume 10 of _ panoramas et synthses [ panoramas and syntheses]_. socit mathmatique de france , paris , 2000 . with a preface by dominique bakry and michel ledoux
. m. talagrand . a new isoperimetric inequality and the concentration of measure phenomenon . in _ geometric aspects of functional analysis ( 198990 )
_ , volume 1469 of _ lecture notes in math .
_ , pages 94124 .
springer , berlin , 1991 . | we give a characterization of transport - entropy inequalities in metric spaces . as an application
we deduce that such inequalities are stable under bounded perturbation ( holley - stroock perturbation lemma ) . |
are the dark matter halos of lsb galaxies `` cuspy '' nfw halos , or are they `` cored '' isothermal halos ?
nfw halos can be fit to the observations , but the cosmological parameters implied by the fits are inconsistent with the standard lcdm picture .
isothermal halos provide much better fits , but they have no cosmological dependence or theoretical basis .
we address this question with new data that is both high resolution and two - dimensional .
we observed 12 lsb galaxies using the densepak integrated field unit on the wiyn telescope at kitt peak .
densepak is comprised of 3 diameter fibers arranged in a 43 @xmath1 28 rectangle .
velocities of the h@xmath2 , [ ] @xmath36584 , [ ] @xmath36717 , and [ ] @xmath36731 emission lines were measured in each fiber .
rotation curves were derived from the two - dimensional velocity fields using the nemo program rotcur .
we combined the densepak rotation curves with previous longslit and hi curves .
we fit the combined data with isothermal and nfw@xmath4 halos .
we also fit an nfw@xmath5 halo which was required to match the velocities at the outer radii of each galaxy , in effect , constraining the concentrations to agree with cosmology .
image of ugc 5750 ; observed velocity field ( km s@xmath6 ) ; minimum disk halo fits to combined data ; combined rotation curves.,title="fig : " ] image of ugc 5750 ; observed velocity field ( km s@xmath6 ) ; minimum disk halo fits to combined data ; combined rotation curves.,title="fig : " ] + image of ugc 5750 ; observed velocity field ( km s@xmath6 ) ; minimum disk halo fits to combined data ; combined rotation curves.,title="fig:",width=181 ] image of ugc 5750 ; observed velocity field ( km s@xmath6 ) ; minimum disk halo fits to combined data ; combined rotation curves.,title="fig:",width=181 ]
we found only one galaxy to be well fit by the nfw halo .
this shows that we would have detected cusps in the other cases had they been present .
future work will include mass modeling and an assessment of systematic effects .
* acknowledgments .
* this research was supported by nsf grant ast 0206078 . | we present high resolution two dimensional velocity fields from integral field spectroscopy along with derived rotation curves for nine low surface brightness galaxies .
this is a positive step forward in terms of both data quality and number of objects studied .
we fit nfw and pseudo - isothermal halo models to the observations .
we find that the pseudo - isothermal halo better represents the data in most cases than the nfw halo , as the resulting concentrations are lower than would be expected for @xmath0cdm . |
in the open quantum system approach@xcite , the evolution of a system interacting with an environment is described in an effective way , that is , the interaction with the environment is incorporated in the description of the evolution of the system . in general , the effects produced by the interaction cause dissipation and irreversibility .
this treatment was originally developed for quantum optics@xcite , in order to take into account the system - reservoir ( environment ) interaction , and it has already been applied to elementary particle systems .
recently , it has also been used in the study of two neutrino oscillations@xcite , modifying the well known oscillation probability , due to the presence of dissipative effects .
one very interesting feature of the oscillation probability in this situation is that even in the case that neutrinos are massless , we can have neutrino flavor conversion@xcite .
it has been pointed out that this mechanism is able to explain well , in the context of two generations , the atmospheric neutrino data collected by the super - kamiokande ( sk ) experiment@xcite , as long as the damping parameter is @xmath2 , @xmath3 being the neutrino energy .
the main goal of this paper is to extend this formalism of mass , mixing and quantum decoherence to three generations .
this is rather well motivated since nowadays the atmospheric and solar neutrino observations@xcite can only be explained by the introduction of some neutrino flavor conversion mechanism which must be understood , specially after the recent impressive sno results@xcite , in terms of three generations .
therefore , this extension will be very useful , because it will permit us to study the decoherence contributions on top of the oscillation induced by mass ( oim ) in a complete three neutrino context .
we have developed this three neutrino formalism using the powerful technique of quantum dynamical semigroups@xcite , which makes our analysis independent of any hypothesis about the interaction between the neutrino system and the environment .
quantum open systems can also be treated using the master equation formalism@xcite , but in this case , a previous knowledge about the interaction with the environment is required .
this is an important point since it is not clearly established which is the origin of the pervasive medium , the most likely possibilities at the moment are quantum gravity effects described by strings at low energy range . in this way ,
our results are broader from the phenomenological point of view .
based on motivations coming from the master equation formalism using the weak coupling limit in two generations@xcite , we have casted the most relevant decoherence effects in a diagonal dissipation matrix , as a result we have obtained manageable probability expressions . additionally , as a phenomenological application of this three neutrino extension for the decoherence phenomena , we have performed a qualitative study to inspect the agreement between the average behaviour of relevant neutrino data and the decoherence solution to the atmospheric neutrino anomaly in two generations , embodied within this enlarged neutrino framework .
we have done this study assuming different alternatives for the introduction of the two generations decoherence solution in our three neutrino scheme .
basically we have divided the study in two parts . on the first part we use only pd to fully describe three neutrino flavor conversions , and on the second we deal with a hybrid case , where we assume decoherence plus oim take place in the @xmath4 sector .
this paper is organized as follows . in sec .
[ sec : qdn ] we describe the quantum dynamical semigroup formalism for a three level quantum system . in sec .
[ sec : neutrinos ] we apply this formalism for the neutrino system , explicitly calculating the survival and conversion probabilities among neutrino flavors under the influence of decoherence effects . in sec .
[ sec : pheno ] we analyze the consistence of the decoherence solution to the atmospheric neutrino problem in the three generation framework . finally in sec .
[ sec : conclusion ] we present our conclusions .
hamiltonian evolution , that is , the time evolution of a physical system described by the schrdinger equation in the case of pure ensembles , or by the liouville equation for mixed ones , is a characteristic of systems isolated from their surroundings .
the time evolution of an isolated quantum system with hamiltonian @xmath5 is given by the continuous group of unitary transformations @xmath6 , where @xmath7 is the time . from the mathematical point of view , the existence of the inverse of the infinitesimal generator @xmath5 , a consequence of the algebraic structure of a group , gives rise to _ reversible processes _ in hamiltonian systems .
we know that the time evolution of a quantum open system is characterized by the presence of dissipative effects which , in turn , give rise to the _ irreversible nature _ of the evolution .
so , if a given family of transformations should be responsible for the evolution in time of a quantum open system , this family will certainly not be a group . a rigorous mathematical treatment of quantum open systems is provided by the so called quantum dynamical semigroups@xcite .
the evolution generated by the operators in these semigroups has the property of being forward in time , as a consequence of the lack of inverse in a semigroup .
physically , this property can be interpreted as the existence of an arrow of time which in turn makes possible the connection with thermodynamics via an entropy . according to ref.@xcite ,
if @xmath8 is the hilbert space of a given open quantum system , and @xmath9 the space of bounded operators acting on @xmath8 , the infinitesimal generator @xmath10 , defined through its action on the density matrix @xmath11 , is given by @xmath12 + \frac{1}{2 } \sum_{j}\left([a_{j},\rho(t ) a_{j}^{\dagger } ] + [ a_{j}\rho(t ) , a_{j}^{\dagger } ] \right ) , \label{gen}\ ] ] where @xmath13 is the `` effective '' hamiltonian of the system , @xmath5 being its free hamiltonian and @xmath14 accounts for possible additional dissipative contributions which can be incorporated to @xmath5 , in other words , which can be put in the hamiltonian form .
@xmath15 is a sequence of bounded operators of @xmath8 ( @xmath16 ) satisfying @xmath17 . the first term in eq .
( [ gen ] ) constitutes the hamiltonian part of the evolution , whereas the second term is responsible for the irreversible ( non - hamiltonian ) nature of an open system evolving in time .
therefore , if we interpret the last expression as an effective equation describing the reduced dynamics of an open system interacting with a certain `` environment '' , the second term in eq .
( [ gen ] ) , in a certain sense , represents the interaction between the open system and the mentioned environment .
however , the generator @xmath18 does not depend on a particular interaction , being constructed based on very general hypothesis about the time evolution , that is , irreversible dynamics , conservation of probability and a less intuitive hypothesis known as complete positivity@xcite . for a n - level system
it is possible to construct an explicit parameterization of eq .
( [ gen ] ) , provided that a suitable basis of @xmath9 , viewed as a vector space , is chosen @xcite . from now on ,
we will restrict ourselves to a three - level system , whose evolution is governed by eq .
( [ gen ] ) and our approach will essentially follow ref.@xcite .
the bounded operators in @xmath9 can be represented by @xmath19 matrices of @xmath20 , which in turn , can be generated by a basis @xmath21 , endowed with the scalar product @xmath22 and satisfying @xmath23 we adopt here the standard basis of hermitian matrices @xmath24 where the @xmath25 are the gell - mann matrices @xcite ) in @xmath26@xcite . ] . using this choice ,
@xmath27 s satisfy the lie algebra @xmath28=i\sum_{k}f_{ijk}f_{k } , \qquad 1\le i , j , k \le 8,\ ] ] where @xmath29 are the structure constants of @xmath30 . expanding the operators in eq .
( [ gen ] ) in the adopted basis we get @xmath31 as pointed out in ref.@xcite , the hermiticity of @xmath15 is a condition that assures the increasing with time of the von neumman entropy @xmath32 , with this choice , the second term on the right hand side of eq .
( [ gen ] ) can be written as @xmath33 + [ a_{j}\rho , a_{j } ] \right\ } = \sum_{\mu,\nu}l_{\mu \nu}\rho_{\mu}f_{\nu } , \label{dissipative}\ ] ] where @xmath34 and in the last equation we have introduced the vectors @xmath35 of @xmath36 with the usual scalar product @xmath37 .
@xmath38 is a real . ] , symmetric matrix , defined according to eqs .
( [ dissipative ] ) and ( [ elements ] ) .
the elements of @xmath38 are not totally arbitrary , but satisfy some relations due to the presence of the scalar product , which in turn , should obey the cauchy - schwartz inequality . introducing the remaining expansions of eq .
( [ expansion ] ) into eq .
( [ gen ] ) we get finally @xmath39 the system of first order differential equations above describes the time evolution of a three - level quantum open system .
such an evolution is governed by the laws of quantum dynamical semigroups .
the physical processes exhibit an arrow of time as a consequence of the monotonical increase of the von neumman entropy as a function of time .
the evolution also allows the interpretation of the eigenvalues of @xmath11 at any instant of time as the probability of finding the open system in the eigenstate associated with the eigenvalue .
as already mentioned , the theoretical approach provided by quantum dynamical semigroups is a very general one to treat open quantum systems in the sense that no explicit hypothesis has to be made about the possible interactions causing the loss of quantum coherence . on the other hand
, there is another approach to deal with systems of the same nature , in which a well defined form of the interaction ought to be provided in order to derive the reduced dynamics of the open quantum system .
this approach is known as the master equation formalism and its rigorous mathematical formulation can be found in a series of papers @xcite .
furthermore , it is possible to show that the master equation formalism applied to an open system weakly coupled to some environment leads to an evolution equation for the reduced density matrix of the open system compatible of that obtained from eq .
( [ gen ] ) @xcite , in other words , in this situation , the master equation and the quantum dynamical semigroup formalism are equivalent . coming back to eq .
( [ differential ] ) , we see that the differential equation for the @xmath40 component and conservation of probability imply @xmath41 .
the remaining differential equations have the form @xmath42 which in the matrix form can be written as , the first being a column vector with components @xmath43 . ]
@xmath44 so that the formal solution is given by @xmath45 if @xmath46 and @xmath47 are the set of eigenvalues and eigenvectors of @xmath48 , respectively , the matrix @xmath49 , with entries @xmath50 , diagonalizes @xmath48 , @xmath51 applying the inverse transformation , @xmath52 , we get @xmath53 so that the complete solution of the system of differential equations ( [ differential ] ) is equivalent to a problem of eigenvalues and eigenvectors . in the case of a matrix of order 8 , with all non null entries ,
the solution is too complicated to allow for direct physical interpretations .
in this section we present an expression for the probability of neutrino flavor conversion assuming that the dynamics responsible by this process is constituted by the already known standard oim mechanism , as well as by dissipative effects according to quantum dynamical semigroups .
the hamiltonian @xmath5 for a free relativistic neutrino with momentum @xmath54 and rest mass @xmath55 is given by @xmath56 so that in the basis of mass eigenstates @xmath57 and we can write for relativistic neutrinos ( @xmath58 ) , @xmath59 the expansion of @xmath5 in the basis @xmath60 is @xmath61 where @xmath62 and @xmath63 .
as already mentioned at the end of the preceding section , an analysis considering the most general form of @xmath48 would certainly not provide information subject to direct physical interpretation . from this point of view , its reasonable to begin the analysis introducing simplifications which allow a direct physical interpretation of the results . of course , these simplifications can not be completely arbitrary , because all the physics depends on them . in this way ,
motivations coming from the formalism of master equation , in the so called weak coupling limit , applied to the problem of neutrino flavor conversion in the two generation case@xcite , lead us to assume a diagonal matrix @xmath38 , that is , @xmath64 = diag(0,-\gamma_{1},-\gamma_{2},-\gamma_{3},-\gamma_{4},-\gamma_{5},-\gamma_{6},-\gamma_{7},-\gamma_{8 } ) , \label{ansatz}\ ] ] where the diagonal elements are given by @xmath65 in the case of two generations , the diagonal form of @xmath38 can be deduced via the master equation formalism in the weak coupling limit , provided that a gas of quanta , satisfying infinite statistics ( for example a d0-brane ) is adopted as the dissipative medium .
these quanta should also be in thermodynamic equilibrium at a finite temperature @xmath66 , where @xmath67 defines an energy scale at which the dissipative effects are believed to become important ( possibly the planck scale if the effects have quantum gravity as their source ) .
the diagonal form is finally obtained if one further imposes the condition of entropy increase for finite @xmath68 , for details see ref.@xcite . we stress that here eq .
( [ ansatz ] ) is just an _ ansatz _ motivated by the results in the two generation case . using this parameterization of @xmath38
we can construct the matrix @xmath48 defined in eq .
( [ diff2 ] ) , and its eigenvalues and eigenvectors can be obtained .
the explicit form of @xmath48 , its eigenvalues and associated eigenvectors can be found in appendix [ appendixa ] .
if at @xmath69 a neutrino of flavor @xmath70 is produced , the probability of its conversion , after a time @xmath7 , to a flavor @xmath71 can be written as @xmath72 = \frac{1}{3}+\frac{1}{2}\sum_{i , k , j}e^{\lambda_{k}t}d_{ik}d_{kj}^{-1}\rho_{j}^{\alpha}(0)\rho_{i}^{\beta}.\ ] ] in a more explicit form , the probability of conversion can be written as @xmath73e^{-\frac{1}{2}(\gamma_{1}+\gamma_{2})t } \nonumber \\ & + & \left[\left ( \rho_{4}^{\alpha}\rho_{4}^{\beta}+\rho_{5}^{\alpha}\rho_{5}^{\beta}\right)\left(\frac{e^{-\frac{\omega_{13}t}{2}}+e^{\frac{\omega_{13}t}{2}}}{2}\right ) \right .
\nonumber \\ & - & \left.\left(\frac{2\delta_{13}\left ( \rho_{4}^{\alpha}\rho_{5}^{\beta}-\rho_{5}^{\alpha}\rho_{4}^{\beta}\right)+\delta\gamma_{45}\left(\rho_{4}^{\alpha}\rho_{4}^{\beta}-\rho_{5}^{\alpha}\rho_{5}^{\beta } \right)}{\omega_{13}}\right)\left(\frac{e^{\frac{-\omega_{13}t}{2}}-e^{\frac{\omega_{13}t}{2}}}{2}\right)\right]e^{-\frac{1}{2}(\gamma_{4}+\gamma_{5})t } \nonumber \\ & + & \left[\left ( \rho_{6}^{\alpha}\rho_{6}^{\beta}+\rho_{7}^{\alpha}\rho_{7}^{\beta}\right)\left(\frac{e^{-\frac{\omega_{23}t}{2}}+e^{\frac{\omega_{23}t}{2}}}{2}\right ) \right .
\nonumber \\ & - & \left.\left(\frac{2\delta_{23}\left ( \rho_{6}^{\alpha}\rho_{7}^{\beta}-\rho_{7}^{\alpha}\rho_{6}^{\beta}\right)+\delta\gamma_{67}\left(\rho_{6}^{\alpha}\rho_{7}^{\beta}-\rho_{7}^{\alpha}\rho_{7}^{\beta } \right)}{\omega_{23}}\right)\left(\frac{e^{-\frac{\omega_{23}t}{2}}-e^{\frac{\omega_{23}t}{2}}}{2}\right)\right]e^{-\frac{1}{2}(\gamma_{6}+\gamma_{7})t } \nonumber \\ & + & e^{-\gamma_{3}t}\rho_{3}^{\alpha}\rho_{3}^{\beta}+e^{-\gamma_{8}t}\rho_{8}^{\alpha}\rho_{8}^{\beta } \bigg\}\ , , \label{final}\end{aligned}\ ] ] where @xmath74 and the definition of the new variables @xmath75 can be found in appendix [ appendixa ] .
it is easy to verify that : eq .
( [ final ] ) exhibits conservation of probability at any instant of time ; reducing the number of dimensions from 3 to 2 , we get the results for decoherence in two generations @xcite ; and in the limit @xmath76 , the standard oim mechanism is recovered .
in this section we will use the just developed three neutrino framework for oim mechanism plus the decoherence effects , to analyze the robustness of the decoherence solution to the atmospheric neutrino problem in two generations @xcite envisaged within this enlarged scheme .
this will be done for different selections of the mixing matrix , first for pd and second for decoherence plus masses and mixing . in order to obtain the probabilities that correspond to the pd case it is enough to eliminate the usual oscillatory terms .
this can be easily done by setting the mixing matrix @xmath77 equal to the unity matrix . applying this to eq .
( [ final ] ) we get the following expressions : @xmath78 where @xmath79 , so consequently @xmath80 despite the fact that mixing is not included , we can observe from these probabilities , that it is possible to have non - null neutrino flavor conversion only taking into account the decoherence phenomena . indeed ,
these results were expected , since the same was obtained in two generations @xcite . the atmospheric neutrino data can be interpreted well using the pd phenomena in the two generation framework @xcite .
thus , we can investigate if the pure decoherence mechanism extrapolated to three generations is still able to explain the current data . a qualitative analysis of the behavior of the above probabilities in the present data context
can clearly demonstrate that this is not so .
the explanation goes as follows .
since we aim to test the decoherence solution to the atmospheric neutrino data in the three generations scheme , we will have to consider the damping parameters with the dependence @xmath81 , in accordance with the two generations assumption .
thus , the probabilities defined in eqs .
( [ decohprob1])-([decohprob5 ] ) will be sensitive only to the ratio @xmath82 ( for c=1 ) . ]
similar to vacuum case for the oim mechanism . here
@xmath83 is the distance between the source of neutrinos and the detector .
note also that these probabilities will not be modified by the presence of matter , since there is no mixing involved .
the conjunction of all of these properties of the probabilities with the fact that @xmath84 , makes the three neutrino pd mechanism incompatible with the actual neutrino data .
one can exemplify this inconsistency by confronting the results of the neutrino experiments chooz @xcite and k2k @xcite with our theoretical predictions .
both experiments have similar @xmath82 , but their results for @xmath85 and @xmath86 are not compatible with our theoretical expectation .
the chooz experiment with @xmath87 observed @xmath88 ) satisfy @xmath89 . ] and the k2k experiment with @xmath90 has observed events compatible with @xmath91 . therefore the relation @xmath92 given by our probabilities , is in contradiction with the results of chooz and k2k .
another example that reinforces this contradiction is the multi - gev sk data sample @xcite in combination with the chooz constraint . in order to illustrate this fact
, we consider the sk normalized @xmath93-like and @xmath94-like event samples , which can be computed as follows @xmath95 where @xmath96 and @xmath97 are the normalized @xmath94-like and @xmath93-like event samples defined as the ratios of the expected number of events considering neutrino conversion over the expected number of events without neutrino flavor conversion , @xmath98 is the original proportion between the muon and electron neutrino fluxes .
we can observe that for ranges of @xmath99 compatible with chooz and corresponding to @xmath100 and @xmath83 is given by @xmath101 , where @xmath102 is the earth radius and @xmath103 is the height of the neutrino production point . ] in the multi - gev sample , @xmath104 @xcite , @xmath105 and @xmath106 .
thus , crossing this information with the chooz bound , we find that @xmath107 ( in fact , for a real mixing matrix , @xmath108 as can be seen from eq .
( [ final ] ) ) and consequently @xmath109 , which is again in conflict with our theoretical prediction . here
we will study two different cases also admitting the presence of the oscillatory terms in the probability expressions , therefore we have here a mixed situation of decoherence @xmath110 oscillation ( d@xmath110o ) . in both cases the oscillation will be included through non - null mixing matrix elements , which connect @xmath111 with @xmath112 or @xmath111 with @xmath113 .
this is chosen in order to preserve the possibility to explain the solar neutrino data through of the oim mechanism .
in addition , another fact in our strategy of analysis will be to adjust the parameters in the @xmath86 , the leading channel for the explanation of the atmospheric data , to be similar to its form in the two generations decoherence solution to the atmospheric neutrino problem .
when we introduce non - zero masses for the neutrinos , two different situations may arise according to the magnitudes of @xmath114 and @xmath115 : either @xmath116 or @xmath117 .
in our analysis we will assume the first situation , where @xmath75 is imaginary so that eq .
( [ final ] ) takes the form @xmath118e^{-\frac{1}{2}(\gamma_{1}+\gamma_{2})t } \nonumber \\ & + & \bigg[\left ( \rho_{4}^{\alpha}\rho_{4}^{\beta}+\rho_{5}^{\alpha}\rho_{5}^{\beta}\right)\cos\left(\frac{|\omega_{13}|t}{2}\right ) \nonumber \\ & + & \left(\frac{2\delta_{13}\left ( \rho_{4}^{\alpha}\rho_{5}^{\beta}-\rho_{5}^{\alpha}\rho_{4}^{\beta}\right)+\delta\gamma_{45}\left(\rho_{4}^{\alpha}\rho_{4}^{\beta}-\rho_{5}^{\alpha}\rho_{5}^{\beta } \right)}{\omega_{13}}\right)\sin\left(\frac{|\omega_{13}|t}{2}\right)\bigg]e^{-\frac{1}{2}(\gamma_{4}+\gamma_{5})t } \nonumber \\ & + & \bigg[\left ( \rho_{6}^{\alpha}\rho_{6}^{\beta}+\rho_{7}^{\alpha}\rho_{7}^{\beta}\right ) \cos\left(\frac{|\omega_{23}|t}{2}\right ) \nonumber \\ & + & \left(\frac{2\delta_{23}\left ( \rho_{6}^{\alpha}\rho_{7}^{\beta}-\rho_{7}^{\alpha}\rho_{6}^{\beta}\right)+\delta\gamma_{67}\left(\rho_{6}^{\alpha}\rho_{7}^{\beta}-\rho_{7}^{\alpha}\rho_{7}^{\beta } \right)}{\omega_{23}}\right)\sin\left(\frac{|\omega_{23}|t}{2}\right)\bigg]e^{-\frac{1}{2}(\gamma_{6}+\gamma_{7})t } \nonumber \\ & + & e^{-\gamma_{3}t}\rho_{3}^{\alpha}\rho_{3}^{\beta}+e^{-\gamma_{8}t}\rho_{8}^{\alpha}\rho_{8}^{\beta } \bigg\}.\end{aligned}\ ] ] in the case @xmath117 , we would have an analogous expression , but with hyperbolic sines and cosines . in this case
we switch on only the non - null mixing matrix elements in the @xmath111-@xmath112 sector , so that the mixing matrix is defined as @xmath119 due to the form adopted for the mixing matrix , oscillatory terms will not appear in the transition probability between @xmath113 and the other flavors , remaining only the decoherence terms for generating non - zero neutrino conversion related to @xmath113
. the survival and conversion probabilities of interest for the case @xmath120 can be explicitly built now and are given by @xmath121e^{-\frac{1}{2}(\gamma_{1}+\gamma_{2})t } \\ & & + \frac{1}{2}\cos^{2}2\theta e^{-\gamma_{3 } t}+\frac{1}{6}e^{-\gamma_{8}t } , \\
p^{\rm{\tiny{\tiny{d\oplus o}}}}_{\nu_{e } \rightarrow \nu_{\mu}}(t)&=&\frac{1}{3}-\frac{1}{2}\sin^{2}2\theta\left[\cos\left(\frac{|\omega_{12}|t}{2}\right)+\frac{\delta\gamma_{12}}{|\omega_{12}|}\sin\left(\frac{|\omega_{12}|t}{2}\right)\right]e^{-\frac{1}{2}(\gamma_{1}+\gamma_{2})t } \nonumber \\ & & -\frac{1}{2}\cos^{2}2\theta e^{-\gamma_{3 } t}+\frac{1}{6}e^{-\gamma_{8}t } , \\
p^{\rm{\tiny{\tiny{d\oplus o}}}}_{\nu_{e } \rightarrow \nu_{\tau}}(t)&=&\frac{1}{3}-\frac{1}{3}e^{-\gamma_{8}t } , \\
p^{\rm{\tiny{\tiny{d\oplus o}}}}_{\nu_{\mu } \rightarrow \nu_{\mu}}(t)&=&\frac{1}{3}+\frac{1}{2}\sin^{2}2\theta\left[\cos\left(\frac{|\omega_{12}|t}{2}\right)+\frac{\delta\gamma_{12}}{|\omega_{12}|}\sin\left(\frac{|\omega_{12}|t}{2}\right)\right]e^{-\frac{1}{2}(\gamma_{1}+\gamma_{2})t } \nonumber \\ & & + \frac{1}{2}\cos^{2}2\theta e^{-\gamma_{3 } t}+\frac{1}{6}e^{-\gamma_{8}t } , \\
p^{\rm{\tiny{\tiny{d\oplus o}}}}_{\nu_{\mu } \rightarrow \nu_{\tau}}(t)&=&\frac{1}{3}-\frac{1}{3}e^{-\gamma_{8}t } , \end{aligned}\ ] ] and analogous expressions can be written for the case @xmath117 , through the substitutions @xmath122 , @xmath123 .
once we have defined these probabilities we will check their consistency with the tendency indicated by the atmospheric neutrino data .
our test will be based upon the definitions for the normalized @xmath93-events and @xmath94-events given in the eqs .
( [ normrat1 ] ) and ( [ normrat2 ] ) .
since our probabilities involving mixing with @xmath111 are going to be applied for the atmospheric neutrino data , matter effects must be considered .
this is because the atmospheric neutrino data involve neutrinos which travel different distances through the earth .
the inclusion of matter effects in the probability expressions is really straightforward in the case where constant matter density is considered @xcite .
basically , we only need to replace @xmath124 ( in the corresponding sines and cosines ) , and @xmath125 where @xmath126 and , @xmath127 where @xmath128 is given by @xmath129 @xmath130 is the fermi constant , @xmath131 the electron number density , @xmath132 the neutrino energy and @xmath133 the matter density . for evaluating how these matter effects affect our probabilities , we will choose the following values for the relevant parameters : @xmath134 , consistent with the large mixing angle ( lma ) solution to the solar neutrino problem @xcite , @xmath135 , consistent with the mean neutrino energy for multi - gev events , and @xmath136 @xcite , this value is one of the lowest values in the earth matter density profile . using these values
we estimate that @xmath137 and @xmath138 .
note that this will happen even faster for higher values of matter density . as a result
, we can write simplified probabilities for this case @xmath139 here we go one step further and assume @xmath140 , @xmath141 thus we mimic completely the two generation expression for decoherence in @xmath1 .
at this point we have @xmath142 , @xmath143 and @xmath144 .
writing eqs .
( [ normrat1 ] ) and ( [ normrat2 ] ) as a function of @xmath145 , we obtain @xmath146 now we can check , in a model independent way , the consistency of these expressions with the atmospheric neutrino observations . to make an analysis consistent with our assumption on the neutrino energy , we will look at the tendencies of multi - gev events , this implies @xmath147 .
the @xmath93-like multi - gev events are consistent with @xmath148 for any neutrino trajectory , so that @xmath149 and consequently , @xmath150 which is in strong disagreement with the behavior of neutrinos coming from below the horizon , since @xmath94-like multi - gev events are indicating in average that @xmath151 . therefore this case is also in conflict with the present data . in this case
we turn on only the mixing matrix elements that connect @xmath111-@xmath113 , which implies the following mixing matrix @xmath152 so that the mass induced oscillation terms do not appear in any transition involving @xmath112 .
then , for the case @xmath120 , we obtain the probability formulae @xmath153 e^{-\frac{1}{2}(\gamma_{4}+\gamma_{5})t } , \\
p^{\rm{\tiny{\tiny{d\oplus o}}}}_{\nu_{e } \rightarrow \nu_{\mu}}(t)&=&\frac{1}{3}-\frac{1}{2}\cos^2\theta e^{-\gamma_{3}t}+\frac{1}{6}(\cos^2\theta-2\sin^2\theta)e^{-\gamma_{8}t } , \\
p^{\rm{\tiny{\tiny{d\oplus o}}}}_{\nu_{e } \rightarrow \nu_{\tau}}(t)&=&\frac{1}{3}+\frac{1}{2}\cos^2\theta\sin^{2}\theta e^{-\gamma_{3}t}+\frac{1}{6}(\cos^2\theta-2\sin^2\theta)(\sin^2\theta-2\cos^2\theta)e^{-\gamma_{8}t } \\ & & -\frac{1}{2}\sin^{2 } 2\theta\left[\cos\left(\frac{|\omega_{13}|t}{2}\right)+\frac{\delta\gamma_{45}}{|\omega_{13}|}\sin\left(\frac{|\omega_{13}|t}{2}\right)\right]e^{-\frac{1}{2}(\gamma_{4}+\gamma_{5})t } , \\
p^{\rm{\tiny{\tiny{d\oplus o}}}}_{\nu_{\mu } \rightarrow \nu_{\mu}}(t)&=&\frac{1}{3}+\frac{1}{2}e^{-\gamma_{3}t}+\frac{1}{6}e^{-\gamma_{8}t } , \\
p^{\rm{\tiny{\tiny{d\oplus o}}}}_{\nu_{\mu } \rightarrow \nu_{\tau}}(t)&=&\frac{1}{3}-\frac{1}{2}\sin^{2}\theta e^{-\gamma_{3}t}+\frac{1}{6}(\sin^2\theta-2\cos^2\theta)^{2}e^{-\gamma_{8}t } , \end{aligned}\ ] ] and once again , the case @xmath154 can be obtained just by substituting the harmonic functions by their hyperbolic partners . before analyzing the probabilities in this case
, we will make some further assumptions .
first , we will take @xmath155 and @xmath156 , both consistent with the lma solution to the solar neutrino problem @xcite , second we will neglect all the decoherence parameters for the @xmath82 at the range of the atmospheric neutrino scale , with the exception of @xmath157 , which is needed in order to make @xmath86 similar to its form in the decoherence solution to the atmospheric neutrino anomaly in two generations .
once we have accepted these assumptions , we can study the compatibility of these probabilities with the current data .
we will not use the atmospheric neutrino data to test the viability of the decoherence solution in the three generations scheme .
this is because the probabilities are written as a function of @xmath158 and @xmath159 , which makes it difficult to extract correct conclusions without introducing refinements to the qualitative analysis we have used up to now .
instead , we will simply look at chooz data .
this will be enough to give us a good idea about the compatibility of the decoherence solution in the three neutrino scheme , through the observations in the channel @xmath160 .
since we are going to use for this study @xmath161 , it is convenient to write it down after the assumptions mentioned above are taken into account @xmath162 we have obtained excluded regions in the plane ( @xmath163 ) , with @xmath164 , following the statistical procedure described in ref .
we must point out that variations in @xmath165 will not affect these excluded regions , since @xmath166 in the chooz range .
the results are shown in fig .
[ fig1 ] , where the points to the right of the contours are excluded .
we observe that the point in this plane which corresponds to the value of the decoherence parameter which best explains the atmospheric neutrino data , @xmath167 gev@xmath168 , as well as @xmath169 which corresponds to the best fit value for the lma solution to the solar neutrino problem , @xmath170 ( @xmath171 ) , denoted by a black dot , is excluded at 99 % c.l . in general , if @xmath172 we can exclude @xmath173 gev@xmath174 at 99 % c.l . in this way we observe that chooz data also highly disfavor this three neutrino scheme , since it excludes a large region of @xmath175 compatible with the atmospheric neutrino solution for values of @xmath169 consistent with the lma solution to the solar neutrino deficit .
under the assumption that the neutrino system can interact with a pervasive environment , we have obtained neutrino probability formulae for three neutrino generations , taking into account quantum dissipative effects coming from the interaction with the medium on top of the oim mechanism .
the damping terms were brought in through the quantum dynamical semigroup formalism .
this approach is very useful , since no _ a priori _ assumption on the form of neutrino - medium interaction has to be made .
some simplifications of the form of the dissipative matrix were adopted based on results in two generations . we have performed a qualitative analysis to test if the two generation decoherence solution to the atmospheric neutrino problem viewed in this hybrid three neutrino framework , can still explain the tendencies of the current experimental neutrino data .
we have analyzed two different cases , the first one considering only pd and the second including a mixture of both conversion mechanisms , that is , decoherence plus oim .
the second case was further subdivided into two cases , according to the choice of mixing matrix : i ) mass and mixing contributions connecting @xmath176 ; ii ) mass and mixing contribution connecting @xmath177 .
we have observed that all of these cases are clearly disfavored by recent relevant experimental neutrino data .
particularly , in the pd case , the fact that @xmath178 is not compatible with the constraint given by chooz combined with sk data ( @xmath97 and @xmath179 ) or with k2k results . for the hybrid case of decoherence plus non - null mixing in @xmath176 , the same prediction
@xmath178 arises which is clearly not supported by sk data . in the case of decoherence plus mixing in @xmath177 , we have made a statistical analysis of chooz data using our theoretical expression for @xmath161 , with some simplifications .
we have obtained that the best fit value for the decoherence solution to the atmospheric neutrino problem , @xmath180 gev@xmath174 , is highly disfavored by data .
values of @xmath181 gev@xmath174 for @xmath158 consistent with the lma solution to the solar neutrino deficit are , in general , excluded at 99 % c.l .. although , the tests we have performed in the three neutrino scheme indicate a disagreement between data and theoretical expectations , this does not mean that dissipative effects can not exist as subleading processes , with the full three neutrino oim as the main mechanism for neutrino flavor conversion .
in fact , the formulae developed here are interesting to be used to help establishing limits on the decoherence parameters or to try to detect their effects using an appropriate three neutrino description .
also it is worth to stress that we have worked here in a simplified situation in which the dissipative matrix is diagonal .
the presence of off - diagonal terms can certainly produce interesting effects on the probability expressions .
our formulae can be easily modified to include these off - diagonal terms .
however , any further qualitative or quantitative analysis in three generations will not be so direct .
this work was supported by conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) and by fundao de amparo pesquisa do estado de so paulo ( fapesp ) . 99 e. b. davies , _ quantum theory of open systems _ , academic press , london ( 1976 ) .
w. h. louisell , _ quantum statistical properties of radiation _
, jonh wiley & sons , new york ( 1973 ) .
f. benatti and r. floreanini , jhep * 0002 * , 32 ( 2000 ) .
f. benatti and r. floreanini , _ phys .
d _ * 64 * , 085015 ( 2001 ) .
a. m. gago , e. m. santos , w. j. c. teves and r. zukanovich funchal , _ phys .
d _ * 63 * , 073001 ( 2001 ) .
a. m. gago , e. m. santos , w. j. c. teves and r. zukanovich funchal , _ phys .
d _ * 63 * , 113013 ( 2001 ) .
e. lisi , a. marrone and , d. montanino , _ phys .
* 85 * , 1166 ( 2000 ) .
kamiokande collaboration , h. s. hirata _
_ , _ phys .
* b 205 * , 416 ( 1988 ) ; _ ibid . _ * 280 * , 146 ( 1992 ) ; y. fukuda _ et al .
_ , _ ibid._*335 * , 237 ( 1994 ) ; imb collaboration , r. becker - szendy _ et al .
* 46 * , 3720 ( 1992 ) ; soudan-2 collaboration , w. w. m. allison _
* b 391 * , 491 ( 1997 ) ; super - kamiokande collaboration , y. fukuda _ et al .
_ , _ phys .
lett . _ * 81 * , 1562 ( 1998 ) ; _ phys .
* b 436 * , 33 ( 1999 ) .
homestake collaboration , k. lande _ et al .
_ , _ astrophys
. j. _ * 496 * , 505 ( 1998 ) ; _ nucl .
* 77 * ( _ proc
. suppl . _ ) , 13 ( 1999 ) ; gallex collaboration , w. hampel _
_ , _ phys .
lett . _ * b 447 * , 127 ( 1999 ) ; gno collaboration , m. altmann _
_ , _ phys .
_ * b 490 * , 16 ( 2000 ) ; sage collaboration , j. n. abdurashitov _
_ , _ phys .
* c 60 * , 055801 ( 1999 ) ; v. n. gavrin , _ nucl .
* 91 * ( _ proc
_ ) , 36 ( 2001 ) ; super - kamiokande collaboration , y. fukuda _ et al .
_ , _ phys .
lett . _ * 81 * , 1158 ( 1999 ) , ( e ) _
ibid._*81 * , 4279 ( 1998 ) ; _ ibid._*82 * , 1810 ( 1999 ) ; _ ibid._*82 * , 2430 ( 1999 ) ; y. suzuki , _ nucl .
* 91 * ( _ proc .
_ ) , 29 ( 2001 ) ; sno collaboration , q. r. ahmad _ et al .
_ , _ phys .
lett . _ * 87 * , 071301 ( 2001 ) .
sno collaboration , q. r. ahmad _ et al .
_ , _ phys .
* 89 * , 011301 ( 2002 ) ; _ ibid._*89 * , 011302 ( 2002 ) .
r. alicki and k. lendi , _ quantum dynamical semigroups and applications _ , lect .
notes phys .
* 286 * , springer - verlag , berlin ( 1987 ) .
g. lindblad , _ commun .
phys . _ * 48 * , 119 ( 1976 ) . v. gorini , a. kossakowski and e. c. g. sudarshan , _ j. math
* 17 * , 821 ( 1976 ) .
m. gell - mann , _ phys .
rev . _ * 125 * , 1067 ( 1962 ) .
f. benatti and h. narnhofer , _ lett .
* 15 * , 325 ( 1988 ) . v. gorini , a. frigerio , m. verri , a. kossakowski and e. c. g. sudarshan , _ rep .
* 13 * , 149 ( 1978 ) .
a. frigerio and v. gorini , _ j. math
* 17 * , 2123 ( 1976 ) .
a. kossakowski , _ rep .
phys . _ * 3 * , 247 ( 1972 ) .
r. s. ingarden and a. kossakowski , _ ann .
_ * 89 * , 451 ( 1975 ) .
e. b. davies , _ commun .
phys . _ * 39 * , 91 ( 1974 ) .
c. favre and p. a. martin , _ helv .
. acta _ * 41 * , 333 ( 1968 ) .
r. dmcke and h. spohn , _
b _ * 22 * , 419 ( 1979 ) . chooz collaboration , m. apollonio _
et al . _ , _ phys .
* b 466 * , 415 ( 1999 ) .
k2k collaboration , j. e. hill _
et al . _ , hep - ex/0110034 .
super - kamiokande collaboration , m. shiozawa , talk at neutrino 2002 , http://neutrino2002.ph.tum.de/. t. kajita and y. totsuka , _ rev .
* 73 * , 85 ( 2001 ) .
super - kamiokande collaboration , s. fukuda _ et al .
_ , _ phys .
* b 539 * , 179 ( 2002 ) .
a. dziewonski and d. anderson , _ phys .
earth planet .
inter . _ * 25 * , 297 ( 1981 ) .
we present here the explicit form of the matrix @xmath48 , its eigenvalues and eigenvectors , as well as the diagonalizing matrix @xmath49 and its inverse defined in eq .
( [ diagonal ] ) . starting from the definition of @xmath48 in eq .
( [ diff2 ] ) , taking as valid the approximation @xmath182@xcite , we can write @xmath183 with the definitions @xmath184 now solving the secular equation @xmath185 , we get @xmath186=\frac{1}{2}\left[-(\gamma_{1}+\gamma_{2})-\omega_{12}\right ] \nonumber \\ \lambda_{2}&=&\frac{1}{2}\left[-(\gamma_{1}+\gamma_{2})+\sqrt{(\gamma_{2}-\gamma_{1})^{2}-4\delta_{12}^{2}}\right]=\frac{1}{2}\left[-(\gamma_{1}+\gamma_{2})+\omega_{12}\right ] \nonumber \\ \lambda_{3}&=&-\gamma_{3 } \nonumber \\ \lambda_{4}&=&\frac{1}{2}\left[-(\gamma_{4}+\gamma_{5})-\sqrt{(\gamma_{5}-\gamma_{4})^{2}-4\delta_{13}^{2}}\right]=\frac{1}{2}\left[-(\gamma_{4}+\gamma_{5})-\omega_{13}\right ] \\
\lambda_{5}&=&\frac{1}{2}\left[-(\gamma_{4}+\gamma_{5})+\sqrt{(\gamma_{5}-\gamma_{4})^{2}-4\delta_{13}^{2}}\right]=\frac{1}{2}\left[-(\gamma_{4}+\gamma_{5})+\omega_{13}\right ] \nonumber \\ \lambda_{6}&=&\frac{1}{2}\left[-(\gamma_{6}+\gamma_{7})-\sqrt{(\gamma_{7}-\gamma_{6})^{2}-4\delta_{23}^{2}}\right]=\frac{1}{2}\left[-(\gamma_{6}+\gamma_{7})-\omega_{23}\right ] \nonumber \\ \lambda_{7}&=&\frac{1}{2}\left[-(\gamma_{7}+\gamma_{8})+\sqrt{(\gamma_{7}-\gamma_{6})^{2}-4\delta_{23}^{2}}\right]=\frac{1}{2}\left[-(\gamma_{6}+\gamma_{7})+\omega_{23}\right ] \nonumber \\ \lambda_{8}&=&-\gamma_{8 } \nonumber\end{aligned}\ ] ] the associated eigenvectors are in turn @xmath187 where the superscript @xmath188 denotes transposition .
one can now construct the diagonalizing matrix @xmath49 and its inverse @xmath189 @xmath190
flavor eigenstates @xmath192 can be written as a function of mass eigenstates @xmath193 through a unitary matrix @xmath77 @xmath194 we can express the density matrix of a flavor state @xmath195 as @xmath196 | -1.1 cm using the open quantum system approach applied to the neutrino system , we derive three generations neutrino probability formulae considering the oscillation induced by mass plus quantum decoherence contributions .
the introduction of these dissipative effects is done through the quantum dynamical semigroup formalism .
in addition to the theoretical interest of the approach , at least from the completeness point of view , this extension of the formalism to the three flavors , provide us with a direct application : we can analyze qualitatively the consistency of the two generation pure decoherence solution to the atmospheric neutrino problem , accommodated within this enlarged scheme , with the mean tendencies observed for some of the current neutrino experimental data .
this study was performed based on different choices of the @xmath0 mixing matrix selected in order to adjust the @xmath1 to the same form it has for the decoherence solution in two generations .
our qualitative tests for decoherence with three neutrinos show a clear incompatibility between neutrino data and the theoretical expectations .
15truecm -0.4truecm 22truecm
1.truecm 1.truecm = |
the measurement of the cabibbo - kobayashi - maskawa ( ckm ) matrix @xcite element @xmath0 is currently one of the important goals of @xmath1 physics .
standard model predictions employ the fundamental parameter @xmath0 as input . as the experiments at the @xmath1 factories are starting to take data , a precise determination of @xmath0 is increasingly vital for testing the standard model .
for example , the standard model prediction for the cp - violating asymmetry in @xmath5 decays depends on the value of @xmath0 .
evidence for the cp - violating asymmetry has already been provided by the cdf collaboration @xcite . as the dedicated experiments are expected to achieve much greater precision ,
an increase in the accuracy of @xmath0 is highly desirable in order to test whether the complex phase of the ckm matrix is the only source of cp violation . on the other hand , despite empirical successes , the standard model is definitely not the final theory of particle physics .
there remain many outstanding issues : electroweak symmetry breaking , fermion masses and mixing , cp violation , replication of families , etc .
a precise determination of @xmath0 is also an important link in pursuit of deeper principles to extend the standard model .
@xmath0 can be determined from both inclusive and exclusive charmless semileptonic @xmath1 meson decays .
the precision of the determination is limited by experimental and theoretical difficulties .
charmless semileptonic @xmath1 decays are a rare process .
the main experimental difficulty in observing signals from @xmath6 processes is the very large background due to @xmath7 whose branching fraction is two orders of magnitude larger . on the theoretical side ,
calculations are needed in order to relate the measured quantity to @xmath0 to extract a value from the data .
strong interaction effects complicate the calculations , causing theoretical uncertainties .
the uncertainties result not just from nonperturbative qcd which is notoriously difficult to calculate , but also from perturbative qcd although in priciple it is calculable in perturbation theory . with the developments in nonperturbative techniques ,
the uncertainty from perturbative qcd due to the truncation of perturbation series in practical calculations can be comparable to that from nonperturbative qcd , indicating that these two sources of uncertainty can be equally important .
two approaches have been developed for a qcd treatment of inclusive decays of heavy hadrons .
one called the heavy quark expansion approach @xcite is based on the operator product expansion .
one called the light - cone approach @xcite is based on the light - cone expansion , using the notion and method of deep inelastic lepton - hadron scattering . in both approaches
the heavy quark effective theory @xcite is exploited as a limit of qcd .
there is important distinction between the two approaches .
the former is a short - distance expansion in local operators of increasing dimension , while the latter is a non - local light - cone expansion in matrix elements of increasing twist .
moreover , the heavy quark expansion approach invokes quark - hadron duality , which assumes rates evaluated at the parton level to be equal to observable rates summed over a sufficient number of hadronic channels .
quark - hadron duality can not be exact @xcite .
the heavy quark expansion approach has to use quark kinematics and can not account for the rate due to the extension of phase space from the parton level to the hadron level .
the rate missing @xcite is just a manifestation of duality violation .
in contrast , the light - cone approach does not rely on the assumption of quark - hadron duality .
rates are evaluated in the light - cone approach using physical phase space at the hadron level .
the light - cone approach was criticized in refs .
however , it has been shown @xcite that that criticism is mistaken .
the validity of the light - cone approach has been tested @xcite experimentally on the lepton energy spectrum in inclusive semileptonic decays of @xmath1 mesons .
it is important to test theoretical approaches experimentally in various ways . to overcome the aforementioned difficulties in the determination of @xmath0 , we proposed @xcite a method of extracting @xmath0 from inclusive charmless semileptonic decays of @xmath1 mesons .
we proposed to measure the lepton pair spectrum , i.e. , the decay distribution of the kinematic variable @xmath8 in the @xmath1-meson rest frame , where @xmath9 is the momentum transfer to the lepton pair and @xmath10 denotes the @xmath1 meson mass .
because most of @xmath3 ( @xmath11 or @xmath12 ) events have a value of @xmath13 beyond the threshold allowed for @xmath4 decay , @xmath14 with @xmath15 being the @xmath16 meson mass , this kinematic requirement provides a powerful tool for background suppression . if nothing else , the decay distribution of @xmath13 is of direct interest from an experimental standpoint .
the sum rule can then be used to extract @xmath0 from the weighted integral of the measured @xmath13 spectrum .
the sum rule follows from the light - cone expansion for inclusive charmless semileptonic decays of @xmath1 mesons and @xmath2-flavored quantum number conservation .
the sum rule is thus independent of phenomenological models .
moreover , the sum rule does not receive any perturbative qcd correction .
therefore , this method is not only experimentally very efficient , but theoretically also quite clean , allowing a precise determination of @xmath0 with a minimal overall ( experimental and theoretical ) error .
the sum rule requires measuring the @xmath13 spectrum in inclusive charmless semileptonic @xmath1 meson decays , weighted with @xmath17 .
the purpose of the present paper is to refine the analysis of @xcite .
we calculate the perturbative qcd correction to the @xmath13 spectrum to order @xmath18 .
we investigate in detail how gluon radiation and hadronic bound state effects affect the shape of the @xmath13 spectrum .
the rest of the paper is organized as follows . in sec .
ii the sum rule for inclusive charmless semileptonic decays of @xmath1 mesons is briefly described .
we discuss how to extract @xmath0 from experiment exploiting the sum rule . in sec .
iii we study the @xmath13 spectrum , including both the leading perturbative and nonperturbative qcd corrections .
the exploration is extended to the weighted @xmath13 spectrum in sec .
iv . finally in sec .
v we present our conclusions .
the sum rule for the decay distribution of @xmath13 in inclusive charmless semileptonic decays of @xmath1 mesons has been derived @xcite from a nonperturbative treatment of qcd based on light - cone expansion .
the large @xmath1 meson mass sets a hard energy scale in the reactions , so that the light - cone expansion is applicable @xcite to inclusive @xmath1 decays that are dominated by light - cone singularities , just as deep inelastic scattering . for inclusive charmless semileptonic decays of @xmath1 mesons , @xmath19 of phase space has @xmath20 gev@xmath21 .
this dominance of the high-@xmath22 region in the whole phase space @xmath23 renders the light - cone expansion especially reasonable . the theoretical expression for the weighted integral of the distribution of @xmath13 @xmath24 is very clean . in the leading - twist approximation of qcd , it was found @xcite that @xmath25 where @xmath26 is the momentum of the decaying @xmath1 meson .
the masses of leptons , the @xmath27 quark and the @xmath28 meson are neglected . because @xmath2-flavored quantum number conservation under strong interactions implies @xmath29 the leading - twist contribution of nonperturbative qcd to the observable @xmath30 in eq .
( [ eq : leading ] ) is precisely calculable from first principles .
the result is the sum rule @xcite @xmath31 the conserved current @xmath32 is not subjected to renormalization and the corresponding anomalous dimension vanishes thanks to the ward - takahashi identity .
hence we recognize that the sum rule ( [ eq : sumrule ] ) does not receive any perturbative qcd correction .
it might seem surprising that one is able to calculate @xmath30 so precisely in qcd when the experimentalist measures hadrons , and it is well known that hadrons can not be understood completely within perturbation theory .
there is a physical way to make clear that we really do not need to know anything about how hadrons are formed in order to predict @xmath30 in inclusive @xmath1 decays .
the observable @xmath30 measures something fundamental namely the @xmath2-flavored quantum number carried by the @xmath1 meson , and is basically determined by the underlying global symmetry of qcd .
it is therefore insensitive to hadronic bound state effects .
a measurement of the observable quantity @xmath30 in inclusive charmless semileptonic decays of @xmath1 mesons will lead to a determination of @xmath0 by use of the sum rule ( [ eq : sumrule ] ) which contains no additional parameter .
this determination of @xmath0 is free of perturbative qcd uncertainties .
meanwhile , the dominant hadronic uncertainty is avoided .
higher - twist corrections of nonperturbative qcd to the sum rule ( [ eq : sumrule ] ) are suppressed by a power of @xmath33 .
once the observable @xmath30 is measured , then eq .
( [ eq : sumrule ] ) yields @xmath0 with small theoretical uncertainty .
moreover , this method is not only theoretically quite clean , but experimentally also very efficient in the discrimination between @xmath34 signal and @xmath35 background , as we shall discuss .
in this section we describe the theoretical prediction for the @xmath13 spectrum .
we explore step by step perturbative and nonperturbative qcd effects on the @xmath13 spectrum .
we take into account first the perturbative qcd correction to order @xmath18 and then the leading nonperturbative qcd correction . ignoring qcd corrections ,
the tree - level @xmath13 spectrum in the free quark decay @xmath6 in the @xmath2-quark rest frame takes the form @xmath37 with @xmath38 where @xmath39 represents the @xmath2-quark mass .
the resulting spectrum is a discrete line at @xmath40 .
this is simply a consequence of kinematics that fixes @xmath13 to the single value @xmath41 , no other values of @xmath13 are kinematically allowed in @xmath6 decays .
this is also the case for @xmath6 processes with virtual gluon emission .
therefore , the @xmath13 spectrum in free quark decays is still a discrete line at @xmath40 , shown in fig . 1
, even if virtual gluon emission occurs
. however , the above kinematic relation no longer holds for free @xmath2-quark decays with gluon bremsstrahlung , and hence the spectrum expands downward below the parton - level end point @xmath40 .
= 9truecm including both virtual gluon emission and gluon bremsstrahlung , the differential decay rate as a function of @xmath13 in the @xmath2-quark rest frame is calculated to order @xmath18 to be @xmath42 \nonumber\\ \nonumber\\ & + & \frac{2\alpha_s}{3\pi}\frac{m_b}{m_b } \bigg [ -2\left(\frac{{\rm ln}r}{r}\right)_+ -\frac{13}{3}\left(\frac{1}{r}\right)_+ \nonumber\\ \nonumber\\ & + & \frac{79}{9}+\frac{407}{36}r-\frac{367}{12}r^2+\frac{59}{3}r^3 -\frac{50}{9}r^4 + \frac{11}{12}r^5-\frac{7}{36}r^6 \nonumber\\ \nonumber\\ & + & \left ( -\frac{2}{3}+\frac{23}{3}r+3r^2-\frac{8}{3}r^3\right ) { \rm ln}r + 2r^2(-3 + 2r){\rm ln}^2 r \bigg ] , \label{eq : free}\end{aligned}\ ] ] where @xmath43 varying in the range @xmath44 , corresponding to the kinematic range @xmath45 at the parton level of quarks and gluons . the distribution @xmath46_+$ ]
is defined to coincide with the function @xmath47 for all values of @xmath48 greater than 0 , and to have a singularity at @xmath49 such that the integral of this distribution with any smooth function @xmath50 gives @xmath51_+ t(r ) = \int_0 ^ 1 dr\ , g(r)[t(r)-t(0 ) ] .
\label{eq : def1}\ ] ] the result for the perturbative @xmath13 spectrum with the qcd radiative correction ) does not receive any perturbative qcd correction because it results from the conserved current . the numerical calculation by integrating eq . ( [ eq : free ] ) with the factor @xmath17 has not reached the absolute zero but yielded a tiny correction ( about @xmath52 ) .
we have checked that this is only a numerical artifact due to the limited accuracy of numerical integration by computer , especially given the rapid oscillation of the integrand near the end point @xmath53 . ] to order @xmath18 is shown in fig . 2 .
the perturbative @xmath13 spectrum is singular at its end point @xmath54 due to infrared divergences .
we observe that gluon bremsstrahlung generates a small tail below @xmath54 .
= 9truecm integrating eq .
( [ eq : free ] ) over @xmath13 yields the total perturbative decay rate @xmath55 .
\label{eq : total}\ ] ] this agrees with the well - known result obtained in @xcite . to calculate the real physical decay distribution in inclusive decays
@xmath3 , perturbative qcd alone is not sufficient .
we must also account for hadronic bound state effects due to the confinement of the @xmath2 quark inside the @xmath1 meson .
we now consider nonperturbative qcd effects on the @xmath13 spectrum . in the framework of the light - cone expansion
, the leading nonperturbative qcd effect is incorporated in the @xmath2-quark distribution function @xcite @xmath56b(y ) \label{eq : def}\ ] ] where @xmath57 denotes path ordering .
the distribution function @xmath58 has a simple physical interpretation : it is the probability of finding a @xmath2-quark with momentum @xmath59 inside the @xmath1 meson with momentum @xmath26 . the real physical spectrum is then obtained from a convolution of the hard perturbative spectrum with the soft nonperturbative distribution function : @xmath60 where the @xmath2-quark momentum @xmath61 in the perturbative spectrum is replaced by @xmath59 .
the analytic result for the perturbative spectrum to order @xmath18 is given in eq .
( [ eq : free ] ) .
the interplay between nonperturbative and perturbative qcd effects has been accounted for since confinement implies that free quarks are not asymptotic states of the theory and the separation of perturbative and nonperturbative effects can not be done in a clear - cut way .
equation ( [ eq : convol ] ) demonstrates that the physical @xmath13 spectrum depends on the distribution function @xmath58 .
the definition of it , eq . ( [ eq : def ] ) , involving the @xmath1-meson matrix element of the non - local @xmath2-quark operators separated along the light cone , makes clear that the distribution function is a nonperturbative quantity .
although a complete calculation of the distribution function in qcd is impossible at present due to our ignorance of nonperturbative qcd , some basic properties of it are known @xcite .
the distribution function is universal in the sense that the same distribution function also summarizes the leading nonperturbative qcd contribution in inclusive radiative @xmath1 decays @xmath62 .
it is gauge invariant and obeys positivity .
it has a support between 0 and 1 and is exactly normalized to unity because of @xmath2-flavored quantum number conservation ) . ] .
it contains the free quark decay as a limiting case with @xmath63 . in the free quark limit , eq .
( [ eq : convol ] ) consistently reproduces the free quark spectrum .
in addition , the mean @xmath12 and the variance @xmath64 of the distribution function were deduced @xcite using operator product expansion and heavy quark effective theory ( hqet ) @xcite : @xmath65 @xmath66 , \label{eq : variance}\ ] ] where @xmath67 and @xmath68 and @xmath69 are the dimensionless hqet parameters of order @xmath70 , which are often referred to by the alternate names @xmath71 and @xmath72 .
the parameter @xmath73 can be extracted from the @xmath74 mass splitting : @xmath75 gev@xmath21 .
the parameter @xmath76 suffers from large uncertainty . the mean value and variance of the distribution function characterize the location of the `` center of mass '' of the distribution function and the square of its width , respectively .
they specify the primary shape of the distribution function . from eqs .
( [ eq : mean ] ) and ( [ eq : variance ] ) we know that the distribution function is sharply peaked around @xmath77 close to 1 and its width of order @xmath78 is narrow , suggesting that the distribution function is close to the delta function form in the free quark limit .
nonperturbative qcd methods such as lattice simulation and qcd sum rules could help determine further the functional form of the distribution function .
the distribution function can also be extracted directly from experiments of inclusive semileptonic @xcite or radiative @xcite decays of @xmath1 mesons .
the universality of the distribution function implies great predictive power : once the distribution function is measured from one process , it can be used to make predictions in all other processes in a model - independent manner .
since these are as yet not done , we perform the calculations using the parametrization @xcite of the distribution function @xmath79^\beta}\theta(\xi ) \theta(1-\xi ) , \label{eq : para}\ ] ] where @xmath80 , @xmath81 , @xmath82 , and @xmath2 are four parameters and @xmath83 is the normalization constant .
the parametrization ( [ eq : para ] ) respects all the known properties of the distribution function , in particular the strong constraints of the sum rules ( [ eq : mean ] ) and ( [ eq : variance ] ) .
the @xmath13 spectrum can be calculated using eqs .
( [ eq : convol ] ) , ( [ eq : free ] ) and ( [ eq : para ] ) . including both the leading nonperturbative and perturbative qcd corrections
, we obtain the @xmath13 spectrum shown in fig .
3 , using @xmath84 , @xmath85 , @xmath86 and @xmath87 . here both values of @xmath80 and @xmath81 for the four - parameter distribution function ( [ eq : para ] ) are preset to be 1 , but in general they need not be integers .
the values of @xmath82 and @xmath2 are then inferred from the sum rules ( [ eq : mean ] ) and ( [ eq : variance ] ) using @xmath88 gev and @xmath89 gev@xmath21 , giving the mean @xmath90 and the variance @xmath91 for the distribution function .
bound - state effects lead to the extension of phase space from the parton level to the hadron level , also stretch the spectrum downward below @xmath41 , and are solely responsible for populating the spectrum upward in the gap between the parton - level end point @xmath40 and the hadron - level end point @xmath92 .
the interplay between nonperturbative and perturbative qcd effects eliminates the singularity at the end point of the perturbative spectrum , so that the physical spectrum shows a smooth behavior over the entire range of @xmath13 , @xmath93 , as in fig .
3 . therefore
, nonperturbative qcd effects play a crucial role in shaping the @xmath13 spectrum .
= 9truecm by integrating eq .
( [ eq : convol ] ) over @xmath13 , we obtain the fraction of @xmath3 events above the charm threshold allowed for the predominant @xmath4 decays , defined as @xmath94 we find that @xmath95 for the spectrum shown in fig .
3 . this result refines our previous result @xcite . the earlier calculation without qcd
radiative corrections found @xmath96 .
gluon bremsstrahlung streches the @xmath13 spectrum downward , and hence gives rise to a decrease of the fraction of @xmath3 events above the charm threshold .
nevertheless , most of @xmath3 events remain above the charm threshold .
thus the kinematic cut on the observable quantity @xmath13 is very efficient in disentangling @xmath3 signal from @xmath4 background .
this efficiency can be explained by the uniqueness of the @xmath13 spectrum : the @xmath13 spectrum stemming from the quark - level and virtual gluon exchange processes would only concentrate at @xmath40 , shown in fig .
1 , solely on kinematic grounds , and gluon bremsstrahlung and hadronic bound state effects smear the spectrum about this point , but most of the decay rate remains at large values of @xmath13 , as revealed by figs . 2 and 3 .
the fraction @xmath97 of course depends on forms of the distribution function . however , we find that @xmath97 is relatively insensitive to forms of the distribution function once the mean and variance of it , which are known from hqet , given by eqs .
( [ eq : mean ] ) and ( [ eq : variance ] ) , are kept fixed .
therefore , the above calculation of @xmath97 can be considered as a typical estimate of the fraction of @xmath3 events above the charm threshold .
the weighted spectrum @xmath98 is more directly relevant for the measurement of the observable @xmath30 in inclusive charmless semileptonic decays of @xmath1 mesons
. measurements of @xmath30 can be obtained from an extrapolation of the weighted spectrum measured above the charm threshold to the full phase space available in @xmath3 decays .
while the normalization of the weighted spectrum given by the sum rule ( [ eq : sumrule ] ) does not depend on the @xmath2-quark distribution function @xmath58 , thus being model - independent , the shape of the weighted spectrum does . in this section
we perform a detailed analysis of the shape of the weighted spectrum , using eqs .
( [ eq : convol ] ) , ( [ eq : free ] ) and ( [ eq : para ] ) .
we first explore the impact of gluon radiation on the shape of the weighted @xmath13 spectrum .
we show in fig .
4 the weighted @xmath13 spectram without and with the qcd radiative correction . it is evident from fig . 4 that the shape of the weighted @xmath13 spectrum receives a significant correction due to gluon radiation .
however , the shape of the spectrum appears to be insensitive to the value of the strong coupling @xmath18 , varied within a reasonable range .
= 9truecm we investigate next the sensitivity of the spectrum to the form of the distribution function , which would reflect the impact of hadronic bound state effects .
for this purpose we choose to use two very different forms @xcite for the distribution function , albeit having the same mean value and variance .
the calculated weighted spectra are shown in fig . 5 , taking into account the qcd radiative correction to order @xmath18 .
the weighted spectrum exhibits a strong dependence on the form of the distribution function , even though the mean value and variance of the distribution functions are the same .
= 9truecm generally , since the quark - level processes , exclusive of gluon bremsstrahlung , generate a discrete line , the shape of the @xmath13 spectrum directly reflects the inner long - distance dynamics of the reaction .
this argument elucidates the strong variation of the weighted spectrum with the form of the distribution function , illustrated in fig . 5 . in fact , ignoring qcd radiative corrections , one obtains from eqs .
( [ eq : convol ] ) and ( [ eq : tree ] ) the weighted spectrum in the @xmath1 rest frame @xmath99 it makes clear that the resulting weighted spectrum is directly proportional to the distribution function . in other words ,
the weighted @xmath13 spectrum is most sensitive to the distribution function .
this salient feature renders the weighted spectrum idealy suited for the direct extraction of the distribution function from experiment . after an observed spectrum @xmath98
is radiatively corrected , it is nothing but the @xmath2-quark distribution function if higher - twist terms are neglected .
extrapolation of the weighted @xmath13 spectrum measured above the charm threshold to the full @xmath13 range requires a theoretical calculation of the rate ratio @xmath100 where @xmath30 is defined in eq .
( [ eq : integral ] ) .
we calculate the ratio @xmath101 using eqs .
( [ eq : sumrule ] ) , ( [ eq : convol ] ) , ( [ eq : free ] ) and ( [ eq : para ] ) , and the theoretical uncertainties are estimated as follows : we study the variation of @xmath101 with respect to the mean value and the variance of the distribution function setting @xmath85 in eq .
( [ eq : para ] ) .
actually , this amounts to the study of the ratio @xmath101 as functions of @xmath39 and @xmath76 , since , essentially , the mean value of the distribution function is determined by the @xmath2-quark mass and its variance is determined by @xmath76 according to the sum rules in eqs .
( [ eq : mean ] ) and ( [ eq : variance ] ) . at present
, the estimated values of the @xmath2-quark mass and @xmath76 vary in the ranges @xmath102 the variation of @xmath39 leads to an uncertainty of @xmath103 in @xmath101 if other parameters are kept fixed . a small uncertainty of @xmath104 in @xmath101 results from the variation of @xmath76 . in other words , the ratio @xmath101 displays a strong dependence on the mean value of the distribution function of the @xmath2 quark inside the @xmath1 meson , but is insensitive to the variance of the distribution function .
we examine the further sensitivity of the rate ratio @xmath101 to the form of the distribution function when keeping the mean value and variance of it fixed , by varying the values of the two additional parameters @xmath80 and @xmath81 in the parametrization ( [ eq : para ] ) .
we estimate that the variation of @xmath101 is @xmath104 if the form of the distribution function is changed but with the same mean value and variance .
we estimate the uncertainty due to the truncation of the perturbative series in eq .
( [ eq : free ] ) by varying the renormalization scale between @xmath105 and @xmath106 .
we find that an uncertainty of @xmath103 in the ratio @xmath101 stems from the renormalization scale dependence .
this analysis implies that at present the theoretical error in the calculation of @xmath101 has two main sources : the value of @xmath39 ( or equivalently , the mean value of the distribution function ) and the renormalization scale dependence . finally , adding all the uncertainties in quadrature we arrive at @xmath107 with an error of @xmath108
we have studied the @xmath13 spectrum in inclusive charmless semileptonic decays of @xmath1 mesons .
the perturbative qcd correction to the spectrum is calculated to order @xmath18 .
the leading nonperturbative qcd effect is calculated using light - cone expansion and heavy quark effective theory .
the @xmath13 spectrum is unique in that the tree - level and virtual gluon exchange processes @xmath6 at the parton level generate a trivial @xmath13 spectrum a discrete line at @xmath109 in the @xmath2-quark rest frame , which is well above the charm threshold @xmath110 .
gluon bremsstrahlung results in a small tail in the lepton pair spectrum below @xmath41 .
bound - state effects lead to the extension of phase space from the parton level to the hadron level , also stretch the spectrum downward below @xmath41 , and are solely responsible for populating the spectrum upward in the gap between the parton - level endpoint @xmath54 and the hadron - level endpoint @xmath111 , smoothing out the singularity in the perturbative spectrum . as a result of these two distinct effects ,
the lepton pair spectrum spreads over the entire physical range @xmath93 .
still , we find that about @xmath112 of the lepton pair spectrum in @xmath3 lies above the charm threshold , @xmath113 .
this kinematic cut is most efficient in the suppression of the background of @xmath1 semileptonic decays into charmed particles .
the observable @xmath30 can be measured by the extrapolation of the weighted spectrum @xmath98 measured above the charm threshold to the entire phase space .
gluon bremsstrahlung and hadronic bound state effects strongly affect the shape of the weighted @xmath13 spectrum . however , the shape of the weighted @xmath13 spectrum is insensitive to the value of the strong coupling @xmath18 , varied in a reasonable range .
the overall picture appears to be that the weighted @xmath13 spectrum is peaked towards larger values of @xmath13 with a narrow width .
the contribution below @xmath114 is moderate and relatively insensitive to forms of the distribution function .
this suggests that extrapolating the weighted @xmath13 spectrum down to low @xmath13 would not introduce a considerable uncertainty in the value of @xmath30 .
quantitatively , our analysis determines the rate ratio for the extrapolation of the weighted @xmath13 spectrum to be @xmath115 with an error of @xmath108 at present .
another interesting use of the weighted @xmath13 spectrum is its utility for directly extracting the @xmath2-quark distribution function .
the universality of the distribution function implies that the distribution function extracted from inclusive semileptonic @xmath1 decays can be used to make predictions in inclusive radiative decays @xmath62 in a model - independent manner and vice versa .
in particular , a measurement of the distribution function from the photon energy spectrum in @xmath62 @xcite would be very useful to improve the measurement of @xmath30 from extrapolation of the weighted @xmath13 spectrum .
an improved knowledge of the form of the distribution function is important for an error reduction in extrapolation .
more precisely , if by direct measurements or theoretical studies the uncertainty in the mean value of the distribution function can be improved from currently @xmath104 to @xmath116 in the future , the resulting uncertainty in the ratio @xmath101 for the extrapolation would decrease from @xmath103 to @xmath104 .
a calculation of the @xmath117 perturbative qcd correction to the @xmath13 spectrum is also important for improving the accuracy of the extrapolation . the physical observable quantity @xmath30 is connected with @xmath0 via the sum rule ( [ eq : sumrule ] ) for semileptonic @xmath1 decay in the leading - twist approximation of qcd .
the sum rule is a fundamental , model - independent prediction of qcd .
note that the sum rule does not rely on the heavy quark effective theory .
no arbitrary parameter other than @xmath0 enters the sum rule . the dominant hadronic uncertainty is avoided in the sum rule .
as important , there is no perturbative qcd modification of this sum rule , so that the potential source of theoretical uncertainty associated with perturbative qcd calculations is totally averted .
the kinematic cut on @xmath13 , @xmath118 , and the semileptonic @xmath1 decay sum rule , eq .
( [ eq : sumrule ] ) , offer an outstanding opportunity for the precise determination of @xmath0 from the observable @xmath30 .
we wish to emphasize that this method is both exceptionally clean theoretically and very efficient experimentally in background suppression .
there remain two types of theoretical uncertainties in the determination of @xmath0 .
first , higher - twist ( or power suppressed ) corrections to the sum rule cause a theoretical error of order @xmath119 on @xmath0 . a quantitative study of higher - twist effects could further reduce this small theoretical uncertainty associated with simply extracting the value of @xmath0 from the measured value of @xmath30 .
second , the extrapolation of the weighted @xmath13 spectrum to low @xmath13 gives rise to a systematic error for the measurement of @xmath30 .
the status of this additional theoretical uncertainty associated with extrapolation , as well as how to improve it , have been discussed above .
eventually , the error on @xmath0 determined by this method would mainly depend on how well the observable @xmath30 can be measured . to measure @xmath30
experimentally one needs to be able to reconstruct the neutrino .
this poses a challenge to experiment .
given the unique potential of determining @xmath0 , we would urge our experimental colleagues to examine the feasibility of the method .
j. chay , h. georgi , and b. grinstein , phys .
b * 247 * , 399 ( 1990 ) ; + i.i .
bigi , n.g .
uraltsev , and a.i .
vainshtein , phys .
b * 293 * , 430 ( 1992 ) ; * 297 * , 477(e ) ( 1993 ) ; + i.i .
bigi , m.a .
shifman , n.g .
uraltsev , and a.i .
vainshtein , phys .
rev . lett . * 71 * , 496 ( 1993 ) ; int .
phys . a * 9 * , 2467 ( 1994 ) ; + a.v .
manohar and m.b .
wise , phys .
d * 49 * , 1310 ( 1994 ) ; + b. blok , l. koyrakh , m.a .
shifman , and a.i .
vainshtein , phys .
d * 49 * , 3356 ( 1994 ) ; * 50 * , 3572(e ) ( 1994 ) ; + m. luke and m.j .
savage , phys .
b * 321 * , 88 ( 1994 ) ; + a.f .
falk , m. luke , and m.j .
savage , phys .
d * 49 * , 3367 ( 1994 ) ; + t. mannel , nucl .
b * 413 * , 396 ( 1994 ) ; + a.f .
falk , z. ligeti , m. neubert , and y. nir , phys .
b * 326 * , 145 ( 1994 ) ; + m. neubert , phys .
d * 49 * , 3392 and 4623 ( 1994 ) ; + t. mannel and m. neubert , phys .
d * 50 * , 2037 ( 1994 ) .
jin and e.a .
paschos , in _ proceedings of the international symposium on heavy flavor and electroweak theory _ ,
beijing , china , 1995 , edited by c.h . chang and c.s .
huang ( world scientific , singapore , 1996 ) , p. 132
; hep - ph/9504375 . | a precise determination of @xmath0 can be obtained exploiting the sum rule for inclusive charmless semileptonic @xmath1-meson decays .
the sum rule is derived on the basis of light - cone expansion and @xmath2-flavored quantum number conservation .
the sum rule does not receive any perturbative qcd correction . in this determination of @xmath0
, there is no perturbative qcd uncertainty , while the dominant hadronic uncertainty is avoided .
moreover , this method is not only theoretically quite clean , but experimentally also very efficient in the discrimination between @xmath3 signal and @xmath4 background .
the sum rule requires measuring the lepton pair spectrum .
we analyze the lepton pair spectrum , including the leading perturbative and nonperturbative qcd corrections . |
growing concerns over the limited supply of fossil - based fuels are motivating intense activity in the search for alternative road transportation propulsion systems . regulatory pressures to reduce urban pollution ,
@xmath0 emissions and city noise have made electric vehicles ( evs ) @xcite and plug - in hybrid vehicles ( phevs ) @xcite a very attractive choice as the alternative to the internal combustion engine ( ice ) @xcite .
electric vehicles in particular , which have zero tailpipe emissions of pollutants , are seen as a very useful tool in reducing urban pollution due to the fact that they deliver energy in a clean form in our cities , as well as reducing the carbon footprint of road transportation generally when combined with clean energy generation .
while the environmental and societal benefits of zero - emission vehicles are evident , their adoption by users has been extremely disappointing .
according to recent reports @xcite , even in europe , where the green agenda is well received , fewer than 12000 electric vehicles were sold in the first half of 2012 ( of which only 1000 of these were sold in the uk ) .
this number represents less than @xmath1 of total new car sales .
these figures are in spite of the fact that many european governments have offered incentives for the purchase of electric vehicles in the form of subsidies and have also invested in enabling infrastructure .
some of the factors hindering the widespread adoption by the general public of electric vehicles can be summarised briefly as follows : * * battery related issues * plug - in vehicles tend to be very expensive , even when subsidised .
a major factor in the cost of such vehicles is the cost of the battery @xcite .
lithium - based batteries are expensive , and while costs are forecast to reduce dramatically over the next few years @xcite , this is currently an important aspect in understanding the sales of electric vehicles . in response to this , renault , and other companies ,
are proposing to lease batteries to the customer to offset some of the battery related costs .
however , even if such initiatives are successful , there are other battery related questions that may hinder the adoption of electric vehicles . a further point concerns whether enough lithium can be sourced to build batteries to construct enough vehicles to replace the existing passenger vehicle fleet .
are we simply substituting one rare resource ( oil ) with another ( lithium ) ? also , the transportation of batteries is not trivial and necessitates special precautions @xcite . finally , most reasonably sized batteries are not capable of realising the range enjoyed by conventional ice based vehicles .
while this latter issue is the subject of much research , battery size and performance currently represents one of the major determinants in the design of electric vehicles today @xcite . * * electromagnetic emissions * a recent issue regarding electric vehicles concerns electromagnetic emissions .
while there is no evidence that em radiation from evs is dangerous , this issue is a focus point for regulatory authorities ( see eu green car programme ) and has been raised by several research agencies @xcite . * * long charging times * charging times for electric vehicles are known to be long @xcite .
an often cited fact by advocates of electric vehicles in response to this is that fast charging algorithms can service average vehicles in about 30 minutes @xcite .
such time - scales may be just about acceptable to a normal car owner .
however , in the presence of queuing , 30 minutes can rapidly become several hours , and push such fast charging stations into the realm of `` not acceptable . ''
thus , it is likely that overnight or workplace charging will be the principal method of vehicle charging for the foreseeable future .
an associated issue in large cities concerns the availability of charging points .
this is especially an issue in cities with large apartment block type dwellings . * * vehicle size * electric vehicles tend to be designed small with limited luggage space to reduce energy consumption .
this is a significant problem for most potential purchasers of vehicles who on occasion would like to transport significant loads using their vehicles . * * range anxiety * one of the most pressing issues in the deployment of evs concerns the issue of range anxiety @xcite .
maximum ranges ( in favourable conditions ) of less than 150 km are not unusual for electric vehicles , and this reduces significantly when air - conditioning or heating is switched on @xcite .
the issue of limited range also exacerbates other issues .
for example , the cost of searching for a parking space at the end of a journey is much higher than for a conventional vehicle ( because the ev s range is so low and therefore energy should not be wasted searching for a parking spot ) .
research is ongoing to address range issues , with much of the current work focussing on new battery types , optimal vehicle charging , vehicle routing , and in - vehicle energy management systems with a view to minimising wastage of energy and thereby increasing vehicle range @xcite .
while all of the above issues are important , as well as some others that we have not mentioned , in this paper , we focus on the pressing issue of * range anxiety * ( see below ) , and to a lesser extent , * vehicle size * by suggesting a flexible vehicle access model to alleviate both of these issues . *
principal contribution of paper :* specifically , a solution is proposed to some of these problems based on a car sharing .
this idea is an embodiment of flexible vehicle access that was first suggested in @xcite and further developed and analyzed in @xcite .
indeed , the timeliness of the idea is evidenced by the fact that since this paper was first submitted for publication and presented @xcite , several automotive manufacturers , including fiat , and volkswagen , have announced plans for a form of car - sharing with similarities in both goals and implementation , to that described here .
the principal difference to these embodiments and that proposed by ibm is that we advocate and describe an * on - demand model * , whereas car manufacturers are suggesting a model that guarantees access for a fixed , but small , number of days .
our contribution in this paper is twofold .
we develop tools to design an on - demand service using ideas from queueing theory and using predictive analytics .
we then demonstrate , in the context of real irish mobility patterns , that such an on - demand service is economical , both in terms of the number of ice vehicles needed , and in terms of real additional cost to vehicle manufacturers .
we point out that in our queueing model it is assumed that extensive trips are planned and the on - demand scheme responds to demand announced on the previous day .
there are , of course , other possible on - demand scenarios .
the assumptions described below pose no fundamental restrictions in this respect and the methods presented here can be extended .
our basic assumption is that the range anxiety problem , along with vehicle cost , are the biggest barriers to the purchase of electric vehicles . while vehicle cost is principally an issue of manufacturing scale and
is likely to come down , range anxiety is an issue that is unlikely to be solved in a cost effective manner in the near future by developments in the vehicle design process . for the purpose of this paper
we consider the term range anxiety to mean the _ angst _ of a vehicle owner that he / she will not have enough range to reach his / her destination without the need for recharging .
roughly speaking , one may consider two types of problems associated with range anxiety .
* the first problem is associated with an inadequate battery level to complete a trip while driving to a destination . *
the second problem is associated with electric vehicles being unsuited to the trip distribution demanded by the user .
for example , a long trip , or vacation ( or indeed a trip where a large luggage load is required ) , are all problematic for a typical electric vehicle . in many cities
the first of these problems can to a large extent be avoided with adequate trip planning .
for example , in cities where single dwelling households with garages / driveways are common , it is possible to charge vehicles adequately overnight to have a full battery charge the next morning . as we shall see later ,
this is , in most cases , adequate for the majority of trips .
since , most uk / irish cities are of this type , we shall assume that overnight charging is possible and focus on the second of the above range anxiety issues .
it is worth noting that the issue of range anxiety has been the subject of the attention of policy bodies and car manufacturers over the past number of years . roughly speaking
, research attention has focussed on three areas : ( i ) better batteries ; ( ii ) optimizing energy management in the vehicles ; and ( iii ) novel energy delivery strategies for electric vehicles .
big efforts on designing better batteries have been made worldwide @xcite . to optimize energy management ,
manufacturers have looked both within the electric vehicle through the management of the vehicle sub - systems , and outside of the vehicle through the use of energy - aware routing and the use of special lanes ( see www.green-cars-initiative.eu/projects/eco-fev ) . finally , several methods of delivering energy to the vehicle have been suggested .
these include battery swapping , as was advocated by betterplace , under - road induction , and fast charging outlets .
these activities have addressed for the most part item ( i ) above ; and largely ignore the inconvenience associated with item ( ii ) .
our car sharing concept closely follows the flexible access suggestion in @xcite in the following manner . 1
when an electric vehicle is purchased , the new ev owner also automatically becomes a member of a car sharing scheme , where a shared vehicle may be borrowed from a common pool on a 24hr basis .
the shared vehicles are large ice - based vehicles suitable for long range travel and with large goods transportation capacity .
we suggest free membership of the scheme , but a pricing model could be implemented to regulate demand on weekends , public holidays , or other occasions when synchronised ( correlated ) demand is likely to emerge , or to regulate emissions .
further , if the shared ice - based vehicles are chosen to be sufficiently high - end , then a further incentive for consumers to purchase electric vehicles is provided .
a number of issues need to be resolved before any such system could be deployed .
these issues reduce to the marginal cost of the system . more specifically
, we wish to determine if such a sharing concept could be deployed giving reasonable quality of service ( qos ) to the electric vehicle owner , without significantly increasing the cost of each vehicle . referring to figure [ fig : concept ] , this amounts to asking whether a reasonable qos can be delivered when @xmath2 , the number of shared ice - based vehicles , is significantly less than @xmath3 , the number of purchased evs . to answer this question
, we consider two scenarios . *
* spontaneous journeys :* if we want to achieve a low probability for the event that an individual requesting a vehicle is not allocated one , how should the proportion @xmath4 be chosen ? *
* planned journeys :* for a fixed @xmath2 and @xmath3 , how many days in advance does a user have to make a reservation so that the probability that the request is declined is lower than some very small constant ? in solving the above problems we shall make the following assumptions . *
a shared vehicle is borrowed for a single day " period and returned to the pool of shared vehicles to be shared again the next day .
in other words , at the beginning of each day " , @xmath2 cars are available for sharing .
* customers collect their vehicles from a number of sharing - sites .
they either park their own ev at the site when collecting the ice vehicle , or travel there using some other form of transport . *
members of our car sharing scheme do not have access to an ice based vehicle other than through the car - sharing scheme .
* members of the car - sharing community are willing to accept the same @xmath5 metric for vehicle access .
_ ii . demographic assumptions _
* long journeys in private cars are rare , meaning that the range of an electric vehicle , even under worst - case conditions ( eg : air conditioning use , traffic congestion , bad weather ) , should be sufficient for most journeys . * most urban dwellings are houses rather than apartments , meaning that there are no structural impediments to overnight charging , and that a full overnight charge should be sufficient to satisfy the needs of most daily mobility patterns .
the validity of these latter assumptions are the subject of the next section . to make a case for the plausibility of assumptions ii.1 and ii.2 we examined publicly available data on contemporary irish mobility patterns .
data for the creation of table [ tab1 ] and figures [ f1 ] , [ f2 ] and [ f3 ] were obtained from the 2009 irish national travel survey ( nts ) microdata file , central statistics office , government of ireland @xcite . in the nts ,
respondents were asked to provide details about their travel for a given ( randomly selected ) 24h period , which roughly corresponded to a day of the week .
table [ tab1 ] shows the percentages of people who drove private vehicles ( over the 24h period that they were queried about ) for cumulative daily distances of greater than 50 km , 75 km and 100 km .
figure [ f1 ] relates to those people who were questioned about their travel over the 24h `` monday '' period ( ie : row two of table [ tab1 ] ) , and depicts number of people versus total distances they drove in private vehicles over that 24h monday period .
figure [ f1 ] illustrates a trend observed in the percentages in table [ tab1 ] ; namely , that longer cumulative journeys over the course of a day were rare , but concerning travel over the other days of the week , were similar in shape to figure [ f1 ] , and have thus been omitted . ] . for respondents who drove cumulative distances greater than 75 km over a 24h period ( see the third column of table [ tab1 ] ) ,
figure [ f2 ] illustrates those hours of a 24h period over which respondents had their vehicles in use ( many vehicles were in use roughly between 8 am and 6 pm ) , and figure [ f3 ] depicts number of respondents versus total time ( out of a 24h period ) their vehicle was in use .
u|ssssample population & 50 km & 75 km & 100 km + monday & 23% & 12% & 7% + tuesday & 23% & 14% & 8% + wednesday & 23% & 14% & 7% + thursday & 26% & 18% & 11% + friday & 26% & 17% & 9% + saturday & 24% & 15% & 9% + sunday & 24% & 17% & 11% + [ tab1 ] 75 km total daily distance ) using their vehicles , versus hour of the day . ]
75 km total daily distance ) , versus total time that their vehicle was in use for over their 24h survey period . ]
the nts data set allows us to make some very general statements which support the assumptions listed above .
as might be anticipated , longer journeys are not as frequent as shorter ones , and vehicles do indeed tend to be in - use for a single day for each user .
thus , the data set appears to be indeed consistent with the assumptions ii.- ii.2 above .
however , the data - set is subject to certain limitations due to its size and the manner in which it was collected , and it is very important to comment on these before proceeding . * first , the data - set does not take into account correlated driving behaviors between days .
however , we assume that drivers making regular long - distance trips would not purchase ev s . consequently , the tail of the distributions following from the data - sets should be representative of the qualitative behavior of drivers making regular short trips , but who infrequently need larger vehicles or vehicles for larger trips . in what follows
we shall assume that the typical ev owner follows a distribution given by these data sets - although clearly from the above discussion this is a simplification as driving behavior varies across ev owners .
* second , it is possible that the car - sharing scheme may affect the driving patters of scheme members ( ev owners ) .
for example , given ice vehicle access , long - distance driving may become more attractive . in the analysis to follow ( in the next section ) , this feedback mechanism is also ignored to keep matters simple . *
the data also indicates that vehicles are `` in - use '' for the entire working day .
we assume that vehicles are charged overnight to full capacity and that it is not necessary for a vehicle to be again charged during the day .
as we have already mentioned , this latter assumption is consistent with cites such as dublin where garage space or a driveway is the norm , and is consistent with charging - times for standard home charge points .
the interested reader is referred to @xcite for further information about driver charging behavior .
given the above discussion it is clear that the presented data provides a plausibility argument in support of assumptions ii.1 and ii.2 subject to the aforementioned limitations of the data - set .
for a further and more detailed discussion of the limits of using such data sets the interested reader is referred to @xcite .
using elementary probability and queueing theory methods , we now pose solutions to the quality of service ( qos ) problems presented in section [ s1 ] . consider a population of @xmath3 electric vehicle owners ( ie : @xmath3 `` users '' ) who occasionally require access to an ice - based vehicle ( ice or icev ) for a non - standard trip ( either a long - range trip or a trip where large load carrying capacity is required ) .
we assume that a user will keep the icev for a full day , based on the driver behaviours described in the previous section .
thus , each day is characterised by the number of users who require an icev on that day .
there is a fleet of @xmath2 icevs available to satisfy this need .
the main question is then to determine the relation between @xmath2 and @xmath3 .
this will be determined by requiring some qos conditions to be met .
for example , a qos condition might be a guarantee on the probability of finding an icev available .
an important assumption in what follows is that all users are willing to accept the same @xmath5 metric for vehicle access .
this is defined by a probability of a scheme member not being able to access a vehicle when required .
similar assumptions are standard in the networking community , and are based on the assumption that a member of the car - sharing scheme will not join the scheme unless the given qos metric is acceptable .
clearly , different levels of service can be provided if groups of users are willing to pay for a higher level of service .
however , to keep matters simple here we assume that drivers who sign up for the scheme are prepared to be served using the same qos metric . in the simplest model
, each user independently requests an icev each day with probability @xmath6 .
based on the data available in table [ tab1 ] we may estimate this probability as @xmath7 , assuming that icevs are used for trips over 100 km .
the assumption that cars are required for a whole day are justified by the data in figures 2 - 4 , which show that the overwhelming majority of cumulative trips over 75 km require a period of 5 hours or more . with the available data we are not able to estimate the proportion of drivers that make long trips on a regular basis , who would not participate in the scheme . on the other hand
it may be expected that some icevs are requested even though the actual trip length turns out to be less than 100 km . in the following
we will use @xmath8 to illustrate the applicability of the proposed scheme .
thus , the number of requests @xmath9 each day is a binomial random variable : x ~bin(n , p ) .
the mean number of requests per day is @xmath10 , and the standard deviation is @xmath11 . in principle
, the number of requests may be anything from @xmath12 to @xmath3 , but for large @xmath3 it is very unlikely that @xmath9 will deviate from the mean by more than a few standard deviations .
this qos condition can be quantified as follows . for each @xmath13 , define q(m ) = p(x > m ) .
then the qos condition could be to find the smallest @xmath2 such that @xmath14 for some specified @xmath15 .
for any given @xmath3 and @xmath6 this can be calculated explicitly using the formula q(m ) = _ k = m+1^n ( n k ) p^k ( 1-p)^n - k .
however it is more useful to get an approximate formula from which the scaling relation can be read off .
for @xmath3 large enough we can use the normal approximation for the binomial , which says that z = is approximately a standard normal random variable .
there is a standard rule of thumb regarding applicability of the normal approximation for the binomial , namely that n 9 ( , ) . using the normal approximation
, we have @xmath16 this readily yields estimates for @xmath2 in order to satisfy a desired qos condition . for example , in order to satisfy the qos condition q(m ) < 0.05 meaning a less than @xmath17 chance of not finding an icev available , it is sufficient to take r 1.65 m n p + 1.65 for example , using the values @xmath18 users and @xmath19 for the probability of a user requesting an icev , this provides a value @xmath20 .
when the number of requests exceeds the number of available icevs , a queue forms and users must wait one or several days until a vehicle becomes available .
it is desirable to keep the probability of long delays small , and this can be achieved by appropriate scaling of @xmath2 with @xmath3 .
this is the subject of this section .
let @xmath21 be the number of outstanding requests at the end of the @xmath22 day , and let @xmath23 be the number of new requests that arrive during the @xmath22 day .
since the number of vehicles is @xmath2 , the relation between these variables on successive days is x_n+1 = \ { 0 , x_n - m } + a_n+1 .
that is , the queue length is reduced by @xmath2 at the start of each day ( but not reduced below zero ) , and is then increased during the day by the number of new requests . the qos condition is to ensure that @xmath21 is unlikely to be large , implying that users are unlikely to have to wait a long time before being assigned an icev .
since @xmath2 users can be serviced each day , in the worst - case a user must wait @xmath24 extra days until service .
we consider the qos condition which guarantees that the probability that any user needs to wait @xmath25 extra days or more is less than @xmath15 , that is p(x_n > k m ) < . by choosing @xmath2
sufficiently large we can guarantee that this probability is small .
1 [ lem : prob - bound ] define = m - n p , ^2 = n p ( 1 - p ) , = .
then for all @xmath26 , p(x_n > k m ) e^-(k-1 ) m ( e^/2 - 1)^-1 . using the bound in lemma [ lem : prob - bound ] we find a sufficient condition to guarantee the qos bound , namely e^-(k-1 ) m ( e^/2 - 1)^-1 < . for a given @xmath25 and @xmath15 we may use this to find a value for @xmath2 needed to meet the qos condition .
for example , using the same values as above @xmath18 , @xmath19 , @xmath27 , and taking @xmath28 ( meaning that the probability that there is a customer who waits more than 4 days is less than @xmath29 ) , we find that the inequality is satisfied whenever @xmath30 .
taking @xmath31 we find @xmath32 , and with @xmath33 we find @xmath34 . for a fleet size of @xmath35 vehicles with @xmath36 @xmath37 and @xmath38 @xmath39 , the behaviour of the bound is depicted in figures [ fig : model2 ] and [ fig : model3 ] , respectively , for various estimates of the probability of a long distance trip .
as can be seen , the bound tends rapidly to zero , indicating that the probability of waiting for a shared vehicle longer than one or two days vanishes rapidly .
_ proof of lemma [ lem : prob - bound ] : _ define w_n = \{0 , x_n - m } , b_n = a_n - m. then the recursion relation becomes w_n+1 = \{0 , w_n + b_n } . as is well - known @xcite , with @xmath40
the solutions of this recursion relation have the same ( marginal ) distributions as the sequence @xmath41 where @xmath42 and w_n = \{0 , b_1 , b_1 + b_2 , } .
so define c_j = _ i=1^j b_i
. then we have for @xmath26 @xmath43 for large @xmath3 we can use the normal approximation to estimate the right side of this inequality .
the arrival numbers @xmath23 are independent and binomial with mean @xmath10 and variance @xmath44 , so in this approximation the variables @xmath45 are also normal , with mean @xmath46 and variance @xmath47 .
thus we find @xmath48 where @xmath49 is a standard normal .
we use the following bound valid for all @xmath50 p(z > b ) e^- b^2/2 to obtain p(c_j > ( k-1 ) m ) e^ - ( ( k-1)m + j ) ^2/2 j ^2 .
we use the bound ( + y ) ^2 2 x y + j y^2 and apply it above with @xmath51 and @xmath52 .
this gives p(x_n > k m ) & & _
j=1^n e^- ( k-1)m /^2 e^ - j /2 .
the sum is geometric , and is bounded by _
e^ - j /2 < _ j=1^ e^ - j /2 = as required .
@xmath53 is the left most curve ; @xmath54 is the right most curve . ]
is the left most curve ; @xmath54 is the right most curve . ]
we now explore the financial model associated with this car sharing model . initially , we will model our calculations based on a shared fleet comprised solely of a single vehicle type : the vw golf class car . in this model
we are essentially comparing the similarly sized vw golf and the nissan leaf to focus solely on examining range anxiety of the kind discussed above .
subsequently , we will model our calculations based on a shared fleet that is constructed to reflect the broader needs of the ev owners ; namely , sometimes a family car is needed , sometimes a large vehicle is needed for transporting goods , and sometimes a smaller car is required for short out of town trips by a single person / couple . in this subsequent model we focus on both range anxiety and a range of vehicle sizes . to do this
we make some simplifying assumptions . instead of modeling the demand for various vehicle types , which is the correct method of analysis , we shall weight various vehicle classes to construct a vehicle fleet .
we assume that these weights reflect demand .
this simplifying model is adopted for two reasons ; to keep analysis simple , and to reflect the fact that data of the type needed to build a multi - dimensional queueing model is not available to us at this time and is the subject of current work in our lab . in both model calculations we use volkswagen ( vw ) vehicles .
note that vw is a relatively high - end marque , and the vehicle fleet could be constructed in a manner that is considerably cheaper .
table [ table : nonlin ] illustrates the composition and pricing of our shared vehicle fleet ( including the single vehicle vw golf price ) .
vehicle purchase prices were sourced from the vw website ( www.volkswagen.ie ) on 23rd september , 2013 . the weighted average vehicle purchase price given
the fleet make - up is 23,308 .
.composition of the shared vehicle fleet . [ cols="<,^,<,^",options="header " , ] [ table5 ] to conclude it is worth recalling some of the assumptions underlying our analysis . *
our analysis ignores synchronized demand ( at weekends or during holiday periods ) .
we argue that a pricing structure could be enforced to breakup this demand and give a degree of qos to the scheme members . *
as we have mentioned , our combined fleet analysis is not based on demand distributions for each vehicle class .
rather , it is a simplified calculation to give the reader an indicative picture of the cost of the scheme .
* we have ignored the fact that the shared vehicles would be in continuous use thereby rendering their value lower than standard gmfv .
however , gmfv mileage restrictions may be offset through gains made when bulk buying 2,000 or 3,500 icev , and/or through negotiating greater mileage restriction caps , and/or by ensuring a reduction in any additional charges beyond the restricted mileage , and so on .
additionally , other depreciation models are easily incorporated into the given framework .
note that depreciation of the entire cost of the shared vehicles over a 3 year period to zero value is also very cost effective when compared with the average cost of government subsidies ( 7.7% for a vw golf in the 100 km model to 15.7% for a weighted average icev in the 75 km model , in comparison with the 23% average subsidy ) .
a fleet of 3,500 vw passat estates in the 75 km model depreciated to zero is more efficient than subsidies ( 19.4% compared to 23% ) . notwithstanding the above facts
, we have shown how any / all of the costs from the two model calculations are significantly less than the government subsidies costs .
this gives governments and/or industry the opportunity to augment and/or replace subsidies with this alternative model to encourage the uptake of ev cars , make ev cars more approachable , and reduce pollution in short journeys .
* simulation :* to conclude our paper we now present a brief simulation to illustrate the performance of our car sharing system .
we simulated demand based on figure 2 , for 365 days , and serviced this demand using our car - sharing model .
users requested an ice ( no accommodation for type of ice was implemented in our simulation ) with a fixed probability of @xmath19 ( to emulate infrequent need to take a long journey ) , independently of each other .
the number of members of the car - sharing program was @xmath55 .
the size of the ice fleet was varied from @xmath56 to @xmath57 .
figure [ fig : model4 ] depicts the percentage of customers who experienced not being able to access a car within @xmath25 days over a 365 day period .
as can be seen , the number of unhappy customers ( rejections ) goes rapidly to zero as @xmath2 approaches 100 .
this is entirely consistent with the conclusions presented above and illustrates that the predictions of the mathematics are upper bounds on performance .
* carbon savings :* finally , to quantify , very roughly , the emissions savings over an ice fleet , we consider the distance distribution depicted in figure 2 under the following assumptions .
we assume that all journeys are made with a single vehicle type traveling at a constant speed .
further , we assume that long distances are those greater than 100 @xmath58 .
then , based on figure 2 , the co@xmath59 saving is roughly @xmath60 using the proposed car - sharing scheme . *
collaborative consumption :*
we believe that the techniques presented in the paper go far beyond electric vehicles .
many sharing schemes ( sometimes called collaborative consumption schemes ) reduce to asking how many devices / units are needed to serve a demand , while at the same time delivering a statistical qos to individual users .
we believe our ideas and models will help in the context of such applications .
in this paper , a solution to a consumer range anxiety problem is studied using a car sharing idea .
the cost of this scheme was shown to be low when compared with current levels of subsidies to electric vehicle manufacturers .
this cost could be reduced further by exploiting car sharing on an hourly basis ( further multiplexing ) and using low - cost vehicles ( as opposed to a premium marque ) .
the authors thank douglas leith and lisa amini for their valuable comments and the excellent comments of the reviewers .
we would also like to thank emanuele crisostomi for noticing an error in figures 5 and 6 .
j. brady and m. omahony , travel to work in dublin .
the potential impacts of electric vehicles on climate change and urban air quality , _ transportaion research part d _ , vol .
16 , no . 2 , 2011 , pp .
188 - 193 .
j. axsen , a. burke and k. kurani , _ batteries for plug - in hybrid electric vehicles ( phevs ) : goals and the state of technology circa 2008 _ , report ucd - its - rr-08 - 14 , institute of transportation studies , university of california , davis , california , usa , 2008 .
m. lowe , s. tokuoka , t. trigg and g. gereffi , _ lithium - ion batteries for electric vehicles : the u.s .
value chain _ ,
research report , center on globalization , governance & competitiveness , duke university , durham , north carolina , usa , october 2010 .
s. amarakoon , j. smith and b. segal , _ application of life - cycle assessment to nanoscale technology : lithium - ion batteries for electric vehicles _ ,
epa 744-r-12 - 00 , u.s .
environmental protection agency , washington , d.c . ,
usa , 2013 .
scenihr ( scientific committee on emerging and newly identified health risks ) , _ possible effects of electromagnetic fields ( emf ) on human health _ , opinion , european commission , brussels , belgium , 21 march 2007 . m. yilmaz and p.t .
krein , review of battery charger topologies , charging power levels , and infrastucture for plug - in electric and hybrid vehicles , _ ieee transactions on power electronics _ ,
28 , no . 5 , 2013 , pp .
2151 - 2169 .
d. christen , s. tschannen and j. biela , highly efficient and compact dc - dc converter for ultra - fast charging of electric vehicles , in _ proceedings of the 15th international power electronics and motion control conference _ , novi sad , serbia , 2012 , pp .
ls5d.3 - 1 to ls5d.3 - 8 .
de oliveira , s. vaughan and j.g .
hayes , range estimation for the nissan leaf and tesla roadster using simplified power train models , in _ proceedings of the 2011 irish transport research network conference _ , university college cork , ireland , 2011 .
king , c. and griggs , w. and wirth , f. , and shorten , r. _ using a car string mode to alleviate electric vehicle range anxiety _ in proceedings of the sixteenth yale workshop on adaptive and learning systems , yale university , new haven , connecticut , usa , june , 2013 | on - demand vehicle access is a method that can be used to reduce types of range anxiety problems related to planned travel for electric vehicle owners . using ideas from elementary queueing theory ,
basic qos metrics are defined to dimension a shared fleet to ensure high levels of vehicle access . using mobility data from ireland
, it is argued that the potential cost of such a system is very low .
* keywords : * electric vehicles ; range anxiety ; on - demand vehicle access |
capacitively coupled plasma ( ccp ) radio frequency discharges are frequently used for applications such as plasma etching or plasma enhanced chemical vapor deposition ( pecvd ) processes in integrated circuit manufacturing .
impact by energetic electrons leads to dissociation and ionization of feedstock gas , and thus creation of reactive radicals and ions .
these applications usually require feedstock gases that are complex mixtures of reactive gases , often electronegative .
one such electronegative discharge is the oxygen discharge .
the oxygen discharge and its mixtures are of significance in various materials processing applications including etching of polymer films , ashing of photoresist , oxidation , and deposition of oxide films .
the oxygen chemistry is complicated due to the presence of metastable atomic and molecular species .
it is in particular the two low lying metastable molecular states designated by @xmath3 and @xmath4 , which are located 0.98 and 1.627 ev above the ground state , respectively , that play an important role .
it is well established that collisions with these metastable states have in many cases larger cross sections and thus higher reaction rates than corresponding collisions with the ground state molecule .
a volume averaged global model is an approach to get a detailed description of the plasma chemistry in low pressure discharges .
global model studies show that the dominant species in the oxygen discharge is the oxygen molecule in the ground state , o@xmath0(x@xmath5 ) , followed by the oxygen atom in the ground state , o(@xmath6p ) .
the singlet metastable states o@xmath0(a@xmath1 ) and o@xmath0(b@xmath2 ) and the metastable atom o(@xmath7d ) are also present in the plasma in significant amounts @xcite .
furthermore , a recent study shows that the o@xmath0(b@xmath2 ) density can overcome the o@xmath0(a@xmath1 ) density in the pressure range below 100 mtorr @xcite .
the o@xmath8-ions are in majority among the positive ions and the o@xmath9-ion density is much smaller and has a sharp decrease for pressures above 4 mtorr .
it is also seen that the negative ion density ratio [ o@xmath10/[o@xmath11 is 5.3 at 1 mtorr , and 1.3 at 100 mtorr @xcite .
such global models studies neglect spatial variations in the plasma parameters as well as the kinetics of the discharge .
they allow us to explore the relative reaction rates for the creation and destruction of particular species over the pressure range of interest and therefore which reactions are of significance . to explore the kinetics of the discharge we apply particle - in - cell monte carlo collision ( pic / mcc ) simulations . in a pic / mcc simulation
the plasma is represented as a collection of macro - particles .
equations of motion are solved for each macro - particle .
the electric and magnetic fields are calculated self - consistently using charge densities and currents produced by these macro - particles .
we use the oopd1 ( object oriented plasma device for one dimension ) code to simulate the discharge .
it has one dimension in space and three velocity components for the particles ( 1d-3v ) .
the oopd1 code is supposed to replace the widely used xpdx1 series ( xpdp1 , xpdc1 and xpds1 ) .
it was developed to simulate various types of plasmas , including processing discharges , accelerators and beams .
the code is written in modular structure and it includes features such as relativistic kinematics and different weights for different species .
earlier we developed a basic oxygen discharge model and benchmarked it against the well known xpdp1 code @xcite .
later we explored adding the singlet metastable molecule o@xmath0(a@xmath12 ) and the metastable oxygen atom o(@xmath7d ) to the reaction set @xcite .
there we demonstrated that the addition of the singlet metastable molecule o@xmath0(a@xmath12 ) has a significant influence on the discharge , such as on the electronegativity , the effective electron temperature , and the electron heating processes for a pressure of 50 mtorr .
we found that it is , in particular , detachment by the metastable molecule o@xmath0(a@xmath12 ) that has a significant influence on the discharge properties .
in a subsequent study we explored how the addition of the singlet metastable molecule o@xmath0(a@xmath12 ) modified the discharge properties in the pressure range 10 500 mtorr @xcite .
we found that at low pressure ( 10 mtorr ) , ohmic ( collisional ) heating in the plasma bulk dominates and the heating mechanism evolves through a phase where both ohmic heating in the bulk and sheath oscillation ( collisionless ) heating contribute , and at higher pressure ( 50 500 mtorr ) , the electron heating occurs almost solely in the sheath region . at the higher pressures , detachment by the metastable singlet molecule o@xmath0(a@xmath12 )
has a significant influence on the electron heating process but at the lowest pressures detachment by o@xmath0(a@xmath12 ) has only a small influence on the heating process . at low pressure ,
the electron energy probability function ( eepf ) is convex and as the pressure is increased the number of low energy electrons increases and the number of higher energy electrons ( @xmath13 ev ) decreases , and the eepf develops a concave shape or becomes bi - maxwellian @xcite .
this contradicts what is observed for a capacitively coupled argon discharge where the eepf evolves from being concave at low pressure to become convex at high pressure @xcite and for the chlorine discharge where the eepf is found to be convex in the pressure range of 10 100 mtorr @xcite . in a more recent study we added the singlet metastable oxygen molecule o@xmath0(b@xmath2 ) to the reaction set @xcite .
we find that including the singlet metastable o@xmath0(b@xmath2 ) further decreases the ohmic heating in the electronegative core and the effective electron temperature in the bulk region .
the effective electron temperature in the electronegative core is found to be less than 1 ev in the pressure range 50 200 mtorr which agrees with recent experimental findings .
furthermore , we have added an energy - dependent secondary electron emission yield for the positive oxygen ions o@xmath8 and o@xmath9 as well as the neutrals o@xmath0 and o @xcite .
we find that including an energy - dependent secondary electron emission yield for o@xmath14-ions has a significant influence on the discharge properties , including decreased sheath width and increased electron density @xcite .
other pic / mcc studies of the oxygen discharge have also indicated a significant role of the singlet metastable state o@xmath0(a@xmath12 ) and in particular the detachment by the o@xmath0(a@xmath12 ) on the overall discharge properties @xcite .
however , these studies did not include a complete reaction set for creation and destruction of the metastable states and the metastables are not treated kinetically . here
, we investigate the spatio - temporal electron heating dynamics in a geometrically symmetric oxygen discharge by 1d-3v pic / mcc simulations .
in particular we explore the influence of the electron detachment from the negative ion o@xmath15 by the singlet molecular metastables o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) on the spatio - temporal electron heating dynamics , the electronegativity , and the effective electron temperature .
we assume a capacitively coupled discharge where one of the electrodes is driven by an rf voltage ( left hand electrode ) while the other ( right hand electrode ) is grounded .
we assume the discharge to be operated at a single frequency of 13.56 mhz , and voltage amplitude of @xmath16 v with an electrode separation of 4.5 cm and a capacitor of 1 f in series with the voltage source .
the simulation grid is uniform and consists of 1000 cells .
the electron time step is @xmath17 s. the simulation was run for @xmath18 time steps or 2750 rf cycles . for the heavy particles we use sub - cycling and the heavy particles
are advanced every 16 electron time steps @xcite .
we assume that the initial density profiles are parabolic .
we consider an oxygen discharge that consists of : electrons , the ground state oxygen molecule o@xmath19x@xmath20 , the metastable oxygen molecules o@xmath0(a@xmath12 ) , and o@xmath0(b@xmath2 ) , the ground state oxygen atom o(@xmath6p ) , the metastable oxygen atom o(@xmath7d ) , the negative oxygen ion o@xmath15 , and the positive oxygen ions o@xmath9 and o@xmath8 .
the basic reaction set and cross sections included in the oopd1 code for the oxygen discharge excluding the metastable states is discussed in an earlier work and includes reactions among electrons , ground state molecules , ground state atoms , negative ions o@xmath15 , and positive ions o@xmath9 and o@xmath14 @xcite .
similarly , the reactions and cross sections involving the metastable oxygen atom o(@xmath7d ) and the metastable oxygen molecule o@xmath0(a@xmath12 ) are discussed by gudmundsson and lieberman @xcite .
the reactions involving the metastable singlet molecule o@xmath0(b@xmath2 ) , that are included in the model , and the energy dependent secondary electron emission yields are discussed by hannesdottir and gudmundsson @xcite .
the combined reaction set discussed in these three publications @xcite is referred to as the full reaction set in the discussion below .
the reactions explored here in particular are detachment by the singlet metastable molecule o@xmath0(a@xmath12 ) @xmath21 and detachment by the metastable o@xmath0(b@xmath2 ) @xmath22 for the detachment by the metastable o@xmath0(@xmath23 ) we use the rate coefficient measured at 400 k of @xmath24 m@xmath25/s by midey et al .
@xcite to estimate the cross section , which is allowed to fall as @xmath26 , where @xmath27 is the energy , to 184 mev and then take a fixed value of @xmath28 m@xmath29 . also we assume that detachment by the metastable o@xmath0(@xmath23 ) leads to the formation of o(@xmath6p ) + o@xmath30 + e , instead of o@xmath31 + e and o + o@xmath32 . the cross section for detachment from o@xmath15 by o@xmath0(b@xmath2 )
is estimated from the rate coefficient of @xmath33 m@xmath6/s , given by @xcite and then we allow it to fall as @xmath26 for @xmath34 mev and then remain constant .
the neutral gas density is much higher than the densities of charged species , so the neutral species at thermal energies ( below a certain cut - off energy ) are treated as a background with fixed density and temperature and maintained uniformly in space .
these neutral background species are assumed to have a maxwellian velocity distribution at the gas temperature ( here @xmath35 = 26 mv ) . in the oxygen discharge ,
the densities of the metastable oxygen molecules o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) , oxygen atom in the ground state o(@xmath6p ) and the metastable oxygen atom o(@xmath7d ) are much larger than the number of charged species .
we assume that the background consists of 4.4 % o@xmath0(a@xmath12 ) , 4.4 % o@xmath0(b@xmath2 ) , 90.6 % o@xmath19x@xmath20 , 0.519 % o(@xmath6p ) and 0.028 % o(@xmath7d ) as in our earlier work @xcite .
the neutrals are treated kinetically as particles if their energy exceeds a preset threshold value .
the threshold values used here for the neutral species are listed in table [ simpar ] .
as a neutral species hits the electrode it returns as a thermal particle with a given probability and atoms can recombine to form a thermal molecule with the same probability .
these wall recombination probabilities as well as the quenching coefficients used for the excited neutral species are listed in table [ simpar ] .
the choice of the quenching coefficients for the molecular metastables and its influence on the discharge properties is discussed towards the end of next section . the secondary electron emission yield for o@xmath8- , o@xmath9-ions , o@xmath0 molecules and o atoms are energy dependent and we used the fits for oxidized metal given elsewhere @xcite .
.the parameters of the simulation , the particle weight , the threshold above which the dynamics of the neutral particles are followed and the wall recombination and quenching coefficients used .
[ cols= " < , < , < , < " , ]
( a@xmath12 ) nor o@xmath0(b@xmath2 ) , ( b ) only detachment by o@xmath0(b@xmath2 ) included , ( c ) only detachment by o@xmath0(a@xmath12 ) included , ( d ) and both detachment by o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) included ( full reaction set ) . [ jdote10 ] ]
the electron power absorption is calculated as @xmath36 , where @xmath37 and @xmath38 are the spatially and temporally varying electron current density and electric field , respectively .
the spatio - temporal behavior of the electron power absorption , for pressures 10 , 50 and 200 mtorr , is shown in figures [ jdote10 ] [ jdote200 ] , respectively , for four cases , ( a ) when the detachment by both the metastable states o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) are neglected , ( b ) when detachment by o@xmath0(a@xmath12 ) is neglected and by o@xmath0(b@xmath2 ) is included , ( c ) when detachment by o@xmath0(a@xmath12 ) is included and by o@xmath0(b@xmath2 ) is neglected , and ( d ) when the detachments by both the metastable states o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) are included in the simulations ( the full reaction set ) .
for each of the figures the abscissa covers the whole inter - electrode gap , from the powered electrode on the left hand side to the grounded electrode on the right hand side .
similarly the ordinate covers the full rf cycle .
the four figures grouped together for each pressure value can have different magnitude scales .
therefore , there can be differences in the four figures , not only qualitatively but also quantitatively . when comparing the figures for each pressure be aware of such subtle differences .
figure [ jdote10 ] ( a ) shows the spatio - temporal behavior of the electron power absorption at 10 mtorr when the detachments by both the metastable states o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) are neglected in the simulation .
the most significant heating is in the sheath region , however there is a significant contribution to the heating in the bulk region . both energy gain ( red and yellow areas ) and energy loss ( dark blue areas )
are evident . as detachment by o@xmath0(b@xmath2 ) only is added to the simulation in figure [ jdote10 ] ( b ) the spatio - temporal heating structure is maintained but the sheath width increases .
note that there is also a change in the scale when detachment by o@xmath0(b@xmath2 ) is added , the peak electron heating in the sheath region decreases .
adding detachment by o@xmath0(a@xmath12 ) only does not lead to decreased sheath with as seen in figure [ jdote10 ] ( c ) .
including both the detachment processes ( the full reaction set ) shows very similar spatio - temporal heating structure as when only o@xmath0(b@xmath2 ) is included in the simulation . in all cases
the electron heating is most significant during the the sheath expansion phase at each electrode ( the red areas ) .
we also observe electron heating during the sheath collapse on the bulk side of the sheath edge while there is cooling ( electrons loose energy ) on the electrode side ( the lower left hand corner and upper center on the right hand side ) .
this is in accordance to what is commonly observed in single frequency capacitively coupled discharges @xcite .
setting the electron emission coefficient to zero gives almost exactly the same spatio - temporal heating structure for all the four cases ( not shown ) .
the spatio - temporal heating structure at 50 mtorr is shown in figure [ jdote50 ] .
when detachment by both the metastable states is neglected there is significant heating in the electronegative core as seen in figure [ jdote50 ] ( a ) ( green and yellow areas ) .
when detachment by the singlet state o@xmath0(a@xmath12 ) only is included , the heating in the electronegative core decreases as seen in figure [ jdote50 ] ( c ) but when detachment by the singlet state o@xmath0(b@xmath2 ) only is included the heating in the electronegative core decreases more or to almost zero as seen in figure [ jdote50 ] ( b ) . including both the detachment processes ( the full reaction set ) shows almost zero heating in the electronegative core as seen in figure [ jdote50 ] ( d ) . in all cases
the electron heating is most significant during the the sheath expansion phase at each electrode ( the red areas ) .
we note oscillations in the heating like those reported by vender and boswell @xcite as we add detachment by the singlet metastable molecules to the simulations .
we also note that cooling in the sheath region during the sheath collapse is always apparent ( note the different scale ) .
( a@xmath12 ) nor o@xmath0(b@xmath2 ) , ( b ) only detachment by o@xmath0(b@xmath2 ) included , ( c ) only detachment by o@xmath0(a@xmath12 ) included , ( d ) and both detachment by o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) included ( full reaction set ) . [ jdote50 ] ] at 200 mtorr when detachment by both the metastable states is neglected there is still a significant heating in the electronegative core as seen in figure [ jdote200 ] ( a ) . when detachment by either o@xmath0(a@xmath12 ) or o@xmath0(b@xmath2 ) is included in the simulation the heating in the electronegative core drops to zero , as seen in figures [ jdote200 ] ( c ) and [ jdote200 ] ( b ) , respectively . as a consequence including both the detachment processes shows almost zero heating in the electronegative core as seen in figure [ jdote200 ] ( d ) . in all cases
the electron heating is most significant during the the sheath expansion phase at each electrode ( the red areas ) .
greb et al .
@xcite find that at 300 mtorr ohmic ( collisional ) heating is the dominant heating mechanism far exceeding collisionless heating .
they see that collisional electron heating has the highest contribution close to the sheath edge , during the sheath expansion and sheath collapse phases .
they also find that collisionless electron heating has higher contribution at the sheath edge during sheath expansion than collisional heating . for collisionless heating
they find high negative electron heating ( or cooling ) at the sheath edge during the sheath collapse phase .
( a@xmath12 ) nor o@xmath0(b@xmath2 ) , ( b ) only detachment by o@xmath0(b@xmath2 ) included , ( c ) only detachment by o@xmath0(a@xmath12 ) included , ( d ) and both detachment by o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) included ( full reaction set ) .
[ jdote200 ] ] figure [ jdote_all ] ( a ) shows the time averaged electron heating profile at 10 , 50 and 200 mtorr for the four cases explored here . at 10 mtorr
the electron heating is mainly ohmic heating in the electronegative core .
this bulk heating decreases only slightly when detachment is by o@xmath0(a@xmath12 ) only is added to the reaction set , but decreases more significantly when detachment by o@xmath0(b@xmath2 ) only is included in the reaction set . having the full reaction set in the simulation gives almost the same electron heating rate profile as when when detachment by o@xmath0(b@xmath2 ) only is included .
note that when we neglect the detachment by the metastables there is effective cooling in the sheath region during one period as seen in figure [ jdote_all ] ( a ) .
this is due to very strong cooling in the sheath collapse phase as seen in figure [ jdote10 ] ( a ) .
the strong heating in the sheath region during the sheath expansion of the rf cycle is not significant enough to overcome this cooling during one period .
when both detachment processes are included these cooling and heating phases balance and there is almost no effective heating in the sheath regions . at 50 mtorr
the ohmic heating in the electronegative core is significant when detachment by the singlet metastables is neglected .
however , ohmic heating in the electronegative core drops to almost zero when either detachment by o@xmath0(a@xmath12 ) or o@xmath0(b@xmath2 ) is included in the simulation .
detachment by o@xmath0(b@xmath2 ) only has slightly stronger influence on the ohmic heating in the electronegative core . at 200 mtorr
there is a significant ohmic heating in the electronegative core when detachment by both the singlet metastables is neglected and it drops to almost zero when either detachment by o@xmath0(a@xmath12 ) or o@xmath0(b@xmath2 ) is included in the simulation .
the peak electron heating in the sheath region is higher at 200 mtorr than at 50 mtorr .
( a@xmath12 ) nor o@xmath0(b@xmath2 ) ( dashed line ) , only detachment by o@xmath0(b@xmath2 ) included ( dotted line ) , only detachment by o@xmath0(a@xmath12 ) included ( dash - dot line ) , and both detachment by o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) included ( full reaction set ) ( solid line ) .
[ jdote_all ] ] ( a@xmath12 ) nor o@xmath0(b@xmath2 ) ( dashed line ) , only detachment by o@xmath0(b@xmath2 ) included ( dotted line ) , only detachment by o@xmath0(a@xmath12 ) included ( dash - dot line ) , and both detachment by o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) included ( full reaction set ) ( solid line ) .
[ jdote_all ] ] the same is seen when exploring the effective electron temperature profile shown in figure [ jdote_all ] ( b ) .
the effective electron temperature profile changes significantly when detachment by singlet metastables is added to the reaction set . at 10 mtorr detachment by o@xmath0(a@xmath12 ) only has small effect on the effective electron temperature while detachment by o@xmath0(b@xmath2 ) only decreases the effective electron temperature significantly . at 50 mtorr detachment by both o@xmath0(a@xmath12 ) and
o@xmath0(b@xmath2 ) lowers the effective electron temperature but detachment by o@xmath0(b@xmath2 ) has more influence . at 200
mtorr including either one of the detachment processes is enough to get the same value of the effective electron temperature as when the full reaction set is used in the simulations .
this low effective electron temperature in the discharge center indicates that the heating of the electrons is effective in the sheath region , but the ohmic ( or collisional ) heating in the discharge center is ineffective .
the effective electron temperature when the full reaction set is assumed in the simulations is in accordance with the measurements of kechkar @xcite which find @xmath39 ev in the pressure range 50 - 200 mtorr in a slightly asymmetric discharge in pure oxygen .
furthermore , pulpytel et al .
@xcite measured the effective electron temperature in an ar / o@xmath0 ( 1:3 ) mixture at 110 mtorr and voltages above 300 v to be around 1 ev .
( a@xmath12 ) nor o@xmath0(b@xmath2 ) ( ) , only detachment by o@xmath0(a@xmath12 ) included ( @xmath40 ) , only detachment by o@xmath0(b@xmath2 ) included ( ) , and both detachment by o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) included ( full reaction set ) ( @xmath41 ) .
[ alpha0 ] ] ( a@xmath12 ) nor o@xmath0(b@xmath2 ) ( ) , only detachment by o@xmath0(a@xmath12 ) included ( @xmath40 ) , only detachment by o@xmath0(b@xmath2 ) included ( ) , and both detachment by o@xmath0(a@xmath12 ) and o@xmath0(b@xmath2 ) included ( full reaction set ) ( @xmath41 ) .
[ alpha0 ] ] , figure [ alpha0 ] ( a ) shows the center electronegativity @xmath42 where @xmath43 is the center negative ion density and @xmath44 is the center electron density .
the negative ions are confined to the plasma bulk due to a potential well , and the negative ion density has a peak near the discharge center .
we see that the center electronegativity decreases with increasing discharge pressure .
stoffels et al .
@xcite measure the electronegativity to be in the range 5 10 and find it to decrease with increased pressure .
this is somewhat higher electronegativity than what we show here .
the pressure dependence contradicts the findings reported by bera et al .
@xcite which apply a pic / mcc hybrid model to a low pressure oxygen discharge , and observe an increase in electronegativity with increased pressure .
as seen in figure [ alpha0 ] ( a ) the electronegativity decreases as we add detachment processes to the discharge model . adding detachment by o@xmath0(b@xmath2 ) has more significant influence on the electronegativity than adding detachment by o@xmath0(a@xmath12 ) .
the average electronegativity is given as @xmath45 where @xmath46 is the o@xmath15-ion density , @xmath47 is the electron density and @xmath48 is the electrode separation .
we see that the average electronegativity shown in figure [ alpha0 ] ( b ) is always somewhat lower than the center electronegativity .
earlier we explored the influence of increasing the singlet molecular metastable o@xmath0(a@xmath12 ) density on the electron heating rate .
we find that increasing the partial pressure of the metastable molecule o@xmath0(a@xmath12 ) in the background does not have much influence on the electron heating except that the ohmic heating in the bulk is closer to zero and the electronegativity is slightly lower when 8.8 % of the discharge pressure is due to the metastable molecule o@xmath0(a@xmath12 ) @xcite . similarly increasing the partial pressure of o@xmath0(b@xmath2 )
decreases the electronegativity even further @xcite .
but for a change in partial pressure of o@xmath0(b@xmath2 ) from 2.2 % to 4.4 % there is not a significant change in electronegativity .
it has been demonstrated by using a 1d fluid model that the electronegativity depends strongly on the o@xmath0(a@xmath12 ) surface quenching probability @xcite .
for this current study we use a quenching probability of 0.007 estimated by sharpless and slanger @xcite for iron . their estimate for aluminum
is @xmath49 .
greb et al .
@xcite argue that increased quenching coefficient leads to decreased o@xmath0(a@xmath12 ) density and thus decreased detachment by the o@xmath0(a@xmath12 ) state and thus higher negative ion density . for iron ryskin and shub
@xcite report a value of 0.0044 and , for aluminum , @xmath50 .
aluminum electrodes would therefore lead to higher singlet metastable densities and lower electronegativity .
we have seen in global model studies that wall quenching can be the main loss mechanism for the singlet metastable state o@xmath0(b@xmath2 ) @xcite . in these studies we assumed the quenching coefficient to be 0.1 , which is the same value as that assumed in this current study .
this assumption is based on the suggestion that the quenching coefficient for the b@xmath2 state is about two orders of magnitude larger than that for the a@xmath12 state @xcite .
we used the one - dimensional object - oriented particle - in - cell monte carlo collision code oopd1 to explore the spatio - temporal evolution of the electron heating mechanism with pressure in a capacitively coupled oxygen discharge including and excluding the detachment by the singlet molecular metastable states .
we find that detachment by the singlet metastable state o@xmath0(b@xmath2 ) has significant effect while o@xmath0(a@xmath12 ) has somewhat less significant influence on the discharge properties .
we conclude that it is essential to include detachment by both the metastable states . including the detachment processes has a strong influence on the effective electron temperature and electronegativity in the oxygen discharge , both of these parameters are significantly lower when the detachment processes are included in the simulation . however , it has to be kept in mind that the wall quenching probability for each of the molecular metastable state has a significant influence on the overall discharge and the actual values of the electronegativity and the effective electron temperature . at 10 mtorr ,
using the full reaction set , the time averaged electron heating displays mainly contribution from ohmic heating in the plasma bulk ( the electronegative core ) and at higher pressures there is no ohmic heating in the plasma bulk and electron heating in the sheath regions dominates .
this work was partially supported by the icelandic research fund grant nos .
130029 and 163086 , and the swedish government agency for innovation systems ( vinnova ) contract no .
2014 - 04876 .
25ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0]$12&12 # 1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , p. ( ) @noop * * , p. ( ) @noop * * , p. ( ) @noop * * , p. ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , p. ( ) @noop * * , p. ( ) @noop * * , ( ) @noop * * , p. ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) `` , '' @noop ph.d .
thesis , @noop * * , ( ) @noop * * , p. ( ) `` , '' @noop ph.d .
thesis , @noop * * , p. ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we use the one - dimensional object - oriented particle - in - cell monte carlo collision code oopd1 to explore the spatio - temporal evolution of the electron heating mechanism in a capacitively coupled oxygen discharge in the pressure range 10 200 mtorr .
the electron heating is most significant in the sheath vicinity during the sheath expansion phase .
we explore how including and excluding detachment by the singlet metastable states o@xmath0(a@xmath1 ) and o@xmath0(b@xmath2 ) influences the heating mechanism , the effective electron temperature and electronegativity , in the oxygen discharge .
we demonstrate that the detachment processes have a significant influence on the discharge properties , in particular for the higher pressures . at 10 mtorr
the time averaged electron heating shows mainly ohmic heating in the plasma bulk ( the electronegative core ) and at higher pressures there is no ohmic heating in the plasma bulk , that is electron heating in the sheath regions dominates . |
pre - main sequence ( pms ) late - type stars are known to be x - ray sources ( see walter et al .
1988 , bouvier 1990 and neuhuser et al .
1995 ) . in the evolution of these stars to the main sequence ( ms ) , there is a state usually defined as post - t tauri stars ( ptts ) .
ptts were first defined by herbig ( 1978 ) as pms stars more evolved than classical t tauri stars ( ctts ) but still contracting to the ms .
given that the stage of ctts is only a small fraction of the total time of contraction of low - mass stars to the ms , ptts should be much more abundant than ctts if star formation has been ongoing for a sufficiently long time . however , it is difficult to find ptts because they do not show spectroscopic or photometric peculiarities which make them easy to detect .
unlike ctts , they do not present significant ir or uv excesses and the h@xmath1 line is not observed as a strong emission line . therefore , the identification of these stars is difficult and relies on the detection of the li i ( 6708 ) absorption line and the chromospheric ca ii ( h & k ) lines in their spectra , as well as on their x - ray detection .
murphy ( 1969 ) first proposed that ptts could be searched as members of young binary systems .
the basis of his idea was that the ms lifetime of high - mass stars is comparable to the contraction timescale of solar - type stars .
hence , there could be binary systems comprised of ms early - type stars physically bound to ptts .
gahm et al .
( 1983 ) carried out photometric and spectroscopic observations of visual double stars with early type primaries .
the derived data , together with @xmath2 observations , allowed lindroos ( 1986 ; l86 hereafter ) to identify 78 likely physical pairs with several ptts candidates as secondaries .
a high lithium abundance and a high chromospheric activity level are necessary ( although not sufficient ) indicators of youth .
martn et al .
( 1992 ) and pallavicini et al .
( 1992 ) carried out optical spectroscopy of the lindroos late - type companions , detecting the li i(6708 ) absorption line and the ca ii ( h & k ) emission lines in the spectra of several ptts candidates .
ray et al .
( 1995 ) took this sample of `` genuine '' ptts and looked for circumstellar matter around them .
for this purpose , they analyzed the iras database ( point source catalogue and faint source catalogue ) and also searched for continuum 1.1 mm emission .
while ir excesses were found for most of the sources , no mm dust continuum was detected ( see also gahm et al . 1994 and jewitt 1994 ) .
llllrrllll hd & sp.type@xmath3 & & sep.@xmath3 & distance@xmath4 & v@xmath5 , v@xmath6 @xmath7 & a@xmath8@xmath9 & clas.@xmath10 + & & ra ( 2000 ) & & dec ( 2000 ) & ( ) & ( pc ) & ( mag ) & ( mag ) + & & ( h ) ( m ) ( s ) & & ( ) ( ) ( ) & & & & + 560 a+b & b9v+g5ve & 00 10 02.20 & & 11 08 44.93 & 7.7 & 100@xmath119 & 5.53,10.37 & 0.00 & cp + 1438 a+b & b8v+f3v & 00 18 42.17 & & 43 47 28.11 & 6.2 & 212@xmath1135 & 6.11,9.7@xmath12 & 0.04 & + 8803 a+b & b9v+f6vp & 01 26 53.55 & & 03 32 08.32 & 6.0 & 160@xmath1125 & 6.43,9.67 & 0.12 & lo + 17543 a+c & b6iv+f8v & 02 49 17.56 & & 17 27 51.51 & 25.2 & 185@xmath1137 & 5.28,10.73 & 0.21 & pp + 23793 a+b & b3v+f3vp & 03 48 16.27 & & 11 08 35.86 & 9.0 & 173@xmath1131 & 5.10,9.41 & 0.08 & lo + 27638 a+b & b9v+g2v & 04 22 34.94 & & 25 37 45.53 & 19.4 & 82@xmath118 & 5.41,8.43 & 0.00 & pp + 33802 a+b & b8v+g8ve & 05 12 17.90 & & -11 52 09.19 & 12.7 & 74@xmath114 & 4.47,9.92 & 0.01 & cp + 35007 a+c & b3v+g3v & 05 21 31.84 & & -00 24 59.36 & 37.6 & 330@xmath1189 & 5.68,11.88 & 0.16 & + 36013 a+b & b2v+f9v & 05 28 45.28 & & 01 38 38.17 & 25.0 & 303@xmath1186 & 6.89,12.49 & 0.04 & + 36151 a+b & b5v+g1v & 05 29 25.4 & & -07 15 39.18 & 48.6 & 370@xmath11122 & 6.69,10.60 & 0.08 & + 36151 a+x & b5v+g7v & 05 29 25.4 & & -07 15 39.18 & 45.0 & 370@xmath11122 & 6.69,11.98 & 0.08 & + 36779 a+b & b2.5v+k5iv & 05 34 03.89 & & -01 02 08.61 & 27.5 & 380@xmath11378 : & 6.24,11.20 & 0.10 & lo + 38622 a+c & b2v+g2v & 05 47 42.91 & & 13 53 58.56 & 24.9 & 245@xmath1154 & 5.27,12.01 & 0.04 & cp + 40494 a+b & b3iv+g8v & 05 57 32.21 & & -35 16 59.80 & 33.8 & 263@xmath1138 & 4.36,12.66 & 0.00 & pp + 43286 a+b & b5iv+g3v & 06 15 30.24 & & 03 57 29.50 & 18.3 & 308@xmath1185 & 6.99,12.38 & 0.06 & + 48425 a+c & b3v+g5v & 06 42 27.57 & & -23 13 57.29 & 34.9 & 368@xmath11122 & 6.9,10.4@xmath13 & 0.00 & + 53191 a+b & a0v+g3v & 07 00 16.73 & & -60 51 45.83 & 17.0 & 207@xmath1125 & 7.74,11.75 & 0.05 & pp + 53755 a+b & b0.5iv+f5iii & 07 05 49.64 & & -10 39 36.28 & 6.2 & 1087 : & 6.49,10.3@xmath13 & 0.65 & + 56504 a+b & b9iv+g6iii & 07 16 00.27 & & -29 29 33.81 & 31.1 & 932@xmath14 & 9.8,9.7@xmath13 & 0.07 & + 60102 a+b & b9.5v+g8v & 07
11 38.41 & & -84 28 09.77 & 16.4 & 206@xmath1125 & 7.54,11.86 & 0.22 & pp + 63465 a+b & b2.5iii+f2v & 07 47 24.99 & & -38 30 40.12 & 10.9 & 380@xmath1180 & 5.08,11.13 & 0.25 & lo + 70309 a+b & b3iv+k2iv & 08 19 05.58 & & -48 11 52.27 & 42.5 & 252@xmath1136 & 6.45,11.28 & 0.12 & lo + 71510 a+c & b3v+g3v & 08 25 31.32 & & -51 43 38.69 & 34.9 & 207@xmath1121 & 5.19,10.77 & 0.01 & lo + 74146 a+b & b5v+f0iv & 08 39 57.59 & & -53 03 17.03 & 16.6 & 131@xmath118 & 5.19,8.66@xmath13 & 0.03 & + 76566 a+b & b3v+g4v & 08 55 19.20 & & -45 02 30.01 & 35.0 & 286@xmath1151 & 6.28,12.64 & 0.00 & lo + 77484 a+b & b9.5v+g5v & 09 02 50.65 & & 00 24 29.54 & 4.4 & 250@xmath1164 & 8.02,12.0 & 0.09 & pp + 86388 a+b & b9v+f5v & 09 55 05.60 & & -69 11 20.31 & 9.2 & 177@xmath1121 & 6.87,9.98 & 0.02 & lo + 87901 a+b & b8v+k0ve & 10 08 22.31 & & 11 58 01.94 & 176.9 & [email protected] & 1.35,8.08 & 0.00 & lo + 90972 a+b & b9.5v+f9ve & 10 29 35.38 & & -30 36 25.43 & 11.0 & 147@xmath1116 & 5.58,9.65 & 0.03 & cp + 104901 a+b & b9ii+f0iie & 12 04 46.98 & & -61 59 48.60 & 23.0 & 980 : & 7.43,8.01@xmath13 & 1.11 & + 106983 a+b & b2.5v+ g8iii & 12 18 26.24 & & -64 00 11.05 & 33.8 & 110@xmath116 & 4.05,12.49 & & lo + 108767 a+b & b9.5v+k2ve & 12 29 51.85 & & -16 30 55.56 & 24.2 & 27@xmath119 & 2.94,8.43 & 0.00 & cp + 109573 a+b & a0+m2.5 & 12 36 01.3 & & -39 52 09.00 & 7.6 & 67@xmath113 & 5.78,13.3 & 0.00 & + 112244 a+b & o9iab+k0iii & 12 55 57.13 & & -56 50 08.90 & 29.1 & 578@xmath11213 & 5.38,11.77 & 0.90 & lo + 112413 a+b & a0iiip+f0v & 12 56 01.67 & & 38 19 06.17 & 19.6 & 34@xmath111 & 2.90,5.60@xmath13 & 0.00 & + 113703 a+b & b4v+k0ve & 13 06 16.70 & & -48 27 47.84 & 11.4 & 127@xmath1112 & 4.72,10.8 & 0.00 & cp + 113791 a+b & b2iii+f7v & 13 06 54.64 & & -49 54 22.49 & 25.1 & 126@xmath1113 & 4.25,9.38 & & cp + 120641 a+b & b8v+f0vp & 13 52 02.98 & & -52 48 05.81 & 18.0 & 333 : & 7.47@xmath13,-@xmath15 & 0.00 & + 123445 a+b & b9v+k2v & 14 08 51.89 & & -43 28 14.80 & 28.6 & 218@xmath1137 & 6.19,12.52 & 0.12 & lo + 127304 a+b & a0v+k1v & 14 29 49.67 & & 31 47 28.28 & 25.8 & 106@xmath118 & 6.07,11.37 & 0.04 & pp + 127971 a+b & b7v+k0v & 14 35 31.48 & & -41 31 02.77 & 26.9 & 109@xmath119 & 5.89,11.22 & 0.08 & lo + 129791 a+b & b9.5v+k5ve & 14 45 56.2 & & -44 52 03.08 & 35.3 & 129@xmath1116 & 6.94,12.93 & 0.26 & cp + 137387 a+b & b3ive+k5iv & 15 31 30.82 & & -73 23 22.53 & 27.0 & 312@xmath1159 & 5.47,11.27 & 0.47 & lo + 138800 a+x & b8iv+kov & 15 40 21.33 & & -73 26 48.07 & 34.0 & 225@xmath1141 & 5.65,12.86 & 0.22 & lo + 143939 a+b & b9iii+k3ve & 16 04 44.49 & & -39 26 04.76 & 8.6 & 167@xmath1127 & 6.98,11.80 & 0.00 &
cp + 145483 a+b & b9v+f3v & 16 12 16.04 & & -28 25 02.29 & 4.6 & 91@xmath118 & 5.67@xmath13,-@xmath15 & 0.25 & + 162082 a+b & b7v+f2v & 17 50 18.10 & & -26 19 33.87 & 10.7 & 314@xmath14 & 8.16,11.05 & 0.63 & lo + 174585 a+b & b2.5v+k2iv & 18 49 45.91 & & 32 48 46.15 & 34.8 & 310@xmath1152 & 5.90,10.89@xmath13 & 0.17 & + 174585 a+c & b2.5v+g0v & 18 49 45.91 & & 32 48 46.15 & 58.7 & 310@xmath1152 & 5.90,10.3@xmath13 & 0.17 & + 180183 a+b & b3v+k05 & 19 18 41.54 & & -56 08 40.90 & 19.4 & 244@xmath1149 & 6.82,11.45 & 0.10 & lo + + * notes : * 1 . taken from pallavicini et al .
( 1992 ) ; 2 .
coordinates of the primary star ; 3 . deduced from hipparcos parallax of the primary star , except those marked with @xmath14 taken from l86 ; 4 . adopted from pallavicini et al .
( 1992 ) except those marked with an asterisk obtained from the simbad database ; 5 . taken from l86
classification taken by pallavicini et al .
( 1992 ) with respect to the presence of the lithium absorption line and caii ( h and k ) chromospheric lines in the spectrum of the secondary : cp ( certainly physical ) , pp ( probably physical ) and lo ( likely optical ) ; 7 .
unknown visual magnitude for the secondary star .
the x - ray emission from lindroos binary systems was first studied by schmitt et al .
( 1993 ) . after the analysis of seven pairs comprised of late b - type stars and later - type companions , the main result was the detection of x - rays from both members of the pair . in the case
of late - type stars it is well - known that they produce x - rays in their hot coronae .
however , this is not the case of late - b type stars .
theoretically , early - type stars between b4 and a7 are not expected to be x - ray emitters : they do not possess the strong winds thought to be responsible of the x - ray emission in o- and early b - type stars ( lucy & white 1980 ) , nor significant convection zones thought to be necessary to sustain a magnetic dynamo to power a corona .
although the detection of x - rays from late - b and early - a type stars have been reported by several authors ( i.e. caillault & zoonematkermani 1989 , schmitt et al .
1993 , berghfer & schmitt 1994 , berghfer et al .
1996 , simon et al .
1995 , panzera et al .
1999 ) , there is no clear mechanism that explains the origin of this emission .
the most accepted explanation is related to the presence of otherwise unknown unresolved late - type companions of these stars .
the aim of this paper is to study the x - ray emission of all l86 binary systems with ptts candidates which were observed by rosat . for this purpose
, we have selected binary systems with late - type stars as secondaries .
we will study the x - ray emission from the ptts candidates as well as the emission from the early - type stars .
the characteristics of the sample are described in sect .
3 provides the details related to the source detection and identification .
the processed x - ray data are analyzed in sections 4 , 5 and 6 .
the conclusions are drawn in sect .
the lindroos catalogue ( l86 ) contains 78 binary systems .
we have selected those binaries in which the secondary member is a f , g , or k - type star ( note that no m - type stars are present in the lindroos catalogue ) .
our final sample consists of 47 systems .
two of them ( hd113791 and hd106983 ) were not included in l86 but in gahm et al .
we have also included the binary system hr4796 ( hd109573 , twa 11 ) , a possible member of the tw hya association ( see webb et al .
1999 ) given that its stellar properties are in agreement with those of the lindroos sample .
this is the only binary system in our sample with an m - type star as a secondary .
note that we have not rejected those pairs that are classified as likely optical pairs by pallavicini et al .
( 1992 ) , having in view that the x - ray emission could provide important information related to the nature of these systems . stellar properties of our binary sample are shown in table 1 .
the name of the source as well as the components of the binary system ( a for the primary , b for the secondary and c or x for companions in multiple systems ) are shown in column 1 .
the spectral types of both stars are given in column 2 .
column 3 shows the optical position of the primary star while column 4 shows the projected separation between both members .
the distance to the primary star , the visual magnitude of both components and the visual extinction to the pair are given in columns 5 , 6 and 7 .
we have also included in column 8 an orientative ` flag ' which is related to the nature of the binary system : according to the spectroscopic survey carried out by pallavicini et al .
( 1992 ) , the systems are classified as lo ( likely optical pair ) , pp ( probably physically bound ) and cp ( certainly physically bound ) .
the x - ray telescope and the instrumentation onboard the rosat satellite are described in detail by trmper ( 1983 ) , pfeffermann et al .
( 1988 ) and david et el .
two main detectors are available : the position sensitive proportional counter ( pspc ) , which can be used either in survey or in pointed mode , and the high resolution imager ( hri ) . we have studied all the available data for our sample , i.e. , pspc ( survey and pointed mode ) and hri data . the source detection and identification
have been carried out using the source detection routines provided by the extended scientific analysis system ( exsas ; zimmermann et al .
1997 ) which are based on a maximum likelihood ( ml ) technique ( cruddace et al . 1988 ) .
for all detected x - ray sources we have looked for optical counterparts to check the reliability of the detection .
we will briefly describe the data in the following subsections .
llrrrrrrr hd & & @xmath16@xmath3 & counts@xmath17 & exp .
time & hr1 & hr2 & ml + & ra ( 2000 ) & dec ( 2000 ) & ( ) & & ( secs . ) & & & + & ( h ) ( m ) ( s ) & ( ) ( ) ( ) & & & & & & + 560 & 00 10 02.73 & 11 08 49.9 & 9.3/15.9 & [email protected] & 565.4 & [email protected] & [email protected] & 239.7 + 1438 & & & & @xmath183.24/3.32 & 458.3 & & & + 17543 & & & & @xmath183.41/2.93 & 332.9 & & & + 23793 & 03 48 16.19 & 11 08 48.6 & 12.8/9.8 & [email protected] & 371.8 & [email protected] & [email protected] & 34.5 + 33802 & 05 12 17.81 & -11 52 01.8 & 7.5/11.0 & [email protected] & 408.0 & [email protected] & [email protected] & 1124.0 + 35007 & & & & @xmath183.26/3.36 & 431.8 & & & + 36151 & & & & @xmath184.64/3.22 & 465.3 & & & + 38622 & 05 47 44.54 & 13 53 15.5 & 49.5/32.1 & 6.0 @xmath113.6 & 458.5 & @xmath190.21 & [email protected] & 9.4 + 40494 & & & & @xmath186.39/5.00 & 674.9 & & & + 43286 & 06 15 26.95 & 03 56 48.8 & 63.8/77.2 & [email protected] & 488.9 & [email protected] & [email protected] & 18.4 + 48425 & & & & @xmath183.49/3.41 & 522.9 & & & + 53191 & & & & @xmath185.74/5.14 & 1236.3 & & & + 56504 & & & & @xmath183.80/3.69 & 406.4 & & & + 60102 & 07 11 36.51 & -84 27 00.6 & 69.2/73.4 & [email protected] & 345.7 & [email protected] & [email protected] & 11.4 + 63465 & & & & @xmath182.01/2.07 & 75.7 & & & + 74146 & & & & @xmath185.31/5.14 & 720.1 & & & + 77484 & 09 02 48.18 & 00 24 17.1 & 39.0/42.1 & [email protected] & 372.4 & [email protected] & @xmath18 - 0.28 & 6.2 + 86388 & & & & @xmath18 3.99/4.00 & 391.3 & & & + 87901 & 10 08 12.72 & 12 00 01.8 & 184.8/13.5 & [email protected] & 393.6 & [email protected] & @xmath18 - 0.67 & 78.0 + 90972 & 10 29 39.77 & -30 35 57.9 & 63.1/73.8 & [email protected] & 132.8 & [email protected] & [email protected] & 6.7 + 104901 & & & & @xmath183.84/4.14 & 361.2 & & + 106983 & & & & @xmath182.16/2.16 & 82.7 & & & + 108767 & 12 29 52.18 & -16 30 48.0 & 9.0/32.6 & [email protected] & 281.3 & [email protected] & [email protected] & 21.7 + 109573 & 12 35 59.5 & -39 52 05.01 & 18.9/17.2 & [email protected] & 285.0 & [email protected] & [email protected] & + 112244 & 12 55 54.00 & -56 50 04.9 & 25.9/19.5 & [email protected] & 108.5 & @xmath190.46 & [email protected] & 7.0 + 112413 & 12 56 01.11 & 38 18 56.2 & 11.7/8.3 & [email protected] & 519.7 & [email protected] & [email protected] & 56.7 + 113703 & 13 06 18.27 & -48 27 44.3 & 16.0/3.7 & [email protected] & 302.7 & [email protected] & [email protected] & 150.7 + 113791 & 13 06 59.79 & -49 54 03.3 & 53.3/33.1 & [email protected] & 343.3 & @xmath190.45 & [email protected] & 15.7 + 127304 & & & & @xmath184.23/2.70 & 592.5 & & + 127971 & 14 35 31.63 & -41 31 08.0 & 5.6/21.3 & [email protected] & 315.0 & @xmath190.26 & [email protected] & 6.8 + 129791 & 14 45 56.17 & -44 52 12.2 & 18.0/19.8 & [email protected] & 327.7 & [email protected] & [email protected] & 62.4 + 137387 & & & & @xmath182.99/3.13 & 248.8 & & & + 138800 & & & & @xmath182.68/2.50 & 227.8 & & & + 143939 & 16 04 44.67 & -39 26
03.0 & 2.8/11.3 & [email protected] & 324.4 & [email protected] & [email protected] & 39.3 + 145483 & 16 12 16.52 & -28 25 30.0 & 7.0/19.39 & [email protected] & 319.7 & [email protected] & [email protected] & 104.9 + 174585 & & & & @xmath182.96/3.10 & 790.1 & & & + 180183 & & & & @xmath18 2.47/1.98 & 139.8 & & &
+ * notes : * + 1 . given the low spatial resolution of the pspc , we can not resolve the two components of the binary system . as a consequence ,
the displacement @xmath16 between the detected x - ray source and the optical positions of the two components of the system are given .
2 . in the case of non - detection ,
upper limits are computed at the positions of both components , a and b , respectively .
l@l@l@l @c@r@r@r @r@r@r hd & ror & & @xmath16 & offaxis & counts & exp .
time & hr1 & hr2 & ml + & number & ra ( 2000 ) & dec ( 2000 ) & ( ) & ( ) & & ( secs . ) & & & + & & ( h ) ( m ) ( s ) & ( ) ( ) ( ) & & & & & & & + 560 & 700503p & 00 10 02.48 & 11 08 41.2 & 5.6/6.8 & 12.32 & [email protected] & 8238.3 & [email protected] & [email protected] & 7713.6 + & 701092p & 00 10 02.47 & 11 08 40.4 & 6.1/6.0 & 12.31 & [email protected] & 8121.8 & [email protected] & [email protected] & 18859.8 + 74146@xmath3 & 200501p & & & & 18.60 & @xmath1869.6/59.7 & 17319.4 & & & + 104901@xmath17 & 201271 & & & & 4.94 & @xmath1816.2/3.94 & 8491.6 & & & + 143939 & 200738p & 16
04 44.66 & -39 26 16.4 & 12.0/13.1 & 7.61 & [email protected] & 1686.6 & [email protected] & [email protected] & 334.2 + * notes : * 1 .
the source is very close to the support structure of the telescope , so we are not sure about the reliability of the upper limit ; 2 .
this source has two pspc observations : 201271p and 201271p-1 .
there are no detections in the individual observations , so we have added up both files to improve the s / n ratio .
the rosat all - sky survey ( rass ) was performed with the pspc .
the diameter of the field of view is 2 and each object is observed up to @xmath20 times separated by @xmath21 minutes , with up to @xmath22 30 sec per scan .
the spectral resolution of the pspc ( 43% at 0.93 kev ) allows spectral analysis in three energy bands : + - _ soft _ = 0.1 to 0.4 kev + - _ hard 1 _ = 0.5 to 0.9 kev + - _ hard 2 _ = 0.9 to 2.0 kev + while in some pointed observations the signal to noise are large enough to carry out detailed spectral analysis , this is not possible for rass data .
note that the rass exposure times of our sample range from 75 sec to 1236 sec .
however , we can obtain spectral information of our sources studying the x - ray hardness ratios ( hr ) defined as follows : + @xmath23 where h1 and h2 are the counts observed in the hard 1 and hard 2 bands , and s are the counts observed in the soft band .
hence , hr values can range from @xmath24 to @xmath25 . if no counts are detected in one of the bands only an upper or lower limit to hr1 and hr2
is available .
neither hr values nor limits are available for undetected stars .
the nominal positional accuracy of the rosat pspc detector in pointing mode is @xmath22 25 at 1 kev ( note that it is energy and off - axis angle dependent ) .
however , this accuracy is reduced to @xmath22 1 in the survey phase .
if the offset between the x - ray source and the optical position exceeds the spatial resolution of the rass data ( @xmath22 1 ) , the identification of the x - ray source with the optical counterpart is doubtful .
hence , for all the rass observations we have to select a maximal displacement between the x - ray and the optical position within which the detections are reliable .
following neuhuser et al .
( 1995 ) , we have taken a value of 40 as the maximal distance between the x - ray detection and the optical counterpart .
the rass data ( detections and upper limits ) of our sample are shown in table 2 .
the name of the source and the position of the x - ray detection are given in columns 1 and 2 .
as mentioned before , the spatial resolution of the pspc is too low to resolve most of the binary systems .
in fact , in most of the cases we have only obtained a single detection displaced from both components of the system .
hence , we show in column 3 the displacements of the x - ray detection with respect to the optical positions of both members of the binary system .
there is only one system in which the projected separation is so large that the x - ray emission can be clearly attributed to the secondary star : hd87901 . the total number of counts in the broad band , the exposure time and the hardness ratios hr1 and hr2 are given in columns 4 , 5 , 6 and 7 . in the case of non - detections we have computed the number of counts at the position of both components , a and b , respectively .
an estimate of the probability of the detection by the ml procedure is given in the last column ; a value of ml=5 corresponds to a 2.7@xmath26 signal over the background .
note that 11 sources listed in table 1 do not appear in table 2 .
most of them ( hd8803 , hd27638 , hd36013 , hd36779 , hd53755 , hd71510 , hd123445 and hd162082 ) are not detected and are located so close to another x - ray sources that a computation of an upper limit is not possible .
two of them ( hd70309 , hd76566 and hd120641 ) show bad quality exposure maps so no x - ray data can be derived .
the same information as in table 2 is given in table 3 for pspc pointed observations .
only four of our binary systems were observed in pspc pointings , with just two of them detected .
columns 1 and 2 provide the name of the system and the rosat observation request ( ror ) number .
the coordinates of the x - ray detection are given in column 3 .
as in the case of the rass data , the spatial resolution is not high enough to detect both components of the binary systems . as a consequence , we show the displacement @xmath16 of the x - ray source with respect to both components of the binary system in column 4 .
column 5 shows the displacement of the source with respect to the center of the image ( off - axis angle ) .
note that the sensitivity of the detector degrades with increasing off - axis angle .
the total number of counts , the exposure time and the hardness ratios , hr1 and hr2 , are given in columns 6 , 7 , 8 and 9 .
the ml coefficient is provided in the last column .
l@l@c@l @r@c@r@r @r@r@r hd & ror & comp .
& & @xmath16 & offaxis & counts & exp .
time & hr & ml + & number & & ra ( 2000 ) & dec ( 2000 ) & ( ) & ( ) & & ( secs . ) & & + & & & ( h ) ( m ) ( s ) & ( ) ( ) ( ) & & & & & & + 560 & 201678h & * e@xmath13 * a+b & 00 10 02.39 & 11 08 40.0 & 5.7/5.0 & 0.33 & [email protected] & 1531.7 & [email protected] & 304.3 + & 201987h & a+b & 00 10 02.67 & 11 08 37.5 & 10.1/7.1 & 0.38 & [email protected] & 2373.5 & [email protected] & 505.7 + 1438 & 201154h , h-1 & a+b & & & & & @xmath1813.51 & 5443.5 & & + 17543 & 201331h & a & 02 49 17.63 & 17 27 47.9 & 3.8 & 0.29 & [email protected] & 5770.1 & @xmath18 - 0.46 & 16.2 + & & c & 02 49 19.34 & 17 27 40.6 & 5.4 & 0.61 & [email protected] & 5767.9 & [email protected] & 75.1 + 27638@xmath27 & 200187 & a & & & & & @xmath184.26 & 3881.6 & & + & & b & & & & & @xmath182.03 & 2312.9 & & + 33802@xmath28 & 200185h & * e * a & 05 12 17.95 & -11 52 05.4 & 3.8 & 0.30 & [email protected] & 2349.8 & & 2500.5 + & & * e * b & 05 12 17.55 & -11 51 56.5 & 4.5 & 0.32 & [email protected] & 2350.1 & & 352.6 + 38622 & 201333h & a & 05 47 42.63 & 13 53 56.2 & 4.5 & 0.16 & [email protected] & 4139.7 & @xmath18 - 0.43 & 33.8 + & & c & 05 47 42.55 & 13 53 31.9 & 2.1 & 0.50 & [email protected] & 4139.0 & @xmath190.54 & 21.3 + 40494 & 201685h & a & 05 57 32.14 & -35 16 59.6 & 0.8 & 0.28 & [email protected] & 4539.2 & @xmath18 - 0.78 & 89.4 + & & b & & & & & @xmath181.59 & 4535.3.8 & & + 53191@xmath4 & 201684h & b & 07 00 14.53 & -60 51 54.4 & 0.7 & 0.41 & 5.9 @xmath112.6 & 4813.5 & & 7.9 + 60102@xmath4 & 201328h & b & 07 11 27.90 & -84 28 10.3 & 2.6 & 0.25 & [email protected] & 2316.0 & [email protected] & 101.8 + 74146@xmath7 & 201447h & a & 08 39 57.41 & -53 03 15.3 & 2.5 & 16.02 & [email protected] & 50034.5 & @xmath18 - 0.68 & 69.9 + & 201682h & a & 08 39 57.78 & -53 03 15.2 & 2.5 & 0.25 & [email protected] & 4680.5 & [email protected] & 24.1 + 86388@xmath4 & 201683h & a+b?,b & 09 55 04.26 & -69 11 28.0 & 11.4/6.4 & 0.08 & [email protected] & 2200.6 & [email protected] & 51.4 + 87901 & 800807h & a & 10 08 22.37 & 11 58 05.8 & 4.0 & 11.59 & [email protected] & 2781.5 & -0.34 @xmath110.07 & 460.0 + & `` & b & 10 08 12.73 & 11 59 50.7 & 2.8 & 14.32 & [email protected] & 2712.7 & [email protected] & 44.6 + & 800807h-1 & a & 10 08 22.53 & 11 58 02.2 & 3.3 & 11.52 & [email protected] & 18262.3 & [email protected] & 3795.3 + & '' & b & 10 08 12.95 & 11 59 47.8 & 1.9 & 14.25 & [email protected] & 17774.6 & [email protected] & 506.1 + & 800880h & a & 10 08 22.10 & 11 58 08.5 & 7.2 & 12.47 & [email protected] & 60144.2 & [email protected] & 9914.8 + & & b & 10 08 12.64 & 11 59 55.7 & 7.8 & 15.23 & [email protected] & 5853.6 & [email protected] & 1560.7 + 90972@xmath29 & 200188h & b & 10 29 34.59 & -30 36 30.4 & 2.0 & 0.32 & [email protected] & 4524.8 & [email protected] & 120.3 + 108767 & 201329h & a & 12 29 52.11 & -16 30 54.3 & 3.9 & 0.34 & [email protected] & 3772.6 & [email protected] & 114.1 + & & b & 12 29 51.07 & -16 31 14.2 & 2.5 & 0.42 & [email protected] & 3775.6 & [email protected] & 384.0 + & 201679h & a & 12 29 52.03 & -16 30 54.7 & 2.7 & 0.35 & [email protected] & 4277.6 & [email protected] & 135.2 + & & b & 12 29 50.97 & -16 31 14.4 & 1.1 & 0.45 & [email protected] & 4279.1 & [email protected] & 278.4 + & 201991h & a & 12 29 51.70 & -16 30 53.3 & 3.1 & 0.39 & [email protected] & 2432.4 & [email protected] & 60.8 + & & b & 12 29 50.67 & -16 31 13.6 & 3.6 & 0.81 & [email protected] & 2431.8 & 0.55@xmath11 0.13 & 104.3 + & 201991h-1 & a & 12 29 51.95 & -16 30 53.7 & 2.3&0.36 & [email protected] & 4218.0 & @xmath180.01 & 68.5 + & & b & 12 29 51.10 & -16 31 13.4 & 3.2 & 0.75 & [email protected] & 4218.3 & [email protected] & 336.3 + 109573 & 702764h & * e * a+b , b ? & 12 35 56.35 & -39 50 15.5 & 7.1/4.7 & 2.69 & [email protected] & 40293.9 & [email protected] & 3439.0 + 113703@xmath29 & 150034h & * e * a+b , b ? & 13 06 18.39 & -48 27 45.2 & 17.0/5.0 & 0.30 & [email protected] & 3042.1 & [email protected] & 412.4 + 113791 & 201680h & a & 13 06 54.9 & -49 54 22.7 & 2.6 & 0.23 & [email protected] & 6102.4 & [email protected] & 358.2 + & & b & 13 06 57.37 & -49 54 27.3 & 2.9 & 0.38 & [email protected] & 6101.9 & [email protected] & 173.6 + 123445 & 200183h & a & 14 08 52.24 & -43 28 09.5 & 6.6 & 0.11 & [email protected] & 4100.6 & [email protected] & 161.1 + & & b & & & & & @xmath30 & 4099.0 & & + 127304 & 201681h & a & & & & & @xmath184.48 & 5158.0 & & + & 201681h & b & 14 29 48.08 & 31 47 22.3 & 5.4 & 0.50 & [email protected] & 5158.4 & & 7.7 + 127971@xmath27 & 200182h & a & 14 35 31.30 & -41 31 05.5 & 3.4 & 0.11 & [email protected] & 5420.1 & [email protected] & 146.8 + & & b & & & & & @xmath181.46 & 12012.1 & & + 129791@xmath27 & 200184h & a & 14 45 57.61 & -44 52 05.6 & 2.5 & 0.11 & [email protected] & 5063.0 & [email protected] & 684.6 + & & b & 14 45 56.21 & -44 52 36.8 & 4.9 & 0.48 & [email protected] & 5061.4 & @xmath190.77 & 60.5 + 143939@xmath29 & 200190h & b & 16 04 43.98 & -39 26 14.9 & 5.1 & 0.21 & [email protected] & 7841.5 & [email protected] & 609.0
+ * notes : * 1 . if both components are detected , coordinates of both x - ray sources are given .
the displacement of the secondary with respect to the detected x - ray source is computed comparing with its optical coordinates ; 2 .
the two stars of the system are not completely resolved so each star is contributing to the count rate of the companion .
the undetected component is so close to the x - ray source that no reliable upper limit can be computed ; 4 .
hd74146 has another observation , 200186h , but with very low exposure time ( t@xmath31 = 357.5 s ) ; * e * for elongated sources ; @xmath13the elongated shape is due to attitude problems in the processed data ; @xmath27 stars previously studied by schmitt et al .
( 1993 ) .
the hri detector allows high resolution imaging of x - rays sources .
the nominal spatial resolution of the detector is 1.7 but a bore - sight correction as much as 10 should be applied .
the projected separations between the members of our binary systems range from 4.4 to 177 .
most of them show separations larger than 10 so , in principle , they can be resolved by the hri . the spectral resolution of the hri is worse than that of the pspc .
however , the hri pulse height distribution can be used to compute a 2-band ( h and s ) hardness ratio .
we have defined the hri hardness ratio as : @xmath32 to obtain the most realiable s- and h - band definitions , several rosat sources known from their pspc pointed observations to be either extremely soft ( hr1 @xmath18 -0.8 ) or extremely hard ( hr1 @xmath19 + 0.8 ) were selected ( @xmath2220 ) . for these sources
the hri hardness ratios were calculated according to eq .
( 2 ) for the following different soft and hard band definitions : s1 = 0 - 5 , h1 = 6 - 15 ; s2 = 0 - 4 , h2 = 5 - 15 ; s3 = 0 - 3 , h3 = 4 - 15 ; and s4 = 0 - 2 , h4 = 3 - 15 , where the numbers give the corresponding hri energy channels . for each of the corresponding four
hri hardness ratios the mean difference to the pspc hardness ratios ( hr1 ) , was computed .
the hri bands for which the selected sources show the smallest deviation to the pspc hardness ratio were s3,h3 .
therefore , the hri bands are defined as the soft band ( s - band ) corresponding to channels 0 to 3 ( e @xmath33 0.3 kev ) and the hard band ( h - band ) corresponding to channels 4 to 15 ( e @xmath34 0.3 kev ) .
this hri hardness ratio shows a smaller dynamical range than the pspc hardness ratio , hr1 , but clearly identifies soft sources with negative values and hard sources with positive values . for more details , see supper et al .
( in prep . )
. table 4 shows the hri detections of lindroos binaries .
most of the hri pointed observations were carried out by one of us ( hz ) as principle investigator ( pi ) .
columns 1 to 3 give the name of the source , the ror number , and the component of the system detected : a+b when both stars are not resolved and there is only a single detection , and a or b ( or c or x for secondaries in multiple systems ) when the members of the binary system have been resolved .
we have also added an e to designate the elongated sources .
the position of the x - ray sources are shown in column 4 .
column 5 shows the difference between the optical and x - ray position .
note that the optical coordinates used as a reference for the b - components are those of the secondaries ( taken from simbad database ) .
column 6 shows the displacement of the sources with respect to the axis of the telescope . finally , the total counts in the broad band , the exposure time , the hardness ratio and the ml coefficient are given in columns 7 , 8 , 9 and 10 .
the upper limits of the undetected sources are also shown .
table 4 includes three binary systems ( hd86388 , hd109573 and hd113703 ) for which it is not clear which of the two members of the system is responsible for the x - ray emission . after the comparison of the optical and the x - ray position
, we believe that the late - type companion is most likely the x - ray emitter in the three cases . this is clearer for the hd86388 system because the x - ray detection is not elongated and it is closer to the optical position of the secondary . in the case of hd109573 and hd113703
the x - ray detections in the broad band are slightly elongated although much closer to the optical position of the secondary star ( see table 4 ) .
hence , we will assume that the x - ray emission comes from the late - type secondaries in the three cases under study . in order to illustrate different observations of the lindroos binary systems ,
we have shown in fig .
1 several hri images of different pairs .
.x - ray luminosities [ cols= " < , < , < , < , < " , ] * notes : * @xmath13 no u - magnitudes available .
we derived them from their spectral types and their v - magnitudes , according to kenyon & hartmann ( 1995 ) .
berghfer et al .
( 1999 ) has shown that the hri detector is sensitive to ultraviolet ( uv ) radiation below 4000 .
the contamination of the uv light to the final count rate is mainly concentrated to pulse height channels 1 - 3 ( @xmath22 our s - band range ) .
our primary stars are bright sources in the uv range so , in principle , the computed s - band rate could be just a response of the detector to the uv light . in order to check the reliability of the s - band counts for the early - type stars , we have estimated the contribution of the photospheric uv light to the total s - band rate .
following berghfer et al .
( 1999 ) : @xmath35 with u being the u - magnitudes of the observed stars .
this equation was deduced considering the emission of the sources from channels 1 to 8 of the hri detector . in order to be conservative ,
we have also computed the s - band rates in these channels .
table 6 shows the results .
column 1 provides the name of the star .
the u - magnitudes , taken from _ the bright star catalogue _ ( hoffleit & jaschek 1991 ) are shown in column 2 . for three sources u - magnitudes are not available , so we deduced them from their spectral types and v - magnitudes following kenyon & hartmann ( 1995 ) .
the s - band rate and the uv rate , both computed in channels 1 - 8 of the detector , are shown in columns 3 and 4 , while their ratio is provided column 5 .
we finally show the derived @xmath36 ratio in the last column , in order to check the reliability of the x - ray detections .
the bolometric luminosities have been computed using the stellar data provided in table 1 and the bolometric corrections from schmidt - kaler ( 1982 ) .
as we can see from table 6 , most of the b - type stars of our sample have s - band rates significantly higher than the estimated uv - rates .
there is only one source for which both values are comparable : hd87901a .
note that the hri observations of this star show very large off - axis values and , according to berghfer et al .
( 1999 ) , eq .
( 1 ) is not reliable for these cases .
however , this system was resolved by the pspc in the rass survey ( see tab .
2 ) and , as we mentioned in sect .
3 , no emission was detected from the early - type star but from the secondary . to further test the reliability of the s - band emission in early - type stars , we have also studied the @xmath36 ratio of these stars to see if it is consistent with the ratio for stars of similar spectral types . according to table 6 ,
most of the stars show ratios consistent with the reported values for early - type stars ( berghfer et al .
there are also four stars with unusual ratios : hd87901 with a lower ratio , and hd123445 , hd127971 and hd129791 , with ratios closer to those found in late - type stars .
given that these three early - type stars also show positive hr , it is possible that they have unresolved late - type companions .
however , in the case of hd87901 the hr is negative and the @xmath36 is lower than in late - type stars , so the x - ray emission can not be related with an unresolved source .
this fact together with its non - detection in the rass make us think that this emission is due to the hri uv leak .
therefore , excluding hd87901a , we can identify the computed s - band rates with intrinsic x - ray emission from the sources .
the previous test allows to confirm the reliability of the computed hri hr s .
given that these hr s are systematically positive for late - type stars and systematically negative for early - type stars , we can conclude that the nature of the x - ray emission is intrinsically different for our two samples .
we will discuss this point more deeply in the following subsection .
the analysis of the hr s suggests that the nature of the x - ray emission is different for our two samples .
a useful way to represent this difference consists of comparing the x - ray luminosities obtained in the two hri bands , s and h , for all of our detected sources .
3 shows this comparison .
note that we have converted the s- and h - band rates into luminosities following the procedure described in sect .
we have not included hd33802 because the binary is not completely resolved and it is difficult to obtain reliable measurements from the individual members ( see fig .
1 ) . as can be seen from fig .
3 , there is a clear separation between early- and late - type stars .
early - type stars are generally softer than the late - type companions .
this result is in agreement with that obtained from the hr s analysis .
note that although most of the early - type stars lie below the 1:1 correlation line , there are three early - type primaries with higher x - ray luminosities in the h - band than in the s - band : hd123445a , hd127971a and hd129791a . as we explained before
, the higher h - band luminosities obtained for the later - type stars are in agreement with their condition of active stars ( note that all are located above or on the 1:1 correlation line ) .
in contrast , the higher h - band luminosities for the early type stars are not easy to explain , given that these stars are not supposed to have a convective zone able to support a corona . because these stars lie in the same part of the diagram as the late - type secondaries , the simplest explanation is to relate them to unresolved late - type companions . after the analysis of the x - ray luminosities in the two bands , we have studied the total x - ray luminosities of the whole sample . given that the x - ray emission typically assumes a particular value for each spectral type , we have plotted the x - ray luminosity against the bolometric luminosity for each star .
the bolometric corrections were adopted from schmidt - kaler ( 1982 ) . only in the case of the m - type star , hd109573b ( hr4796b ) ,
the bolometric correction was taken from kenyon & hartmann ( 1995 ) . as shown in fig .
4 , there is a clear separation between late- and early - type stars .
late - type stars lie in the region of @xmath37 with x - ray luminosities varying between @xmath3827.5 and 31 . on the other hand ,
the early - type primaries show lower x - ray luminosities that decrease from b1 to a0 spectral types . however , the three late - b type stars which clearly deviate in fig . 3 also deviate in this figure , showing a higher x - ray emission than expected for their spectral types : b9 , b7 and b9.5 .
we have compared these results with those previously obtained by schmitt et al .
the x - ray luminosities derived for their sample of 7 lindroos systems agree with our results except in one case : hd113703 .
while these authors identify the single hri x - ray detection with the early - type primary , we think it is most probably related to the late - type secondary . as we discussed in sect .
3 , the difference between the x - ray detection and the optical position is smaller for the late - type secondary in both the broad and hard band images . at this point we must also remark that the young star hd109573b ( hr4796b ) shows an x - ray luminosity higher than that reported by jura et al .
these authors compute the x - ray luminosity from the same hri image but using an ecf that corresponds to the pspc detector , although these detectors have different sensitivities .
moreover , they take a mean ecf from neuhuser et al . ( 1995 ) which is deduced from a rosat survey on taurus , a star forming region where most of the stars show visual extinctions larger than those in twa . therefore , we think that the value listed in table 5 is more realistic since it takes into account the ecf from the hri detector and a negligible absorption to the source .
one of our binary systems , hd560 , was detected but unresolved by the rosat hri ( see fig .
because this system is comprised of a b9 primary and a g5 type companion , we have made a simple test to confirm the results obtained for hd123445a , hd127971a and hd129791a : although we certainly know that hd560 is a binary system , we have supposed hd560 to be a single b9 star and not a pair .
therefore , if we assume that the single x - ray detection corresponds to a b9-type star and we plot this source into fig .
4 ( starred symbol ) , the result is that hd560 lies in the same part of the diagram as the three sources with unresolved late - type candidates .
hence , this test strengthens the idea of unresolved late - type companions in these late - b type stars .
we have compared our sample of primary stars with a sample of ms b - stars taken from the _ the rass catalogue of optically bright ob - type stars _ ( berghfer et al . , 1996 ) in figure 4 . in principle , our primary stars are in good agreement with the sample of b - type stars with comparable x - ray luminosities .
note , however , the large scatter in the x - ray values at @xmath39 35.5 ( b9 stars ) .
hd123445a , hd127971a and hd129791a lie in this region of the diagram with x - ray luminosities ranging from @xmath40 to @xmath41 erg / s . the wide range of x - ray luminosities found in late - b stars from berghfer s sample together with the position of our late - b primary stars in fig .
4 , suggests the possibility of having unresolved late - type companions in late - b type stars with @xmath4229.5 .
although it is beyond the scope of this paper , it would be interesting to check if the ms late - b type stars from berghfer et al .
( 1996 ) with highest x - ray luminosities are suspected to have late - type unresolved companions . as we mentioned before
, the @xmath43 ratio is generally similar for stars of the same spectral type . as an example
, a `` canonical '' relation of @xmath44 has been reported for o- and early b - type stars ( harnden et al .
1979 , long & white 1980 , pallavicini et al .
1981 , sciortino et al .
given that the @xmath45 ratio is characteristic for each spectral type , we have finally compared this value with the hri hr s computed in sect .
lllllllllll hd & * sp.t . * & * log l@xmath46 * & * ew li i@xmath3 * & & * rv@xmath4 * & + & & & & f(12@xmath47 m ) & f(100@xmath47 m ) & & x - rays & li i & ir - ex . & rv + & & ( erg / s ) & ( ) & ( jy ) & ( jy ) & & & & & + 560 b ? & g5 & 30.56 & 0.290 & 0.278 & 1.037 & e & + & + & + & + + 17543 c & f8 & 30.17 & 0.12 & 0.269 & 1.089 & & + & + & + & - + 27638 b & g2 & @xmath1828.49 & 0.152 & 0.375 & 2.014 & e & - & + & + & + + 33802 b & g8 & 30.03 & 0.318 & 0.508 & 2.718 & e : & + & + & + & ( + ) + 38622 b & g2 & 29.99 & [email protected]@xmath48 & 0.263 & 3.056 & e & + & + & + & + + 40494 b & g8 & @xmath1828.57 & [email protected]@xmath13 & 0.417 & 0.618 & & - & + & + & ? + 53191
b & g3 & 29.53 & [email protected]@xmath13 & & & & + & + & ? & ?
+ 60102 b & g8 & 30.62 & [email protected]@xmath13 & & & & + & + & ? & ?
+ 86388 b ? & f5 & 29.42 & [email protected]@xmath13 & & & _ e _ & + & + & ? & + + 87901 b & k0 & 28.76 & 0.017 & & & e : & + & + & ? & ( + ) + 90972 b & f9 & 30.11 & 0.147 & 0.253 & 1.657 & _ e : _ & + & + & + & ( + ) + 108767 b & k2 & 28.94 & 0.175 & 2.563 & 0.497 & e & + & + & - & + + 109573 b & m2.5 & 29.75 & 0.55@xmath49 & & & & + & + & ? & ?
+ 113703 b ? & k0 & 30.57 & 0.367 & 0.296 & 1.555 & _ e _ & + & + & + & + + 113791 b & f7 & 29.98 & [email protected]@xmath13 & 0.452 & 1.581 & _ e _ & + & + & + & + + 123445 & k2 & @xmath1829.21 & @xmath500.04@xmath13 & & & & - & - & ? & ?
+ 127304 b & k1 & 28.89 & [email protected]@xmath48 & 0.153 & 0.379 & e & + & + & - & + + 127971 b & k0 & @xmath1827.90 & @xmath510.030@xmath13 & & & & - & - & ? & ?
+ 129791 b @xmath52 & k5 & 30.63 & 0.230 & ? & ? & & + & + & ? & ?
+ 143939 b @xmath52 & k3 & 30.50 & 0.400 & ? & ? & & + & + & ? & ?
+ * notes : * 1 .
most of the li i ew values have been adopted pallavicini et al .
these measurements were derived from high - resolution spectra ( resolution of 0.1 ) , except those marked with an asterisk , which were obtained from low - resolution spectra ( resolution of 2 ; the data marked with a double asterisk are taken from martn et al . , 1992 ( disp . of 0.22 / pix ) ; @xmath49 adopted from webb et al .
( 1999 ) ; 2 .
the ir data are taken from ray et al .
( 1995 ) ; the two sources marked with a @xmath52 are located close to the galactic plane and this seems to be the reason of not being detected by iras ; 3 . if similar radial velocities have been measured for both members of the pair it is indicated by an _e_ ( if taken from gahm et al . 1983 ) and by an e ( if taken from martn et al .
if the measurements are doubtful , it is indicated by a : ( see gahm et al . , 1982 and martin et al . , 1992 for more details ) ; 4 .
we have summarized the data of the table in this column using the symbols + , - and ?
for positive , negative and unknown properties respectively .
the ( + ) symbols indicate doubtful properties .
5 shows a clear separation between the stars of our sample .
a boundary line drawn at @xmath53 allows us to classify our stars in soft and hard x - ray emitters .
while the softer x - ray emitters lie close to the @xmath54 ratio reported for o- and early - b type stars , those with positive hr , are spread over a wide range of @xmath43 values . among the latter , we can also distinguish two groups .
on one hand , we find the late - type secondaries lying at the top right corner of the figure . on the other hand ,
the three late - b type stars with possible unresolved late - type companions occupy a band with @xmath43 ranging from -4.5 to -6 .
note that these stars clearly deviate from the group of early - type primaries at the left bottom corner of the figure .
if we include hd560 in fig .
5 making the same assumptions as in fig .
4 , ie . assuming that it is a single b9 star responsible for the detected x - ray emission
, we can see that the source also lies in the same region as hd123445a , hd127971a and hd129791a . as a conclusion
, we can confirm that b - type stars in the lindroos systems under study generally show a decrease in the x - ray luminosity for decreasing spectral types ( from b0 to b9 ) .
the @xmath45 ratio is in agreement with their spectral types and it is well - correlated with their negative hr s .
however , there are three sources , hd123445a , hd127971a and hd129791a which show @xmath55 values higher than those reported for earlier b - type stars . moreover ,
when comparing the @xmath45 ratio with the computed hri hr s , these three late - b type stars clearly deviate from the sample of primary stars , showing values closer to those reported for the late - type secondaries .
therefore , these three late - b type stars are suspected to have unresolved late - type companions .
as mentioned in the introduction , the late - type secondaries of lindroos systems have been studied in several spectral ranges . in particular ,
the optical and ir data provide clear evidences of youth among these stars . in table 7
we ahow some of the main observational properties of these late - type secondaries .
columns 1 , 2 and 3 provide the name of the source , its spectral type and the x - ray luminosity derived from this work .
column 4 shows the equivalent width ( ew ) of the li i absorption line while ir data from the _ iras _ satellite are provided in columns 5 and 6 .
a ` flag ' related to the measured radial velocity of the pair is given in column 7 .
we have finally summarized all these data in the last columns of the table in order to isolate those late - type stars with evidences to be physically bound to the early - type primaries . apart from the measured radial velocity of the pair and the li i ew we have also considered other indicators of youth like the x - ray emission and the measured ir excesses . as we can see from table 7 ,
all the late - type secondaries in binary systems with similar radial velocities show indications of youth . for the stars with no radial velocity measurements ,
we have found pairs with clear evidences of youth and hence , most probably physically bound to their primaries : hd17543 , hd53191 , hd60102 , hd129791 and hd143939 . hd123445b and hd127971b do not seem to be bound to their primaries because they lack indicators of youth . finally , hd40494b is not detected in x - rays but it shows a strong li i absorption line ( as hd27638 ) .
therefore , it is not obvious how to classify this source .
radial velocity measurements would be convenient to confirm the nature of this pair . according to lindroos ( 1985 ) ,
most of these stars show ages lower than 70 myr but , as discussed in l86 , the uncertainties associated with the age determination are large .
note that the ages of the lindroos systems were first determined from uvby@xmath0 photometry of the primaries ( lindroos 1985 ) and making use of the evolutionary tracks and isochrones by hejlesen ( 1980 ) .
a new estimation of the ages using modern isochrones plus new optical and ir photometrical data will be postponed to a later publication . instead of studying the ages of the systems we have considered the li i ( 6708 ) ew and we have related it to the derived x - ray luminosities .
as shown in table 7 , the li i ew has been directly measured in most of the lindroos secondaries , so we can study if there is a relation between both parameters .
we have plotted in fig .
6 the li i ew versus the effective temperature of our stars . note that these effective temperatures have been derived from their spectral types according to the conversion given by kenyon & hartmann ( 1995 ) .
we have also considered the upper envelopes to the li i ew of two young clusters , the pleiades ( @xmath22 125 myr ) and ic2602 ( @xmath22 35 myr ) , so that those secondary stars with li i ew stronger than in the pleiades or in ic2602 are younger than these clusters ( data from soderblom et al .
1993 , randich et al .
1997 and stauffer et al .
1997 ) .
6a shows that most of the strongest x - ray emitters are located above the pleiades upper envelopes to the li i. this suggests that a large fraction of the lindroos secondaries are younger than 125 myr , with 4 of them showing ages lower than 35 myr . among these 4 stars
, we can clearly distinguish hd109573b ( hr4796b ) in the right middle part of the figure , with a high value of the li i ew .
its x - ray emission is not as large as this coming from the other three sources above the ic2602 li i envelope .
the different spectral type as well as the different evolutionary state could explain this difference .
6a also allows us to confirm the nature of the hd123445 and hd127971 pairs .
the secondary stars lie at the bottom of the diagram with upper limits in both , the li i ew and the x - ray emission .
two more stars show non - detections in x - rays although they lie very close to the upper envelope of the pleiades : hd27638 and hd40494 . as we have discussed above , the former
is probably bound to the primary ( see table 7 ) while the nature of the later one is not still clear . as a conclusion , we can say that those stars with strong x - ray emission seem to be younger than ic2602 .
note , however that although the uncertainties associated with the li i ew measurements are generally low ( see table 7 ) , for the stars earlier than @xmath22k0 the li i ew does not allow us to reach any conclusion about their ages because f- and g - type pms together with zams stars still have their initial li content . to study the last issue in detail , we have converted the li i ew into lithium abundances , n(li ) making use of non - lte curves of growth from pavlenko & magazz ( 1996 ) and assuming a surface gravity of @xmath56 . note that we have only considered high - resolution data to make this conversion ( see table 7 ) .
we have plotted this value against the effective temperature of our stars in fig .
6b . as in the left panel of the figure , we have found strong x - ray emitters above the pleiades and ic2602 data .
note that most of the lindroos secondaries ( 11 ) show lithium abundances close to 3.0 which is consistent with low mass pms stars .
we have reported the x - ray emission from lindroos binary systems observed with rosat .
most of the stars from this sample are detected in the rass and several binary systems were observed with the pspc and the hri . after the analysis of the data , the main conclusions are : * 1 . * both
, early- and late - type stars , show the same distribution of the x - ray luminosity function ( xlf ) .
both samples are statistically similar with a probability of a 0.8 .
when we repeat the analysis without including stars that could be contaminating the samples , that is , without considering most certainly unbound secondaries , early - type primaries with possible unresolved late - type companions and one unreliable x - ray detection , the result does not change . the probability is reduced to 0.4 but it is large enough to statistically confirm the similarity of both samples . * 2 . * a careful study of the hri hardness ratios of our sample allows to discriminate between soft and hard x - ray emitters .
late - type stars always show positive hr while most of the early - type stars show negative values . moreover ,
if we compare the x - ray luminosity in the two energy bands of the hri , we can see that there is a clear separation between the two samples .
late - type stars show higher h - band than s - band x - ray luminosities .
this result can be explained in terms of an energetic corona and activity episodes commonly reported in these stars . in the case of early - type stars ,
most of them present higher s - band than h - band luminosities . however , there are three x - ray sources formally identified as early - type primaries which show h - band luminosities comparable to those of late - type stars : hd123445a , hd127971a and hd129791a .
the @xmath57 diagram shows a clear separation among sources of different spectral types .
this separation is even more prononced when the @xmath58 ratio is plotted vs. the hri hardness ratio .
while early - type stars , which generally show negative hr s , lie close to the @xmath59 `` canonical '' relation , late - type stars show positive hr s and present higher values of this ratio ( @xmath60 ) . for those particular cases of early - type stars with positive hr s
, we have seen that they display x - ray luminosities comparable to those of the late - type companions .
when they are plotted into the @xmath58 - hri hr diagram , they lie closer to the late - type stars than to the b - type group . all these evidences
make us conclude that hd123445a , hd127971a and hd129791a are good candidates to have unresolved late - type companions .
* 5 . * the computed x - ray luminosities together with the li i ( 6708 ) ew and abundances deduced for the lindroos late - type secondaries have revealed this group of stars to be a good sample of ptts candidates .
a strong x - ray emission is reported for the youngest ones .
although we have no reliable measurements of their ages , most of the lindroos secondaries seem to be younger than 125 myr ( when compared to the pleiades ) with three of them showing ages lower than 35 myr ( ic2602 cluster ) . also hd109573b ( hr4796b )
is located above these two clusters showing a high x - ray emission .
this study also has allowed us to confirm the optical nature of the hd123445 and hd127971 pairs . in this sense
, two more systems are classified as doubtful : hd87901 and hd40494 .
the former lacks clear indicators of youth while hd40494 , although not detected in x - rays , may be a young star bound to its primary .
we would like to thank t. hearty , m. fernndez , j. alves , s. dbereiner and b. knig for their assistance and their useful comments .
we are very grateful to t. berghfer for providing his data .
some archived rosat observations investigated here were performed by t. berghfer , j. krautter , j. puls , t. simon and j. staubert as pi s .
the rosat project is supported by the max - planck - society and the german government ( dlr / bmbf ) .
berghfer , t.w . ,
schmitt , j.h.m.m . ,
cassinelli , j.p . , 1996 , a&as 118 , 481 berghfer , t.w . , schmitt , j.h.m.m . ,
danner , r. , cassinelli , j.p .
, 1997 , a&as 322 , 167 berghfer , t.w . ,
schmitt , j.h.m.m .
, hnsch , m. , 1999 , a&a 342 , l17 cruddace , r.g . ,
hasinger , g.r . ,
schmitt , j.h.m.m . , 1988 , _ the application of a maximun likelihood analysis to detection of sources in the rosat data_. in : murtagh , f. and heck , a. ( eds ) , eso conf . and workshop proc
28 , astronomy from large databases , garching , p.177 | we present a study of the x - ray emission from binary systems extracted from the lindroos catalogue ( lindroos 1986 ) based on the rosat all - sky survey as well as rosat pspc and hri pointings .
the studied sample consists of visual binary systems comprised of early - type primaries and late - type secondaries .
the ages of the systems were determined by lindroos ( 1985 ) from uvby@xmath0 photometry of the primaries .
these ages range between 33 and 135 myr , so if the late - type secondaries are physically bound to the early - type primaries , they could be post - t tauri stars ( ptts ) .
we have found strong x - ray emission from several secondaries .
this fact together with their optical and ir data , make them _ bona fide _ ptts candidates .
we have also detected x - ray emission from several early - type primaries and , in particular , from most of the late - b type stars .
because their hri hardness ratios are similar to those from resolved late - type stars , the presence of an unresolved late - type companion seems to be the cause of this emission . |
0 the study of the matter distributions in the borromean nucleus @xmath3 and the neutron rich isotope @xmath4 is an actual problem discussed in many papers ( see , e.g. , in @xcite and refs . therein ) .
investigations give a reason to expect long tails ( halo ) of the neutron densities .
the first result on the enhancement of the strong interaction radii @xmath5 for @xmath6 was deduced in @xcite from measurements of their interaction cross sections with different target nuclei at fixed energy 800 mev / nucleon .
the same conclusion was made in @xcite about the @xmath0he @xmath7-radii .
later in @xcite , the model densities in the form of a sum of two gaussian functions @xcite were fitted to the data and a conclusion for a significant neutron halo in the outer region of these nuclei was made .
as to the differential elastic scattering cross sections , they were studied in collisions of @xmath0he with protons , and the matter distributions of these nuclei were tested using different phenomenological and theoretical models .
elastic scattering angular distributions of @xmath8 , @xmath9 ( @xmath10 ) and @xmath8 , @xmath11 ( @xmath12 ) were calculated in @xcite and @xcite , respectively , by means of the real part of the optical potentials obtained microscopically by using the realistic m3y - paris effective interaction and taking into account the neutron and proton densities of helium isotopes from @xcite and also derived from the cluster - orbital shell - model approximation ( cosma ) @xcite,@xcite,@xcite,@xcite .
it was shown in @xcite that elastic scattering is a good tool to distinguish between different density distributions . the results of @xcite were compared also with those of the alpha - core approach with the fully nonlocal effective interactions @xcite and the no - core model based on the large space shell - model calculations ( lssm ) @xcite,@xcite,@xcite . elastic scattering was recently studied also in @xcite .
at the same time , small experimental information on the energy dependence of the total reaction cross sections of @xmath6 with nuclei is available . generally , from such data for the well - known densities of the target - nuclei , one can get additional constraints on the parameters of the tested models of the halo nuclei .
below , we present our calculations of the total reaction cross - sections and compare them with the respective data on interactions of @xmath6 with @xmath13 at energies e / nucleon @xmath14 10@xmath15 50 mev .
following @xcite we start with the expression for the total reaction cross - sections in the framework of the optical - limit model of the high - energy multiple - scattering theory : + @xmath16 here , the eikonal phase for the scattering of the projectile nucleus ( @xmath17 ) by the target nucleus ( @xmath18 ) is given in @xcite by the expression + @xmath19 + for the isospin averaged total nucleon - nucleon cross - section @xmath20 we use the parametrization from @xcite .
the vectors @xmath21 are in the impact parameter @xmath22 plane perpendicular to the @xmath23- axis along the relative momentum @xmath24 of the two scattered nuclei .
the thickness densities are given as + @xmath25 where the bare density distributions @xmath26 of point nucleons in respective nuclei depend on their coordinates in the center - of - mass frames .
the function @xmath27 is defined by the form of the nucleon - nucleon interaction amplitude . to present the phase integral in one - dimensional form , one uses the two - dimensional fourier - bessel transformation of all functions in the integrand @xmath28 where @xmath29 this results in : @xmath30
the original high - energy scattering theory uses only the transversal component @xmath31 of the total momentum transfer @xmath32 and neglects its longitudinal component @xmath33 .
here @xmath34 , and @xmath35 is the momentum of the relative motion .
so , the theory is applied to the scattering angles @xmath36 , where @xmath37 is approximately the sum of the radii of both the nuclei@xmath38 .
so , in eq.([2.6 ] ) the use of @xmath39 allows one to take the 3-dimensional fourier transforms of the functions @xmath26 and @xmath27 instead of the 2-dimensional transforms of their profile functions @xmath40 and @xmath41 .
we remind that the form factors of the point @xmath42 and nuclear @xmath43 densities obey the relation @xmath44 where the nucleon form factor @xmath45 is suggested to be the proton one and has a dipole form which is transformed to the gaussian function at small @xmath46 : @xmath47 at the same time the form @xmath48 of the nn - scattering amplitude is known to be @xmath49 the squared values of the @xmath7 radii of the nucleon and the nn - interaction , known from the experimental data as @xmath50 and @xmath51 , almost coincide .
it was concluded in @xcite that a small difference between these @xmath7 radii does not influence visibly the calculated total reaction cross sections , and , thus , one can use approximately @xmath52 . then taking into account eqs .
( [ 2.7])-([2.9 ] ) we get the phase ( [ 2.6 ] ) in a simple form @xmath53 the form factors @xmath54 can be calculated by using the data on nuclear density distributions . here
, we use the realistic density in the form of the symmetrized fermi density distribution @xmath55},\ ] ] and the respective form factor obtained in @xcite ( see also @xcite ) in the form @xmath56 , \qquad \rho_0={3a\over 4\pi r^3}\,{\left}[1+{\left}({\pi a\over r}{\right})^2{\right}]^{-1}.\ ] ] here @xmath37 and @xmath57 are the radius and diffuseness parameters , respectively . now , taking different models for the projectile nuclei we can test them by comparison of the calculated cross sections with the existing experimental data . note that if for the densities of the projectile and the target nucleus one takes the gaussian functions and uses ( [ 2.9 ] ) for @xmath58 , then the integration in ( [ 2.6 ] ) can be performed explicitly @xcite , and one obtains the result of @xcite frequently used by experimentalists analyzing experimental data @xmath59 where @xmath60 . in the case of the zero - range approximation @xmath61 . for the repulsive coulomb field
the trajectory of the incident nucleus deflects from the scattering center .
this effect can be taken into account replacing he impact parameter @xmath22 in the phase @xmath62 t by the distance of the closest approach @xmath63 in the coulomb field @xmath64 where @xmath65 is the half - distance of the closest approach at @xmath66 . in our work
, we use models for the projectile nuclei which are known in a table form and , thus , we apply expression ( [ 2.10 ] ) in the calculations where the density @xmath67 of the target nucleus @xmath68 is taken in a realistic form ( [ 2.11 ] ) with @xmath69 and @xmath70 from @xcite . the coulomb distortion effect is also taken into account by using ( [ 2.14 ] ) .
we apply the model for the s - p and s - d shell nuclei , developed in ref .
@xcite ( m - model ) , to calculate the @xmath71he densities .
the densities of these projectile nuclei are used to study the effect of the jastrow - type short range nn - correlations ( src ) , on the total reaction cross - section @xmath71he @xmath72si measured at different energies @xcite . in the m - model , a general expression for the one - body density matrix of @xmath73 , @xmath74-@xmath17 and
@xmath74-@xmath75 shell nuclei @xcite has been obtained using the factor cluster expansion method ( see ref.@xcite and refs . therein ) .
that expression depends on the harmonic oscillator ( ho ) parameter @xmath76 ( @xmath77 ) , the occupation numbers @xmath78 of the various states and the correlation parameter @xmath79 that comes from the jastrow type correlation function @xmath80 , \quad r_{ij}=|{\bf r_i}-{\bf r_j}|\ ] ] which introduces short range correlations
. the effect of correlations introduced by the function @xmath81 becomes large when the correlation parameter @xmath79 becomes small and vice versa .
the expression of @xmath82 ( in the two body approximation for the cluster expansion ) has the form @xmath83,\quad r_b = r / b\ ] ] where @xmath84 , the uncorrelated density distribution corresponding to the slater determinant ho total wave function , comes from the one - body term of the cluster expansion of the one - body density matrix and the term @xmath85 comes from the two - body term .
their expressions as well as the expression of the normalization factor @xmath86 are given in @xcite .
similar expressions for the form factor and the momentum distribution have also been given in ref .
@xcite .
the parameters @xmath76 , @xmath79 and @xmath78 can be determined , for each nucleus , by fit to the experimental form factors and momentum distributions . as such data are not available for @xmath71he we will use the data of the total reaction cross - section and the @xmath7-radius .
that is , we involve the parameters @xmath76 , @xmath79 and the four occupation number @xmath87 , @xmath88 , @xmath89 and @xmath90 in the fitting procedure .
we used as experimental data the well known results of tanihata et al @xcite ( t - model ) for the density distributions obtained by fit the total reaction cross - section of @xmath71he @xmath91c at 800 mev in calculation with the model discussed in sec .
1 , using for the target nucleus the gaussian form @xmath92 fitted to the realistic density in the outer region and using for the projectile nucleus the harmonic - oscillator form @xmath93 where @xmath94 , and @xmath95 , @xmath96 are the fitting parameters within the t - model .
expression ( [ 3.3 ] ) and the corresponding expression of the @xmath7-radius have been used as the `` experimental '' data of the projectile nuclei @xmath71he .
the values of the parameters @xmath76 , @xmath79 and @xmath78 which have been found by a least squares fit of the theoretical neutron and proton density distributions ( expression ( [ 3.2 ] ) ) to the `` experimental '' ones with the constrain that the two densities have the same @xmath7-radius are given in table i. in the case of proton densities there is one free parameter , the ho parameter .
for that reason the proton density is taken to be the same for both models . in order to study the effect of src on the density and the total cross section we examined two cases . in the first case we have taken the neutron occupation numbers to be the ones given in table i , we have given to the correlation parameter @xmath79 various values from @xmath97@xmath98 to @xmath99@xmath98 and for each value of @xmath79 we determined the parameter @xmath76 so that the neutron @xmath7-radius to be the same with the one obtained within the t - model . in figure 1 we present the point density distribution for @xmath100he calculated in this case .
the bold curve is the bundle of eight figures .
it is seen that the neutron densities of the m - model have lower tails at @xmath101fm compared to the one of the t - model . at the same time , we see that the cross - sections are in good coincidence .
this means that the main contribution comes from the region @xmath102fm , i.e. , up to @xmath103 equal to about two @xmath7 radius of the @xmath100he nucleus ( the @xmath7-radius of the mater distribution is 2.331 fm for both model ) . from a physical point of view
, the increase of the src at small @xmath79 enhances the nuclear radius , but in the considered case one can think that this effect is compensated by reducing the ho parameter @xmath76 which results in `` squeezing '' the wave functions of the basis .
for that reason we considered the second case where the calculations were made for various values of the parameter @xmath79 in the range @xmath104 to @xmath105@xmath98 while the ho parameter @xmath76 had the value 1.8249 fm and the parameters @xmath78 were taken as before from table i. the results are presented in fig .
the calculations in ref.@xcite showed that the changes of the values of @xmath79 in that region give very significant changes in the form factors and the momentum distributions of nuclei .
in the present paper , we study the total cross - sections and as a result , we obtained the corresponding spread of the @xmath7 radius of the neutron density for @xmath100he in the limit @xmath106fm .
however , this does not noticeably influence the cross - sections . once again , the cross - sections of the m- and the t - models are in coincidence , while the @xmath7 radii are slightly different .
the same conclusions can be made for @xmath107he nucleus from figs .
3 and 4 , where the results of the calculations within the m - model are presented . in fig .
3 , one can see the bunch of nine curves for the neutron densities when the pairs of the parameters @xmath79 and @xmath76 which give the @xmath7-radius @xmath108fm are in the regions @xmath109@xmath98 and @xmath110fm .
the occupation numbers are the ones given in table i. for all the neutron densities the @xmath7-radii are in coincidence with the t - model @xmath111fm .
the matter density radius is also the same in both models @xmath112fm .
the values of the parameters of the curves in fig .
4 are : @xmath113@xmath98 , @xmath114fm and @xmath78 are the same as in fig .
3 . the obtained values of the @xmath7 radii of the neutron densities are in the region @xmath115fm .
we see that in both figures the slopes of the densities from the m - model begin to differ from that of the t - model at about @xmath116fm , where the densities fall down by about two orders of magnitude comparing to the values in the nuclear center .
their difference in the range of twice the @xmath7-radius does not affect the total cross - sections .
.the values of the parameters @xmath76 , @xmath79 and @xmath78 , for the neutron and proton densities of @xmath71he , obtained by fit of the expression ( [ 3.2 ] ) to expression ( [ 3.3 ] ) with the constrain that the two densities to give the same @xmath7-radius . [ cols="<,^,^,^,^,^,^,^",options="header " , ]
one should mention that in the m - model there were no calculations of the binding energies and the structure of excited states of these nuclei .
however , large - scale shell - model calculations ( lssm ) of the wave functions and characteristics of light exotic nuclei , such as @xmath117 and @xmath118 , were reported by several groups , e.g. .
they are usually performed by using interactions obtained directly from nn g - matrices with various nn - interactions as their base . in @xcite ,
the wave functions for @xmath117 were calculated within a complete 4@xmath119 model space by using the g - matrix with interaction from @xcite .
all calculations were performed by the shell - model code oxbash @xcite .
for @xmath120 the harmonic - oscillator single - particle wave functions were used , while for @xmath117 they were the woods - saxon ones .
it was concluded in @xcite that @xmath3 corresponds to halo nucleus , while @xmath4 is a neutron skin nucleus . in ( and refs .
therein ) , the k - harmonic method was developed and calculations were based on the assumption of the @xmath121 state for each of the valence neutrons related to the alpha - core center ( cosma - model ) .
the admixtures of the 3-body forces is included to get the proper value of the binding energy . in fig.5
, we show the matter density distributions for @xmath3 obtained from the calculations in the framework of the cosma - model @xcite ( dash - dots ) and the lssm - model @xcite ( dots ) .
the m- and t - model densities are shown by the solid bunch and the dashed curve , respectively .
first of all , one can see that in the outer region the cosma- and lssm - model densities have extended tails which are followed by exponential asymptotics . here
, the dotted curve exceeds the dash - dotted one which results in the enhancement of the corresponding cross section .
it is seen that both the cross sections are about @xmath122 above those calculated by the m- and t - models and the experimental data .
the respective @xmath7 matter radii are @xmath123 ( for the dash - dotted curve ) and @xmath124 ( for the dotted curve ) . in this connection
we refer to @xcite where the coupling of elastic and elastic breakup channels were shown to play an important role in processes with weakly bound nuclei , and this can diminish the cross section obtained without accounting for the additional channels .
recently , the theory of the coupled channels was developed in to study the breakup reactions of halo nuclei . in general , the implied optical - limit model needs higher order nn - scattering terms to be included as had been done in the multiple - scattering theory .
for example , it was shown in @xcite that these terms provide a negative interference and thus diminish the @xmath125 differential elastic cross section calculated by taking into account only the single - scattering term .
the in - medium effect , which changes the free nn - scattering cross section @xmath20 in the phase ( [ 2.2 ] ) can be included as well .
as shown qualitatively in @xcite , this effect can decrease the total nucleus - nucleus reaction cross sections by about 10% .
all the mentioned problems regarding the mechanism of the nucleus - nucleus interaction are still under investigations of the current theory .
however , our study of the total reaction cross - sections of the @xmath6 interactions with the @xmath68 target leads to some general conclusions which , we hope , will not be affected by a possible further modification of the theory . * the @xmath7 radii are more sensitive nuclear characteristics than the total reaction cross section with respect to changes of the density distributions .
the radius depends strongly on variations of the density in the nuclear interior while the behaviour of the total reaction cross - section depends mainly on the peculiarities of the nuclear surface and weaker on the central nuclear region where strong absorption takes place .
* the previous item explains why variations of the densities due to src do not affect the total cross sections while they are significant in calculations of the form factors and of the single - particle momentum distributions in nuclei at large momentum transfers . *
the fitting procedure in our work showed that there is not a @xmath126-shell contribution to the wave functions of @xmath3 and @xmath4 .
* in the case of the harmonic - oscillator basis , the gaussian type of the wave function asymptotics does not affect significantly the cross sections and the latter are formed by taking into account the interior region of the projectile nucleus up to about only two @xmath7 nuclear radii .
otherwise , if one gets a basis for realistic wave functions , which have an exponential asymptotics , the region of the main contribution becomes larger and achieves the value of about four and larger times the @xmath7 nuclear radius .
* acknowledgments * + the authors would like to thank dr.s.ershov for providing the @xmath3 matter density distribution used in investigations in .
e.v.z . was supported by the rfbr under grant no.03 - 01 - 00657 .
000 -1 mm r.f.casten , b.m.sherill , _ prog.part.nucl.phys . _
* 45 * , s171 ( 2000 ) ; _ see also a special issue of nucl.phys._ * a693 * , no.1 - 2 ( 2001 ) .
i. tanihata et al . ,
_ phys.lett . _
* b160 * , 380 ( 1985 ) .
i. tanihata et al . , _ phys.rev.lett . _ * 55 * , ( 1985 ) 2676 .
i. tanihata et al . , _ phys.lett . _ * b289 * , 261 ( 1992 ) .
zhukov , a.a .
korsheninnikov , and m.h .
smedberg , _ phys.rev . _ * c50 * , r1 ( 1994 ) .
m. avrigeanu , g.s .
anagnostatos , a.n .
antonov , j. giapitzakis , _
phys.rev . _ * c62 * , 017001 ( 2000 ) .
n. anantaraman , h. toki , and g. bertsch , _ nucl.phys .
_ * a398 * , 279 ( 1983 ) .
r. satchler and w.g .
love , _ phys.rep .
_ * 55 * , 183 ( 1979 ) ; g.r .
satchler , _ direct nuclear reactions _
( clarendon , oxford , 1983 ) .
dao t. khoa and w. von oertzen , _ phys.lett . _ * b304 * , 8 ( 1993 ) ; _ ibid _ * b342 * , 6 ( 1995 ) ; dao t. khoa , w. von oertzen , and h.g .
bohlen , _ phys.rev . _ * c49 * , 1652 ( 1994 ) ; dao t. khoa , w. von oertzen , and a.a . ogloblin , _ nucl.phys . _ * a602 * , 98 ( 1996 ) ; dao t. khoa and g.r .
satchler , _ nucl.phys .
_ * a668 * , 3 ( 2000 ) .
korsheninnikov , e.a .
kuzmin , a. ozawa et al . , _
_ * a616 * , 189 ( 1997 ) .
korsheninnikov , e.yu .
nikolskii , c.a .
bertulani et al . , _ nucl.phys .
_ * a617 * , 45 ( 1997 ) .
zhukov , b.v .
danilin , d.v .
fedorov et al . , _ phys.rep .
_ * 231 * , 151 ( 1993 ) .
a. lagoyannis , f .
auger , a. musnamarra et al . ,
_ phys.lett .
_ * b518 * , 27 ( 2001 ) .
k. amos , in _ proc .
reaction mechanisms , varenna , 2000 , _ ed .
ettore gadioli , ( ricerca scientifica ed educazione permanente , milano , 2000 ) , p.51 .
s. karataglidis , b.a .
brown , k. amos , and p.j .
dortmans , _ phys.rev . _ * c55 * , 2826 ( 1997 ) .
alkhazov et al . , _ nucl.phys . _
* a712 * , 269 ( 2002 ) ; r.e .
warner et al . , _ phys.rev . _
* c52 * , r1166 ( 1995 ) ; i.v .
kuznetsov et al . , _
( phys.at.nucl . ) _ * 65 * , 1609 ( 2002 ) ; v.k .
lukyanov , b. slowinski , e.v .
zemlyanaya , `` nucleus - nucleus reaction cross - sections calculated for realistic nuclear matter distributions within the glauber - sitenko approach '' , jinr preprint p4 - 2003 - 80 , dubna , 2003 ; http://arxiv.org/abs/nucl-th/0308079 .
glauber , _ lectures on theor .
_ , * 1 * , ( interscience , new york , 1959 ) .
sitenko , _
ukr.fiz.j . _ * 4 * , 152 ( 1959 ) .
w. czyz , l.s .
maximon , _
ann.of phys .
_ * 52 * , 59 ( 1969 ) .
s. charagi and g.gupta , _ phys.rev .
_ * c41 * , 1610 ( 1990 ) .
lukyanov , e.v .
zemlyanaya , _
int.j.mod.phys . _ * e10 * , no.3 , 169 ( 2001 ) .
petkov , v.k .
lukyanov , yu.s .
pol , _ yad.fiz . _ * 9 * , 349 ( 1969 )
grypeos , c.g .
koutroulos , v.k .
lukyanov , a.v .
shebeko , _ fiz.elem.chast . &
at.yad._ * 32 * , 1494 ( 2001 ) ; m.e .
grypeos , c.g .
koutroulos , v.k .
lukyanov , a.v .
shebeko , _
_ * g32 * , 779 ( 2001 ) .
phys.rev . _ * c11 * , 1203 ( 1975 ) .
burov , d.n .
kadrev , v.k .
lukyanov , yu.s .
pol,_yad.fiz . _ * 61 * , 595 ( 1998 ) ; v.v .
burov , d.n .
kadrev , v.k .
lukyanov , yu.s .
pol , _ phys.at.nucl . _ * 61 * , 525 ( 1998 ) .
massen , ch.c .
moustakidis , _ phys.rev . _ * c60 * , ( 1999 ) 0240005 ( 1999 ) ; ch.c .
moustakidis , s.e .
_ phys.rev . _ * c64 * , 014314 ( 2000 ) .
clark , _ prog.part.nucl.phys . _
* 2 * , 89 ( 1979 ) .
p. navratil and b.r.barrett , _ phys.rev .
_ * c54 * , 2986 ( 1996 ) .
p. navratil and b.r .
barrett , _ phys.rev . _ * c57 * , 3119 ( 1998 ) .
s. karataglidis , p.g .
hansen , b.a .
brown , k. amos , and p.j .
dortmans , _ phys.rev.lett .
_ * 79 * , 1447 ( 1997 ) ; s. karataglidis , b.a .
brown , k. amos , and p.j .
dortmans , _ phys.rev .
_ * c55 * ) , 2826 ( 1997 . s. karataglidis , p.j .
dortmans , k. amos , and c. bennhold , _ phys.rev . _ * c61 * , 024319 ( 2000 ) .
d.c . zheng , b.r .
barrett , j.p .
vary , w.c .
haxton , and c .- l .
song , _ phys.rev . _ * c52 * , 2488 ( 1995 ) .
oxbash - msu , the oxford - buenos aires - michigan state university - shell model code , a. etchegoyen , w.d.m .
rae , and n.s .
godwin , msu version by b.a .
brown , 1986 ; b.a .
brown , a. etchegoyen , and w.d.m .
rae , msucl & nbsp ; report no .
524 , 1986 ( unpublished ) .
al - khalili , j.a .
tostevin , and i.j .
thompson , _ phys.rev . _ * c54 * , 1843 ( 1996 ) .
ershov , t. rogde , b.v .
danilin , et al . ,
_ phys.rev . _ * c56 * , 1483 ( 1997 ) .
ershov , b.n .
danilin , j.s .
phys.rev . _
* c64 * , 064609 ( 2001 ) . | calculations of the @xmath0he@xmath1si total reaction cross sections at intermediate energies are performed on the basis of the glauber - sitenko microscopic optical - limit model .
the target - nucleus density distribution is taken from the electron - nucleus scattering data , and the @xmath0he densities are used as they are derived in different models .
the results of the calculations are compared with the existing experimental data .
the effects of the density tails of the projectile nuclei as well as the role of shell admixtures and short - range correlations are analyzed .
* testing @xmath0he density distributions by calculations of + total reaction cross - sections of @xmath0he+@xmath2si * + [ .3 cm ] v.k .
lukyanov , e.v .
zemlyanaya + _ joint institute for nuclear research , dubna 141980 , russia
_ + s.e .
massen , ch.c .
moustakidis + _ aristotle university of thessaloniki , greece _
+ a.n .
antonov + _ institute for nuclear research and nuclear energy , sofia 1784 , bulgaria _ + g.z .
krumova + _ university of rousse , rousse 7017 , bulgaria _ |
it is generally considered that a full understanding of the masses and associated mixings of the quarks and leptons is one of the chief unsolved problems in elementary particle physics .
while a great deal of experimental knowledge about quark masses and mixings has been available for quite a long time , it is only in the last few years that very detailed information on lepton masses and mixings has been found from a series of remarkable experiments involving neutrino oscillations ( see for some examples , refs ) .
these results can be expected to provide valuable clues toward the solution of this mass and mixing problem .
one such clue is the fact that the leptonic mixing matrix is now known to be somewhat close to what is called the `` tribimaximal '' form ; actually a number of interesting discussions of this form have been presented over a period of years . for immediate convenience
this form may be read from eq.([r ] ) below and it can be seen that one column has three equal entries while another has two entries of equal magnitude with the third zero . such a structure is natural in the context of permutation symmetry since it is a characteristic one which brings the basis of the defining representation of @xmath0 ( the permutation group on three objects ) to the basis in which it is decomposed into two and one dimensional irreducible pieces . of course
, @xmath0 is natural in the context of the standard model of elementary particles , which contains three parallel families of fermions .
it has also been often discussed in the literature .
however , it seems that this symmetry can not be exact ; otherwise there would be two exactly degenerate neutrinos , which is not the case .
it is possible that the situation may be analogous to another three flavor theory- the old @xmath1 model in which the three flavors turned out to represent the u , d and s quarks rather than the three families we now know about .
an initial @xmath1 flavor symmetry for the then unknown strong interaction hamiltonian was assumed to be broken , according to the gell mann okubo hypothesis @xcite , as : @xmath3 where @xmath4 is @xmath1 invariant and @xmath5 breaks the symmetry , leaving the iso - spin subgroup invariant .
actually , another symmetry breaking term , which breaks the iso - spin subgroup but preserves a different @xmath6 subgroup is also required . a possible analogy to the @xmath0 symmetry case might be to postulate that the standard electroweak lagrangian density has a decomposition like @xmath7 where , on the right hand side , the first term is assumed to be @xmath0 invariant , the second term only preserves one @xmath8 subgroup of @xmath0 and the third term preserves a different @xmath8 subgroup of @xmath0 .
there is a presumption that the first term is the dominant one . in the present paper
we will concentrate on the fully @xmath0 invariant term , @xmath9 of eq.([s3breaker ] ) .
we must specify what are the first approximation fermion masses and the first approximation mixing matrices for which we are aiming . in the cases of the charged leptons and the up and down quarks , the first approximation to the masses
is conventionally taken to mean assigning the heaviest fermion of each of the three families ( @xmath10 ) a non - zero mass and setting the masses of the others to zero .
considering that the quark ckm angles are all small it seems reasonable to consider the charged fermion mixing matrices be just the unit matrices in first approximation . on the other hand we would like to take the tribimaximal form as the first @xmath0 invariant approximation to the neutrino mixing matrix .
the question of what is a good first approximation to the spectrum of neutrino masses is perhaps not so clear .
we will argue that an approximately degnerate neutrino spectrum is the most reasonable choice .
our job will be to demonstrate that the same @xmath0 symmetry can give rise to different patterns of masses and mixings for the electrically charged fermions and the neutrinos . in this paper we will work completely in the framework of the electroweak theory , postponing any possible inclusion in a grand unified theory .
we will assume only a minimal number of fermion fields so that the neutrinos will be initially taken to be majorana type .
on the other hand we wo nt fully specify the higgs fields in advance .
we will also show that the model can be modified for the case of dirac neutrinos . in section ii we will give the needed notation for the fermion fields and
shall show how the permutation group , @xmath0 may act on them .
section iii first contains a discussion of the form of the neutrino mass matrix under the assumption of @xmath0 invariance .
it is shown how it can be diagonalized by a matrix which is , up to a diagonal phase matrix and a possible rotation in a two dimensional subspace , the same as the tribimaximal form .
then , an argument is given that the largest part of the solution space for the three neutrino masses derived from the known mass differences corresponds to almost complete degeneracy .
also in section iii , we note that this can easily be achieved by making use of the freedom to have complex constants in the mass matrix .
this results in a physical majorana phase in the mixing matrix when the neutrinos are introduced as majorana particles .
section iv contains a discussion of the mass matrix of the charged leptons when it is assumed that these particles acquire mass using just the standard higgs field .
there is no problem with the @xmath0 invariance allowing for a mass spectrum in which only the @xmath11 lepton is massive , as desired .
however , there is a problem with getting a trivial charged lepton mixing matrix , as needed to realize the tribimaximal form for the overall lepton mixing matrix . to solve this problem one might imagine backing up and allowing various row and column permutations for both the neutrino and charged lepton mixing matrices ( there are 36 x 36 possibilities ) and possible two dimensional rotations in the two dimensional degenerate subspaces of each in the hope that the product comes out to be of tribimaximal form .
this hope is not realized .
we give a simple proof that the tribimaximal form can not be obtained in this manner . in section
v , we point out a possible alternative treatment of the charged lepton masses and mixings which seems able to give in the first approximation , the desired trivial mixing matrix and non - zero mass only for the tau lepton . instead of requiring the right handed charged leptons to transform under @xmath0 we consider them to be singlets under @xmath0 so that the relevant discrete group becomes just @xmath12 . to construct an @xmath0 invariant yukawa term with the higgs fields
now requires that we introduce three different higgs doublets , one associated with each fermion family .
furthermore , the higgs potential ( which may involve other possible higgs fields too ) should allow for spontaneous breakdown of the @xmath0 symmetry in order to obtain the desired mass spectrum and trivial mixing .
as an application of the first approximation model with three approximately degenerate neutrinos we calculate , in this section , the leptonic factor for neutrinoless beta decay .
it depends on the value of the degenerate neutrino masses and the single majorana phase in the model .
the up and down quark masses and mixings are briefly discussed in section vi .
since we want to have masses only for the b and t quarks and trivial mixing in first approximation we use exactly the same model for the yukawa terms as in the case of the charged leptons , just mentioned .
note that only the left handed quark fields , not the right handed ones , are assigned to transform under @xmath0 . in section vii
, we briefly discuss the form of the characteristic matrices which reflect the invariance under any @xmath8 subgroup of @xmath0 .
these may be considered as plausible perturbations to give higher approximations to this model .
we count the number of parameters corresponding to the sum of the @xmath0 matrix and either one perturbation matrix or two perturbation matrices invariant under different @xmath8 subgroups . finally .
section viii contains a brief summary and discussion .
for a general orientation it may be useful to start from the part of the electroweak theory which contains just the minimal set of fundamental fermions and their interactions with the @xmath13 x @xmath14 gauge fields : @xmath15 where a=(1,2,3 ) are the generation indices , @xmath16 are the appropriate covariant derivatives for each term and @xmath17 , \hspace{.5 in } q_a= \left [ \begin{array}{c } u_{la } \\
d_{la } \end{array } \right ] .
\label{leftdoublets}\ ] ] the notation is conventional except that we are denoting @xmath18 as the two component neutrino field belonging to generation a. we will need the charged current leptonic weak interaction terms contained in this equation : @xmath19 where @xmath20 is the weak coupling constant and @xmath21 is the charged intermediate vector boson field .
the hatted fields correspond to mass eigenstates and are related to the original ones as : @xmath22 here , @xmath23 and @xmath24 are 3x3 unitary matrices .
finally the leptonic mixing matrix is defined as @xmath25 it is clear that the part of the lagrangian given in eq.([firstl ] ) ( which seems to be the most securely established ) possesses a global @xmath26 symmetry- one @xmath27 symmetry for each term .
subgroups of this very large group are candidates for extra symmetries which might be imposed when adding the higgs interaction terms and potential .
discrete symmetries have the feature that they do not necessarily imply the existence of new goldstone bosons when spontaneous breakdown to @xmath28 occurs . the discrete symmetry which is staring us in the face
is the generational permutation symmetry @xmath0 and has been discussed by many people .
of course , there are a number of possibilities for implementing this symmetry
. the simplest would be to have all the 5 @xmath0 s act in the same way on all the fermions . in the following
we will discuss other possibilities too .
the specific symmetry transformations take the form : @xmath29 where the permutation matrices
@xmath30 are the 6 matrices : @xmath31,\hspace{.3 in } s^{(12)}= \left [ \begin{array}{ccc } 0&1 & 0 \\ 1&0&0\\ 0&0&1 \end{array } \right],\hspace{.3 in } s^{(13)}= \left [ \begin{array}{ccc } 0&0&1 \\
0&1&0\\ 1&0&0 \end{array } \right ] , \nonumber \\ s^{(23)}&= & \left [ \begin{array}{ccc } 1&0&0 \\ 0&0&1\\ 0&1&0 \end{array } \right],\hspace{.3 in } s^{(123)}= \left [ \begin{array}{ccc } 0&0&1 \\ 1&0&0\\ 0&1&0 \end{array } \right],\hspace{.3 in } s^{(132)}= \left [ \begin{array}{ccc } 0&1&0 \\
0&0&1\\ 1&0&0 \end{array } \right ] , \label{s3matrices}\end{aligned}\ ] ] for our purpose here it will be sufficient to use this ( defining ) reducible representation of @xmath0 .
the simplest way to give masses to the three 2-component majorana neutrino fields is to add to the theory , in addition to the usual complex higgs doublet , a complex higgs triplet : @xmath32 . \label{triplethiggs}\ ] ] a vacuum expectation value for @xmath33 will then generate the neutrino mass terms : @xmath34 to begin with , @xmath35 is an arbitrary symmetric ( but not necessarily real ) 3x3 matrix . requiring @xmath0 invariance of the lagrangian demands invariance of this term under the transformation @xmath36 , where s stands for any of the permutation matrices in eq .
( [ s3matrices ] ) .
this requires : @xmath37=0 , \label{commutator}\ ] ] for all six matrices @xmath30 . by explicitly evaluating the commutators we obtain the solution : @xmath38+\beta \left [ \begin{array}{ccc } 1&1&1\\ 1&1&1\\ 1&1&1 \end{array } \right ] \equiv \alpha { \bf 1}+\beta d .
\label{solution}\ ] ] @xmath39 and @xmath40 are , in general complex numbers while @xmath41 is usually called the democratic " matrix .
we remark that the result of eq.([solution ] ) would be the same even for the case of dirac neutrinos , where the initial matrix is not required to be symmetric .
the form of eq .
( [ mnu ] ) indicates that the correct diagonalization " of @xmath35 should be : @xmath42 where @xmath43 is a real , diagonal matrix and @xmath23 is a unitary matrix .
let us perform this diagonalization while making contact with the historical background of the subject .
it is easy to verify that @xmath35 may be brought to diagonal ( but not necessarily real ) form by a real orthogonal matrix , @xmath44 as : @xmath45 .
\label{complexeigenvalues}\ ] ] the real matrix , @xmath44 is typically chosen as : @xmath46 \equiv \left [ \begin{array}{ccc } \vec{r}_1&\vec{r}_2&\vec{r}_3 \end{array } \right ] , \label{r}\ ] ] wherein we have explicitly shown how @xmath44 is constructed out of the three eigenvectors ( columns in @xmath44 ) of @xmath35 . in order to bring @xmath47 to real diagonal form we first define the ( majorana - type ) phases , @xmath48 and @xmath11 as : @xmath49 and the diagonal matrix @xmath50 . \label{pmatrix}\ ] ] now we note that the choice @xmath51 yields , as desired , real positive neutrino masses : @xmath52 .
\label{realeigenvalues}\ ] ] most previous workers on models of this type have considered @xmath47 to be a real matrix , which results in a little simplification .
let us discuss this case first .
it is clear from a purely mathematical standpoint , that the choice of matrix @xmath44 has a large amount of arbitrariness .
first , there is nothing a priori wrong with permuting the three eigenvectors , @xmath53 in eq .
( [ r ] ) in any way .
second , there is nothing wrong with permuting the three rows of @xmath44 in any way .
in fact , there is a larger freedom of making an arbitrary rotation in the two dimensional subspace of the degenerate eigenvectors @xmath54 and @xmath55 .
this arbitrariness leaves @xmath56 invariant up to its column placement .
this structure is a reflection of the possibility of decomposing the three dimensional representation of @xmath0 into its irreducible two and one dimensional pieces . to illustrate the rotational freedom , we may use , instead of @xmath44
, @xmath57 = \left [ \begin{array}{ccc } \vec{r}_1&\vec{r}_2&\vec{r}_3 \end{array } \right ]
\left [ \begin{array}{ccc } cos\xi&0&-sin\xi \\ 0&1&0 \\ sin\xi&0&cos\xi \end{array } \right ] .
\label{rotation}\ ] ] notice that @xmath56 is left invariant under this rotation .
furthermore , the matrix element , @xmath58 is no longer zero for general @xmath59 . the particular choice of @xmath44 written in eq .
( [ r ] ) is called the tribimaximal " matrix .
it seems to agree with present experimental indications when it is identified with the lepton mixing matrix , @xmath60 in eq .
( [ k ] ) .
( of course , the identification of @xmath44 with @xmath60 requires the assumption that @xmath24 is at least approximately the unit matrix . )
especially , the @xmath61 element is known to be small and could be zero
. however this model can not be exactly correct because it leads to two degenerate neutrino masses , which is ruled out since both @xmath62 ( from solar neutrino experiments ) and @xmath63 ( from atmospheric neutrino experiments ) are non - zero and not equal to each other .
thus the @xmath0 symmetry can only be a kind of first approximation to nature .
nevertheless it may be a good first approximation and hence interesting to study in detail . that is the point of view we will take here . in @xcite ,
harrison and scott have argued that , even if one accepts the predicted two - fold degeneracy , the @xmath0 model is still not a good first approximation because the two degenerate masses belong to columns 1 and 3 rather than columns 1 and 2 .
masses 1 and 2 are expected to be closer to each other because the solar neutrino mass difference is a lot smaller than the atmospheric neutrino mass difference . here
we would like to point out that this conclusion can be altered if we allow for complex values of @xmath39 and @xmath40 . in that case
, we can easily achieve a situation in which all three neutrinos are degenerate .
this is a valid first approximation to the neutrino spectrum for a relatively large range of neutrino masses .
one may notice this fact by adopting a typical choice @xcite of best fit neutrino mass differences : @xmath64 the uncertainty in these numbers is roughly 25 per cent .
from these two numbers we may - with a two fold ambiguity due to the unknown sign of @xmath65- determine masses @xmath66 and @xmath67 for a given choice of @xmath68 .
we denote the type i solution to be the case when @xmath68 is greater than @xmath66 and @xmath67 and the type ii solution to be the case when @xmath68 is less than @xmath66 and @xmath67 .
note that there is an upper limit on @xmath68 corresponding to the cosmological bound from structure formation @xcite : @xmath69 the results are shown as table [ typicalsolutions ] , parts of which have already appeared in refs .
it should be remarked that the number of digits given for the neutrino masses are only for comparing them with each other .
they clearly do not reflect the uncertainties of the experimental determinations in eq.([massdifferences ] ) .
.typical solutions for @xmath70 as @xmath68 is lowered from about the highest value which is experimentally reasonable . in the type i solutions @xmath68 is the largest mass while in the type ii solutions @xmath68 is the smallest mass . [ cols="^,^,^,^",options="header " , ] in the large range 0.3 ev @xmath71 0.15ev , degeneracy of the three neutrino masses is a good first approximation to the spectrum .
it is clearly less good in the lower range 0.1 ev @xmath71 0.001 ev . incidentally , it is seen that only in the small range just above @xmath720.0517 ev does one have anything like a usual hierarchy .
an amusing feature is that the viability of @xmath0 symmetry as a good first order approximation selects a range of favored neutrino masses . actually , at first glance it may not be apparent how eq .
( [ realeigenvalues ] ) allows for a degenerate mass spectrum . to achieve an
approximately degenerate mass spectrum we impose : @xmath73 where @xmath74 is a real number which may be as small as one likes .
a picture of this equation with @xmath75 , where @xmath39 and @xmath76 are displayed as 2 dimensional vectors making up the equal sides of an isosceles triangle is given in fig.[triangle ] .
the angle , @xmath77 between them must obey ( to first order in @xmath74 ) : @xmath78 it is interesting to contrast this solution with one in which the coefficient @xmath40 in eq.([solution ] ) vanishes . then the three neutrino masses are obviously degenerate but @xmath44 in eq.([complexeigenvalues ] ) can be any orthogonal matrix .
however , in the present case , for any finite @xmath74 no matter how small , @xmath44 can have at most a two dimensional degenerate subspace.we will see that there is a physical effect which persists for small @xmath74 .
since only experiment can tell us eventually whether neutrinos are of majorana or dirac type , it seems worthwhile to show that the same results may be obtained in case the neutrinos are of dirac type . in that case
we have the two fields @xmath79 and @xmath80 so the mass term becomes @xmath81 .
the diagonalization equation which replaces eq.([correct ] ) is : @xmath82 if , for example , both @xmath83 and @xmath80 transform as triplets under the same @xmath0 , we will again have the same form of @xmath47 as given in eq .
( [ solution ] ) .
the bi - diagonalizing matrices then become @xmath84 where @xmath44 is given in eq.([r ] ) and @xmath85 is given in eq.([pmatrix ] ) .
this replaces eq.([defineu ] ) so the neutrino contribution to the overall lepton mixing matrix is just the same . in this case , however , the extra phases , @xmath85 in @xmath86 may be transformed away .
there is no majorana phase and , of course , no neutrinoless double beta decay .
nevertheless , the complex phase plays a role in the mass eigenvalues , allowing all three neutrinos to be degenerate with non zero @xmath40 .
the charged leptons may simply obtain their masses using the conventional complex doublet higgs field , @xmath87 , \label{usualhiggs}\ ] ] in the lagrangian term , @xmath88 where @xmath89 and @xmath90 . here ,
@xmath91 is the pre - diagonal charged lepton mass matrix .
this matrix may , in general , be brought to a real diagonal form @xmath92 by the bi - unitary transformation , @xmath93 where , @xmath94 this leads to the mass diagonal , hatted , fields : @xmath95 so far we havent discussed the imposition of @xmath0 symmetry on @xmath91 . because the fields @xmath18 and @xmath96 both belong to the same @xmath13 doublet we should , to maintain the @xmath13 symmetry , require @xmath96 to transform in the same manner as @xmath18 .
however @xmath97 is not required to do so .
nevertheless that is the obvious first guess so let us initially require invariance of the lagrangian under both transformations in eq .
( [ symmtransf ] ) .
then we must also impose @xmath98=0 , \label{meform}\ ] ] for all six permutation matrices .
as before , this has the solution , @xmath99 where @xmath41 is the previously defined democratic matrix while @xmath100 and @xmath101 are two complex numbers .
@xmath92 will then have have the doubly degenerate real positive eigenvalue @xmath102 as well as the non degenerate eigenvalue @xmath103 .
we define phases @xmath104 and @xmath105 by , @xmath106 if these phases are used to define a diagonal matrix , @xmath107 of the form , @xmath108 \label{pprime}\ ] ] we can satisfy eq.([biunitary ] ) with the identifications , @xmath109 where @xmath110 is a real orthogonal matrix of the generalized ( as discussed around eq .
( [ rotation ] ) ) tribimaximal type .
we remark that it is not necessary for the ordering of the eigenvalues in @xmath92 ( and correspondingly in eq.([pprime ] ) to be the same as in @xmath43 .
as discussed above , we can also permute the rows of @xmath110 and make an arbitrary rotation in the degenerate subspace . a reasonable first approximation to the charged lepton mass spectrum
would seem to consist of a single massive @xmath111 and two massless ( or small mass ) others ( @xmath112 and @xmath113 ) .
this can easily be achieved by setting @xmath100 to zero and identifying the third eigenvalue to be the non - degenerate one . however , it is not so easy to get the correct lepton mixing matrix which , using eq.([k ] ) , takes the form , @xmath114 note that the phase matrix @xmath115 can be eliminated by a rephasing of the field @xmath116 , which sits to the left of it in the physical interaction term , eq.([lcc ] ) .
the question is whether the product @xmath117 can have anything like the tribimaximal form .
ideally , one might like to keep @xmath44 as the tribimaximal matrix and arrange for @xmath110 to approximate the unit matrix .
this could be achieved if the complex number @xmath101 in eq .
( [ solveme ] ) could be made zero .
however we have seen that it is necessary for @xmath100 to be zero in order to give a realistic first approximation to the charged lepton mass spectrum .
clearly , having @xmath118 is not viable .
then , @xmath110 can not be the unit matrix but must be of generalized tribimaximal type .
it is not expected that @xmath110 would be the same as @xmath44 since we have seen that a different ordering of mass eigenvalues is appropriate . in general
there are 36 x 36 discrete possibilities for the matrix @xmath119 corresponding to 6 possible row exchanges and six possible column exchanges for each and also the possibilities of two independent rotations in the two degenerate subspaces . at first , it seems to be a daunting task to study them .
nevertheless we will demonstrate a simple way to see that there are no acceptable solutions that would give a resulting @xmath60 anything like the tribimaximal form desired . for this task
it is convenient to represent each of the relevant matrices as a single row , having elements which are column vectors as in eqs .
( [ r ] ) and ( [ rotation ] ) . to anchor the notation we introduce unit vectors : @xmath120 , \hspace{.15 in } \vec{n}_2=\left [ \begin{array}{c } 0\\1\\0 \end{array } \right ] , \hspace{.15 in } \vec{n}_3=\left [ \begin{array}{c } 0\\0\\1 \end{array } \right ] .
\label{unitvectors}\ ] ] now the matrices @xmath121 $ ] and @xmath122 $ ] may be written as row vectors with indexed column vector elements according to : @xmath123 from the first of these equations we may write @xmath124 . substituting this into the second of eqs .
( [ compactnotation ] ) we find @xmath125 , which may be more compactly written as : @xmath126 here @xmath127 is the non - phase part of the lepton mixing matrix .
in other words , the lepton mixing matrix is simply displayed as the transformation between the columns of @xmath44 and @xmath110 .
the key point for our present purpose is that one of the @xmath128 and one of the @xmath129 must be : @xmath130 .
\label{invariantvector}\ ] ] this is seen to be the case for any generalized tribimaximal matrix which diagonalizes permutation invariant mass matrices in eq .
( [ rotation ] ) .
it is also clear that an arbitrary permutation of rows in the diagonalizing matrix does not change this vector . without any loss of generality
let the invariant vector in eq.([invariantvector ] ) occur as @xmath131 and @xmath56 .
then eq.([ktilde ] ) becomes , @xmath132 since the three vectors on the right hand side are linearly independent , we conclude that @xmath133 and in addition @xmath134 . similarly , the inverse relation @xmath135 yields again @xmath133 and also @xmath136 .
thus the matrix @xmath137 must have the form , @xmath138 , \label{ksofar}\ ] ] where the x s stand for elements yet to be determined . since @xmath137 is a real orthogonal matrix we can finally write : @xmath139 , \label{finally}\ ] ] where @xmath140 and @xmath141 stand for the sine and cosine of some angle .
the key point is that there are four zero elements of the matrix and one independent parameter .
therefore it represents , up to some possible reflections and relabeling of axes , a rotation in a two dimensional subspace .
this structure clearly does not depend on the choices of @xmath129 and @xmath142 we chose to identify with the invariant vector of eq.([invariantvector ] ) .
the tribimaximal matrix of eq.([r ] ) and the generalized tribimaximal mixing matrix of eq.([rotation ] ) can not be obtained from this too simple form . to sum up the work so far
, we have seen that @xmath0 symmetry based on the transformations properties for @xmath143 and @xmath97 as given in eq.([symmtransf ] ) has some very encouraging features for a first approximation to lepton properties . first , the neutrino mass spectrum is consistent with three approximately degenerate masses .
in addition , the predicted neutrino mixing matrix is consistent with the tribimaximal matrix , as experiment suggests .
also , the charged lepton mass pattern can be made consistent with a heavy @xmath111 and zero mass @xmath113 and @xmath112 .
things would be fine if the charged lepton mixing matrix could be made consistent with the unit matrix so that , as in eq.([k ] ) , the lepton mixing matrix , @xmath60 could be identified with the tribimaximal form ( up to the possibility of majorana phases ) .
unfortunately , we have just shown that this is not possible .
we will now investigate a modification of the field transformation properties which looks able to solve this problem .
for clarification we remark that the problem above would not have arisen if we had not assumed the restriction on the @xmath0 invariant form of the charged lepton spectrum that only the @xmath11 particle has mass .
if we could tolerate a first approximation in which @xmath144 are all degenerate , then the choice of @xmath145 in eq.([solveme ] ) would work .
however that does not seem physically reasonable .
as mentioned around eq.([symmtransf ] ) , there is some freedom to modify the type of the @xmath0 transformations we assume . in the last section we assumed that @xmath143 and @xmath97 transform in the same way so we were identifying the @xmath12 and @xmath146 transformations .
now let us assume that @xmath97 does not transform under @xmath12 ; i.e. , it transforms as a singlet under @xmath12 .
then , in order to construct an invariant charged lepton mass term we need to assume that there are three higgs doublets which belong to the ( reducible ) defining representations of @xmath12 : @xmath147 , \label{phia}\ ] ] where @xmath148 .
the charged lepton masses may then arise from a term , @xmath149 where @xmath150 is a matrix of coupling constants whose form will be restricted by the @xmath12 symmetry ( the @xmath146 symmetry is not being implemented now ) .
the charged lepton mass matrix is : @xmath151 so that the charged lepton mass term in the lagrangian looks conventional : @xmath152 as before , the @xmath12 invariance demands that the three matrices , @xmath153 have the forms : @xmath154 wherein the six quantities @xmath155 and @xmath156 are all undetermined complex numbers .
then the predicted form for @xmath91 is : @xmath157 , \label{bigme}\ ] ] where we have denoted the sum of the three vacuum values as @xmath158 .
our goal is to see if eq.([bigme ] ) can yield a mass spectrum with a single massive charged lepton and the unit matrix for @xmath24 , the charged lepton mixing matrix defined in eq.([uandw ] ) .
this can be simply achieved if first , the three arbitrary constants @xmath159 are all set to zero together with the two arbitrary constants @xmath160 and @xmath161 .
second , since just the bottom row remains , we must have a spontaneous breakdown structure in the higgs potential which results in the vanishing of the expectation values @xmath162 and @xmath163 but not @xmath164 .
the @xmath111 mass then becomes , in the first approximation limit , @xmath165 .
the higgs potential appropriate for this model may be rather complicated since many additional higgs fields could very well be present .
one characteristic term illustrating how a basic permutation invariant quadratic form appears is : @xmath166 \left[\phi_e^{\dagger}(\epsilon\delta_{ef } + \zeta{d_{ef}})\phi_f - k_2\right]^*+\hspace{.15 in } \cdots , \label{higgsv}\ ] ] where @xmath167,@xmath168,@xmath74 and @xmath169 are constants while @xmath170 stands for elements of the democratic matrix .
so it seems that by using , instead of a single higgs doublet , three higgs doublets linked to the three leptonic families by @xmath12 invariance , we can obtain a suitable first approximation to the lepton mass spectrum and to the leptonic mixing matrix @xmath171 . specifically , the charged leptons have a trivial mixing matrix and only the heaviest one is massive . on the other hand , in this same approximation ,
the three neutrino masses are approximately degenerate and have a mixing matrix which can be taken to be of the tribimaximal form , eq.([defineu ] ) or its generalization to include a rotation in the two dimensional degenerate subspace . associated with this model in the case where the neutrinos are formulated as majorana particles , is the existence of the majorana phase @xmath172 defined in eq.([psi ] ) . to illustrate the importance of this cp violating majorana phase we calculate the characteristic leptonic factor @xmath173 for neutrinoless double beta decay in the present model with three approximately degenerate neutrinos having mass , @xmath174 .
the general formula is : @xmath175 where the lepton mixing matrix elements are obtained by identifying @xmath60 with @xmath23 given in eqs.([defineu ] ) , ( [ pmatrix ] ) and ( [ r ] ) .
this yields with @xmath74 negligible , @xmath176 it is easy to check that this result would not change if we computed the matrix elements @xmath177 using the rotated form of @xmath44 given in eq.([rotation ] ) rather than the one in eq.([r ] ) .
clearly @xmath173 obeys the inequality , @xmath178 the present experimental bound on this quantity is @xcite , @xmath179 which is not much greater than the range of m for which the approximate degeneracy of all three neutrinos holds .
thus this kind of model may be tested in the next few years .
the up and down quark mass matrices clearly have a reasonable first approximation in which the heaviest of each is massive while the other two have no mass . in the same approximation
it is reasonable to have a unit mixing matrix .
then the situation for each is the same as for the charged leptons , as just discussed .
it thus seems natural to use the same setup as for the charged leptons , with three higgs fields transforming under @xmath12 .
the quark mass terms in the lagrangian would take the form , @xmath180 which yields the up and down quark mass matrices as : @xmath181 the predictions of @xmath0 invariance are @xmath182 as in the charged lepton case the first approximation to the quark mass spectrum yields , @xmath183 while the other quark masses remain zero .
for completeness we mention that the up and down mass matrices in the general case are brought to diagonal forms by the bi - unitary transformations : @xmath184 this leads to the diagonal ( hatted ) states , @xmath185 and the hadronic part of the charged current weak interaction , @xmath186 here @xmath187 is the quark mixing ( or ckm ) matrix . in the first approximation under consideration @xmath188
next let us briefly consider the perturbations which might be employed to improve the first approximation fermion spectra and mixing matrices .
for this purpose , we first summarize the general form of the @xmath0 invariant matrices which appear in eqs .
( [ solution ] ) , ( [ solveme ] ) , ( [ gc ] ) and ( [ aandb ] ) as : @xmath189 . \label{s3type}\ ] ] this matrix contains two complex parameters . as we have seen the non - reality plays an important role in obtaining the proposed first approximation to the neutrino mass spectrum
of course , the parameters will be different for fermions of each electric charge .
the assumption of a grand unified theory would give additional relations among them .
note also that the @xmath0 invariant matrix and possible perturbations of it directly correspond to the the mass matrix for the assumed majorana neutrinos but not for the charged leptons and quarks . in the latter cases ,
the effects of the three higgs mesons being assumed must be folded in as in eqs.([3phime ] ) and ( [ quarkmasses ] ) .
an additional difference is that the matrix for the majorana neutrino masses must be a symmetric one .
if we assume that the perturbations are invariant under only an @xmath8 subgroup of @xmath0 , there are three choices .
these correspond to two element subgroups containing , besides the identity element , @xmath190 , @xmath191 or @xmath192 .
requiring , a general matrix , @xmath193 to commute with these separately gives , respectively the @xmath8 invariant forms : @xmath194 , \hspace{.2 in } \left [ \begin{array}{ccc } a''&c''&b '' \\
\\ b''&c''&a '' \end{array } \right ] , \hspace{.2 in } \left [ \begin{array}{ccc } e'''&d'''&d ' '' \\ c'''&a'''&b ' '' \\ c'''&b'''&a ' '' \end{array } \right ] .
\label{s2matrices}\ ] ] each of these possible perturbation matrices contains five complex parameters . in case of application to the majorana neutrino masses , there will be only four complex parameters since we have the corresponding symmetric structures : @xmath195 , \hspace{.2 in } \left [ \begin{array}{ccc } a''&c''&b '' \\
\\ b''&c''&a '' \end{array } \right ] , \hspace{.2 in } \left [ \begin{array}{ccc } e'''&c'''&c ' '' \\ c'''&a'''&b ' '' \\ c'''&b'''&a ' '' \end{array } \right ] .
\label{s2majoranamatrices}\ ] ] the simplest perturbation scheme would consist of the original form , eq.([s3type ] ) plus one of the three matrices in eq.([s2matrices ] ) [ or eq.([s2majoranamatrices ] ) for the majorana neutrino case ] .
each of the three possible sums can be seen to have five [ four ] complex parameters , as one can absorb the two @xmath0 invariant matrix parameters in the parameters of the perturbing matrix .
presumably , one of these three possibilities would be the best .
the perturbation scheme which might be the closest analog of the generalized gell - mann okubo type mentioned in the introduction would contain the @xmath0 invariant part together with two of the matrices in eq.([s2matrices ] ) [ eq.([s2majoranamatrices ] ) ] .
it can be seen that one would get the same number , eight [ six ] , of complex parameters regardless of which set of two is selected .
in fact , regardless of which set is chosen , the resulting sum of three matrices would be the same . in the case of the majorana neutrino matrix
, the six complex parameters already comprise the most general matrix . for the charged fermion matrices ,
nine parameters are needed for the most general possibility ; however , all of the perturbation matrices satisfy the relation : @xmath196 thereby eliminating one complex parameter .
clearly , there is no point in considering a perturbation made as the sum of all three matrices in eq.([s2matrices ] ) .
the most practical second approximation would seem to involve choosing just one of the matrices of eq.([s2matrices ] ) [ eq.([s2majoranamatrices ] ) ] . a promising choice of perturbation , where the @xmath8 subgroup includes the element @xmath192 has been investigated by a number of authors @xcite .
for a long time the permutation invariance , @xmath0 of the three known generations has been considered a very natural assumption for understanding the non - trivial spectrum of fundamental fermion masses and their associated mixings .
recently , experimental data on neutrino oscillations have pointed to a leptonic mixing matrix which is of the so called tribimaximal form ; this is in essence the transformation from the natural three dimensional basis of @xmath0 to the irreducible basis and therefore can be regarded as another motivation for this group .
we have attempted to spell out in detail a possible way in which this attractive @xmath0 permutation invariance can be consistently implemented in the standard electroweak theory .
we started by reviewing the fact that exact @xmath0 invariance is not correct and suggesting that the situation might be analogous to an older one where a symmetry group was very plausible but the exact dynamics were unknown
the su(3 ) symmetry of the three quark model .
a similar proposal is that the @xmath0 invariance holds for a dominant piece but that there are other ( presumably relatively small ) pieces which preserve different @xmath8 subgroups of @xmath0 . in this paper we concentrated on looking at the first approximation of exact @xmath0 invariance .
we showed , by comparison with results of analysis of neutrino oscillation experiments ( see table i ) that there is a large range of allowed neutrino mass sets where near degeneracy of all three neutrino masses is a very good approximation .
so we regarded a near degenerate neutrino mass spectrum together with a tribimaximal type neutrino mixing matrix as the @xmath0 invariant form to be obtained . at the same time we considered that the @xmath0 invariant structures for the charged leptons and quarks should contains masses for the ( @xmath197 ) states and zero masses for the others .
the @xmath0 invariant mixing matrices for the charged fermions are all considered to be just the unit matrix ( no mixing to leading order ) .
it is not completely trivial to obtain a model which achieves this desired spectrum . to get it
, we included the features of i ) majorana type cp violation , ii ) @xmath0 invariance for only the left fermions , iii ) inclusion of three higgs doublets linked to the three generations and transforming under @xmath0 and iv ) spontaneous breakdown of @xmath0 due to only one of those three doublets developing a vacuum value .
the motivation is given in section iv in which it is shown that , very generally , one can not get a tribimaximal form for the overall lepton mixing matrix together with the given mass spectrum of the charged leptons in the simplest way one would imagine .
a physical prediction of the model when the neutrinos are assumed to be of majorana type is obtained for the neutrinoless double beta decay factor @xmath198 already in the @xmath0 invariant limit ( see eq .
( [ ourmee ] ) ) .
this gives a bound which should be testable in the relatively near future if the three neutrinos belong to the energy range where they are approximately degenerate .
one natural extension of this approach would be to look at its embedding in a grand unified theory .
the next step would be to investigate in detail the case when a perturbation breaking the symmetry down to an @xmath8 subgroup is included .
as discussed in section vii , this would introduce two additional complex parameters for the neutrino mass matrix .
-.5 cm we would like to thank lincoln wolfenstein for a helpful comment on the manuscript and together with samina masood and susumu okubo for helpful discussions . the work of r.jora and j.schechter has been supported in part by the us doe under contract no .
de - fg-02 - 85er 40231 .
s.n is supported by national science foundation ( nsf ) grant no .
phy-0354401 .
in addition to the papers just cited , earlier references are s. pakvasa and h. sugawara , phys .
b * 73 * , 61 ( 1978 ) ; * 82 * , 105 ( 1979 ) ; e. derman and h.s.tsao , phys . rev .
d * 20 * , 1207 ( 1979 ) and y. yamanaka , h. sugawara and s. pakvasa phys .
d * 25 * , 1895 ( 1982 ) .
m. maltoni , t. schwetz , m. a. tortola and j. w. f. valle , phys .
d * 68 * , 113010 ( 2003 ) . for
the choice @xmath199 see a. de gouvea , a. friedland and h. murayama , phys .
b490 * , 125 ( 2000 ) . for recent reviews
see , for examples , s. pakvasa and j. w. f. valle , special issue on neutrino , proc .
indian natl .
part a * 70 * , 189 ( 2004 ) ; arxiv : hep - ph/0301061 and v. barger , d. marfatia and k. whisnant , int . j. mod
e * 12 * , 569 ( 2003 ) ; arxiv : hep - ph/0308123 ; r. n. mohapatra and a. y. smirnov , to appear in ann .
science , * 56 * ( 2006 ) .
s. nasri , j. schechter and s. moussa , phys . rev . *
d70 * , 053005 ( 2004 ) ; s.s .
masood , s. nasri and j. schechter , phys .
d * 71 * , 093005 ( 2005 ) ; int .
a * 21 * , 517 ( 2006 ) .
graphical representation of this information may be found in s. m. bilenky , c. giunti , j.a .
grifols and e. masso , phys . rep . *
379 * , 69 ( 2003 ) .
t. fukuyama and h. nishiura , hep - ph/9702253 ; r. n. mohapatra and s. nussinov , phys . rev . *
d 60 * , 013002 ( 1999 ) ; e. ma and m. raidal , phys . rev . lett . * 87 * , 011802 ( 2001 ) ; c. s. lam , hep - ph/0104116 ; t. kitabayashi and m. yasue , phys.rev . *
d67 * 015006 ( 2003 ) ; w. grimus and l. lavoura , hep - ph/0305046 ; 0309050 ; y. koide , phys.rev . *
d69 * , 093001 ( 2004);y .
h. ahn , sin kyu kang , c. s. kim , jake lee , hep - ph/0602160 ; a. ghosal , hep - ph/0304090 ; w. grimus and l. lavoura , phys .
b * 572 * , 189 ( 2003 ) ; w. grimus and l. lavoura , j. phys .
g * 30 * , 73 ( 2004 ) .
w. grimus , a. s.joshipura , s. kaneko , l. lavoura , h. sawanaka , m. tanimoto , hep - ph/0408123 ; r. n. mohapatra , jhep , * 0410 * , 027 ( 2004 ) ; a. de gouvea , phys.rev .
* d69 * , 093007 ( 2004 ) ; r. n. mohapatra and w. rodejohann , phys . rev . *
d 72 * , 053001 ( 2005 ) ; t. kitabayashi and m. yasue , phys .
b 621 * , 133 ( 2005 ) ; r. n. mohapatra and s. nasri , phys .
d * 71 * , 033001 ( 2005);r .
n. mohapatra , s. nasri and h. b. yu , phys.lett .
b * 615 * , 231 ( 2005 ) . | the form of the leptonic mixing matrix emerging from experiment has , in the last few years , generated a lot of interest in the so - called tribimaximal type .
this form may be naturally associated with the possibility of a discrete permutation symmetry ( @xmath0 ) among the three generations . however , trying to implement this attractive symmetry has resulted in some problems and it seems to have fallen out of favor .
we suggest an approach in which the @xmath0 holds to first approximation , somewhat in the manner of the old @xmath1 flavor symmetry of the three flavor quark model . it is shown that in the case of the neutrino sector , a presently large experimentally allowed region can be fairly well described in this first approximation .
we briefly discuss the nature of the perturbations which are the analogs of the gell - mann okubo perturbations but confine our attention for the most part to the @xmath0 invariant model .
we postulate that the @xmath0 invariant mass spectrum consists of non zero masses for the @xmath2 and zero masses for the other charged fermions but approximately degenerate masses for the three neutrinos .
the mixing matrices are assumed to be trivial for the charged fermions but of tribimaximal type for the neutrinos in the first approximation .
it is shown that this can be implemented by allowing complex entries for the mass matrix and spontaneous breakdown of the @xmath0 invariance of the lagrangian . |
since their first detection by cobe , the cmb temperature fluctuations have become an essential tool for constraining cosmological parameters . from the beginning of 2000 , new experiments have released data set of good quality up to the third acoustic peak .
better constraints have been obtained on several cosmological parameters .
nevertheless , it has been shown that even with precise measurements of the power spectrum , it is nearly impossible to distinguish models with the same physical parameters on the last scattering surface . basically , some degeneracies are inherent to the cmb . we consider in this work x
ray clusters as an independent way for constraining cosmological parameters .
in the present analysis we consider the data from cobe , boomerang , maxima and dasi .
we analysed the likelihood of inflationary models with @xmath0 , @xmath1 as free cosmological parameters .
this corresponds to @xmath2 models tested.to derive the likelihood values for the models we considered , we used the method developed in bartlett et al .
( 2000 ) , and already used in le dour et al .
( 2000 ) .
the results are presented as two - dimensional contours plots of the likelihood projected onto various parameters planes .
this 2-d presentation has the advantage of showing clearly all the degeneracies between parameters .
the left contour plots of figure 1 show some degeneracies between our parameters .
basically , @xmath3 , @xmath4 and @xmath5 are strongly degenerated if only cmb data are used in parameter estimation .
this means that one should consider another source of constraints , independent from cmb observations , for a better cosmological parameter estimation .
we choose x - ray clusters as additional constraints .
clusters are interesting object for which both the luminous baryonic mass ( x - ray emitting intracluster gas ) and the total gravitating mass can be determined .
therefore , an upper limit on the fraction baryon , @xmath6 , can be estimated .
results are not necessary in agreement allowing values between @xmath7 and @xmath8 . in this work ,
we consider the constraints given in sadat and blanchard 2001 @xcite for the baryon fraction : @xmath9 .
the aim of this contribution being to show that clusters may help for breaking degeneracies shown by cmb parameters estimations , we did not consider constraints given by other groups .
a better consideration of these different results on x - ray clusters will be found in douspis et al .
( 2001b ) .
given the constraints of cmb and x - ray clusters , one could combined the two by multiplying the corresponding likelihood . the right figure of fig .
1 shows the constraints from the combined analysis in the same plane as the left figure .
we can see that the degeneracies are broken .
confidence intervals are now determined for @xmath3 and @xmath4 . for both parameters ,
low values are preferred : @xmath10 and @xmath11 at 99% cl . due to inherent degeneracies in the cmb
it is nearly impossible to specify some of the cosmological parameter ; `` cross constraints '' are then necessary .
this work is thus a preliminary view of the kind of constraints one would obtain in a near future , using last x ray satellite data , and in a less near future the cmb satellite data ( map and planck ) . | we present the results of a combined analysis of cosmic microwave background ( cmb ) and x - ray galaxy clusters baryon fraction to deduce constraints over 6 inflationnary cosmological parameters .
such a combination is necessary for breaking degeneracies inherent to the cmb .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
it is not commonly known , but nevertheless well established since the work of khalfin in 1958@xcite , extended by numerous other authors in the intervening years ( i.e. @xcite that for times long compared to the characteristic decay time of a metastable quantum state , the decay of such states is no longer described by an exponential , but rather by a power law .
one of the reasons that this interesting fact is perhaps not more well recognized is that the characteristic time - scale determined for departures from exponential decay in several sample systems is generally so long that for all intents and purposes all unstable states in any finite system will have decayed before this onset would become relevant .
however , if one generalizes the results from quantum mechanics to quantum field theory then the nature of false vacuum decay can altered , with potentially significant implications for cosmology , as we describe here .
we begin by exploring the generic features of late - time decay , examining the transition from quantum mechanics to quantum field theory in order to explore those implications that are independent of the details of the relevant potential or model , in order to consider their impact upon the dynamics of cosmological false vacuum decay .
we find that the time for departures from exponential growth should depend logarithmically on ratio of the energy gap and the decay width with a prefactor of order unity and then explore the cosmological implications of this .
several open questions are then raised , including whether gravitational effects may alter the long time non - exponential tail in vacuum decay , and also whether cosmological observation within a given metastable universe may alter its lifetime .
in 1958 khalfin @xcite gave a general argument for why the exponential decay of an unstable state breaks down at late times by employing the payley - wiener theorem @xcite to show that if the energy distribution function is bounded below , then its fourier transform , the survival amplitude , necessarily leads to a power - law decay .
( notably the breit - wigner form of the energy distribution function is not bounded and does lead to exponential decay ) .
a key question of course is when the deviation from exponential decay will begin . while this is in general a model - dependent question , it turns out , as we demonstrate by utilizing three different @xcite , but particularly useful formalisms of deriving the power law decay at large times , that the overall time - scale has various model - independent features .
first , following @xcite we consider an initial unstable state , @xmath0 , for which the survival probability @xmath1 is given by @xmath2 where @xmath3 is the hamiltonian .
@xmath1 is equal to the square of the survival amplitude @xmath4 which may be written as the fourier transform of the energy distribution function , @xmath5 @xmath6 here we will assume that the energy distribution function is bounded below so that there exists a lowest @xmath7 such that @xmath8 for @xmath9 and thus the paley - wiener theorem applies which will insure a non - exponential decay at large times .
the survival amplitude @xmath4 can be represented using a contour integral in the complex @xmath10-plane ( see figure 1 ) : @xmath11 where @xmath12 is given in terms of the resolvent @xmath13 as @xmath14 .
there is a cut along the spectrum of @xmath3 which is taken for simplicity to be given by @xmath15 along the real axis . upon analytic continuation to the second sheet , @xmath16 , which is analytic in the cut plane and has no zeros there , contains poles which give rise to the exponential decay law as well as corrections which lead to the power law decay at large time .
the resonance condition requires a pole at @xmath17 , with @xmath18 .
physically @xmath19 is the energy difference between the unstable state and the state into which it decays . in order to relate this expression for the survival amplitude to previous studies ,
it is simpler to work with the inverse of @xmath16 , @xmath20 .
@xmath21 may be written in terms of dispersion relations whose exact form depends on the asymptotic behavior of @xmath22 .
it is useful to frame this problem in terms of a formalism that will later be useful for thinking about field theory .
one thus can consider a density of states factor @xmath23 defined in terms of @xmath22 : @xmath24 is a phase - space weight factor . now
@xmath4 may be written as @xmath25 then using that @xmath26 ( which is also true in the relativistic case ) @xmath27 one can see that @xmath28 at large times ( and the survival probability goes as @xmath29 ) .
to return to the key issue of when the turnaround between exponential and power - law behavior , one must be more specific in calculating @xmath4 .
a variable change @xcite simplifies the @xmath4 integration . defining @xmath30 , then due to the resonance condition and the jump discontinuity across the branch cut
, @xmath21 has poles on the second sheet at @xmath31 ( where @xmath18 ) which , in terms of @xmath32 occur approximately at @xmath33 where @xmath34 one can then show the zero structure of @xmath21 ( or the pole structure of @xmath4 ) more clearly as @xmath35 , which places all model dependence into the function @xmath36 . using this , @xmath4
can now be written as a sum of an exponential decay piece as well as a background integral term . in @xcite two specific models for @xmath36 were studied , and interestingly in both cases the crossover time was precisely the same , and given by : @xmath37 the fact that the logarithm of the ratio @xmath38 appears in this result is not surprising , as these are the only two dimensional parameters that appear in the general problem and the crossover time will be based on a comparison of the magnitude of an exponential decay rate and a power law decay rate .
the question that remains is to what extent the pre - factor arises from model - dependent features , and how things may generalize in quantum field theory . in order to explore this ,
it is useful to examine the same result within other formalisms , . in @xcite
a hamiltonian approach was taken where one writes the full hamiltonian as a sum of terms , and considers a basis of incoming states that diagonalize part of the hamiltonian , treating the remainder as a perturbation . in this case by expanding a wavefunction in this diagonal basis one can calculate the time dependence of the coefficients . the time dependence is in turn given by the integral over matrix elements between the initial state and all states with nearby energy , with the time dependence of the amplitude being given by : @xmath39 where @xmath40 introducing the function @xmath41 which is the fourier transform of @xmath42 @xmath43 the survival amplitude of a state is controlled solely by the integral of the function @xmath41 .
a specific example for @xmath41 appropriate for a particle in three dimensions then allows an explicit calculation of the cross - over time @xcite which is comparable to that given in the previous analysis , and in particular again depends logarithmically on @xmath44 .
presenting these results in this way makes the decay rate calculation reminiscent of a wkb calculation , where an integral over the action of the barrier between states plays a key role , and where semi - classical steepest descent estimates are often used in the calculation of false vacuum decay in field theory .
this approach , at least in the context of a simple quantum mechanical model , was used to calculate non - exponential decay at large times in @xcite . for the simple case of a particle trapped in a square potential well the euclidean ( imaginary time in minkowski space )
kernel was calculated and at large times was shown to exhibit a power law behavior @xmath45 , which could then used to approximate the cross - over time with precisely the same prefactor as that given in ( [ large time ] ) . with a euclidean time formalism for calculating late time
decay one can straightforwardly extrapolate to the case of false vacuum decay in field theory .
we now address various issues associated with this extrapolation .
\(i ) the examples above involved single particles in one or three dimensions and one may thus question whether the power law dependence derived there is appropriate in a field theoretical context . to address this question
one can make use of a heuristic analysis first applied to false vacuum decay @xcite , which reproduces the above result and also illuminates the origin of the power law dependence in a way that directly applies to the field theory calculation .
for tunneling from a barrier in one dimension there is an exact solution to the time - dependent schrodinger equation , but it is not normalizable , and is therefore only an approximation to the true physical wavefunction .
a wavepacket originally centered at the local minimum of the potential energy will initially exponentially decay , but the outgoing wavepacket to the right will spread as it evolves .
thus there will be some backflow of probability at the origin .
the spread will grow linearly with time @xmath46 , and the density of the packet , extrapolating to @xmath47 dimensions will fall like @xmath48 .
in the field theory case one extrapolates to an infinite number of degrees of freedom .
however , as long as the euclidean bounce solution has only one negative eigenvalue @xcite , the number of effective degrees of freedom associated with the wave - packet spread effectively reduces to the single particle case , with the the appropriate dimensions in the late time behavior of decay being the dimensionality of space .
\(ii ) the effect of going from quantum mechanics to field theory is to include , among other things , the need to consider volume factors so that one calculates survival probabilities per unit volume per unit time . in the standard false vacuum decay calculation , for example
, the fact that euclidean time solutions ( instantons ) can be placed throughout the full volume of space time is of importance .
however , since the same volume factors should appear in both the estimate of the early and late time decay rate estimates , it is reasonable to assume that the crossover time estimate derived above will apply in this case to survival probabilities per unit volume . in particular , recognizing that both the exponential and non - exponential decay rates depend upon the same integral and phase space factors as described above , we infer that the result in ( [ large time ] ) should give a quantitative estimate of the crossover time in which false vacuum decay rate per unit volume converts from exponential to power law , namely : @xmath49 this implies that the survival probability ( per unit volume ) at this time would be given by : @xmath50 we shall use this estimate in the discussions that follow .
typically for laboratory systems one finds that @xmath51 so that ( [ largish time ] ) implies the exponential suppression in the number of decaying states is large , typically involving a reduction by @xmath52 . as a result ,
this phenomena is generally unobservable .
however , it is well known in inflationary cosmology that even exponentially suppressed processes can be significant . for example , in eternal inflation , even though regions of false vacua are assumed to decay exponentially , gravitational effects cause space in a region that has not decayed to itself grow exponentially fast .
thus most of space can , at any time , be inflating , even well after the canonical time in which one would consider , in the absence of gravity , the phase transition to have been completed .
the result we describe here appears to exacerbate this situation considerably .
those false vacuum regions which do survive well beyond the canonical decay time , up to the time given in ( [ largisher time ] ) , may effectively never decay , as within these regions the survival probability per unit volume will fall as a power law , while the volume of the regions will grow exponentially . in the long term ,
then , space will become dominated by precisely these regions .
if this is the case , the most probable regions of space to find oneself within , even long after the canonical false vacuum decay time would be metastable regions .
indeed , the fact that survival probability only falls as a power law at late times implies that one can have eternal inflation scenarios even when the decay rate is comparable or even somewhat larger than the inflationary expansion rate . in such scenarios
the volume of space associated with false vacuum will initially decrease , but at late times , when the power law kicks in , the volume of space in false vacua may quickly begin to increase again . because the ratio of metastable energy to decay rate appears in the crossover time , the absolute survival probability at late times for any specific region of inflating false vacuum will increase as the vacuum energy decreases for a given fixed decay rate .
( because , in the wkb approximation the decay rate depends primarily upon the integral of the action under the barrier it is not implausible that this decay rate can be a slowly varying function of the energy gap . )
while probabilities in eternal inflation scenarios are notoriously difficult to normalize , this argument suggests that the survival probability may be largest in those metastable regions with the smallest allowed vacuum energy .
is it therefore coincidence that we currently find ourselves in a universe with extremely small vacuum energy ?
one can not answer this question unambiguously without knowledge of the underlying theory but since exponential expansion has only begun in our universe relatively recently , if our vacuum is metastable we are not likely to exist in the late time regime . specifically , given the inferred vacuum energy today , if we assume the ultimate ground state vacuum energy is zero , then @xmath53 ev . if our vacuum were metastable , the time at which power law decay probability would set in would depend upon the unknown decay rate .
if the canonical decay rate for our vacuum is less than about twice the present hubble constant , @xmath54 ev , so that vacuum decay would generically not occur until after exponential expansion began ( and after finite temperature effects would presumably no longer stabilize the false vacuum ) , the onset of late time power law decay will not occur until the universe is at least 30 times its current age .
( if the decay rate were much faster , then we would generically expect the fraction of space which would survive as a function of time would have been decreasing dramatically over cosmic time before the current vacuum energy began to dominate , which does not seem likely given the size of the currently observable universe . )
these considerations provide a small modicum of hope .
if our universe is metastable , the expected survival time may be not much larger than the current hubble age .
however , we might take solace in the possibility that if our region is one of the rare late - time survivors , vacuum decay will occur over an exponentially decreasing fraction of our space as time proceeds
. questions of the long - term survival of universe aside , these arguments do suggest that maximizing the volume of space contained in metastable vacua at late times will favor those regions with the smallest vacuum energies .
alternatively , it would appear that a late time transition to a true vacuum state with large negative energy is disfavored .
if , for example , the canonical decay rate were comparable to the present age of the universe , and the energy gap to the true vacuum were @xmath55 gev , the exponential suppression factor would exceed @xmath56 before a period of power law decay would set in .
this argument disfavor string landscape scenario where the ultimate vacuum state is anti - de sitter and anthropic arguments are used to explain our present configuration .
the possibility that macroscopic quantum mechanical considerations may impact upon the very existence of our own universe at late times is an intriguing one .
as we have seen , at the very least they may require a dramatic alteration of probability considerations in chaotic eternal inflationary universe scenarios , and may explain why universes with unnaturally small vacuum energies may be favored at late times , even as these arguments would tend to disfavor ultimate decay into a state with large negative vacuum energy .
however , these considerations also lead us into unexplored realms associated with macroscopic quantum mechanics .
the first , which we plan to investigate , involves the question of whether late - time power law time dependence persists when gravitational effects in vacuum decay @xcite are taken into account . the heuristic argument regarding spreading of the wavefunction following barrier penetration , leaving a power law tail at the origin as the source of late time behavior suggests that the exponential expansion of space accompanying metastability may suppress this term and thus produce a faster falloff .
the second consideration is even more interesting .
have we ensured , by measuring the existence dark energy in our own universe , that the quantum mechanical configuration of our own universe is such that late time decay is not relevant ?
put another way , what can internal observations of the state of a metastable universe say about its longevity ?
is supported in part by the doe and nasa , and thanks vanderbilt university for hospitality while this work was begun .
we thank tom weiler for making us aware of power law late - time decay , and alan guth , glenn starkman , tom kephart and robert scherrer for useful conversations .
khalfin , _
. jetp _ * 6*,1053 ( 1958 ) r. paley and n. wiener , `` fourier transforms in the complex domain '' _ am .
soc . pub .
_ , * xix * , ( 1934 ) c.b .
chiu , e.c.g .
sudarshan , and b. misra , _ phys .
rev . _ * d16 * , 520 ( 1977 ) a. peres , _ annals of physics _ , * 129 * , 33 , ( 1980 ) a. patrascioiu _ phys .
* d24 * , 496 , ( 1981 ) we thank alan guth for pointing out this heuristic argument . c. g. callan , s. coleman , _ phys . rev _ * d16 * , 1762 , ( 1977 ) s. coleman , f. de lucia _ phys . rev _ * d21 * , 3305 , ( 1980 ) | we describe here how the late time behavior of the decaying states , which is predicted to deviate from an exponential form , while normally of insignificant consequence , may have important cosmological implications in the case of false vacuum decay .
it may increase the likelihood of eternal inflation , and may help explain the likelihood of observing a small vacuum energy at late times , as well as arguing against decay into a large negative energy ( anti - de sitter space ) , vacuum state as has been motivated by some string theory considerations .
several interesting open questions are raised , including whether observing the cosmological configuration of a metastable universe can constrain its inferred lifetime . |
it is well known that supersymmetric models with conserved @xmath2-parity contain one new stable particle which is a candidate for cold dark matter ( cdm ) @xcite .
there are very strong constraints , however , forbidding the existence of stable or long lived particles which are not color and electrically neutral .
strong and electromagnetically interacting lsps would become bound with normal matter forming anomalously heavy isotopes . indeed , there are very strong upper limits on the abundances , relative to hydrogen , of nuclear isotopes @xcite , @xmath3 for 1 gev @xmath4 1 tev .
a strongly interacting stable relic is expected to have an abundance @xmath5 with a higher abundance for charged particles .
there are relatively few supersymmetric candidates which are not colored and are electrically neutral .
the sneutrino @xcite is one possibility , but in the mssm , it has been excluded as a dark matter candidate by direct @xcite and indirect @xcite searches .
in fact , one can set an accelerator based limit on the sneutrino mass from neutrino counting , @xmath6 44.7 gev @xcite . in this case , the direct relic searches in underground low - background experiments require @xmath6 20 tev @xcite . another possibility is the gravitino which is probably the most difficult to exclude .
i will concentrate on the remaining possibility in the mssm , namely the neutralinos .
there are four neutralinos , each of which is a linear combination of the @xmath7 , neutral fermions @xcite : the wino @xmath8 , the partner of the 3rd component of the @xmath9 gauge boson ; the bino , @xmath10 , the partner of the @xmath11 gauge boson ; and the two neutral higgsinos , @xmath12 and @xmath13 . in general , neutralinos can be expressed as a linear combination @xmath14 the solution for the coefficients @xmath15 and @xmath16 for neutralinos that make up the lsp can be found by diagonalizing the mass matrix ( w^3 , b , h^0_1,h^0_2 ) ( cccc m_2 & 0 & -g_2 v_1 & g_2 v_2 + 0 & m_1 & g_1 v_1 & -g_1 v_2 + -g_2 v_1 & g_1 v_1 & 0 & - + g_2 v_2 & -g_1 v_2 & - & 0 ) ( c w^3 + b + h^0_1 + h^0_2 ) where @xmath17 are the soft supersymmetry breaking u(1 ) ( su(2 ) ) gaugino mass terms .
@xmath18 is the supersymmetric higgs mixing mass parameter and since there are two higgs doublets in the mssm , there are two vacuum expectation values , @xmath19 and @xmath20 .
one combination of these is related to the @xmath21 mass , and therefore is not a free parameter , while the other combination , the ratio of the two vevs , @xmath22 , is free .
the most general version of the mssm , despite its minimality in particles and interactions contains well over a hundred new parameters .
the study of such a model would be untenable were it not for some ( well motivated ) assumptions .
these have to do with the parameters associated with supersymmetry breaking .
it is often assumed that , at some unification scale , all of the gaugino masses receive a common mass , @xmath23 .
the gaugino masses at the weak scale are determined by running a set of renormalization group equations .
similarly , one often assumes that all scalars receive a common mass , @xmath24 , at the gut scale ( though one may wish to make an exception for the higgs soft masses ) .
these too are run down to the weak scale .
the remaining supersymmetry breaking parameters are the trilinear mass terms , @xmath25 , which i will also assume are unified at the gut scale , and the bilinear mass term @xmath26 .
there are , in addition , two physical cp violating phases which will not be considered here .
the natural boundary conditions at the gut scale for the mssm would include @xmath18 , the two soft higgs masses ( @xmath27 and @xmath28 ) and @xmath26 in addition to @xmath23 , @xmath24 , and @xmath25 . in this case , by running the rges down to a low energy scale , one would predict the values of @xmath29 , @xmath22 , and the higgs pseudoscalar mass , @xmath30 ( in addition to all of the sparticle masses ) . since @xmath29 is known , it is more useful to analyze supersymmetric models where @xmath29 is input rather than output .
it is also common to treat @xmath22 as an input parameter .
this can be done at the expense of shifting @xmath18 ( up to a sign ) and @xmath26 from inputs to outputs .
when the supersymmetry breaking higgs soft masses are also unified at the gut scale ( and take the common value @xmath24 ) , the model is often referred to as the constrained mssm or cmssm .
once these parameters are set , the entire spectrum of sparticle masses at the weak scale can be calculated .
the relic abundance of lsp s is determined by solving the boltzmann equation for the lsp number density in an expanding universe .
the technique @xcite used is similar to that for computing the relic abundance of massive neutrinos @xcite .
the relic density depends on additional parameters in the mssm beyond @xmath31 , and @xmath22 .
these include the sfermion masses , @xmath32 , as well as @xmath30 , all derived from @xmath24 , @xmath25 , and @xmath23 . to determine the relic density it is necessary to obtain the general annihilation cross - section for neutralinos . in much of the parameter space of interest ,
the lsp is a bino and the annihilation proceeds mainly through sfermion exchange . because of the p - wave suppression associated with majorana fermions , the s - wave part of the annihilation cross - section is suppressed by the outgoing fermion masses .
this means that it is necessary to expand the cross - section to include p - wave corrections which can be expressed as a term proportional to the temperature if neutralinos are in equilibrium .
unless the neutralino mass happens to lie near near a pole , such as @xmath33 @xmath34 or @xmath35 , in which case there are large contributions to the annihilation through direct @xmath36-channel resonance exchange , the dominant contribution to the @xmath37 annihilation cross section comes from crossed @xmath38-channel sfermion exchange .
annihilations in the early universe continue until the annihilation rate @xmath39 drops below the expansion rate given by the hubble parameter , @xmath40 . for particles which annihilate through approximate weak scale interactions
, this occurs when @xmath41 .
subsequently , the relic density of neutralinos is fixed relative to the number of relativistic particles .
as noted above , the number density of neutralinos is tracked by a boltzmann - like equation , = -3r r n - v ( n^2 - n_0 ^ 2 ) [ rate ] where @xmath42 is the equilibrium number density of neutralinos . by defining the quantity @xmath43
, we can rewrite this equation in terms of @xmath44 , as = m _ ( 4 45 ^3 g_n n ) ^-1/2 v ( f^2 - f_0 ^ 2 ) [ rate2 ] where @xmath45 is the number of relativistic degrees of freedom . the solution to this equation at late times ( small @xmath46 )
yields a constant value of @xmath47 , so that @xmath48 . the final relic density expressed as a fraction of the critical energy density
can be written as @xcite _ h^2 1.9 10 ^ -11 ( t_t_)^3 n_f^1/2 ( gev ax_f + 12 b x_f^2 ) [ relic ] where @xmath49 accounts for the subsequent reheating of the photon temperature with respect to @xmath50 , due to the annihilations of particles with mass @xmath51 @xcite .
the subscript @xmath47 refers to values at freeze - out , i.e. , when annihilations cease .
the coefficients @xmath52 and @xmath53 are related to the partial wave expansion of the cross - section , @xmath54 .
( [ relic ] ) results in a very good approximation to the relic density expect near s - channel annihilation poles , thresholds and in regions where the lsp is nearly degenerate with the next lightest supersymmetric particle @xcite .
the preferred range of the relic lsp density is provided by data on the cosmic microwave background ( cmb ) , which have recently been used to obtain the following 95% confidence range : @xmath55 @xcite .
values much smaller than @xmath56 seem to be disfavoured by earlier analyses of structure formation in the cdm framework , so we restrict our attention to @xmath57 .
however , one should note that the lsp may not constitute all the cdm , in which case @xmath58 could be reduced below this value . on the upper side ,
it is preferable to remain very conservative , in particular because the upper limit on @xmath58 sets the upper limit for the sparticle mass scale .
here , @xmath59 is used , while being aware that the lower part of this range currently appears the most plausible .
[ fig : rough ] illustrates qualitatively the regions of the @xmath60 plane favoured by lep limits , particularly on @xmath61 , @xmath62 and cosmology .
electroweak symmetry breaking is not possible in the dark - shaded triangular region in the top left corner , and the lightest supersymmetric particle would be charged in the bottom right dark - shaded triangular region .
the experimental constraints on @xmath61 and @xmath63 exert pressures from the left , as indicated , which depend on the value of @xmath22 and the sign of @xmath18 .
the indication of a deviation from the standard model in @xmath64 disfavours @xmath65 and large values of @xmath24 and @xmath23 for @xmath66 .
the region where @xmath67 falls within the preferred range is indicated in light shading , its exact shape being dependent on the value of @xmath68 . as discussed later in more detail , in addition to the ` bulk ' region at low @xmath24 and @xmath23
, there are a number of regions at large values of @xmath23 and/or @xmath24 where the relic density is still compatible with the cosmological constraints . at large values of @xmath23 , the lighter stau ,
becomes nearly degenerate with the neutralino and coannihilations between these particles must be taken into account @xcite . for non - zero values of @xmath25
, there are new regions for which @xmath69 coannihilations are important @xcite . at large @xmath22 , as one increases @xmath23 , the pseudo - scalar mass , @xmath30 begins to drop so that there is a wide funnel - like region ( at all values of @xmath24 ) such that @xmath70 and s - channel annihilations become important @xcite .
finally , there is a region at very high @xmath24 where the value of @xmath18 begins to fall and the lsp becomes more higgsino - like .
this is known as the ` focus point ' region @xcite .
also shown are the position of several benchmark points chosen for detailed phenomenological study @xcite .
let us first focus on the ` bulk ' region in the cmssm for @xmath71 and @xmath72 shown in fig .
[ fig : sm ] @xcite .
the light shaded region correspond to .
the dark shaded region has @xmath73 and is excluded .
the light dashed contours indicate the corresponding region in @xmath67 if one ignores the effect of coannihilations . neglecting coannihilations
, one would find an upper bound of @xmath74 on @xmath75 , corresponding to an upper bound of roughly @xmath76 on @xmath77 .
coannihilations are important when there are several particle species @xmath78 , with different masses , and each with its own number density @xmath79 and equilibrium number density @xmath80 . in this case
@xcite , the rate equation ( [ rate ] ) still applies , provided @xmath81 is interpreted as the total number density , n _
i n_i , [ n ] @xmath42 as the total equilibrium number density , n_0 _ i n_0,i , [ neq ] and the effective annihilation cross section as _ eff
v_rel_ij n_0,i n_0,j n_0 ^ 2 _ ij v_rel .
[ sv2 ] in eq .
( [ rate2 ] ) , @xmath82 is now understood as the mass of the lightest sparticle under consideration .
note that this implies that the ratio of relic densities computed with and without coannihilations is , roughly , @xmath83 where @xmath84 and sub- and superscripts 0 denote quantities computed ignoring coannihilations .
the ratio @xmath85 , where @xmath86 .
for the case of three degenerate slepton nlsps , @xmath87 and @xmath88 .
the effect of coannihilations is to create an allowed band about 25 - 50 @xmath89 wide in @xmath24 for @xmath90 , which tracks above the @xmath91 contour . along the line @xmath92 , @xmath93 , from ( [ eq : r ] )
@xcite . as @xmath24 increases , the mass difference increases and the slepton contribution to @xmath94 falls , and the relic density rises abruptly .
a larger view of the @xmath95 parameter plane is shown in fig . [
fig : uhm ] @xcite . included here
are the most important phenomenological constraints ( shown schematically in figure [ fig : rough ] ) .
these include the constraints on the mssm parameter space that are provided by direct sparticle searches at lep , including that on the lightest chargino @xmath96 : @xmath97 103.5 gev @xcite , and that on the selectron @xmath98 : @xmath99 99 gev @xcite .
another important constraint is provided by the lep lower limit on the higgs mass : @xmath100 gev @xcite in the standard model . since @xmath61 is sensitive to sparticle masses , particularly @xmath101 , via loop corrections ,
the higgs limit also imposes important constraints on the cmssm parameters , principally @xmath23 . the constraint imposed by measurements of @xmath102
@xcite also excludes small values of @xmath23 .
these measurements agree with the standard model , and therefore provide bounds on mssm particles , such as the chargino and charged higgs masses , in particular . typically , the @xmath102 constraint is more important for @xmath65 , but it is also relevant for @xmath72 , particularly when @xmath68 is large . the latest value of the anomalous magnetic moment of the muon reported @xcite by the bnl e821 experiment is also taken into account .
the world average of @xmath103 now deviates by @xmath104 from the standard model calculation of @xcite using @xmath105 data , and by @xmath106 from the standard model calculation of @xcite based on @xmath107 decay data .
other recent analyses of the @xmath105 data yield similar results . on some of the subsequent plots ,
the formal 2-@xmath108 range @xmath109 is displayed .
as one can see , the region preferred by @xmath110 overlaps very nicely with the ` bulk ' region for @xmath95 and @xmath72 . as noted above ,
another mechanism for extending the allowed cmssm region to large @xmath82 is rapid annihilation via a direct - channel pole when @xmath111 @xcite . since the heavy scalar and pseudoscalar higgs masses decrease as @xmath22 increases , eventually @xmath112 yielding a ` funnel ' extending to large @xmath23 and @xmath24 at large @xmath68 , as seen in fig .
[ rd2c50 ] .
the difficulty and necessary care involved in calculations at large @xmath22 were discussed in @xcite . for related cmssm calculations
see @xcite .
another way to see the effects of increasing @xmath22 is plot look at the @xmath113 plane for fixed @xmath24 .
this is shown in fig .
[ newones ] . as one can see ,
there is a charged @xmath114 lsp in the upper right corner for both values of @xmath24 , though the area is reduced when @xmath24 is increased as is expected .
one also sees the cosmological region moving up in @xmath22 as @xmath24 is increased . in b ) , one also sees the well defined coannihilation tail . here , the importance of @xmath115 at large @xmath22 is clearly displayed .
if one relaxes the unification condition for the two soft higgs mass , @xmath27 and @xmath28 , one could decide to make @xmath18 and @xmath30 inputs and instead making the higgs soft masses outputs .
this is referred to as the cmssm with non - universal higgs masses or nuhm @xcite .
the attractive feature of radiative symmetry breaking in the cmssm is maintained by using the electroweak minimization conditions to solve for @xmath27 and @xmath28 , given values of @xmath18 and @xmath30 at the weak scale .
the electroweak symmetry breaking conditions may be written in the form : @xmath116 and @xmath117 where @xmath118 and @xmath119 are loop corrections @xcite and @xmath120 .
the known radiative corrections @xcite @xmath121 and @xmath122 relating the values of the nuhm parameters at @xmath123 to their values at @xmath124 are incorporated : @xmath125 solving for @xmath126 and @xmath127 , one has @xmath128 and @xmath129 which are used to perform the numerical calculations @xcite .
it can be seen from ( [ m1 ] ) and ( [ m2 ] ) that , if @xmath30 is too small or @xmath18 is too large , then @xmath130 and/or @xmath131 can become negative and large .
this could lead to @xmath132 and/or @xmath133 , thus triggering electroweak symmetry breaking at the gut scale .
the requirement that electroweak symmetry breaking occurs far below the gut scale forces one to impose the conditions @xmath134 as extra constraints , which we call the gut stability constraint .
the nuhm parameter space was recently analyzed @xcite and a sample of the results found is shown in fig .
while much of the cosmologically preferred area with @xmath65 is excluded , there is a significant enhancement in the allowed parameter space for @xmath72 . as usual , there are dark ( red ) regions where there is one or more charged sparticle lighter than the neutralino @xmath50 so that @xmath50 is no longer the lsp .
first , there are ` shark s teeth ' at @xmath135 gev , @xmath136 gev in panel ( a ) of fig .
[ muma ] where the @xmath137 is the lsp . at small @xmath138 , particularly at small @xmath30 when the mass difference @xmath139 is small ,
the @xmath140 mass is driven small , making the @xmath141 the lsp . at
even smaller @xmath138 , however , the lightest neutralino gets lighter again , since @xmath142 when @xmath143 .
in addition , there are extended dark ( red ) shaded regions at large @xmath138 where left - handed sleptons become lighter than the neutralino .
however , the electron sneutrino @xmath144 and the muon sneutrino @xmath145 ( which are degenerate within the approximations used here ) have become joint lsps at a slightly smaller @xmath138 .
since the possibility of sneutrino dark matter has been essentially excluded by a combination of ` @xmath146
counting ' at lep , which excludes @xmath147 gev @xcite , and searches for cold dark matter , which exclude heavier @xmath148 weighing @xmath149 tev @xcite , we still demand that the lsp be a neutralino @xmath50 .
the darker ( dark blue ) shaded regions are where the sneutrinos are the lsps , and therefore excluded . in the strips adjacent to the @xmath150 lsp regions ,
neutralino - sneutrino coannihilation is important in suppressing the relic density to an acceptable level @xcite .
the thick cosmological region at smaller @xmath18 corresponds to the ` bulk ' region familiar from cmssm studies .
the two ( black ) crosses indicate the position of the cmssm points for these input parameters . extending upward in @xmath30 from this region
, there is another light ( turquoise ) shaded band at smaller @xmath138 . here
, the neutralino gets more higgsino - like and the annihilation to @xmath151 becomes important , yielding a relic density in the allowed range . for smaller @xmath138 ,
the relic density becomes too small due to the @xmath152 coannihilations .
for even smaller @xmath138 ( @xmath153 gev ) many channels are kinematically unavailable and we are no longer near the @xmath154 and @xmath21 pole . as a result the relic density may again come into the cosmologically preferred region . however , this region is excluded by the lep limit on the chargino mass .
the unshaded regions between the allowed bands have a relic density that is too high : @xmath155 . however , the @xmath156 and @xmath157 coannihilation and bulk bands are connected by horizontal bands of acceptable relic density that are themselves separated by unshaded regions of low relic density , threaded by solid ( blue ) lines asymptoting to @xmath158 gev .
these lines correspond to cases when @xmath159 , where direct - channel annihilation : @xmath160 is important , and suppresses the relic density @xcite creating ` funnel'-like regions . in panel ( b ) , a larger value of @xmath23 is chosen . in this case
, the previous region excluded by the neutral lsp constraint at large @xmath138 , migrates to larger @xmath138 .
the ` shark s teeth ' for moderate @xmath138 grow , reaching up to @xmath161 gev .
these arise when one combines a large value of @xmath23 with a relatively small value of @xmath24 , and one may find a @xmath162 or even a @xmath98 lsp .
the large value of @xmath23 also keeps the rate of @xmath163 under control unless @xmath30 is small .
the chargino constraint is similar to that in panel ( a ) , whereas the @xmath61 constraint is irrelevant due to the large value of @xmath23 .
finally , the gut mass - squared positivity constraint now allows larger values of @xmath164 gev . it is also interesting to examine the @xmath24 - @xmath23 plane in the nuhm @xcite . in fig .
[ fig : m12m0]a , there is a bulk region at @xmath165 gev to 350 gev , @xmath166 gev to 150 gev . as in the cmssm , the @xmath167 is the lsp in the bigger area at larger @xmath23 , and there are light ( turquoise ) shaded strips close to these forbidden regions where coannihilation suppresses the relic density sufficiently to be cosmologically acceptable .
further away from these regions , the relic density is generally too high .
however , for larger @xmath23 there is another suppression , discussed below , which makes the relic density too low . at small @xmath23 and
@xmath24 the left handed sleptons , and also the sneutrinos , become lighter than the neutralino .
the darker ( dark blue ) shaded area is where a sneutrino is the lsp .
the near - vertical dark ( black ) dashed and light ( red ) dot - dashed lines in fig .
[ fig : m12m0 ] are the lep exclusion contours @xmath168 gev and @xmath169 gev respectively . as in the cmssm case
, they exclude low values of @xmath23 , and hence rule out rapid relic annihilation via direct - channel @xmath154 and @xmath170 poles .
the solid lines curved around small values of @xmath23 and @xmath24 bound the light ( pink ) shaded region favoured by @xmath171 and recent analyses of the @xmath105 data . 0.5 in a striking feature in fig .
[ fig : m12m0](a ) when @xmath172 gev is a strip with low @xmath67 , which has bands with acceptable relic density on either side .
the low-@xmath67 strip is due to rapid annihilation via the direct - channel @xmath173 poles which occur when @xmath174 gev , indicated by the near - vertical solid ( blue ) line .
the analogous rapid - annihilation strips occur in the cmssm but at larger @xmath22 as seen in fig .
[ rd2c50 ] .
there , they are diagonal in the @xmath60 plane , reflecting a cmssm link between @xmath24 and @xmath30 that is absent in this implementation of the nuhm .
the right - hand band in fig . [ fig : m12m0](a ) with acceptable @xmath67 is broadened because the neutralino acquires significant higgsino content , and the relic density is suppressed by the increased @xmath151 production . as @xmath23 increases
, the neutralino becomes almost degenerate with the second lightest neutralino and the lighter chargino , and the @xmath175 coannihilation processes eventually push @xmath176 when @xmath177 gev .
the two dark ( black ) dash - dotted lines in fig .
[ fig : m12m0](a ) indicate where scalar squared masses become negative at the input gut scale for one of the higgs multiplets .
one of these gut stability lines is near - vertical at @xmath178 gev , and the other is a curved line at @xmath179 gev , @xmath180 gev .
panel ( b ) of fig .
[ fig : m12m0 ] is for @xmath181 gev and @xmath182 gev .
we notice immediately that the heavy higgs pole and the right - hand boundary of the gut stability region move out to larger @xmath183 gev , respectively , as one would expect for larger @xmath30 . at this value of @xmath30 , the right side of the rapid annihilation ( ` funnel ' )
strip has disappeared , due to enhanced chargino - neutralino coannihilation effects .
these two examples serve to demonstrate that the @xmath60 plane may look rather different in the cmssm from its appearance in the cmssm for the same value of @xmath22 . in particular , the locations of rapid - annihilation funnels and @xmath184 coannihilation regions are quite model - dependent , and the gut stability requirement may exclude large parts of the @xmath60 plane . because the lsp as dark matter is present locally ,
there are many avenues for pursuing dark matter detection .
direct detection techniques rely on an ample neutralino - nucleon scattering cross - section .
the effective four - fermion lagrangian can be written as @xmath185 however , the terms involving @xmath186 , and @xmath187 lead to velocity dependent elastic cross sections .
the remaining terms are : the spin dependent coefficient , @xmath188 and the scalar coefficient @xmath189 .
contributions to @xmath188 are predominantly through light squark exchange .
this is the dominant channel for binos .
scattering also occurs through z exchange but this channel requires a strong higgsino component .
contributions to @xmath189 are also dominated by light squark exchange but higgs exchange is non - negligible in most cases .
[ fig : sicontours ] displays contours of the spin - independent cross section for the elastic scattering of the lsp @xmath50 on protons in the @xmath190 planes for ( a ) @xmath191 , ( b ) @xmath192 @xcite .
the double dot - dashed ( orange ) lines are contours of the spin - independent cross section , and the contours @xmath193 pb in panel ( a ) and @xmath194 pb in panel ( b ) are indicated .
the lep lower limits on @xmath61 and @xmath195 , as well as the experimental measurement of @xmath115 for @xmath65 , tend to bound the cross sections from above , as discussed in more detail below . generally speaking , the spin - independent cross section is relatively large in the ` bulk ' region , but falls off in the coannihilation ` tail ' . also , we note also that there is a strong cancellation in the spin - independent cross section when @xmath65 @xcite , as seen along strips in panel ( a ) of fig . [
fig : sicontours ] where @xmath172 gev . in the cancellation region ,
the cross section drops lower than @xmath196 pb . all these possibilities for suppressed spin - independent cross sections are disfavoured by the data on @xmath64 , which favour values of @xmath23 and @xmath24 that are not very large , as well as @xmath72 , as seen in panel ( b ) of fig .
[ fig : sicontours ] .
thus @xmath64 tends to provide a lower bound on the spin - independent cross section .
[ fig : decimation](a ) illustrates the effect on the cross sections of each of the principal phenomenological constraints , for the particular case @xmath95 @xmath72 .
the solid ( blue ) lines mark the bounds on the cross sections allowed by the relic - density constraint @xmath197 alone . for any given value of @xmath23
, only a restricted range of @xmath24 is allowed .
therefore , only a limited range of @xmath24 , and hence only a limited range for the cross section , is allowed for any given value of @xmath82 .
the thicknesses of the allowed regions are due in part to the assumed uncertainties in the nuclear inputs .
these have been discussed at length in @xcite . on the other hand , a broad range of @xmath82
is allowed , when one takes into account the coannihilation ` tail ' region at each @xmath198 and the rapid - annihilation ` funnel ' regions for @xmath199 .
the dashed ( black ) line displays the range allowed by the @xmath200 constraint alone . in this case ,
a broader range of @xmath24 and hence the spin - independent cross section is possible for any given value of @xmath82 .
the impact of the constraint due to @xmath61 is shown by the dot - dashed ( green ) line . comparing with the previous constraints
, we see that a region at low @xmath82 is excluded by @xmath61 , strengthening significantly the previous _ upper _ limit on the spin - independent cross section .
finally , the dotted ( red ) lines in fig . [ fig : decimation ] show the impact of the @xmath64 constraint .
this imposes an upper bound on @xmath23 and hence @xmath82 , and correspondingly a _ lower _ limit on the spin - independent cross section .
this analysis is extended in panel ( b ) of fig .
[ fig : decimation ] to all the values @xmath201 and we find overall that @xcite @xmath202 for @xmath72 .
( @xmath203 is the spin - dependent cross - section not shown in the figures presented here . ) as we see in panel ( b ) of fig .
[ fig : decimation ] , @xmath61 provides the most important upper limit on the cross sections for @xmath204 , and @xmath115 for larger @xmath22 , with @xmath64 always providing a more stringent lower limit than the relic - density constraint .
the relic density constraint shown is evaluated at the endpoint of the coannihilation region . at large @xmath22
, the higgs funnels or the focus - point regions have not been considered , as their locations are very sensitive to input parameters and calculational details @xcite . the results from a cmssm and mssm analysis @xcite for @xmath205 and 10 are compared with the most recent cdms @xcite and edelweiss @xcite bounds in fig
. [ cdms ] .
these results have nearly entirely excluded the region purported by the dama @xcite experiment . the cmssm prediction @xcite is shown by the dark shaded region , while the nuhm case @xcite is shown by the larger lighter shaded region .
i conclude by showing the prospects for direct detection for the benchmark points discussed above@xcite .
[ fig : dm ] shows rates for the elastic spin - independent scattering of supersymmetric relics , including the projected sensitivities for cdms ii @xcite and cresst @xcite ( solid ) and genius @xcite ( dashed ) . also shown
are the cross sections calculated in the proposed benchmark scenarios discussed in the previous section , which are considerably below the dama @xcite range ( @xmath206 pb ) .
indirect searches for supersymmetric dark matter via the products of annihilations in the galactic halo or inside the sun also have prospects in some of the benchmark scenarios @xcite .
i would like to thank j. ellis , t. falk , a. ferstl , g. ganis , y. santoso , and m. srednicki for enjoyable collaborations from which this work is culled .
this work was supported in part by doe grant de - fg02 - 94er40823 at minnesota .
j. rich , m. spiro and j. lloyd - owen , _
phys.rep . _ * 151 * , 239 ( 1987 ) ; + p.f .
smith , _ contemp.phys . _ * 29 * , 159 ( 1998 ) ; + t.k .
hemmick et al .
, , 2074 ( 1990 ) .
t. yanagata , y. takamori , and h. utsunomiya , , 1231 ( 1993 ) .
j. r. ellis , t. falk , k. a. olive and m. schmitt , phys .
b * 388 * , 97 ( 1996 ) [ arxiv : hep - ph/9607292 ] ; + the lep collaborations , the lep electroweak working group , and the sld heavy flavour and electroweak groups , cern - ep-2000 - 016 .
j. r. ellis , t. falk and k. a. olive , phys .
b * 444 * , 367 ( 1998 ) [ arxiv : hep - ph/9810360 ] ; + j. r. ellis , t. falk , k. a. olive and m. srednicki , astropart .
* 13 * , 181 ( 2000 ) [ erratum - ibid . *
15 * , 413 ( 2001 ) ] [ arxiv : hep - ph/9905481 ] .
m. e. gmez , g. lazarides and c. pallis , phys .
* d61 * , 123512 ( 2000 ) [ hep - ph/9907261 ] ; phys .
* b487 * , 313 ( 2000 ) [ hep - ph/0004028 ] ; nucl .
b * 638 * ( 2002 ) 165 [ arxiv : hep - ph/0203131 ] ; + r. arnowitt , b. dutta and y. santoso , hep - ph/0102181 . c. boehm , a. djouadi and m. drees , phys .
d * 62 * , 035012 ( 2000 ) [ arxiv : hep - ph/9911496 ] ; + j. ellis , k.a . olive and y. santoso , arxiv : hep - ph/0112113 .
m. drees and m. m. nojiri , _ phys .
d _ * 47 * , 376 ( 1993 ) ; + h. baer and m. brhlik , _ phys .
d _ * 53 * , 597 ( 1996 ) ; and _ phys .
d _ * 57 * 567 ( 1998 ) ; + h. baer , m. brhlik , m. a. diaz , j. ferrandis , p. mercadante , p. quintana and x. tata , _ phys . rev .
d _ * 63 * , 015007 ( 2001 ) ; + a. b. lahanas , d. v. nanopoulos and v. c. spanos , _ mod .
a _ * 16 * 1229 ( 2001 ) .
j. ellis , t. falk , g. ganis , k. a. olive and m. srednicki , hep - ph/0102098 .
j. l. feng , k. t. matchev and f. wilczek , phys .
b * 482 * , 388 ( 2000 ) [ hep - ph/0004043 ] .
m. battaglia et al . , _ eur .
phys . j. c _ * 22 * 535 ( 2001 ) .
s. heinemeyer , w. hollik and g. weiglein , comput .
commun .
* 124 * ( 2000 ) 76 [ arxiv : hep - ph/9812320 ] ; + s. heinemeyer , w. hollik and g. weiglein , eur
. phys .
j. c * 9 * ( 1999 ) 343 [ arxiv : hep - ph/9812472 ] .
j. r. ellis , t. falk , g. ganis and k. a. olive , phys .
d * 62 * , 075010 ( 2000 ) [ arxiv : hep - ph/0004169 ] . j. r. ellis , k. a. olive and y. santoso , new j. phys .
* 4 * ( 2002 ) 32 [ arxiv : hep - ph/0202110 ] .
lep higgs working group for higgs boson searches , opal collaboration , aleph collaboration , delphi collaboration and l3 collaboration , _
search for the standard model higgs boson at lep _
, aleph-2001 - 066 , delphi-2001 - 113 , cern - l3-note-2699 , opal - pn-479 , lhwg - note-2001 - 03 , cern - ep/2001 - 055 , arxiv : hep - ex/0107029 ; _ searches for the neutral higgs bosons of the mssm : preliminary combined results using lep data collected at energies up to 209 gev _ , lhwg - note-2001 - 04 , aleph-2001 - 057 , delphi-2001 - 114 , l3-note-2700 , opal - tn-699 , arxiv : hep - ex/0107030 .
alam et al . , [ cleo collaboration ] , _ phys .
lett . _ * 74 * , 2885 ( 1995 ) ; + as updated in s. ahmed et al .
, cleo conf 99 - 10 ; belle collaboration , belle - conf-0003 , contribution to the 30th international conference on high - energy physics , osaka , 2000 ; + see also k. abe _ et al .
_ , [ belle collaboration ] , [ arxiv : hep - ex/0107065 ] ; + l. lista [ babar collaboration ] , [ arxiv : hep - ex/0110010 ] ; + c. degrassi , p. gambino and g. f. giudice , _ jhep _ * 0012 * , 009 ( 2000 ) ; + m. carena , d. garcia , u. nierste and c. e. wagner , _ phys .
b _ * 499 * , 141 ( 2001 ) ; + p. gambino and m. misiak , nucl . phys .
b * 611 * ( 2001 ) 338 ; + d. a. demir and k. a. olive , phys .
d * 65 * , 034007 ( 2002 ) .
g. w. bennett _ et al . _ [ muon g-2 collaboration ] , phys . rev
. lett .
* 89 * ( 2002 ) 101804 [ erratum - ibid .
* 89 * ( 2002 ) 129903 ] [ arxiv : hep - ex/0208001 ] .
m. davier , s. eidelman , a. hocker and z. zhang , arxiv : hep - ph/0208177 ; see also k. hagiwara , a. d. martin , d. nomura and t. teubner , arxiv : hep - ph/0209187 ; f. jegerlehner ( unpublished , as reported in m. krawczyk , arxiv : hep - ph/0208076 .
a. b. lahanas , d. v. nanopoulos and v. c. spanos , phys .
b * 518 * ( 2001 ) 94 [ arxiv : hep - ph/0107151 ] ; + v. barger and c. kao , phys .
b * 518 * , 117 ( 2001 ) [ arxiv : hep - ph/0106189 ] ; + l. roszkowski , r. ruiz de austri and t. nihei , jhep * 0108 * , 024 ( 2001 ) [ arxiv : hep - ph/0106334 ] ; + a. djouadi , m. drees and j. l. kneur , jhep * 0108 * , 055 ( 2001 ) [ arxiv : hep - ph/0107316 ] ; + h. baer , c. balazs and a. belyaev , jhep * 0203 * , 042 ( 2002 ) [ arxiv : hep - ph/0202076 ] . v. berezinsky , a. bottino , j. r. ellis , n. fornengo , g. mignola and s. scopel , astropart .
* 5 * , 1 ( 1996 ) [ arxiv : hep - ph/9508249 ] ; + p.
nath and r. arnowitt , phys .
d * 56 * , 2820 ( 1997 ) [ arxiv : hep - ph/9701301 ] ; + m. drees , m. m. nojiri , d. p. roy and y. yamada , phys .
d * 56 * , 276 ( 1997 ) [ erratum - ibid .
d * 64 * , 039901 ( 1997 ) ] [ arxiv : hep - ph/9701219 ] ; + m. drees , y. g. kim , m. m. nojiri , d. toya , k. hasuko and t. kobayashi , phys .
d * 63 * , 035008 ( 2001 ) [ arxiv : hep - ph/0007202 ] .
j. r. ellis , t. falk , g. ganis , k. a. olive and m. schmitt , phys .
d * 58 * ( 1998 ) 095002 [ arxiv : hep - ph/9801445 ] .
v. d. barger , m. s. berger and p. ohmann , phys .
d * 49 * ( 1994 ) 4908 [ arxiv : hep - ph/9311269 ] .
w. de boer , r. ehret and d. i. kazakov , z. phys .
c * 67 * ( 1995 ) 647 [ arxiv : hep - ph/9405342 ] .
m. carena , j. r. ellis , a. pilaftsis and c. e. wagner , nucl .
b * 625 * ( 2002 ) 345 [ arxiv : hep - ph/0111245 ] .
l. e. ibanez , c. lopez and c. munoz , nucl .
b * 256 * ( 1985 ) 218 .
t. falk , k. a. olive , l. roszkowski and m. srednicki , phys .
b * 367 * ( 1996 ) 183 [ arxiv : hep - ph/9510308 ] ; + a. riotto and e. roulet , phys .
b * 377 * ( 1996 ) 60 [ arxiv : hep - ph/9512401 ] ; + a. kusenko , p. langacker and g. segre , phys .
d * 54 * ( 1996 ) 5824 [ arxiv : hep - ph/9602414 ] ; + t. falk , k. a. olive , l. roszkowski , a. singh and m. srednicki , phys .
b * 396 * ( 1997 ) 50 [ arxiv : hep - ph/9611325 ] .
j. r. ellis , a. ferstl and k. a. olive , phys .
b * 532 * , 318 ( 2002 ) .
j. r. ellis , a. ferstl and k. a. olive , phys .
b * 481 * , 304 ( 2000 ) . j. r. ellis , a. ferstl and k. a. olive , phys . rev .
d * 63 * , 065016 ( 2001 ) .
j. r. ellis and k. a. olive , phys .
b * 514 * , 114 ( 2001 ) [ arxiv : hep - ph/0105004 ] .
d. abrams _ et al .
_ [ cdms collaboration ] , arxiv : astro - ph/0203500 .
r. jakob , arxiv : hep - ph/0206271 . | the supersymmetric extension to the standard model offers a promising cold dark matter candidate , the lightest neutralino .
i will review the phenomenological and cosmological constraints on the supersymmetric parameter space and discuss the prospects for the detection of this candidate in both accelerator and direct detection searches .
0.0 in -0.5 in 12m_1/2 hep - ph/0211064 + tpi minn02/45 + umn th2117/02 + november 2002 * cdm in supersymmetric models * keith a. olive@xmath0 @xmath1_theoretical physics institute , school of physics and astronomy , + university of minnesota , minneapolis , mn 55455 _ |
behaviour in field theories with nonlocal interactions in the time - like momentum region is the central concern of this paper .
nonlocal quantum field theory has a long history .
initially it was invented as an attempt at solving the problem of uv divergences .
however after the development of renormalisation techniques for local qft the study of nonlocal interactions became a field of academic interest .
nevertheless the challenge of constructing self - consistent and mathematically rigorous quantum field theories with nonlocal interactions and their phenomenological application have been attracting permanent attention . as a result of this long activity of several decades the general principles of quantisation of nonlocal field theories have been formulated .
various issues related to quantisation , unitarity of the @xmath0-matrix , realisation of causality , the validity of froissart - type bounds at high energy and the physical interpretation of nonlocal fields can be found for instance in @xcite .
for the present paper the essential result is that for a theory with typical growth of amplitudes in the complex momentum plane and @xmath1 the scale of nonlocality the froissart type bound on the total cross - section @xcite , @xmath2 @xmath3 can be derived assuming * unitarity of the @xmath0-matrix , * analyticity of amplitudes on the martin - lehmann ellipse .
these two requirements are fulfilled in local as well as in nonlocal qft with the form factors being entire analytical functions of momenta . despite this upper bound on the total - cross section , practically useful procedures for calculating observables at high energy in nonlocal qft
have been not formulated .
thus , since amplitudes are exponentially growing within perturbation theory , the common wisdom is that nonlocal theories necessarily display pathological behaviour at high energy , which is usually considered as a fundamental drawback of nonlocal qft models @xcite .
the aim of the present paper is to give an example of such a practical calculation for the total annihilation cross section in a confining nonlocal model exhibiting a regge spectrum of bound states . in order to decode this statement of the problem we first have to explain its context .
an essential disadvantage of nonlocal theories has always been seen in the functional ambiguity in the choice of form factors related to the nonlocal character of interactions . though requirement of unitarity restricts the form factors to be entire analytical functions of momenta , an ambiguity within this class of form factors is unavoidable as far as nonlocality is introduced as a fundamental property of the theory .
an unexpected source for a nonlocal character of quantum fields has been pointed out by leutwyler @xcite who has noticed that the propagator of the massless charged scalar field in the presence of background ( anti-)self - dual homogeneous gauge field takes the form @xmath4 here @xmath5 is euclidean momentum , and scale @xmath1 is related to the strength of the background gauge field .
the nonlocal form factor @xmath6 appears here due to the presence of strong nonperturbative gauge fields , and thus its appearance itself and particular form can be related to the properties of the physical vacuum of the system . for vanishing field strength @xmath7 , and the usual scalar propagator
is recovered .
such an observation was immediately recognised as potentially interesting in application to the confinement problem in qcd @xcite . for physical interpretation
it is important that the nonperturbative background gauge field eliminates the pole in the momentum space propagator of a charged field rendering it an entire analytical function .
such an analytical property of the propagator can be regarded as dynamical confinement of the charged field since it has no particle interpretation @xcite .
it is also important that such well established perturbative properties of qcd as asymptotic freedom at short distances are unaffected by the nonlocality generated by such background gluon fields .
we have here an example of a theory defined by a local classical lagrangian but with nonlocal behaviour of the quantised fields appearing as a consequence of a nonperturbative gauge field configuration .
the ( anti-)self - dual homogeneous gluon field itself can be seen as a highly idealised analytically tractable realisation of more complicated and realistic nonperturbative gauge configurations responsible for confinement ( for instance , the domain model @xcite is a specific attempt at realising a more detailed picture of the qcd vacuum in this spirit ) . turning to confinement specifically now , this phenomenon is more than just the absence of the coloured quark - gluon states in the physical spectrum of qcd , but the mechanism by which colourless bound states are present in the spectrum of qcd .
one of the first connections between confinement and the hadron spectrum was achieved by the string picture , which provided a simple explanation for the pattern of regge trajectories from a mechanism of confinement . however , the string explanation for regge trajectories is not unique .
another , purely field theoretical approach to bound state formation is provided by the hadronisation procedure in the qcd functional integral which is closely connected with bethe - salpeter approach to the description of relativistic bound states @xcite .
it has been shown within the bosonisation procedure that nonlocality in quark propagators similar to eq .
( [ scal ] ) generated by the homogeneous ( anti-)self - dual field in qcd leads to asymptotically linear regge trajectories @xcite .
in fact this approach turned out to be successful in application to wide variety of meson spectra light , heavy - light mesons and heavy quarkonia @xcite .
the specific chiral properties of quark eigenmodes are similar to those in an instanton background and are of crucial importance in the description of light mesons in this context . in its most clean and refined form the role of nonlocality of the type eq .
( [ scal ] ) in generating a linear regge spectrum of bound states has been exposed in @xcite where the method of @xcite has been improved and applied to a scalar field model . unlike genuine qcd with all its complications ,
this model allows for analytical solution of the bethe - salpeter equation in the one - boson exchange approximation .
the lagrangian of the model has the same structure as that of the well - known wick - cutkosky model @xcite @xmath8 the wick - cutkosky model corresponds to a standard real massless scalar field @xmath9 and massive charged scalar field @xmath10 .
it was invented as a prototype theory for studying the relativistic bound state problem in quantum electrodynamics .
it turned out that the choice of euclidean momentum space propagators of the form @xmath11 leads to a bethe - salpeter equation which is analytically soluble in the one - boson exchange approximation generating a regge spectrum of relativistic bound states @xcite with mass - squared @xmath12 which is linear both in radial number @xmath13 and angular momentum @xmath14 .
thus the model given by eqs.([lagr ] ) and ( [ props ] ) can be seen as a soluble prototype of a confining theory , here with confined fundamental fields @xmath9 and @xmath10 and a regge spectrum of relativistic bound states representing the physical particle spectrum .
it is remarkable that the effective action for the composite fields describing the bound states can also be derived analytically within the bosonisation approach , thus enabling further calculations of physical quantities such as form - factors , decay widths , and scattering cross - sections .
the more realistic choice @xmath15 where nonlocality appears as an ( exponentially small at short distances ) correction to the local massless propagators , allows a variational approximate solution of the corresponding bethe - salpeter equation and displays an asymptotically linear regge spectrum @xcite . in this paper
we consider the total annihilation cross section in the model of eqs.([lagr ] ) and ( [ props ] ) using the method suggested in @xcite , where a similar problem has been studied in application to the t hooft model .
the total cross section for annihilation of a lepton - anti - lepton pair into hadrons via a scalar `` photon '' was evaluated in @xcite by means of continuation of breit - wigner formulae in the complex plane using the fact that the physical spectrum of the t hooft model consists of a regge spectrum of hadronic resonances with the finite widths . in our case
the bound states form a physical regge spectrum of resonances and the effective `` hadron '' action is known explicitly .
the scattering problem can be formulated in a standard way for these hadrons and a unitary @xmath0-matrix can be defined .
the total cross section is calculated by means of the optical theorem , namely via the imaginary part of the correlator of two scalar currents @xmath16 as in the thooft model the crucial point is to incorporate into the calculation of the rhs of eq.([polar ] ) the finite widths of the resonances of the physical spectrum .
as explained below , the imaginary part of the correlator is calculated in the approximation based on an extended breit - wigner formula .
this approximation is valid if resonances are narrow enough , which is satisfied in the model under consideration for the seven lowest resonances corresponding to the energy interval @xmath17 .
it should be stressed that our task is not verification of the asymptotic inequality eq.([bound ] ) . at asymptotically high energy
the number of resonances involved in the calculation of the total cross - section grows rapidly , their width becoming so large that they strongly overlap . in this regime
an approximation based on the breit - wigner formula is not valid and methods in the spirit of the statistical bootstrap @xcite should be more appropriate .
as already stated , we intend rather to give a particular example demonstrating a computation of physical quantities at relatively high energies within a nonlocal model with confinement of fundamental fields and a regge spectrum of bound states .
the final result for our estimate of the spectral density is given in the upper plot of fig.[fig : sigma ] .
we conclude that in a nonlocal theory with confined fundamental fields and exponentially growing amplitudes , taking into account the physical bound states can suppress unphysical growth of observable quantities at higher energies .
the finite width of these resonances is of particular importance .
exponential growth of the widths is a particularly distinctive feature of the analytic confinement scenario .
it leads to the convergence of the breit - wigner type series for the spectral function . in a broader context , as indicated in a recent paper @xcite , exponentially growing widths of hadronic resonances can be important for understanding the phenomenology related to the quark - gluon plasma : it eliminates the divergence of the thermodynamic functions above the hagedorn temperature and is able to successfully describe lattice qcd data for the energy density .
we discuss this issue briefly in the last section .
our starting point is the euclidean functional integral @xmath18={\cal n}\int { \cal d}\phi{\cal d}\phi{\cal d}\phi^\dagger \exp\left\{\int d^4x \left [ { \cal l}(x ) - \phi^\dagger(x)\phi(x ) i(x)\right ] \right\},\end{aligned}\ ] ] where the lagrangian @xmath19 is defined by eqs.([lagr ] ) and ( [ props ] ) , and the external current @xmath20 @xmath21 represents interaction with the scalar lepton current @xmath22 by virtue of scalar particle ( analogous to photon ) exchange reflected in the propagator @xmath23 .
the propagators @xmath24 and @xmath25 given in eq.([props ] ) are entire functions such that no particles can be associated with the fields @xmath10 and @xmath9 .
these fields represent fluctuations localised in space and time .
a typical space - time size of fluctuations is set by a confinement scale @xmath1 .
the physical particle spectrum of the system can be recovered by means of the `` bosonisation '' procedure applied to the functional integral eq.([fint0 ] ) . for a detailed description of this procedure for this case
we refer the reader to @xcite .
essentially , bosonisation requires solving the bethe - salpeter equation in the one - boson exchange approximation , which can be done analytically due to the simple gaussian form of propagators .
the main steps are as follows . integrating out the field @xmath9 one arrives at a quartic interaction of the scalar fields @xmath10 : @xmath26 = \frac{g^2}{2 } \int d^4x_1 d^4 x_2 \phi^{\dagger}(x_1)\phi(x_1 )
d(x_1-x_2 ) \phi^{\dagger}(x_2)\phi(x_2).\ ] ] introducing a complete orthonormal set of functions @xmath27 , corresponding to the radial quantum number , total momentum and magnetic numbers @xmath28 , the four point interaction can be rewritten as an infinite sum of products of a non - local currents @xmath29 with @xmath30 one introduces auxiliary fields @xmath31 with a subsequent gaussian integration over the field @xmath10 . as a result the original functional integral ( [ fint0 ] ) turns out to be identically rewritten in terms of composite fields @xmath31 representing collective excitations in the system @xmath32 \right\}\end{aligned}\ ] ] where the dimensionless coupling constant @xmath33 , and @xmath34 = - { \rm { tr } } [ \ln(1- g\psi_{\cal q } v_{\cal q } s-
i s ) + \frac{g^2}{2 } \psi_{\cal q } v_{\cal q } s \psi_{\cal q ' } v_{\cal q ' } s ] \ ] ] describes interactions between composite fields from which the quadratic term in the expansion of the logarithm is subtracted and added to the original quadratic part , @xmath35 the nontrivial requirement is that the basis @xmath27 be chosen such that the self - energy @xmath36 and hence the quadratic part of the effective action is diagonal in the quantum numbers @xmath37 , @xmath38 bound state masses are then real solutions to @xmath39 diagonalisation of the self - energy is equivalent to the solution of the bethe - salpeter equation in the one - boson exchange approximation . in the momentum representation propagators and vertices involved in the effective action
have the form @xcite as a function of radial quantum number .
deviation from strictly exponential behaviour is due to the polynomial @xmath40 in @xmath41 . ]
@xmath42 , \nonumber\\ & = & ( -1)^{n+l } \lambda c_{n l } t^l_{\mu_1 , \dots , \mu_l } ( k ) l^{l+1}_n(a k^2 ) e^{-b k^2}\end{aligned}\ ] ] with @xmath43 , @xmath44 , @xmath45 and @xmath46 here the angular part of the vertex is given by @xmath47 , the irreducible tensors of the euclidean rotation group @xmath48 , @xmath49 with @xmath50 being gegenbauer polynomials .
the radial part corresponds to the laguerre polynomials @xmath51 , with @xmath52 calculation of the self - energy gives @xmath53 and the square of the bound state masses read @xmath54\ ] ] which manifests a linear regge spectrum .
finally the fields should be rescaled @xmath55 in order to ensure the correct residue of the propagator at the mass pole .
the coupling constant @xmath56 is defined as a dimensionless quantity .
thus the final form of the functional integral for composite fields is @xmath57 \right\}.\end{aligned}\ ] ] the original coupling constant @xmath58 enters only the quadratic part of the effective action , while the remaining terms contain the effective coupling constant @xmath56 .
the formal functional integral ( [ fint2 ] ) can be used to define a unitary nonlocal theory for the interacting fields @xmath31 by defining the appropriate gaussian measure for its computation .
namely , one rewrites @xmath59 as @xmath60 , \\ & & 1 -\alpha e_{\cal q}(m^2_{\cal q})=0 . \nonumber\end{aligned}\ ] ] this decomposition exists since @xmath59 is an entire analytical function .
if one uses now the first term of the rhs of eq.([measure ] ) to define the standard free field gaussian measure for computation of the functional integral then the theory of interacting composite fields @xmath31 so defined is unitary since all nonlocal form factors in the action are entire analytical functions , and the only singularities which appear in the @xmath0-matrix are related to the physical particles associated with fields @xmath31 . in particular , this theory satisfies the conditions required for the derivation of the high energy bound eq.([bound ] ) with @xmath61 .
such a `` free field '' gaussian measure is appropriate for computing the processes in which all composite particles are almost on - shell . in this case
@xmath62=o((m^2_{\cal q}+p^2)^2)$ ] is small and can be considered as a perturbation .
according to the optical theorem the total cross section for the inclusive process @xmath64 @xmath65 , namely the imaginary part of the correlator of two scalar currents @xmath66 in terms of the effective action in eq.([fint1 ] ) the spectral density is given by the imaginary part of diagrams with one , two and generally @xmath13 intermediate `` hadrons '' as shown in fig.[fig : diagr2]a - c . since the propagator of the constituent field @xmath10 ( thin solid line ) and the vertices @xmath67 ( see eq.([vertex ] ) ) are entire analytical functions no contribution to the imaginary part of diagrams in fig.[fig : diagr2 ] can come from the constituent field loops .
in particular , the diagram in fig.[fig : diagr2]d has no imaginary part and does not contribute to the spectral density . ) ] in the breit - wigner approximation of the resonance propagator , the contribution of diagram in fig.[fig : diagr2]a to the spectral density takes the form @xmath68 } { [ ( m^2_{n0}-s)]^2+[\varepsilon+\gamma_{n}(m^2_{n0})]^2},\end{aligned}\ ] ] where @xmath69 is the width of the resonance , which is equal to zero for the lowest state .
it should be noted that the expansion analogous to that in @xmath70 in the t hooft model or @xmath71 is represented in the model under consideration via an expansion in the number of loops which include internal resonance propagators .
further calculations will be based on this decomposition over `` hadronic '' loops . at this point
we take this expansion as a way of classifying the various contributions , and will discuss its validity in more detail shortly . as for the leading order contributions in @xmath70 in the t hooft model , the bound states in the nonlocal model are stable in the zeroeth order of this expansion over `` hadronic '' loops . in this approximation , namely with decay widths neglected in eq.([densitybw ] ) , the spectral density is simply an infinite equidistant sequence of delta functions .
the lowest order contribution to the widths of resonances in this loop expansion are the two - particle decays .
numerically the widths for few lowest resonances are shown in fig.[fig : width - n ] .
one can see that the width is exponentially growing with @xmath13 , which ensures convergence of the series for the spectral density in eq.([densitybw ] ) for any @xmath72 since @xmath73 for @xmath74 .
the exponential growth of the decay width is precisely due to nonlocality .
this should be compared with the behaviour @xmath75 and @xmath76 in the t hooft model or string based picture of confinement .
we comment in the discussion section on whether available experimental data can distinguish between these two qualitatively different dependencies . summing the convergent series in eq.([densitybw ] )
gives the spectral density shown in fig.[fig : rhobw ] .
qualitatively it has a reasonable form and decreases at large @xmath72 .
the advantage of the breit - wigner approximation for the resonance propagator is that it explicitly ensures the correct analytical properties : only physical singularities in the complex @xmath72-plane are present in the breit - wigner propagator . however , there are three important factors at large @xmath72 which are missed in the spectral density calculated through the breit - wigner propagator with inclusion of two - particle decay widths .
first of all decays to three and more particles become important at large @xmath72 . accounting for them would broaden the resonances and lower the spectral density .
the second contribution comes from the diagrams with two or more intermediate bound states ( see fig.[fig : diagr2]b , c ) which would increase the spectral density at large @xmath72 .
these two factors indicate that eq.([densitybw ] ) can not give a truly reliable result for asymptotically large @xmath72 .
at most it provides a hint at the tendency of @xmath77 to decrease at larger @xmath72 .
the third factor is that the breit - wigner approximation is valid , strictly speaking , only in the vicinity of a resonance pole and if the resonance is narrow enough .
this means for @xmath72 located away from the resonance position the contribution of the @xmath13-th resonance to the series eq.([densitybw ] ) might not be reflected correctly . in terms of the effective action discussed in the previous section the breit - wigner approximation for
the propagator corresponds to the `` free field '' gaussian measure in the functional integral with the addition of the on - shell widths of resonances .
it is clear that corrections coming from the second term in the rhs of eq.([measure ] ) are not small for @xmath78 being off - shell .
certain resummations are necessary . as a function of the energy variable @xmath72 ( upper plot ) and the contributions @xmath79 of few lowest radial numbers to the total spectral density ( lower plot ) ] the effective action at our disposal offers an extension of the breit - wigner representation , which can account for such resummations just mentioned via inclusion of the energy dependence of the self - energy @xmath80 , the transition amplitude @xmath81 and the imaginary part decay width @xmath82 .
this extension of the spectral density takes the form @xmath83}{[1-\alpha e_{n0}(s)]^2+[\varepsilon+\tilde\gamma_{n0}(s)]^2 } , \nonumber\\ & & \tilde\gamma_{n0}=\frac{16\pi^2\alpha}{h_{n0}\lambda^2}\gamma_{n0}\end{aligned}\ ] ] where the summation spans only radial excitations with @xmath84 because of conservation of orbital momentum which is a good quantum number in this problem .
the transition amplitude @xmath81 is given by the diagram in fig.[fig : diagr1]b .
simple calculation gives @xmath85 the quantity @xmath82 is the total width of two - particle decays of a resonance with orbital momentum @xmath84 , radial number @xmath13 and squared - energy @xmath72 .
details of the calculation of @xmath82 are given in the appendix .
as a function of energy variable @xmath72 for the few lowest resonances . ] numerical results for @xmath82 are given in fig.[fig : width ] , which shows its exponential growth with @xmath72 , and @xmath86 for large @xmath72 .
this inequality results in the behaviour of the separate terms @xmath79 shown in the lower plot in fig.[fig : sigma ] .
summing up these terms gives the spectral density shown in the upper plot in fig.[fig : sigma ] .
qualitatively this spectral density reproduces the character of the simple breit - wigner approximation , fig.[fig : rhobw ] .
the oscillations seen in @xmath79 in fig.[fig : sigma ] additional to the resonance peaks are reflections of two features .
the first is the corresponding oscillations in @xmath87 in fig.[fig : width ] which are due to the opening of new decay channels at certain values of @xmath72 .
the second , and more important , cause is the functional structure of the amplitude @xmath88 ( see appendix ) : essentially it is a product of a polynomial and exponent in @xmath72 . again
the spectral density shows several sharp peaks corresponding to the lowest narrow resonances , then it has a maximum and a tendency to decrease at higher energy .
it should be stressed that @xmath77 is limited by the region of not very large values of @xmath72 because of the same as for @xmath89 two reasons ( decays into more than two final states and diagrams with two or more intermediate resonances ) .
however unlike the plain breit - wigner approximation the extended formulae for the propagator should be more appropriate for the values of @xmath72 away from resonance position although only in the vicinity of the real axis in the complex @xmath72-plane .
the reason preventing it from being a good approximation in the whole complex plane is hidden in the property of entire functions which have infinitely many zeroes in the complex plane .
in addition to the desirable physical pole corresponding to the resonance , the extended propagator has artificial unphysical poles for complex values of @xmath72 .
as soon as these poles are sufficiently far from the real axis and do not noticeably affect the form of @xmath77 the approximation eq .
( [ density ] ) can be considered appropriate . in the absence of the term with @xmath90 in eq .
( [ density ] ) and using eq .
( [ enl ] ) , the unphysical singularities would appear at @xmath91 for any nonzero integer @xmath92 .
the addition of @xmath87 shifts all poles to the complex plane , and a cursory numerical check reveals that the artificial singularities produced by the resummation are located sufficiently far from the real axis .
in summary , we have presented the spectral density in the breit - wigner approximation for a nonlocal model with confined constituent fields but a regge spectrum of bound states .
the two - particle decay widths have been computed and incorporated into the evaluation of the spectral density .
the main observations are : firstly , due to rapid growth of resonance widths with growing radial number the formal sum over resonances ( which represents the spectral density ) is convergent ; secondly , apart from the expected delta - function like peaks corresponding to the lowest narrow resonances the spectral density in this approximation decays at larger values of the energy variable @xmath72 as shown in fig.[fig : rhobw ] .
the breit - wigner approximation roughly reproduces the behaviour of the resonance propagator as a function of @xmath72 away from the resonance position . in order to improve this a resummation of the series over the effective `` bound state - constituent field ''
coupling constants @xmath56 are necessary , what we have called an `` extended '' breit - wigner approximation . the spectral density computed by means of such
a resummation shows qualitatively the same behaviour as the breit - wigner spectral function
there are several peaks corresponding to the lowest narrow resonances and a tendency to decay at higher energies .
as already discussed , the resummation introduces artificial unphysical singularities .
however , in the particular case under consideration these unphysical singularities are located sufficiently far from the real @xmath72-axis .
insofar as we consider the resonance propagator in the vicinity of the real axis , the approximation based on resummation of higher orders in @xmath56 is quite reliable .
it should be noted that , in the context of nonlocal quantum field theory , the problem of unphysical singularities appearing in the complex @xmath72-plane is akin to the landau pole problem in local quantum field theory : a partial resummation of the perturbation series is generally hard to achieve without introducing unphysical singularities , which applies both to local and nonlocal models .
there are two aspects to this issue .
the practical way to manage is to look for summation prescriptions which are appropriate in the restricted region of the complex momentum plane relevant to a given problem ; this is what we have pursued in the present work .
a more fundamental approach to the problem would require formulation of the general principles and methods of summation which would avoid the appearance of unphysical singularities . as mentioned , the asymptotic dependence of two particle decay widths on the bound state quantum number in the nonlocal model @xmath93 is drastically different from the dependence in the thooft model @xmath94 , and semi - classical string based estimations of @xcite where @xmath95 .
the computation in the t hooft model is similar to ours in that decays of excitations to lower states at the same trajectory are considered , while the computation of @xcite applies to all possible decays .
it is interesting to take a look at experimental data on decay widths in the hope to distinguish between these qualitatively different situations .
unfortunately the available data seem to be inconclusive in this respect .
the problem is that for the purpose of extracting asymptotic dependence of the width on radial or orbital number one should consider only decay modes of a given excited state to states lower on the same regge trajectory . in the real multiflavour world , decays of excited states are dominated by the modes into final states which include ground state mesons from regge trajectories of other flavour octet states .
a typical example is the family of orbital excitations of @xmath96 , @xmath97 with the total widths ( using @xcite and data available through the 2003 partial update ) @xmath98 looking as well fitting linear in @xmath14 behaviour of the width ( though @xmath99 expected in the domain model @xcite would also fit but be implausible with the limited range of data ) .
however the problem is that decay modes which contribute to the widths of these resonances all include @xmath100 , @xmath101 , @xmath102 , @xmath103 mesons which obviously do not belong to the trajectory of @xmath96 .
so , this approximately linear dependence in @xmath14 does not contain the information which we are interested in .
it can be strongly influenced by chiral dynamics , and its description requires a model with both chiral symmetry and confinement implemented simultaneously ( for instance the model of @xcite has a chance to provide such a suitable framework for calculating partial widths known from experiment and enabling thus a detailed analysis ) .
the clearest case would be the regge family of radial excitations of a pseudoscalar meson like @xmath104 , however the widths and especially the partial widths for different decay modes are not known for these mesons with the required accuracy .
these two examples are quite typical , and we conclude that it is not yet possible to distinguish between the two asymptotic behaviours on the basis of experimental data . it should be stressed that the nonlocal model , despite the exponential asymptotics for the width , indicates very small partial widths of the few lowest resonances for decays into lower states on the same trajectory due to the polynomial pre - exponent , and can considerably deviate from strictly exponential form as can be seen from fig .
[ fig : width ] .
these partial modes would anyway be screened by other modes in the multiflavour real world .
a further signature of exponential behaviour for larger @xmath105 states would be that observable trajectories can not be `` long '' , since resonances with radial or orbital numbers bigger than some critical @xmath106 ( @xmath107 in fig .
[ fig : width ] ) or @xmath108 are too wide to be observable .
such a sharp cut - off of the trajectories would be typical for exponential dependence , but not for linear or square root behaviour .
the property that the known regge trajectories , for example , for unflavoured mesons cut - off at about @xmath109 gev may be an indication of this feature .
the crucial role of the finite widths in modulating the behaviour of the cross - section for growing energy can be understood in terms of the energy being dissipated into the creation of unstable particles , which decay into more stable particles and so on . in this respect ,
the mechanism is close in spirit to the hagedorn mechanism for a maximal temperature in hadron - hadron collisions @xcite where one has no recourse to the optical theorem and thus the machinery of the statistical bootstrap is necessary _ a priori_. in our case , this framework would be useful once the narrow width approximation has broken down .
in this context we mention that exponentially growing decay widths can impact on the deconfinement phase transition in particular the persistence of hadronic properties beyond the critical temperature .
for example in @xcite , an ansatz involving @xmath110 for the width contribution to the spectral function of a hadron gas at temperatures greater than the hagedorn temperature , @xmath111 , reproduces the characteristic slow approach to the stefan - boltzmann ideal quark - gluon gas result seen in lattice calculations @xcite .
interestingly , the hagedorn temperature scale , traditionally argued to be determined by the mass of the lightest hadron , is commensurate with the confinement scale which figures prominently in analytical confinement and the exponentially growing decay widths discussed in this paper : @xmath112 .
the possibility that the confinement scale is more significant in fixing the hagedorn temperature , especially in the chiral limit , has been argued recently @xcite .
further study of the consequences of analytic confinement to high temperature behaviour is appropriate .
finally , we have not addressed the relation between our study to the idea and formalism of `` quark - hadron duality '' . with the purely gaussian propagators treated in this article
, such an investigation would be pointless .
however this problem becomes well - formulated if one considers the more realistic propagators of the form eq .
( [ scal ] ) .
this propagator has the same short distance behaviour as the local one and nonlocal corrections are exponentially small in the deep euclidean region . with such propagators
one can address the issue of quark - hadron duality within a model with dynamical confinement of constituent fields , in a spirit close to that of @xcite .
the width for two - particle decay for a state with the time - like four momentum squared @xmath113 decaying to particles with masses @xmath114 and @xmath115 is given by the standard equation @xmath116 the modulus of the amplitude is independent of the momenta .
thus for the width for the decay of the state with angular momentum @xmath84 and radial number @xmath13 into the final states with masses @xmath114 and @xmath115 and quantum numbers @xmath117 and @xmath118 respectively , we get @xmath119 where @xmath120 is the two - particle phase space @xmath121 the total width is then @xmath122 the amplitude @xmath123 which should be computed is given by the diagram fig.[fig : diagr1]c . for arbitrary @xmath124 and polarisations ,
the corresponding loop integral takes the form ( all momenta below represent dimensionless ratios , e.g. @xmath125 ) @xmath126 since the orbital momentum of the incoming state is zero , the final states are required to be on - shell , and the angular momentum is a conserved quantum number in this problem , we need to extract from this amplitude the contribution of final states with total angular momentum equal to zero . thus we need the term with the angular momentum @xmath84 in the decomposition of the amplitude @xmath127 over the irreducible tensors @xmath128 with @xmath129 .
the required contribution is proportional to the trace of @xmath127 for @xmath130 @xmath131 where @xmath132 is defined as a dimensionless quantity .
taking into account eqs.([vertex ] ) and ( [ gegen ] ) we arrive at @xmath133 \nonumber\\ & \times&l^1_n[aq^2 ] l^{l+1}_{n_1}[a ( q+(p - k_1)/2)^2 ] l^{l+1}_{n_2}[a(q - k_1/2)^2 ] c_l^{(1)}[(q+(p - k_1)/2)(q - k_1/2 ) ] \nonumber \\ & \times & \exp\left\ { -b q^2 - b(q+(p - k_1)/2)^2 - b(q - k_1/2)^2 -(q+p/2)^2 - ( q+p/2-k_1)^2 - ( q - p/2)^2 \right\},\end{aligned}\ ] ] where @xmath50 are gegenbauer polynomials . the gaussian integral over the loop momentum @xmath134 can be performed analytically . the most convenient way to do
this is to use a generating function for laguerre polynomials @xmath135 both sides of eq.([ampl ] ) are multiplied by @xmath136 and summed over @xmath137 and the result of integration is expanded in a power series in each of @xmath138 and @xmath139 .
the coefficients of expansion gives the amplitudes @xmath140 .
the gegenbauer polynomials are substituted in explicit form .
this procedure can be easily implemented in _
mathematica _ or _
maple_. finally the amplitude @xmath140 as a function of @xmath72 is a product of a polynomial in @xmath72 and an exponential of @xmath72 .
the degree of the polynomial depends on @xmath13 , @xmath141 , @xmath142 and @xmath14 .
the total two - particle decay width of the resonance with four momentum squared @xmath72 , zero orbital momentum and radial number @xmath13 entering eq.([density ] ) is then @xmath143 where all states with nonzero phase space are summed .
this work was supported by the university of adelaide under the small research grants scheme and partially by the grant rfbr 04 - 02 - 17370 .
ack is supported by a research fellowship of the australian research council .
s.n.n was supported by the dfg , contract sm70/1 - 1 .
we thank garii efimov , rajeev bhalerao , tony thomas , rod crewther , johann rafelski and frieder lenz for clarifying discussions and ian cloet for helpful assistance .
david blaschke is thanked for attracting our attention to the work @xcite . 9 n.n .
meiman , jetp , 47 ( 1964 ) 1966 , a.m. jaffe , phys .
lett . , 17 ( 1966 ) 661 .
fainberg , and m.a .
soloviev , theor.math.phys.93 ( 1992 ) 1438 ; _ ibid _ ann .
113 ( 1978 ) 421 ; v.ya .
fainberg , and m.z .
yoffa , nuovo cim . a , 5 ( 1971 ) 275 ; _ ibid _ jetp , 56 ( 1969 ) 1644 .
_ ibid _ commun.math.phys . , 57 ( 1977 ) 149 ; v.a .
alebastrov , and g.v .
efimov , commun .
phys . , 31 ( 1973 ) 1 ; _ ibid _ commun . math
. phys . , 38 ( 1974 ) 11 ; g.v .
efimov , and o.a .
mogilevsky , nucl .
b44 ( 1972 ) 541 ; g.v .
efimov , `` nonlocal interactions of quantised fields '' , nauka , moscow ( 1977 ) ( in russian ) ; kh .
namsrai , `` nonlocal quantum field theory and stochastic quantum mechanics '' , d. reidel publishing company , dordrecht / boston / lancaster / tokyo ( 1986 ) . g.v .
efimov , theor .
phys . , 128 , 1169 ( 2001 ) .
moffat , phys . rev .
d 41 ( 1991 ) 1177 ; d. evens _
et al _ , phys . rev .
d 43 ( 1991 ) 499 .
g. kleppe and r.p .
woodard , nucl .
b388 ( 1992 ) 81 ; _ ibid _ ann . phys .
221 ( 1993 ) 106 .
h. leutwyler , phys .
b96 ( 1980 ) 154 .
efimov , m.a .
ivanov , `` the quark confinement model of hadrons '' , iop , bristol ( 1993 ) .
feynman , a.r .
hibbs , `` quantum mechanics and path integrals '' , new york , mcgraw - hill , 1965 .
efimov , extended version of the talk at the xii international conference `` selected problems of modern physics '' , dubna , june 8 - 11 , 2003 , to appear in phys . part .
35 ( 5 ) , 2004 .
r. alkofer , and l. von smekal , phys .
353 ( 2001 ) 281 .
a. jain , and s.d .
joglekar , hep - th/0307208 .
j. finjord , nucl .
b 194 ( 1982 ) 77 ; _ ibid _ nucl .
b 222 ( 1983 ) 507 ; a. i. milshtein and y. f. pinelis , z. phys .
c 27 ( 1985 ) 461 .
kalloniatis , s. n. nedelko , phys .
d 64 ( 2001 ) 114025 ; _ ibid _ phys . rev .
d 66 ( 2002 ) 074020 .
kalloniatis , s. n. nedelko , phys .
d 69 ( 2004 ) 074029 .
j. praschifka , c.d .
roberts , r.t .
cahill , phys .
d 36 ( 1987 ) 209 ; h. reinhardt , phys .
b 244(1990 ) 316 - 326 ; d. ebert , h. reinhardt , nucl.phys .
b 271 ( 1986 ) 188 . g.v .
efimov , s.n .
nedelko , phys . rev .
d 51 , 176 ( 1995 ) .
burdanov , g.v .
efimov , s.n .
nedelko , s.a .
solunin , phys . rev .
d 54 , 4483 ( 1996 ) ; _ ibid _ hep - ph/9806478 ( 1998 ) .
efimov , g. ganbold , phys.rev.d65 ( 2002 ) 054012 .
wick , phys .
96 ( 1954 ) 1124 ; r.e .
cutkosky , _ ibid _ 1135 . b. blok , m. shifman , d - x .
zhang , phys.rev.d57 ( 1998 ) 2691 .
r. hagedorn , suppl .
cim . 3 , 147 ( 1965 ) .
a. casher , h. neuberger , s. nussinov phys.rev .
d 20 ( 1979 ) 179 k. hagiwara et al .
( particle date group ) , phys .
d 66 , 010001 ( 2002 ) .
see 2002 mass - width data available at : http://pdg.lbl.gov/rpp/mcdata/ and partial 2003 data available on the web page .
blaschke , k.a .
bugaev , nucl - th/0311021 .
f. karsch , k. redlich , a. tawfik , eur .
j. c29 ( 2003 ) 549 .
p. blanchard , s. fortunato , h. satz , eur .
j. c34 ( 2004 ) 361 . | we study the role of finite widths of resonances in a nonlocal version of the wick - cutkosky model .
the spectrum of bound states is known analytically in this model and forms linear regge tragectories .
we compute the widths of resonances , calculate the spectral density in an extension of the breit - wigner _ ansatz _ and discuss a mechanism for the damping of unphysical exponential growth of observables at high energy due to finite widths of resonances . |
a quick search through the nasa / ads abstract service indicates that more than @xmath0 5000 papers were published over the past five years on agns ! clearly , a comprehensive review of this literature is well beyond the scope of the present article .
the present review highlights important results from recent large - scale ground - based surveys and space missions , and describes the consequences of these results on our understanding of radio - quiet agns .
the standard paradigm for radio - quiet agns is described in 2 along with a few key supporting observations .
important new constraints on this standard picture have been derived from recent uv / optical , infrared , and x - ray surveys ; these are discussed in 3 . in 4 , the nature of the connection between the supermassive black hole , host galaxy , circumnuclear starburst , and agn is briefly reviewed taking into account recent new data .
the issue of agn fueling / triggering is addressed in 5 .
a summary of the outstanding issues is given in 6 along with suggestions for future avenues of research .
2.6 in figure 1 is an idealized representation of the inner structure of radio - quiet agns .
direct observational support for this picture comes from the detection in a few radio - quiet agns of h@xmath1o - masing disk - like structures in orbit around central masses of a few @xmath2 10@xmath3
10@xmath4 m@xmath5 on scales of 0.1 1.0 pc ( @xmath0 10@xmath6 @xmath7 , where @xmath8 is the gravitational radius ) .
the most convincing cases so far are ngc 4258 ( miyoshi et al .
1995 ) , ngc 1068 ( greenhill et al . 1996
; greenhill & gwinn 1997 ; gallimore et al .
1997 , 2001 ) , and ngc 4945 ( greenhill , moran , & herrnstein 1997 ) .
a survey by braatz , wilson , & henkel ( 1997 ) failed to detect h@xmath1o masers in seyfert 1s , perhaps indicating a less favorable face - on disk geometry in these objects ( small @xmath9 along the line of sight ) , as expected in the unification model of seyfert galaxies ( fig .
indirect evidence for accretion disk in radio - quiet agns comes from the presence of broad ( @xmath10 70,000 km s@xmath11 in mcg06 - 30 - 15 ) , skewed and redshifted fe k@xmath12 lines in seyfert 1s .
this line probes the purported accretion disk within a few @xmath7 of the black hole ( e.g. , tanaka et al . 1995 ; iwasawa et al .
1996 , 1999 ; nandra et al . 1997 ; guainazzi et al . 1999 ; wilms et al . 2001 ; lee et al . 2002 ; fabian et al . 2002 ; turner et al .
2002 ; see contribution by fabian in this volume ) .
the presence of large - scale ( photo-)ionization cones in radio - quiet agns is further indirect evidence for an inner disk structure in these objects ( e.g. , haniff , wilson , & ward 1988 ; pogge 1989 ; tadhunter & tzvetanov 1989 ; wilson & tzvetanov 1994 ; mulchaey et al .
1996a , 1996b ; pogge
2000 ; veilleux 2002 ; kinkhabwala et al .
2002 ) .
historically , spectropolarimetric studies of radio - quiet agns have been the main driver behind the unification model of seyfert galaxies ( e.g. , antonucci & miller 1985 ; miller , goodrich , & matthews 1991 ) .
but recent spectropolarimetric surveys indicate that only @xmath0 30 50% ( _ i.e. _ not all ) of seyfert 2 galaxies harbor hidden broad - line regions ( hblrs ; e.g. , tran 2001 ) .
seyfert 2s with hblrs tend to have larger radio - to - fir flux ratios and warmer dust temperatures than those without ( e.g. , miller & goodrich 1990 ; kay 1994 ; heisler , lumsden , & bailey 1997 ; moran et al .
2000 ; tran 2001 ) .
these trends have been interpreted by some to imply the existence of two separate classes of seyfert 2s .
however , several authors have been quick to point out the importance of the orientation - dependent selection criteria used in many of these studies ( e.g. , antonucci 2001 ; alexander 2001 ; lumsden et al .
2001 ; gu & huang 2002 ) .
biases in luminosity , starburst contribution , and covering factor are necessarily present in these studies , so caution should be used when interpreting the results .
recent results from near- and mid - infrared spectroscopy of radio - quiet agns have provided further support for the unification model .
obscured blrs have been detected at @xmath0 2 4 @xmath13 m in a number of uv- and infrared - selected seyfert 2 galaxies ( e.g. , veilleux , goodrich , & hill 1997a ; veilleux , sanders , & kim 1997b , 1999b ; lutz et al . 2002 ) .
clavel et al .
( 2000 ) also note that the weaker 7-@xmath13 m continuum and larger equivalent widths of the pah feature in seyfert 2s relative to seyfert 1s can be explained if the continuum in the seyfert 2s is more strongly extinguished ( @xmath14 mag ) .
studies at near- and mid - infrared wavelengths can only probe down to @xmath9 @xmath15 few @xmath2 10@xmath16 @xmath17 , _ i.e. _ considerably smaller than the expected column density of the dusty torus in agn .
hard x - ray observations have extended the range of column densities up to @xmath0 10@xmath18 @xmath17 .
maiolino et al .
( 1998 ) and risaliti , maiolino , & salvati ( 1999 ) have used bepposax , asca , and ginga data to show that seyfert 2 galaxies generally have larger obscuration than seyfert 1s , as expected in the standard unification model . however
, these observations do not put any constraints on the exact geometry and location of the obscuring material in seyfert 2s ( e.g. , inner disk / torus _ vs. _ galaxy - scale circumnuclear material ? ) .
optical , uv , and x - ray observations of agns indicate that the simplest form of the standard picture ( fig .
1 ) is not tenable and must be modified to account for dependences on agn luminosity .
optical spectroscopy of infrared - selected agns ( in which orientation - dependent biases are less important ) indicates that the ratio of seyfert 2s to seyfert 1s decreases significantly with infrared luminosity ( e.g. , veilleux et al .
1995 ; veilleux , kim , & sanders 1999a ) . the well - known decrease of the c iv @xmath191549 equivalent widths with increasing luminosity ( _ i.e. _ the uv baldwin effect ; e.g. , osmer & shields 1999 ; espey & andreadis 1999 ) is another manifestation of this strong luminosity dependence .
the fe k@xmath12 line and compton scattering hump also are weaker in qsos than in seyferts ( the x - ray baldwin effect ; e.g. , nandra et al .
finally , uv and x - ray narrow absorption lines are rarely seen in the spectra of quasars but not so in seyferts ( e.g. , turner et al .
1993 ; mathur et al . 1999
; nicastro et al . 1999 ; kaastra et al . 2000 , 2002 ; kaspi et al .
all of these observations can be explained in the context of the unification model if the opening angle , or location of the inner edge , of the disk / torus / wind structure is allowed to vary with luminosity ( e.g. , lawrence 1991 ; hill et al .
1996 ; elvis 2000 ) .
the discussion in this section focusses on moderate - to - high luminosity agns .
the reader should refer to reviews by vron - cetty & vron 2000 , ho ( 2002 ) , barth ( 2002 ) and contributions in this volume by filippenko and ho for discussions of recent results on low - luminosity agns ( llagns ) .
radio - loud agns are reviewed by urry in this volume so radio surveys are not discussed here .
several new surveys have improved our knowledge of agn demography as a function of redshift ; three of them are discussed here .
the results from the hamburg / eso bright qso survey ( he s ; e.g. , wisotzki et al . 2000 ) indicate that the local qso density is @xmath0 50% larger than previously thought ( e.g. , bright quasar survey of schmidt & green 1983 ) .
the 2df qso redshift survey ( e.g. , boyle et al . 2000 ) and the sloan digital sky survey ( sdss ; e.g. , fan et al .
2001 ) have nicely confirmed that the qso density rises steeply up to @xmath20 2 and decreases beyond @xmath21 . at the time of writing , sdss has revealed four quasars at @xmath22 .
this set of high-@xmath23 qsos should at least double in size by the time the survey is completed .
these objects will provide important new constraints on the epoch of qso / galaxy formation ( see , e.g. , haiman & cen 2002 ) .
the two micron all sky survey ( 2mass ) has uncovered a large population of red ( @xmath24 ) agns at relatively small redshifts ( median @xmath23 @xmath0 0.25 ; cutri et al .
2001 ; see also gregg et al .
approximately 75% of these sources are previously unidentified agns whose space density ( @xmath0 0.5 deg@xmath25 ) is comparable to that of optical / uv selected agns of the same ir magnitudes . a large fraction ( @xmath0 80% ) of these objects are type 1 agns .
chandra follow - up observations of some of these agns indicate that all of them are x - ray faint , with the reddest being the faintest in x - rays ( wilkes et al .
interestingly , these broad - lined agns show significant absorption ( several of them have @xmath26 @xmath17 ) and may be important contributors to the cosmic x - ray background ( cxb ; cf . 3.3 ; see also webster et al .
1995 ; benn et al . 1998 ; whiting et al .
2001 for analyses of similar pks radio - selected red qsos ) .
as mentioned in 2 , x - ray observations generally provide support for the unification model of seyfert galaxies . however , there are a number of important exceptions .
these mismatches in the optical hard x - ray classification fall into three broad categories : ( 1 ) several broad - line agns show significant absorption in the x - rays ( e.g. , seyferts and qsos : fiore et al .
2001 , 2002 , wilkes et al . 2002 ; balqsos : mathur et al .
2001 , gallagher et al .
( 2 ) deep x - ray surveys have revealed a large population ( @xmath0 40 60% ) of x - ray bright sources with no obvious optical agns ( e.g. , barger et al .
2001 , 2002 ; tozzi et al . 2001 ; stern et al .
2002b ; fiore et al .
( 3 ) a number of seyfert 2 galaxies show no obvious x - ray absorption ( e.g. , ptak et al .
1996 ; bassani et al .
1999 ; pappa et al .
2001 ; panessa & bassani 2002 ) . the first category of objects may be explained in the context of the unification model if dust near the agn is different from galactic ( e.g. , different dust / gas ratio , metallicity , grain size ; veilleux et al .
1997a ; maiolino et al .
2001 ) or is not co - spatial with the ( neutral and ionized ) material producing the x - ray absorption ( e.g. , veilleux et al . 1997a ) .
x - ray bright sources with no obvious optical agns may be powered through inefficient adaf - like accretion with @xmath27 , or perhaps the apparent lack of optical agns in these sources is simply due to strong dilution effects by the host galaxy light ( e.g. , moran , filippenko , & chornock 2002 ) .
finally , narrow - line agns with no obvious x - ray absorption have often been shown to be ( nearly ) compton thick when observed in the hard x - rays .
half of all nearby optically selected seyfert 2s may fall in this category [ e.g. , risaliti et al .
1999 ; well - known examples include ngc 1068 , ngc 4945 ( done , madejski , & smith 1996 ) , and circinus ( matt et al .
luminous examples also exist , although they may be less common : e.g. , qso-2s ( norman et al . 2002 ; crawford et al . 2002 ; stern et al .
2002a ) , x - ray detected extremely red objects ( eros ) with @xmath28 ( e.g. , alexander et al .
2002 ; mainieri et al .
2002 ) , and ultraluminous infrared galaxies ( e.g. , ngc 6240 ; iwasawa & comastri 1998 ; vignati et al . 1999 ) .
the improved sensitivity of the current x - ray facilities has allowed to search for possible evolutionary effects in the x - ray spectra of quasars and agns over a broad redshift range .
brandt et al . ( 2002 ) and mathur , wilkes , & ghosh ( 2002 ) recently failed to find convincing evidence for a redshift dependence of the optical x - ray slope in qsos out to @xmath29 .
bechtold et al . (
2002 ) come to a different conclusion using a larger comparison sample . a significant population of buried type 2 agns with peak emissivity around @xmath30 ( rather than @xmath31 for type 1 agns ; 3.1 ) appears to be needed to explain the properties of the cxb and the redshift dependence of the number density of x - ray sources in the deep chandra and xmm surveys ( e.g. , rosati et al .
2002 ; franceschini , braito , & fadda 2002 ) .
these results may imply that type 1 and 2 agns follow different evolutionary paths , in disagreement with the agn unification model .
note , however , that the recent discovery of a large population of strongly absorbed type 1 agns by wilkes et al .
( 2002 ) may affect these conclusions .
ground - based and @xmath32 observations of the stellar and gas kinematics near the center of `` normal '' ( inactive ) galaxies have revealed a close connection between the mass of the supermassive black hole at the center of each of these objects and the mass of the spheroidal component ( e.g. , kormendy & richstone 1995 ; faber et al .
1997 ; magorrian et al . 1998 ; gebhardt et al . 2000 ; kormendy & gebhardt 2001 ; merritt & ferrarese 2001 ; tremaine et al .
2002 ) and perhaps also the mass of the dark matter halo ( ferrarese 2002 ) .
the results of reverberation mapping in agns suggest that the @xmath33 relations found in inactive galaxies also apply to active galaxies ( e.g. , wandel , peterson , & malkan 1999 ; kaspi et al .
2000 ; peterson & wandel 2000 ; laor 2001 ; onken & peterson 2001 ; mclure & dunlop 2002 ; ferrarese et al . 2001 ; although see krolik 2001 ) .
these results suggest a causal connection between spheroid / galaxy formation , black hole growth and agn activity ( e.g. , efstathiou & rees 1988 ; small & blandford 1992 ; haehnelt & rees 1993 ; chokshi 1997 ; silk & rees 1998 ; haiman & loeb 1998 ; fabian 1999 ; kauffmann & haehnelt 2000 ; burkert & silk 2001 ; adams , graff , & richstone 2001 ) .
this tight smbh
host galaxy connection is discussed in more detail by urry , peterson , and merritt in this volume .
a strong connection also exists between starbursts and agns ( veilleux 2001 ; also ward this volume ) .
nuclear starbursts appear to be present in several seyfert galaxies , based on the strength of uv and optical absorption and emission features from young / intermediate - age stars in the nuclear spectra of these objects ( e.g. , terlevich , diaz , & terlevich 1990 ; heckman et al .
1997 ; gonzalez delgado et al .
1998 , 2001 ) .
the presence of extended soft thermal x - ray emission in seyferts also supports this scenario ( e.g. , levenson , weaver , & heckman 2001 ) .
claims that starbursts are more common in seyfert 2s than in seyfert 1s have been made for many years ( e.g. , rodriguez - espinosa , rudy , & jones 1986 ; pier & krolik 1993 ; maiolino et al . 1995 ; nelson & whittle 1996 ; gonzalez delgado et al .
2001 ; gu , dultzin - hacyan , & de diego 2001 ) , but orientation - dependent selection criteria may seriously bias some of these samples and cause the apparent seyfert 1 / seyfert 2 dichotomy .
the nuclear starbursts in active galaxies may have a very strong impact on the evolution of the host galaxy and the agn itself .
the stellar winds and supernovae associated with the nuclear starburst may deposit sufficient amounts of mechanical energy at the centers of these objects to severely disrupt the gas phase of these systems and result in large - scale galactic winds .
a well - known example of starburst - driven wind in an active galaxy is ngc 3079 .
detailed optical , radio , and x - ray studies of this object have revealed the presence of a powerful galactic wind that is strongly interacting with the ambient material of the host galaxy ( e.g. , veilleux et al .
1994 ; cecil et al . 2001 ; cecil , bland - hawthorn , & veilleux 2002 ) .
the wind event is clearly disturbing the distribution of gas within the central kpc of this object and may therefore also affect the feeding of the agn
. the frequency of starburst - driven winds in agns is poorly constrained ( see veilleux & rupke 2002 for a discussion of a promising search technique ) , but if common these winds may provide negative feedback on the fueling mechanisms of agns ( e.g. , rupke , veilleux , & sanders 2002 ) .
the broad range in luminosity of agn ( @xmath34 10@xmath35 10@xmath36 l@xmath5 ) implies accretion rates of order @xmath37 0.001 100 m@xmath5 yr@xmath11 ( assuming a radiative efficiency of 10% in rest mass units ) ; the required accretion rates are very small for llagns but quite substantial for qsos .
the requirements on the fueling / triggering processes for agns are therefore highly dependent on the agn luminosity .
local processes are sufficient to power llagns and seyferts , but galactic - scale phenomena may be required to explain powerful qsos . a broad range of processes including galaxy interactions and mergers , large - scale and nuclear bars , and nuclear gaseous spirals
have been proposed to account for the fueling of seyferts and llagns ( see combes 2000 and the contribution by combes in this volume ) .
the lack of obvious excess of companions and mergers seems to rule out the possibility that galaxy interactions and mergers are solely responsible for triggering these objects ( e.g. , fuentes - williams & stocke 1988 ; dultzin - hacyan et al .
1999 ; de robertis , yee , & hayhoe 1998 ; virani , de robertis , & vandalfsen 2000 ) .
a slight ( 2.5-@xmath38 ) excess of bars appear to be present among seyferts ( e.g. , knapen , shlosman , & peletier 2000 ; laine et al .
2002 ; knapen this volume ; although see mulchaey & regan 1997 ; regan & mulchaey 1999 ; martini et al .
2001 ) , but this result can not explain seyfert activity in galaxies without bars .
nuclear ( 0.1 1 kpc ) gaseous spirals have been detected in most seyferts but they also appear to be present in several normal galaxies and therefore can not be the only reason for the seyfert activity ( e.g. , ford et al .
1994 ; dopita et al .
1997 ; laine et al . 1999 , 2001 ; martini & pogge 1999 ; regan & mulchaey 1999 ; pogge & martini 2002 ; emsellen this volume ) .
most likely , these various processes help replenish a reservoir of fuel on scales of @xmath0 100 pc or larger , but other processes are needed to carry the material down to sub - pc scales ( e.g. , gas instabilities , stellar ejecta , magnetic fields ) . unfortunately , very little is known observationally on scales @xmath37 10 pc .
the origin of the activity in high - luminosity objects is perhaps less ambiguous than in low - luminosity objects .
hosts of luminous qsos generally appear to be elliptical galaxies ( e.g. , dunlop et al .
2002 ; although see mcleod & mcleod 2001 ) , but signs of interaction are seen in several qsos , especially in those with infrared excess ( e.g. , surace , sanders , & evans 2001 ; canalizo & stockton 2000 , 2001 ) .
several of these objects contain large quantities of molecular gas ( e.g. , evans et al .
2001 ) , suggestive of a gas - rich merger origin .
ultraluminous infrared galaxies ( uligs ) have long been suspected to be the progenitors of optical quasars based on the fact that nearly all uligs show signs of recent galaxy mergers and starburst / agn activity , but only recently has there been a large enough homogeneous sample of uligs to look carefully at this question .
recent optical , near - infrared , and mid - infrared spectroscopic surveys of the 1-jy sample of 118 uligs indicate that @xmath0 30% ( 50% ) of the objects with log[l@xmath39/l@xmath5 ] @xmath41 12.0 ( 12.3 ) are powered predominantly by a quasar rather than a starburst ( e.g. , kim , veilleux , & sanders 1998 ; genzel et al .
1998 ; veilleux , kim , & veilleux 1999a ; lutz , veilleux , & genzel 1999 ) .
a recent morphological analysis of the 1-jy sample indicates strong trends between infrared luminosity and colors , quasar activity , and merger phase ( veilleux , kim , & sanders 2002 ) .
all uligs in the 1-jy sample are found to be in the pre - merger or final merger phase . about @xmath0 70% of the extreme uligs with log[l@xmath39/l@xmath5 ]
@xmath42 12.5 are single - nucleus advanced mergers .
all ( most ) seyfert 1s ( 2s ) in the sample are also advanced mergers .
the @xmath43- and @xmath44-band profiles in 73% of the single - nucleus advanced mergers are well fitted over @xmath43 = 4 12 kpc by an elliptical - like @xmath45 law ( see also scoville et al .
2000 ; cui et al .
these elliptical - like hosts follow the same @xmath43-band @xmath46 @xmath47 relation as normal ellipticals , suggesting that these objects may eventually become intermediate - luminosity ( 1 3 l@xmath48 ) elliptical galaxies if they get rid of their excess gas or transform the gas into stars .
the hosts of uligs show a broad range in luminosity ( mean @xmath0 2 l@xmath48 ) which overlaps with that of qso hosts . the @xmath43 @xmath44 colors of ulig hosts ( mean @xmath0 3.0 ) are also similar to those of qso hosts and normal ellipticals .
the average half - light radius of uligs is 4.8 @xmath49 1.4 kpc at @xmath43 and 3.5 @xmath49 1.4 kpc at @xmath44 , similar to the qso host sizes measured at @xmath50 by mcleod & mcleod ( 2001 ) but slightly smaller than those measured at @xmath43 by dunlop et al .
the reason for this apparent discrepancy between the two qso datasets is not known .
overall , these results support the scenario in which the ulig is the result of the merger of two @xmath0 l@xmath48 disk galaxies which eventually evolves into an elliptical - like galaxy with a powerful agn .
there are obviously exceptions to this scenario and this is why a large sample of objects like the 1-jy sample is needed to draw statistically meaningful conclusions .
the following is a list of outstanding issues and unanswered questions : * there is strong general support for the agn unification model ( fig .
1 ) , but there are important exceptions : ( a ) not all seyfert 2s show hblrs . this may be a real effect or it could be due to instrument sensitivity ( _ i.e. _ not enough `` mirrors '' to scatter the blr emission back into our line of sight ) .
( b ) optical and x - ray classifications are not always consistent with each other .
these classification mismatches may be explained by non - standard dust near agns , inefficient adaf - like accretion , dilution effects by galaxy light , or compton thickness . *
the luminosity dependence of the type 2 / type 1 agn ratio and the uv and x - ray baldwin effects of the emission and absorption lines in agns require that the geometry of the disk / torus / wind structure depends on agn luminosity .
the exact dependence relies on knowing the fraction of obscured type 2 agns , which is still subject to large uncertainties . *
recent chandra investigations of high - redshift quasars come to conflicting conclusions regarding possible evolutionary effects of the sources of energy and absorption .
the situation at low redshift is less ambiguous : the properties of the cxb and results from deep chandra and xmm surveys appear to require a significant population of buried type 2 agns with peak emissivity around @xmath30 , rather than @xmath31 for unobscured type 1 agns .
these results may imply that type 1 and 2 agns follow different evolutionary paths , in contradiction with the agn unification model .
the recent discovery of a large population of partially obscured type 1 agns at low redshifts may affect these conclusions .
* there is now strong evidence that the host galaxy smbh connection originally found in inactive galaxies also applies to active galaxies .
emission - line methods to test this connection at high redshifts are promising but will undoubtedly be less accurate ( e.g. , nelson 2000 ; mclure & jarvis 2002 ; vestergaard 2002 ) .
* there appears to be a symbiotic relation between starbursts and agns .
the origin of this relation is not known .
the surveys suggesting a greater occurrence of ( circumnuclear ) starbursts in seyfert 2s than in seyfert 1s are often strongly affected by selection biases .
feedback from starburst - driven galactic winds may be one of many side effects of this tight starburst
agn connection .
this wind phenomenon is particularly important in understanding agn activity at high redshift .
* minor mergers , bars , and nuclear spirals may all combine to bring the fuel down to @xmath0 100 pc in seyferts and llagns , but it is not clear what happens next .
there is now strong evidence that gas - rich mergers are able to trigger some qsos at low redshifts after undergoing a ulig phase , but there is very few observational constraints on the formation process for the bulk of qsos that were formed at @xmath51 .
the author congratulates the conference organizers for a very successful meeting .
some of the work discussed in this review was done in collaboration with d .- c .
kim , d. b. sanders , j. bland - hawthorn , g. cecil , p. l. shopbell , and d. s. rupke .
this work was supported by nasa / ltsa grant nag 56547 , nsf / career grant ast-9874973 , and a cottrell scholarship awarded by the research corporation .
adams , f. c. , graff , d. s. , & richstone , d. o. 2001 , apj , 551 , l31 alexander , d. m. 2001 , mnras , 320 , l15 alexander , d. m. , et al .
2002 , aj , 123 , 1149 antonucci , r. 2001 , in iau coll .
184 , agn surveys , eds .
green , e.ye .
khachikian , and d.b .
sanders , asp conf .
series , in press ( astro - ph/0110343 ) antonucci , r. r. j. , & miller , j. s. 1985 , apj , 297 , 621 barger , a. j. , et al .
2001 , aj , 121 , 662 . 2002 , aj , 124 , 1839 barth , a. j. 2002 , in issues in unification of active galactic nuclei , asp conf .
series 258 , eds .
r. maiolino , a. marconi , and n. nagar , 147 bassani , l. , et al . 1999 ,
apjs , 121 , 473 bechtold , j. , et al .
2002 , apj , in press ( astro - ph/0204462 ) benn , c. r. , et al .
1998 , mnras , 295 , 451 boyle , b. j. , et al .
2000 , mnras , 317 , 1014 braatz , j. a. , wilson , a. s. , & henkel , c. 1997 , apjs , 110 , 321 brandt , w. n. , et al .
2002 , apj , 569 , l5 burkert , a. , & silk , j. 2001 , apj , 554 , l151 canalizo , g. , & stockton , a. 2000 , aj , 120 , 1750 . 2001 , apj , 555 , 719 cecil , g. , bland - hawthorn , j. , & veilleux , s. 2002 , apj , 576 , 745 cecil , g. , et al .
2001 , apj , 555 , 338 chokshi , a. 1997 , apj , 491 , 78 clavel , j. , et al .
2000 , a&a , 357 , 839 combes , f. 2000 , lectures given at gh advanced lectures on the starburst - agn connection , inaoe , eds .
d. kunth , i. aretxaga crawford , c. s. , et al .
2002 , mnras , 333 , 809 cui , j. , et al .
2001 , aj , 122 , 63 cutri , r. , et al .
2001 , in new era of wide field astronomy , asp conf .
series 232 , eds .
r. clowes , a. adamson , and g. bromage , 78 de robertis , m. m. , yee , h. k. , & hayhoe , k. 1998 , apj , 496 , 93 done , c. , madejski , g. m. , & smith , d. a. 1996 , apj , 463 , l63 dopita , m. a. , et al .
1997 , apj , 490 , 202 dulzin - hacyan , d. , et al .
1999 , in galaxy interactions at low and high redshift , iau symp .
# 186 , 329 dunlop , j. s. , et al .
2002 , , in press ( astro - ph/0108397 ) efstathiou , g. , & rees , m. j. 1988 , mnras , 230 , 5p elvis , m. 2000 , apj , 545 , 63 espey , b. , & andreadis , s. 1999 , in quasars and cosmology , asp conf .
series 162 , eds .
g. ferland and j. baldwin , 351 evans , a. s. , et al .
2001 , apj , 521 , l107 faber , s. m. , et al .
1997 , aj , 114 , 1771 fabian , a. c. 1999 , mnras , 309 , 39 fabian , a. c. , et al . 2002 ,
mnras , 335 , l1 fan , x. , et al .
2001 , aj , 122 , 2833 ferrarese , l. 2002 , apj , 578 , 90 ferrarese , l. , et al .
2001 , apj , 555 , l79 fiore , f. , et al .
2001 , mnras , 327 , 771 . 2002 , in issues of unifications of agns , asp . conf .
series 258 , eds .
r. maiolino , a. marconi , and n. nagar , 205 ford , h. c. , et al .
1994 , apj , 435 , l27 franceschini , a. , braito , v. , & fadda , d. 2002 , mnras , 335 , l51 fuentes - williams , f. , & stocke , j. t. 1988 , aj , 96 , 1235 gallagher , s. c. , et al .
2002 , apj , 567 , 37 gallimore , j. f. , baum , s. a. , & odea , c. p. 1997 , nature , 388 , 852 gallimore , j. f. , et al .
2001 , apj , 556 , 694 gebhardt , k. , et al .
2000 , apj , 539 , l13 genzel , r. , et al .
1998 , apj , 498 , 579 gonzalez delgado , r. m. , et al .
1998 , apj , 505 , 174 gonzalez delgado , r. m. , heckman , t. , & leitherer , c. 2001 , apj , 546 , 845 greenhill , l. j. , et al .
1996 , apj , 472 , l21 greenhill , l. j. , & gwinn , c. r. 1997 , ap&ss , 248 , 261 greenhill , l. j. , moran , j. m. , & herrnstein , j. r. 1997 , apj , 481 , l23 gregg , m. d. , et al .
2002 , apj , 564 , 133 gu , q. , dultzin - hacyan , d. , & de diego , j. a. 2001 , rmxaa , 37 , 3 gu , q. , & huang , j. 2002 , apj , 579 , 205 guainazzi , m. , et al . 1999 ,
a&a , 341 , l27 haehnelt , m. g. , & rees , m. j. 1993 , mnras , 263 , 168 haiman , z. , & cen , r. 2002 , apj , 578 , 702 haiman , z. , & loeb , a. 1998 , apj , 503 , 505 haniff , c. a. , wilson , a. s. , & ward , m. j. 1988 , apj , 334 , 104 heckman , t. m. , et al .
1997 , apj , 482 , 114 heisler , c. a. , lumsden , s. l. , & bailey , j. a. 1997 , nature , 385 , 700 hill , g. j. , goodrich , r. w. , & depoy , d. l. 1996 , apj , 462 , 163 ho , l. c. 2002 , in agn surveys , proceedings of iau coll . 184 , eds .
r. f. green , e. y. khachikian , and d. b. sanders , in press iwasawa , k. , & comastri , a. 1998 , mnras , 297 , 1219 iwasawa , k. , et al .
1996 , mnras , 282 , 1038 iwasawa , k. , et al .
1999 , mnras , 306 , l19 kaastra , j. s. , et al .
2000 , a&a , 354 , l83 . 2002 , a&a , 386 , 427 kaspi , s. , et al .
2000 , apj , 533 , 631 . 2002 , apj , 574 , 643 kauffmann , g. , & haehnelt , m. 2000 , mnras , 311 , 576 kay , l. e. 1994 , apj , 430 , 196 kim , d .-
c . , veilleux , s. , & sanders , d. b. 1998 , apj , 508 , 627 . 2002 , apjs , 143 , 000 ( astro - ph/0207373 ) kinkhabwala , a. , et al . 2002 ,
apj , 575 , 732 knapen , j. h. , shlosman , i. , & peletier , r. f. 2000 , apj , 529 , 93 kormendy , j. , & gebhardt , k. 2001 , in the 20th texas symposium on relativistic astrophysics , aip conf .
# 586 , eds .
h. martel and j. c. wheeler , 363 kormendy , j. , & richstone , d. 1995 , araa , 33 , 581 krolik , j. h. 2001 , apj , 551 , 72 laine , s. , et al .
1999 , apj , 511 , 709 . 2001 , mnras , 324 , 891 . 2002 , apj , 567 , 97 laor , a. 2001 , apj , 533 , 677 lawrence , a. 1991 , mnras , 252 , 586 lee , j. c. , et al .
2002 , apj , 570 , l47 levenson , n. a. , weaver , k. a. , & heckman , t. m. 2001 , apj , 550 , 230 lumsden , s. , et al .
2001 , mnras , 327 , 459 lutz , d. , veilleux , s. , & genzel , r. 1999 , apj , 517 , l13 lutz , d. , et al .
2002 , a&a , in press ( astro - ph/0209475 ) magorrian , j. , et al .
1998 , aj , 115 , 2285 maiolino , r. , et al .
1995 , apj , 446 , 561 . 1998 , a&a , 338 , 781 . 2001 , a&a , 365 , 37 mainieri , v. , et al .
2002 , a&a , 393 , 425 martini , p. , & pogge , r. w. 1999 , aj , 118 , 2646 martini , p. , et al .
2001 , apj , 562 , 139 mathur , s. , elvis , m. , & wilkes , b. j. 1999 , apj , 519 , 605 mathurs , s , wilkes , b. j. , & ghosh , h. 2002 , apj , 570 , l5 mathur , s. , et al .
2001 , apj , 551 , l13 matt , g. , et al . 1999 ,
a&a , 341 , l39 mcleod , k. k. , & mcleod , b. a. 2001 , apj , 546 , 782 mclure , r. j. , & dunlop , j. s. 2002 , mnras , 331 , 795 mclure , r. j. , & jarvis , m. j. 2002 , mnras , in press ( astro - ph/0204473 ) merritt , d. , & ferrarese , l. 2001 , apj , 547 , 140 miller , j. s. , & goodrich , r. w. 1990 , apj , 355 , 456 miller , j. s. , goodrich , r. w. , & matthews , w. g. 1991 , apj , 378 , 47 miyoshi , m. , et al .
1995 , nature , 373 , 127 moran , e. c. , filippenko , a. v. , & chornock , r. 2002 , apj , 579 , l71 moran , e. c. , et al .
2000 , apj , 540 , l73 mulchaey , j. s. , & regan , m. w. 1997 , apj , 482 , l135 mulchaey , j. s. , wilson , a. s. , & tsvetanov , z. 1996a , apjs , 102 , 309 . 1996b , apj , 467 , 197 nandra , k. , et al . 1997 , apj , 488 , l91 nelson , c. h. 2000 , apj , 544 , l91 nelson , c. h. , & whittle , m. 1996 , apj , 465 , 96 nicastro , f. , et al . 1999 ,
apj , 512 , 184 norman , c. l. , et al .
2002 , apj , 571 , 218 onkel , c. a. , & peterson , b. m. 2002 , apj , 572 , 746 osmer , p. s. , & shields , j. c. 1999 , in quasars and cosmology , asp conf .
163 , eds .
g. ferland and j. baldwin , 235 panessa , f. , & bassani , l. a&a , in press ( astro - ph/0208496 ) pappa , a. , et al .
2001 , mnras , 326 , 995 peterson , b. m. , & wandel , a. 2000 , apj , 540 , l13 pier , e. a. , & krolik , j. h. 1993 , apj , 418 , 673 pogge , r. w. 1989 , apj , 345 , 730 pogge , r. w. , & martini , p. 2002 , apj , 569 , 624 pogge , r. w. , et al .
2000 , apj , 532 , 323 ptak , a. , et al .
1996 , apj , 459 , 542 regan , m. w. , & mulchaey , j. s. 1999 , aj , 117 , 2676 risaliti , g. , maiolino , r. , & salvati , m. 1999 , apj , 522 , 157 rodriguez - espinosa , j. m. , rudy , r. j. , & jones , b. 1986 , apj , 309 , 76 rosati , p. , et al .
2002 , apj , 566 , 667 rupke , d. s. , veilleux , s. , & sanders , d. b. 2002 , apj , 570 , 588 schmidt , m. , & green , r. f. 1983 , apj , 269 , 352 scoville , n. z. , et al .
2000 , aj , 119 , 991 silk , j. , & rees , m. j. 1998 , a&a , 331 , l1 small , t. a. , & blandford , r. d. 1992 , mnras , 259 , 725 stern , d. , et al . 2002a , apj , 568 , 71 stern , d. , et al .
2002b , aj , 123 , 2223 surace , j. a. , sanders , d. b. , & evans , a. s. 2001 , , 122 , 2791 tadhunter , c. , & tsvetanov , z. 1989 , nature , 341 , 422 tanaka , y. , et al .
1995 , nature , 375 , 659 terlevich , e. , diaz , a. i. , & terlevich , r. 1990 , mnras , 242 , 271 tozzi , p. , et al .
2001 , apj , 562 , 42 tran , h. d. 2001 , apj , 554 , l19 tremaine , s. , et al .
2002 , apj , 574 , 740 turner , t. j. , et al .
1993 , apj , 419 , 123 . 2002 , apj , 574 , l123 urry , c. m. , & padovani , p. 1995 , pasp , 107 , 803 veilleux , s. 2001 , in starburst galaxies : near and far , eds . l. tacconi and d. lutz .
heidelberg : springer - verlag , 2001 , p.88 . 2002 , in extragalactic gas at low redshift , eds .
mulchaey and j.t .
stocke , asp conf .
series 254 , 313 veilleux , s. , kim , d .- c . , & sanders , d. b. 1999a , apj , 522 , 113 . 2002 , apjs , 143 , 000 ( astro - ph/0207401 ) veilleux , s. , goodrich , r. w. , & hill , g. j. 1997a , apj , 477 , 631 veilleux , s. , & rupke , d. s. 2002 , apj , 565 , l63 veilleux , s. , sanders , d. b. , & kim , d
1997b , apj , 484 , 92 veilleux , s. , sanders , d. b. , & kim , d .-
1999b , apj , 522 , 139 veilleux , s. , et al .
1994 , apj , 433 , 48 . 1995 , apjs , 98 , 171 vron - cetty , m. p. , & vron , p. 2000 , a&a rev . , 10 , 81 vestergaard , m. 2002 , apj , 571 , 733 vignati , p. , et al . 1999 ,
a&a , 349 , l57 virani , s. n. , de robertis , m. m. , & vandalfsen 2000 , aj , 120 , 1739 wandel , a. , peterson , b. m. , & malkan , m. a. 1999 , apj , 526 , 579 webster , r. , et al .
1995 , nature , 375 , 469 whiting , m. t. , webster , r. l. , & francis , p. j. 2001 , mnras , 323 , 718 wilkes , b. j. , et al .
2002 , apj , 564 , l65 wilms , j. , et al .
2001 , mnras , 328 , l27 wilson , a. s. , & tzvetanov , z. 1994 , aj , 107 , 1227 wisotzki , l. , et al .
2000 , a&a , 358 , 77 | the current paradigm for radio - quiet agns is reviewed , taking into account new results from recent large - scale surveys carried out from the ground and from space .
topics include structure of the central engine , agn demography , fueling / triggering processes , and connection between the supermassive black hole , host galaxy , circumnuclear starburst and agn .
dependences on agn power and lookback time are pointed out in the discussion .
suggestions for future avenues of research are mentioned in the last section . |
the minimal standard model ( sm ) is a very successful theory of the elementary particles known today .
but it should not be the final theory , because it has too many unknown parameters to be put by hand .
most of these parameters are in the fermion part of the theory .
we do nt know the origin of the quarks and leptons , as well as how to determinate their mass and number theoretically .
we have to get their information all from experiments .
there is still not a successful theory which can unify all the four basic interactions .
perhaps our _ elementary particles _ have substructure and we need a more fundamental theory than sm .
this is beyond the scope of our current experiments . on the other hand , the recent measurement of the muon anomalous magnetic moment by the experiment e821 @xcite disagrees with the sm expectations at more than 1.6@xmath6 - 3@xmath6 @xcite level .
moreover , there are convincing evidences that neutrinos are massive and oscillate in flavor @xcite .
it seems to indicate the presence of new physics .
the discovery of neutrino oscillation has been one of the most exciting experimental results in the recent years @xcite . as the experimental data accumulating in the super - kamiokande collaboration @xcite , more results published by sno @xcite , k2k @xcite and chooz @xcite experiments .
these experiments give possible signals of new physics beyond sm . in this paper , we consider the sequential fourth generation model to estimate the possible contributions to the decay @xmath7 from the exchange of the fourth up - type quark @xmath2 . from the point of phenomenology , there is a realistic question one may ask : what is the number of the fermion generations or weather there are additional quarks or leptons other than the known three generations .
the present experiments can tell us there are only three generation of fermions with @xmath8 neutrinos whose mass is smaller than @xmath9 @xcite , but the experiments do nt exclude the existence of other kind of additional generation , such as the fourth generation , with a @xmath10 neutrino , i.e. @xmath11 @xcite .
latest studies in the electroweak sector allow the existence of a fourth generation with heavy dirac neutrino @xcite .
many authors studied models which extend the fermions part , such as vector - like quark models @xcite , sterile neutrino models @xcite and the sequential four generation standard model ( sm4 ) @xcite . here
we consider the sequential fourth generation , non -supersymmetric model , which includes an up - like quark @xmath2 , a down - like quark @xmath12 , a lepton @xmath13 , and a heavy neutrino @xmath14 besides the known three generation of fermions in the sm .
the properties of these new fermions are the same as their corresponding counterparts of other three generations except their masses and ckm matrix elements .
many attempts have been made recently about the fourth generation @xcite and the experimental searches of the fourth generation particles @xcite . in our previous papers @xcite
, we investigated the rare @xmath15 meson decays with the fourth generation @xcite , @xmath16 in @xmath17 systems and @xmath18 in sm4 @xcite .
we got some interesting results , such as the new effects of the 4th generation particle on the meson decays and cp violation .
we also got the constraints of the fourth generation ckm matrix factors @xcite . recently
, these decays in fourth generation were considered again . in ref .
@xcite , the phases of the fourth generation ckm matrix factors , @xmath19 , have been included .
the exclusive b decays with the fourth generation were investigated in ref .
these rare decays are very interesting to study the new physics effects of the fourth generation . in this paper
, we would like to investigate the rare decay @xmath20 with the fourth generation . as a loop - induced flavor changing neutral current ( fcnc ) process the inclusive decay ( at quark level )
@xmath21 is in particular sensitive to contributions from those new physics beyond the sm .
there is a vast interest in this decay . on the experimental side , only upper limits ( @xmath22 ) on the branching ratios of @xmath23 decay
are currently available @xmath24 which are roughly two orders above the sm predictions @xcite , @xmath25 within the sm , the electroweak contributions to @xmath26 and @xmath27 decays have been calculated some time ago @xcite , the leading - order qcd corrections and the long - distance contributions were evaluated recently by several groups @xcite .
the new physics corrections were also considered , for example , in the two - higgs doublet model @xcite and the supersymmetric model @xcite .
there is also study concerning other exclusive channels @xcite . in a previous paper @xcite , we calculated the technicolor corrections to the rare decays @xmath28 in the qcd factorization approach .
the great progress in theoretical studies recently encourage us to do more investigations about this decay in the sequential fourth generation model .
this paper is organized as the following : in section 2 , we present the formulae for the decays @xmath7 in sm and the fourth generation model .
we also obtain the constraints of the fourth generation ckm matrix factors , @xmath29 and @xmath30 from experiments in this section . in section 3 ,
we give the numerical analysis of the branching ratios in the fourth generation model .
the conclusions are in the last section .
same as in our last paper @xcite , up to the corrections of order @xmath31 , the effective hamiltonian for @xmath32 at scales @xmath33 is just the one for @xmath34 and takes the form @xmath35\ , , \label{eq : heff}\ ] ] where in view of @xmath36 , we have neglected the term proportional to @xmath37 . here
@xmath38 are the usual four - fermion operators whose explicit form is given below .
the last two operators in the eq.([heff_at_mu ] ) , characteristic for this decay , are the _
magnetic penguin _ operators . the complete list of operators is given as follows @xmath39 it is the magnetic @xmath40-penguin operator @xmath41 , which plays the crucial role in this decay
. the effective hamiltonian for @xmath42 is obtained from eqs.([eq : heff]-[eq : o8 g ] ) by the replacement @xmath43 . in the case of the fourth generation
there is an additional contribution to @xmath44 from the virtual exchange of the fourth generation up quark @xmath45 . in fig.1 , we draw the relevant feynman diagrams which contribute to the decays @xmath46
. the wilson coefficients of the dipole operators are given by @xmath47 where @xmath48 represent the contributions of @xmath45 to the wilson coefficients .
we recall here that the ckm coefficient corresponding to the @xmath49 quark contribution , i.e. , @xmath50 , is factorized in the effective hamiltonian as given in eq .
( [ heff_at_mu ] ) .
the formulae for calculating the wilson coefficients @xmath51 are the same as in the sm except exchanging @xmath45 quark for @xmath49 quark . in sm4
, the quark mixing matrix can be written as , @xmath52 where @xmath53 and @xmath54 are the elements of the @xmath55 ckm mixing matrix of the sm4 , which now contains nine parameters , i.e. , six angles and three phases .
the rest elements without prime in the matrix are the usual three generation ckm matrix elements .
the current experimental bounds of @xmath56 and @xmath57 can be found in ref .
the factor @xmath56 can be constrained by the decay @xmath58
@xcite @xmath59 or @xmath60 considering the unitarity conditions of the ckm matrix , @xmath61 and taking the average values of the sm @xmath62 ckm matrix elements from ref .
@xcite , we obtain , @xmath63 in the following numerical calculation , we take the values @xmath64 and @xmath65 , for illustration . for the ckm factor @xmath57 , we can get its constraint from the present experimental value of @xmath66 @xcite . also considering the unitarity conditions of the ckm matrix , @xmath67
we finally get @xmath68 in the next section , we will take the values @xmath69 , @xmath70 , @xmath71 , @xmath72 and @xmath73 in numerical calculation respectively .
we first calculate the decay @xmath5 ( the formulas in the case of @xmath75 can be obtained just by the replacement of @xmath43 ) . for this exclusive decay of @xmath76 meson
, we need to deal with both of the long distance and short distance qcd corrections in the @xmath76 meson side . applying the factorization assumption
, we shall adapt a phenomenological approach where the long distance effects are replaced by a few non - perturbative parameters .
in other words we simply evaluate the hadronic matrix element of @xmath77 from one particle irreducible ( 1pi ) and one particle reducible ( 1pr ) feynman diagrams , relying on a phenomenological model . before doing so
, it is important to note that @xmath78 is apparently non - local due to internal @xmath79 or @xmath80 quark propagator . to handle these non - local terms
, one observes that the @xmath79 quark inside the @xmath76 meson carries most of the meson energy , and its four velocity can be treated as equal to that of @xmath76 . hence one may write @xmath79 quark momentum as @xmath81 where @xmath82 is the common four velocity of @xmath79 and @xmath76 . with this parametrization
, we have @xmath83 where the second equation is based on a constituent picture @xcite that @xmath79 and @xmath84 quarks share the total energy of @xmath76 quark is @xmath85 as @xmath86 denotes the momentum of @xmath80 quark in @xmath87 . ]
. therefore @xmath85 should be taken as the four momentum of the constituent @xmath84 quark . with eq.([pp ] ) , @xmath78 is readily made local .
we then compute the amplitude for @xmath5 using the following relations @xmath88 where @xmath89 is the @xmath76 meson decay constant which is about @xmath90 mev according to recent lattice qcd calculations @xcite .
the total amplitude is now separated into a cp - even and a cp - odd part @xmath91 we find that @xmath92 with @xmath93 , @xmath94 , and @xmath95 where @xmath96,\nonumber \\ a_c & = & ( c_1+c_3-c_5)n_c+(c_2+c_4-c_6 ) , \nonumber \\ a_s & = & a_b={1\over 4}\left[(c_3+c_4-c_5)n_c+(c_3+c_4-c_6)\right].\end{aligned}\ ] ] the functions @xmath97 , @xmath98 and @xmath99 are defined by @xmath100 where @xmath101 , and @xmath102 the decay width for @xmath5 is simply @xmath103 to obtain numerical results , we have set light quark masses to zero and used @xcite @xmath104 , @xmath105 and @xmath106 .
furthermore , we take @xmath107 , @xmath108 and @xmath109 .
the numerical values for wilson coefficients @xmath110 evaluated in sm at @xmath111 are listed in table [ t2 ] .
we first consider the decay mode @xmath112 .
figs.2 , 3 and 4 show the branching ratio @xmath113 @xmath114 the mass of @xmath2 and the fourth generation ckm matrix factors @xmath115 and @xmath116 , respectively . from these figures , we find that the sm4 predictions for the branching ratio @xmath113 are rather different from that of the standard model .
the decay rates can be enhanced by about 1 - 2 orders relative to the sm prediction in the reasonable range of @xmath117 and @xmath30 .
this implies that the new physics signal of the fourth generation model may come from the decay of @xmath112 .
the future experiment of lhcb in cern can test it . from fig.2 , we find that the branching ratio of @xmath112 increase with the mass of @xmath2 for the fourth generation ckm matrix factor , @xmath118 and @xmath119 . when @xmath120 , the curve increase rapidly with @xmath117 .
but when @xmath121 , the branching ratio changes very slowly .
because the fourth generation ckm matrix factor , @xmath30 , is small in the latter case .
it can be seen from eq .
( [ ceff4 ] ) that the new physics contribution is proportional to the size of the ckm factor @xmath30 , thus it is very important in the new physics contribution with @xmath117 . from figs.3 and 4 , one can also see that , the branching ratio of @xmath112 decease with @xmath115 but increase with @xmath116 , since @xmath115 is negative and gives the opposite contributions as sm to the coefficients @xmath122 ( see eq .
( [ ceff4 ] ) ) . for the decay @xmath123 , it is easy to find from figs.5 and 6 that the new physics contribution to the branching ratio is tiny
. it is shown that it is not suitable to test the new physics signals from the 4th generation in this channel .
the prediction of the branching ratio in the 4th generation model is between @xmath124 and @xmath125 , when we take the reasonable values of @xmath117 ( from 200 gev to 1000 gev ) and @xmath29 ( see eq . ( [ vtd ] ) ) .
this numerical result is fully in the range of the sm prediction ( see eq . ( 4 ) ) .
the reason is that the experimental constraint of the 4th generation ckm matrix element factor @xmath29 is very stringent ( only at @xmath126 level , see eq.([vtd ] ) ) .
we also find that the curves change little along with the variations of @xmath117 and @xmath29 .
the reason is also from the tiny @xmath29 .
we can see that the branching ratios of both @xmath112 and @xmath112 decays strongly depend on the size of the related mixing elements .
as a summary , the size of new physics contributions to the rare decays of @xmath7 from the fourth generation strongly depends on the values of the masses of the new heavy up quark @xmath2 and the 4th generation ckm matrix elements , @xmath30 and@xmath29 .
the results in the 4th generation model are quite different from the sm case for decay @xmath112 , its branching ratio can be enhanced by about 1 - 2 orders of magnitude
. a large branching ratio of @xmath127 decay measured in experiment may be a new physics signal of the 4th generation model .
but for decay @xmath123 , because of the tiny values of 4th generation ckm matrix element , @xmath3 , the new physics contribution to @xmath4 is very small and can not be distinct from the sm prediction .
huo acknowledges the support from the chinese postdoctoral science foundation.from the natural sciences and engineering research council of canada .
l acknowledges the support by national science foundation of china under grants no .
90103013 and 10135060 . and z.j .
xiao acknowledges the support by the national natural science foundation of china under grants no .
10075013 and 10275035 , and by the research foundation of nanjing normal university under grant no .
214080a916 .
u. chattopadhyay , a. corille and p. nath , hep - ph/0202275;m . knecht and a. nyffeler , hep - ph/0111058 ; m. knecht , a. nyffeler , m. perrottet and e. de pafael , phys .
lett . * 88 * , 071802 ( 2002 ) ; m. hayakawa and t. kinsoshita , hep - ph/0112102 ; i. blokland , a. czarnecki and k. melnikov , phys .
* 88 * , 071803 ( 2002 ) ; j. bijnens , e. pallante and j. prades , hep - ph/0112255 . g.s .
abrams et al . , mark ii collab . , phys .
* 63 * ( 1989 ) 2173 ; b. advera et al .
, l3 collab .
b * 231 * ( 1989 ) 509;i .
decamp et al . , opal collab . , _ ibid .
_ , * 231 * ( 1989 ) 519 ; m.z .
akrawy et al . , delphi collab . , _ ibid .
_ , * 231 * ( 1989 ) 539 ; c.caso et al .
, ( particle data group ) , eur .
j. * c 3 * ( 1998 ) 1 . v. barger , y.b .
dai , k. whisnant and b.l .
young , hep - ph/9901380 ; r.n .
mohapatra , hep - ph/9702229 ; s. mohanty , d.p .
roy and u. sarkar , hep - ph/9810309 ; s.c .
gibbons,_et al_. , phys . lett .
* b430 * ( 1998 ) 296 ; v. barger , k. whisnant and t.j .
weiler , phys .
b427 * , ( 1998 ) 97 ; v. barger , s. pakvasa , t.j . weiler and k. whisnant ,
phys . rev .
* d58 * ( 1998 ) 093016 .
hou , a. soni and h. steger , phys . lett . *
b192 * ( 1987 ) 441 ; w.s .
hou , r.s .
willey and a. soni , phys .
bf 58 , 1608 ( 1987 ) ; erratum , _ ibid _ * 60 * , 2337 ( 1987 ) ; j.f .
gunion , douglas w. mckay , h. pois , phys .
b * 334 * ( 1994 ) 339 ; phys .
d * 51 * ( 1995 ) 201 .
lin , j. liu , y .- p .
yao , phys .
64 ( 1990 ) 1498 ; phys .
d42 ( 1990 ) 2314 ; mod .
lett . a6 ( 1991 ) 1333 ; h. simma , d. wyler , nucl .
b344 ( 1990 ) 283 ; s. herrlich , j. kalinowski , nucl .
b381 ( 1992 ) 501 .
bosch and g. buchalla , _ the double radiative decays @xmath129 in the heavy quark limit _ ,
jhep 0208 ( 2002 ) 054 ; s.w .
bosch , _ exclusive radiative decays of b mesons in qcd factorization _ , ph.d . thesis , hep - ph/0208203 . | we calculate the branching ratios of @xmath0 decay in the sequential fourth generation model .
we find that the theoretical values of the branching ratios , @xmath1 , including the contributions of @xmath2 , are much different from the minimal standard model ( sm ) predictions .
the new physics contribution can provide one to two orders enhancement to the standard model prediction of @xmath1 .
but due to the tiny values of 4th generation ckm matrix element @xmath3 , the new physics effect on @xmath4 is very small and can not be distinct from the sm prediction .
it is shown that the decay @xmath5 can give the test of new physics signals from the 4th generation model . |
following the discovery in 1961 of @xmath8 meson in @xmath0 reactions , the measurements of forward - backward asymmetry in @xmath9 suggested the existence of a rho - like resonance in the @xmath2-wave amplitudes , later referred to as @xmath10 scalar meson @xcite .
the cern measurements of @xmath11 and @xmath12 on polarized targets in 1970 s cofirmed the existence of @xmath10 scalar meson .
evidence for a narrow @xmath10 was found in amplitude analyses of @xmath11 at 17.2 gev / c @xcite and in @xmath12 at 5.98 and 11.85 @xcite .
additional evidence for @xmath10 came from the amplitude analysis of the itep data on @xmath7 on polarized target at 1.78 gev / c @xcite .
the @xmath2-wave amplitudes and intensities from all these analyses are surveyed in ref .
these findings were controversial because the measurements of @xmath13 at cern in 1972 found no evidence for a rho - like meson in the @xmath2-wave amplitudes @xcite . in 2001 ,
e852 collaboration at bnl reported high statistics measurements @xmath13 at 18.3 gev / c @xcite . once again , there was no evidence for a rho - like resonance in the @xmath14 @xmath2-wave amplitudes .
the resolution of the puzzle of @xmath10 resonance came from our recent high resolution amplitude analysis of the cern data on @xmath7 for dipion masses 580 - 1080 mev at low momentum transfers @xcite where @xmath2- and @xmath3-wave amplitudes dominate .
the analysis shows that the rho - like resonance in the @xmath2-wave transversity amplitudes arises entirely from the contribution of the @xmath15 resonance to the data component in the expression for the moduli of the @xmath2-wave amplitudes .
the @xmath2-wave transversity amplitudes are nearly either in phase or 180@xmath16 out of phase with the resonating @xmath3-wave transversity amplitudes .
these facts allow us to identify @xmath10 with @xmath15 .
there is a pronounced dip at 980 mev in the moduli of @xmath3-wave amplitudes @xmath17 that is identified with @xmath18 resonance leading to @xmath6 mixing in both @xmath2- and @xmath3-wave amplitudes .
a model independent determination of helicity amplitudes from the transversity amplitudes shows a @xmath15 peak in the @xmath2-wave single flip helicity amplitude @xmath19 @xcite .
the phase of the amplitude @xmath20 is close to the phase of @xmath15 breit - wigner amplitude @xcite . identifying @xmath10 with @xmath15
explains why there is no rho - like resonance observed in @xmath13 since there is no @xmath3-wave amplitude in this process .
however , the @xmath6 mixing appears to violate rotational and lorentz symmetry .
if this symmetry were violated in strong interactions we should observe a dependence of @xmath15 width @xmath21 on its helicity @xmath22 . in our analysis
@xcite we demonstrate the independence of @xmath21 on the helicity confirming the rotational and lorentz symmetry of strong interactions .
this suggests that the observed @xmath6 mixing arises from a new kind of interaction independently involved in the pion creation process .
the observed @xmath6 mixing can be naturally explained when we assume that the produced partial waves are not isolated spin states but interact with a common environment .
we can see how this might work using the following mechanical analogy .
consider a pendulum of mass @xmath23 and length @xmath24 oscillating with a natural frequency @xmath25 , and an ensemble of several other penduli with various masses @xmath26 and lengths @xmath27 .
when the other penduli are isolated from the oscillating pendulum they do not oscillate .
however , when all the penduli are attached to a common rod all penduli begin to oscillate with the same resonant frequency although with different amplitudes .
the interaction of the penduli with a common environment - the rod - allows a resonance from one pendulum to `` leak '' into the other penduli which have different natural frequencies .
similarly we can imagine that the partial waves produced in @xmath7 process interact with a quantum environment which allows resonances from one partial wave amplitude to `` leak '' into other partial wave amplitudes .
then the pion creation process is no longer an isolated system but behaves as an open quantum system described by a non - unitary dynamics .
a non - unitary evolution law must evolve the produced final state @xmath1 into an observed final state @xmath28 with new spin mixing partial wave amplitudes .
but how do we know that such quantum environment actually exists ? and
why its interaction with the @xmath29 process should lead to the @xmath6 mixing ?
the basic assumption of the @xmath2-matrix theory is that the minkowski spacetime is empty .
it contains no environment with which particle scattering processes could interact .
in part i. of this work @xcite we put forward a hypothesis that the physical space is not empty but contains a universal quantum environment that can interact with some particle scattering processes .
the new interaction can not change the identity and the four - momenta of the final state particles produced by the @xmath2-matrix dynamics in order to leave the particle scattering dynamics open to the experimental observation .
this means the new interaction must be a pure dephasing interaction that modifies phases of the amplitudes and nothing else .
this is a new kind of interaction outside of the standard model . to test our hypothesis we used unitarity of the @xmath2-matrix to obtain information on the relative phases of the measured partial wave transversity amplitudes @xmath30 and @xmath31 .
unitary evolution law evolves pure initial states into pure final states .
this fact implies that the relative phases are constrained to @xmath32 in a disagreement with the measured phases in all analyses of @xmath0 processes .
an amplitude analysis of the @xmath2- and @xmath3-wave subsystem in @xmath7 assuming these unitary phases is in disagreement with the cern data on polarized target at least at @xmath33 level .
we conclude that the cern measurements of @xmath7 and @xmath12 on polarized targets support the hypothesis of the existence of the quantum environment and its dephasing interaction with some particle scattering processes .
if the quantum environment is to be an integral part of the nature then its dephasing interactions with particle scattering processes can not contradict standard model .
the two aspects of the nature must be mutually consistent .
what are the cosequences of this consistency ?
the dephasing interaction evolves the produced @xmath2-matrix final state @xmath1 into an observed final state @xmath28 .
the non - unitary evolution law is described by kraus representation . in this paper
we show that the consistency of the pure dephasing interaction with the standard model in @xmath0 processes requires that it be a dipion spin mixing interaction . in what we call spin mixing mechanism the observed partial wave amplitudes
are a unitary transform of the corresponding @xmath2-matrix partial wave amplitudes .
the elements of the spin mixing matrix are forward scattering amplitudes of the partial waves with initial and final dipion spins @xmath4 and @xmath5 , respectively .
helicities are conserved in this scattering .
these amplitudes are matrix elements of kraus operators describing the non - unitary interaction with the environment .
the theory predicts @xmath6 mixing in the @xmath2-and @xmath3-wave amplitudes in @xmath7 . the predicted moduli and relative phases of the mixed amplitudes are in an excellent qualitative agreement with the experimental results @xcite .
the theory correctly predicts the absence of @xmath34 mixing .
the spin mixing of @xmath2-matrix partial wave amplitudes to form new observable partial wave amplitudes is a new phenomenon outside of standard model and represents a genuine new physics .
we discuss kraus representation and kraus operators in section ii .
the non - unitary evolution of the produced final state @xmath1 into an observed final state @xmath28 is introduced in section iii . in section iv .
we use the purity of the dephasing interaction to derive the form of spin mixing mechanism and define kraus partial wave helicity amplitudes which are mixtures of the @xmath2-matrix partial wave amplitudes . in section v. the requirements of lorentz symmetry , conservation of total angular momentum , @xmath3-parity and total isospin impose consistency constraints on the spin mixing mechanism and kraus helicity amplitudes . in section vi .
self - consistency requirements constrain the spin mixing mechanism to a simple unitary transform of the @xmath2-matrix amplitudes . in section vii .
we relate the observed effective amplitudes to the kraus amplitudes and demonstrate the conservation of probability by the dephasing intaraction . in section viii .
we discuss resonance mixing in @xmath7 below 1400 mev .
we use spin mixing mechanism for the @xmath2- and @xmath3-waves to present qualitative predictions for the moduli and relative phases of the spin mixing @xmath2- and @xmath3-wave amplitudes below 1080 mev which we compare with two analyses of experimental data .
section ix . reviews the determination of the dimension @xmath35 of the hilbert space of the environment . in section
x. we identify the decoherence free and decohering amplitudes below 1400 mev in @xmath7 and briefly discuss our new analysis using spin mixing mechanism @xcite . in section xi .
we use coleman - mandula no - go theorem to show that the dimension @xmath36 and that the amplitudes in the decoherence free subspace must mix spins while the decohering amplitudes can not mix spins to protect the lorentz symmetry of the @xmath2-matrix .
the spontaneous violation of the lorentz symmetry observed in the measured amplitudes is thus consistent with lorentz symmetry of the standard model .
we conclude in section xii . with a summary and a brief discussion of the physical nature of the quantum environment .
the interaction of an open quantum system @xmath37 with a quantum environment @xmath38 is assumed to be described by a unitary operator @xmath39 .
the initial state @xmath40 of the combined system is prepared in a separable state @xmath41 .
the interaction evolves @xmath40 into a final state @xcite @xmath42 unitary evolutions are entangling quantum operations and the joint state @xmath43 is not a separable but an entangled state of the final system @xmath44 and the environment @xmath38 .
the observer can perform measurements on the systems @xmath37 and @xmath44 but can not perform direct measurements on the environment @xmath38 . after the transformation @xmath39
, the system @xmath44 no longer interacts with the environment @xmath38 .
the quantum state of system @xmath44 is then fully described by reduced density matrix @xmath45 in a sense that we can calculate average values @xmath46 of any observable @xmath47 .
the trace in ( 2.2 ) is over the interacting degrees of freedom of the environment @xmath48 . in general , the reduced state @xmath49 is no longer related to the initial state @xmath50 by a unitary transformation @xmath51 .
instead it is given by kraus representation @xcite @xmath52 where we assumed that @xmath53 is a pure state .
the evolution operators @xmath54 are called kraus operators .
they are acting on the hilbert spaces @xmath55 and @xmath56 of the systems @xmath37 and @xmath44 , respectively , and can be unitary or non - unitary . for trace preserving maps
the kraus operators must satisfy completness relation @xmath57 to be physically meaningful , a linear map @xmath58 must be completely positive in order to preserve the positivity of all probabilities . a linear map @xmath59 is completely positive if and only if it has the form of the kraus representation @xcite . the operator @xmath59 is also called superscattering operator . the quantum states @xmath60 of the environment form an orthonormal basis in hilbert space @xmath61 with a finite dimension @xcite @xmath62 when the initial state of the environment is a mixed state @xmath63 the diagonal elements @xmath64 and @xmath65 the reduced density matrix ( 2.2 ) then takes the form @xmath66 where @xmath67 . in order for the reduced density matrix ( 2.8 ) to be completely positive
it must acquire the form of the kraus representation ( 2.3 ) . for arbitrary states
@xmath68 this will happen if and only if the unitary evolution operator @xmath39 conserves the quantum numbers of the quantum states @xmath60 @xmath69 where @xmath70 .
then ( 2.8 ) reads @xmath71 with the replacement @xmath72 we recover the form ( 2.3 ) .
we shall refer to @xmath73 as kraus density matrices .
the final state @xmath49 is a mixed state even when the initial state @xmath50 is a pure state . for trace preserving maps
the kraus operators @xmath74 satisfy a completness relation similar to ( 2.4 ) @xmath75 a trace preserving kraus representation is called bistochastic if it leaves invariant the maximally mixed state of the system @xmath76 the kraus representation ( 2.10 ) is a bistochastic map when the kraus operators @xmath74 are unitary . in that case
the map ( 2.10 ) is referred to as a random external field @xcite .
when @xmath77 the effect of the non - unitary interaction with the environment described by kraus representation is a modification ( dephasing ) of the phases of the quantum state @xmath78 . in dissipative dephasing interactions
the system @xmath2 and the environment @xmath38 exchange energy - momentum .
there is no exchange of energy - momentum or angular momentum in pure dephasing interactions which effect only a change of phases .
a very important concept of decoherence free subspace was introduced in ref .
it describes a decoupling of a subsystem @xmath79 of the system @xmath2 from the environment . in this case
all kraus operators are equal to a unitary operator @xmath80 and the kraus representation is reduced to a unitary evolution law for that subsystem @xmath81 .
unitary @xmath2-matrix gives rise to a unitary evolution law @xmath82 the initial state has a general form @xmath83 are the initial four - momenta , helicities and quantum numbers of the state . in the following
we suppress the labels @xmath84 and @xmath85 .
the completness relation reads @xmath86 the first sum in ( 3.3 ) is over all allowed final states @xmath87 , the second sum is over final state helicities @xmath88 and the integration is over the entire phase space of final state momenta @xmath89 .
the symbol @xmath90 labels the quantum numbers of the state @xmath87 . using the completness relation
the density matrix @xmath91 has an explicit form @xmath92 where the diagonal terms @xmath93 are the produced @xmath2-matrix states with final particle states @xmath94 and @xmath95 are the off - diagonal terms of @xmath91 .
we assume that particle scattering processes are not isolated events in the universe as implied by the unitary evolution law but behave as open quantum systems interacting with a universal quantum environment of quantum states @xmath96 .
if the quantum environment and its interaction with particle scattering processes are to be an integral part of the nature then they must be fully consistent with the standard model .
this is possible if this new kind of particle interaction does not mix with the fundamental interactions of the standard model and thus does not participate in the scattering or decay process itself .
then the interaction with the quantum states @xmath96 can only evolve the state @xmath91 produced by the @xmath2-matrix dynamics into a new observed state @xmath97 .
the observed state @xmath97 is described by the non - unitary kraus representation @xmath98 where @xmath99 it is the observed states @xmath28 with final particle states @xmath94 which are the object of experimental measurements .
the states @xmath96 and thus the probabilities @xmath100 are poincare invariant .
the kraus operators @xmath101 are unitary to preserve in ( 3.8 ) the completness relations ( 3.3 ) .
the observed states @xmath28 must communicate the information about the scattering dynamics encoded in the @xmath2-matrix state @xmath1 . the kraus operators therefore must preserve the identities of all final state particles of the produced state @xmath1 including the four - momenta of individual particles .
this means that we need to concern ourselves only with the projective measurements of the states @xmath28 as the projections of @xmath102 vanish .
there is no exchange of four - momentum and angular momentum between the environment and the produced state @xmath1 .
the total four - momentum and total angular momentum are conserved by the kraus operators .
the interaction of the produced state @xmath1 with the quantum environment is a non - dissipative pure dephasing interaction .
the consistency of this interaction with the symmetries and conservation laws of the standard model imposes constraints on the kraus operators and their matrix elements .
how do we observe the changes of the @xmath2-matrix amplitudes in the observed amplitudes from the measurements of the density matrix @xmath28 ? to answer this question consider three complex functions @xmath103 .
their relative phases @xmath104 and @xmath105 satisfy the phase condition @xmath106 and the equivalent cosine condition @xmath107 @xmath2-matrix amplitudes @xmath108 are complex valued functions that satisfy the cosine condition ( 3.11 ) .
another fundamental property of the @xmath2-matrix amplitudes is that they do not mix resonances of different spins @xmath109 and @xmath110 in the bilinear terms @xmath111 and @xmath112 .
to be physically meaningfull the observed density matrix @xmath28 and the kraus densities @xmath113 must have the same bilinear structure as the @xmath2-matrix density @xmath1 .
then the measured bilinears @xmath114 and @xmath115 carry the same quantum numbers as the corresponding @xmath2-matrix bilinear terms . from the measured bilinear terms we can determine three cosines @xmath116 . since the observed amplitudes are not @xmath2-matrix amplitudes
, these cosines will , in general , violate the cosine condition .
for the same reason in some processes certain bilinear terms will exhibit spin mixing of resonances . only in the case of a decoherence free subspace involving at least three amplitudes
will the observed cosines satisfy the cosine condition ( 3.11 ) .
if there is no spin mixing observed in these amplitudes then they can be identified with @xmath2-matrix amplitudes .
this is the case of all two - body scattering or decay processes .
these processes as well as free single particles thus do not interact with the quantum environment .
the pion creation process @xmath117 is the very simplest particle scattering process where we can observe non - unitary evolution and the spin mixing interaction . with four - momenta @xmath118
the initial and final particle states of the process are described by state vectors @xmath119 and @xmath120 where @xmath121 and @xmath122 and @xmath123 are the helicities of the target and recoil nucleon , respectively .
the direction of @xmath124 in dipion center - of - mass system is descibed by @xmath125 .
target nucleon spin density matrix is @xmath126 where @xmath127 is target polarization .
the produced final state is given by the @xmath2-matrix unitary evolution law @xmath128 where @xmath129 and where we have used the phase space relation @xcite @xmath130 in the completness relation .
the angular @xmath2-matrix amplitudes are matrix elements @xmath131 prior to the interaction of the produced final state with the quantum environment there is no measurement and thus no measurement projection of this state to @xmath132 ) .
the angular state @xmath133 can be expanded in terms of spherical harmonics @xcite @xmath134 where @xmath135 and @xmath136 are the dipion spin and helicity , respectively . integrating over @xmath137 and @xmath138 and using the relation @xmath139 we find @xmath140 the partial wave @xmath2-matrix amplitudes are matrix elements @xmath141 in general
, the final state density matrix ( 4.7 ) is a mixed state of entangled dipion and recoil nucleon helicity helicity states @xmath142 and @xmath143 with different dipion spins @xmath135 . equation ( 4.7 ) can be written in a physically important form @xmath144 where @xmath145 and @xmath146 are superpositions of partial waves @xmath147 given by @xmath148 we now use the kraus representation ( 3.7 ) to determine the nature and the form of the pure dephasing interaction with the quantum environment in @xmath0 processes .
the unitary kraus operators @xmath101 act on the superposions ( 4.11 ) of the partial waves @xmath149 by rotating the dipion spin states @xmath150 .
applying the completness relation ( 3.3 ) from left to @xmath151 we find @xmath152 since the kraus operators preserve the four - momenta and the identity of the final state particles .
using the identity @xmath153 the equation ( 4.12 ) then acquires the form @xmath154 this relation shows the interaction of the produced state @xmath1 with the environment is a dipion spin mixing interaction . its effect is to produce a new superposition of partial waves @xmath155 @xmath156 are kraus partial wave amplitudes @xmath157 which explicitely show the mixing of @xmath2-matrix partial wave amplitudes with different dipion spins @xmath135 and helicities @xmath136 .
we refer to the relation ( 4.16 ) as spin mixing mechanism . applying the relation ( 4.13 ) to the combination @xmath158 into an angular form @xmath159 where the kraus angular amplitudes @xmath160 similarly we find for @xmath161 @xmath162 @xmath163 where the kraus amplitudes @xmath164 @xmath165 the kraus representation ( 3.7 ) then reads @xmath166 where @xmath167 the equations ( 4.23 ) have a formal resemblance to the equation ( 4.1 ) for the @xmath2-matrix final state @xmath1 .
it is evident from the definitions of kraus partial wave and angular amplitudes ( 4.16 ) , ( 4.19 ) and ( 4.21 ) that these amplitudes can not be written as matrix elements of a unitary operator in contrast to the @xmath2-matrix partial wave amplitudes if not all spin states @xmath168 are allowed and ( 4.13 ) does not apply . in that case
the kraus partial wave amplitudes shall violate the unitarity conditions on relative phases of the @xmath2-matrix partial wave amplitudes @xcite . as a result they are complex valued functions with non - unitary phases .
spins @xmath135 are resticted by the conditions ( 5.6 ) . following a projective measurement of the final state @xmath28 into a state with definite four - momenta @xmath169 and the direction @xmath170 of the pion
@xmath124 , the kraus representation ( 4.22 ) takes the form @xmath171 suppressing the four - momenta in the initial and final state vectors @xmath172 and @xmath173 , the kraus density matrices read @xmath174 with @xmath175 we have from ( 4.13 ) @xmath176 where @xmath177 are the new kraus angular amplitudes . following formally the steps employed in the case of the @xmath2-matrix spin formalism
@xcite we can redefine the kraus density matrices ( 4.25 ) in terms of new kraus helicity amplitudes @xmath178 the normalization factor @xmath179 where @xmath180 is the dipion mass , @xmath181 is the center - of - mass energy squared , @xmath182 is pion momentum in dipion center - of - mass , @xmath183 and @xmath184 are the energy dependent part of the phase space and flux , respectively @xcite .
the redefined matrices ( 4.25 ) then read @xmath185 angular expansion of kraus helicity amplitudes @xmath186 defines kraus partial wave helicity amplitudes with a definite dipion spin and helicity @xmath187
so far we have suppressed isospin labels of the produced pions . in strong interactions pions
are considered as identical particles .
the ensuing bose - einstein statistics for two - pion states requires that @xmath188 where @xmath189 is the total dipion isospin @xcite .
only @xmath190 is allowed for identical pions . for different pions
@xmath2-matrix partial waves with @xmath191 vanish while for identical pions amplitudes with @xmath192 vanish . labeling the isospin states @xmath193
, the spin mixing mechanism for kraus partial wave helicity amplitudes reads @xmath194 where @xmath195 are the redefined @xmath2-matrix partial wave helicity amplitudes @xcite .
the isospin of the kraus partial wave helicity amplitudes is defined to be @xmath193 although for different pions these amplitudes mix isospin as the result of spin mixing .
kraus partial wave amplitudes with @xmath192 still vanish for identical pions due to the bose - einstein statistics for pion spin .
we refer to the matrix elements @xmath196 as dephasing amplitudes . they describe a forward two - body scattering of dipion states and
recoil nucleons @xmath197 with the kraus operators @xmath101 taking on the role of scattering operators .
as in the case of @xmath2-matrix scattering , there is a priori no restriction on the final spin @xmath4 and helicity @xmath22 . to be consistent with the standard model , the dephasing interaction must be lorentz invariant .
that means that all kraus operators must be lorentz invariant .
the rescattering process of the dipions with the recoil nucleon must thus conserve the total four - momentum and total angular momentum .
the dephasing amplitudes obviously conserve the total four - momentum .
they must depend on the lorentz invariants of the four - momenta @xmath198 and @xmath199 , i.e. the total energy @xmath181 and dipion mass @xmath200 . to examine their conservation of the total angular momentum
we go into the center - of - mass system in the initial and final states of the process ( 4.33 ) where the dipion has 3-momentum q and direction @xmath201 .
then @xcite @xmath202 where @xmath203 and @xmath204 .
the expansion of the dephasing amplitudes defines partial wave amplitudes in @xmath205 which conserve the total angular momentum @xmath206 , @xmath207 and which are independent of @xmath26 due to rotational invariance .
then @xmath208 where @xmath209 we see that the dephasing amplitudes are non - zero only for @xmath210 we can write the conservation of total angular momentum @xmath211 in the form @xmath212 where @xmath27 and @xmath213 are the initial and final orbital momenta while @xmath37 and @xmath44 are the initial and final total spins . since the four - momenta of the particles in the initial and final states are equal we have @xmath214 . for total spins we have @xmath215 and @xmath216 .
then @xmath217 implies that the only possible values of the spin @xmath135 are @xmath218 we require that the kraus helicity amplitudes ( 4.32 ) be consistent with the standard model , in particular with the conservation of @xmath3-parity , total four - momentum and total angular momentum , total isospin , and with the conservation of probability imposed by the @xmath2-matrix .
these requirements will impose constraints on the matrix elements of the kraus operators that ensure the dipion spin mixing interaction is a pure dephasing interaction . to ensure that all observers can communicate the same information about the production process and
its interaction with the quantum environment the kraus partial wave helicity amplitudes ( 4.32 ) must transform as two - body helicity @xmath2-matrix amplitudes @xcite under lorentz transformations .
this in turn requires that the spin mixing mechanism is lorentz invariant .
let @xmath219 be a homogeneous lorentz transformation from a frame @xmath220 to a frame @xmath221 .
the transformation law of the helicity amplitudes is obtained from @xmath222 in ( 5.7 ) @xmath223 and @xmath224 is the matrix representing wigner rotation @xmath225 .
here @xmath226 where @xmath227 is a boost lorentz transformation along the @xmath228-axis followed by a rotation through euler angles @xmath229 . with lorentz symmetry of the @xmath2-matrix and kraus operators , the transformation laws for the @xmath2-matrix and dephasing amplitudes
read @xcite @xmath230 @xmath231 @xmath232 @xmath233 where we suppressed the isospin labels for the sake of brevity . combining these results in the expression ( 4.32 ) for the kraus helicity amplitudes and using @xmath234 we obtain @xmath235
@xmath236 this is the required transformation law for the kraus helicity amplitudes which also shows that the spin mixing mechanism is lorentz invariant .
the requirement of @xmath3-parity conservation implies that the kraus helicity amplitudes satisfy the same parity relations as the corresponding @xmath2-matrix helisity amplitudes , namely @xcite @xmath237 the relation ( 4.32 ) then requires that @xmath238 @xmath239 since the dephasing interaction preserves the four - momenta of the all final state particles , the total four - momentum is obviously conserved by kraus helicity amplitudes . to show they conserve the total angular momentum we consider the @xmath2-matrix and kraus amplitudes as functions of the c.m.s .
scattering angle @xmath240 .
then the partial wave expansion of the final two - body state in @xmath2-matrix and kraus amplitudes reads @xmath241 where @xmath242 and @xmath243 are the total angular mometum and its @xmath228-axis component , and @xmath244 , @xmath245 . assuming @xmath246 , or @xmath247
, we can write @xmath248 which ensures the conservation of the total angular momentum by the kraus helicity amplitudes .
note that the equality @xmath246 follows from the equation ( 5.5 ) , i.e. from the lorentz symmetry of the kraus operators .
the conservation of the total isospin in kraus helicity amplitudes follows from the fact that the isospin identities of the final state particles are preserved by the dephasing interaction .
consider @xmath7 process .
for @xmath249 isospin state @xmath250 while for @xmath251 the isospin state @xmath252 is a combination of states @xmath253 and @xmath254 .
since the total initial isospin is conserved by the @xmath2-matrix amplitudes in both cases , then it must be conserved by the dephasing amplitudes with the states @xmath255 and @xmath256 as well .
this propagation of total isospin in the spin mechanism ensures isospin conservation by kraus helicity amplitudes .
unitarity of the kraus operators requires @xmath257 the conservation of the total angular momentum constraint ( 5.5 ) requires that @xmath258 for arbitrary @xmath259 . for @xmath260
the unitarity relation is a null identity .
the measurements of @xmath0 processes on polarized targets are best analyzed in terms of transversity amplitudes with definite @xmath261-channel naturality . to define kraus transversity
amplitudes we first must define kraus helicity amplitudes with definite @xmath261-channel naturality @xcite .
the unnatural exchange amplitudes are defined by @xmath262 while the natural exchange amplitudes are given by @xmath263 where @xmath264 and @xmath265 are nucleon helicities .
both amplitudes satisfy the same parity relations ( 5.12 ) as the helicity amplitudes .
however to express @xmath266 and @xmath267 in terms of the @xmath2-matrix amplitudes with definite naturality we need another constraint @xmath268 which from the @xmath3-parity relations ( 5.13 ) implies @xmath269 with the resulting expressions we calculate kraus unnatural and natural exchange transversity amplitudes defined by @xcite @xmath270 for unnatural exchange amplitudes and with a similar definition @xmath271 for the natural exchange amplitudes . in ( 6.5 ) and ( 6.6 ) @xmath272 is the target nucleon transversity .
the kraus transversity amplitudes satisfy parity relations @xmath273 for @xmath274 note that @xmath275 so that @xmath276 .
starting with the spin mixing mechanism for kraus helicity amplitudes ( 4.32 ) and using the parity relations ( 5.12 ) , ( 6.3 ) and ( 6.4 ) , the spin mixing mechanism for the kraus unnatural and natural transversity amplitudes reads @xmath277 @xmath278 where @xmath279 and @xmath280 are the @xmath2-matrix unnatural and natural exchange transversity amplitudes , respectively , and where @xmath281 we have omitted isospin labels in ( 6.10 ) for the sake of brevity . from the unitarity of the kraus operators @xmath101 follows the unitarity of matrices @xmath282 in ( 6.8 ) @xmath283 the expressions ( 6.8 ) and ( 6.9 ) satisfy the @xmath3-parity relations ( 6.7 ) .
we now impose the conditions @xmath284 and @xmath5 that follow from the lorentz symmetry of the kraus operators .
the expressions ( 6.8 ) and ( 6.9 ) then read @xmath285 @xmath286 with a change @xmath287 the expressions ( 6.12 ) and ( 6.13 ) read @xmath288 @xmath289 these expressions must satisfy the @xmath3-parity relations ( 6.7 ) for any @xmath22 such that @xmath290 . equating the second sum in the equations ( 6.12 ) and ( 6.13 ) with the second sum in ( 6.14 ) and ( 6.15 ) , respectively ,
we obtain relations that impose spin mixing on the @xmath2-matrix amplitudes .
these constraints are therefore not physical and we must require that @xmath291 then the spin mechanism for the kraus transversity amplitudes takes a simple form for any @xmath290 @xmath292 @xmath293 where we have restored the isospin labels and used the constraint ( 5.6 ) from the lorentz symmetry of kraus operators . the lorentz conditions ( 5.5 ) also require @xmath294 with the only non - zero dephasing amplitudes given by the spin mixing mechanism ( 6.17 ) and ( 6.18 ) , the spin mixing mechanism for the kraus helicity amplitudes ( 4.32 ) reads @xmath295 where @xmath296 we have arrived at a self - consistent form of the spin mixing mechanism for kraus amplitudes that arises from their consistency with the standard model .
the consistency of the kraus transversity amplitudes with lorentz symmetry of the kraus operators results in the breaking of the unitarity of the kraus operators into a block unitarity with each block characterized by the helicity @xmath22 .
the self - consistency constraint requires that the dephasing interaction with the environment conserves both the dipion helicity and the recoil nucleon helicity and transversity .
assuming unitary @xmath2-matrix we used the evolution equation @xmath297 to develop in ref @xcite a spin formalism to express the measured final state @xmath298 in terms of spin density matrix elements @xmath299 which are bilinear combinations of helicity amplitudes @xmath300 or transversity amplitudes @xmath301 and @xmath302 . on the other hand , the data analyses using this formalism lead to the conclusion that @xmath303 is given by a non - unitary kraus representation ( 4.24 ) .
self - consistency requires that the kraus representation preserves the experimental form of the angular intensities describing the final state density matrix and the form of the expressions for density matrix elements in terms of transversity amplitudes . in this section
we show that such invariance holds provided that the kraus partial wave amplitudes satisfy the same parity relations as the @xmath2-matrix amplitudes due to the conservation of @xmath3-parity .
the form ( 4.25 ) of each term @xmath304 in the kraus representation ( 4,24 ) is the same as the form of the @xmath2-matrix final state in terms of @xmath2-matrix amplitudes @xmath305 we can thus apply the spin formalism develped in ref .
@xcite to each such term separately
. then each term @xmath304 in ( 4.29 ) has the following form in terms of angular intensities @xmath306 with a decomposition of the intensities @xmath307 in terms of component intensities @xmath308 the components of intensities @xmath309 have angular expansion @xmath310 where the matrix elements @xmath311 are expressed in terms of kraus partial wave helicity amplitudes @xmath312 as the result of linearity of the kraus representation , the observed final state density matrix has the same form as in the case of the @xmath2-matrix @xmath313 where @xmath314 to bring @xmath315 to the experimental @xmath2-matrix form we need to bring to this form the intensities @xmath309 . to this
end the density matrix elements @xmath311 must satisfy the same parity relations as the corresponding matrix elements in the @xmath2-matrix theory .
this in turn requires that the kraus partial wave amplitudes satisfy the same parity relations as the corresponding @xmath2-matrix partial wave amplitudes . with the parity relation ( 5.12 )
the components @xmath309 then have the desired form @xcite @xmath316 for @xmath317 and the form @xmath318 for @xmath319 . from ( 7.7 )
it follows that the measured intensities @xmath315 will retain the same form of angular expansions provided that the measured density matrix elements @xmath320 are redefined as @xmath321 the measured density matrix elements are thus environment - averaged elements of the kraus density matrices .
l|r @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 $ ] + [ table i. ] the kraus density matrix elements @xmath354 are linear combinations of bilinear terms of kraus transversity amplitudes which are the same as those for the @xmath2-matrix elements and which are given in the table i. the measured density matrix elements @xmath320 are given by ( 5.10 ) in terms of elements of kraus density matrices . because this relation is linear the measured elements @xmath320 will retain the form given in table i. with moduli and bilinear terms of transversity amplitudes now replaced by environment - averaged moduli and bilinear terms of kraus amplitudes , respectively .
the expressions for the measured bilinear terms for any two transversity amplitudes @xmath355 and @xmath356 read @xmath357 where we have omitted the spin labels of the amplitudes for the sake of clarity .
note that the symbol @xmath358 does not stand for a product of two complex functions @xmath355 and @xmath359 but communicates which amplitudes @xmath360 are being averaged .
the averaged bilinear terms can be written in terms of effective transversity amplitudes @xmath361 where @xmath362 and @xmath363 are the moduli of the effective amplitudes and @xmath364 and @xmath365 are their correlations . with this form of measured bilinear terms we can no longer associate complex functions @xmath355 and @xmath356 with the moduli @xmath362 and @xmath363 , respectively , and identify the phases @xmath366 and @xmath367 as relative phases of these complex functions . in general the measured correlations violate the necessary trigonometric identity with @xmath368 . since the phases @xmath366 are not relative phases of complex functions
, they violate the phase condition @xmath106 and the equivalent cosine condition @xmath107 which hold for any three complex valued functions @xmath103 . to better understand the meaning of the measured moduli and correlations we define kraus vectors in the hilbert space @xmath61 @xmath369 and a scalar product @xmath370 comparing ( 5.16 ) with ( 5.11 ) and ( 5.12 ) we find latexmath:[\[\begin{aligned } \nonumber \nonumber provide a new physical interpretation of the measured ( averaged ) bilinear terms : instead of associating the measured effective amplitudes with complex functions we must associate them with complex kraus vectors ( 5.15 ) .
the measured bilinear terms of effective amplitudes are real or imaginary parts of scalar products of their kraus vectors .
it should be noted that the bilinear terms of kraus amplitudes @xmath360 carry the quantum numbers of the state vectors @xmath60 which define the interacting degrees of freedom of the quantum environment .
the bilinear terms of effective amplitudes are mixtures of these quantum numbers .
the unitarity of the dephasing amplitudes and the spin mixing mechanism ( 6.17),(6.18 ) imply that for each @xmath372 and @xmath272 @xmath373 depends on the dipion mass @xmath23 .
the observed moduli for transversity amplitudes are given by latexmath:[\[\begin{aligned } \nonumber ( 7.19 ) that the observed differential cross - section is equal to that given by the @xmath2-matrix amplitudes latexmath:[\[\frac{d^2 \sigma}{dmdt } = \sum \limits_{j \lambda,\tau } |u^j_{\lambda \tau}|^2 + |n^j_{\lambda \tau}|^2 = \sum \limits_{j \lambda,\tau } embodies the conservation of probability of a particle reaction in dephasing interaction with the environment .
the spin mixing mechanism ( 6.20 ) for the helicity amplitudes implies the same conservation of the differential cross - section .
we note that the the observed polarized target assymetry is similarly given by the @xmath2-matrix amplitudes @xmath376 recall that unitary evolution law of the @xmath2-matrix imposes relative phases @xmath377 on such transversity amplitudes where @xmath378 are integers @xcite .
then the equations ( 7.23 ) imply @xmath379 the cern - cracow - munich ( ccm ) measurements of @xmath7 at 17.2 gev / c @xcite measured moments @xmath380 ( unpolarized target ) , @xmath381 ( polarization component perpendicular to the scattering plane ) and @xmath382 ( polarization component in the scattering plane ) .
these moments are linear combinations of spin density matrix elements @xmath383 and @xmath384 , respectively , corresponding to different spins @xmath385 .
the itep measurements of @xmath7 at 1.78 gev / c @xcite and the cern - saclay measurements of @xmath386 at 5.98 and 11.85 gev / c @xcite measured directly the spin density matrix elements @xmath383 and @xmath384 for the @xmath2- and @xmath3-wave subsystem . to determine the observed production amplitudes ccm used a @xmath387 minimization method @xcite .
the moments are expressed in terms of the transversity amplitudes and these theoretical formulas are fitted to the measured momements .
all @xmath387 fits found two solutions @xmath388 for the @xmath2- and @xmath3-wave amplitudes @xmath389 below 980 mev ( @xmath390 threshold ) and a single solution for @xmath2-,@xmath391 and @xmath392-waves above 980 mev .
the cosines of the relative phases are treated as free parameters and are not constrained by the cosine conditions ( 3.11 ) in the minimization process .
the combinations @xmath393 of the density matrix elements of the @xmath2- and @xmath3-wave subsystem form two separate systems of equations for the amplitudes with transversity `` up '' ( sign + ) and `` down '' ( sign - ) .
each system is analytically solvable provided the cosines of the relative phases in each system satisfy the cosine conditions ( 3.11 ) which leads to a cubic equation for each @xmath394 .
the two real physical solutions of each cubic equation allow to determine two solutions @xmath395 for the @xmath2- and @xmath3-wave amplitudes @xmath389 below 980 mev .
the analytical method has been used to determine the initial values in the @xmath387 minimization analyses @xcite .
it has been fully employed in the itep analysis @xcite and in the monte carlo analyses of the ccm data on @xmath7 at 17.2 gev / c and the cern - saclay data on @xmath386 at 5.98 and 11.85 gev / c @xcite . in monte carlo
analyses the amplitudes are calculated for each sampling of the error volume .
unphysical solutions are rejected and the physical solutions are used to determine the average values of the amplitudes and their errors .
the resulting amplitudes are genuine complex valued functions .
kraus amplitudes are complex valued functions .
this suggests that the @xmath2- and @xmath3-wave analytical amplitudes may be related to the kraus amplitudes .
since the monte carlo amplitude analysis selects data that satisfy the cosine conditions , this data must describe a decoherence free subspace . in this case
there is only one set of @xmath2- and @xmath3-wave kraus amplitudes equal to one of the two analytical solutions . in our new analysis of the cern data using spin mixing mechanism @xcite we show that spin mixing mechanism selects the solution 2 of the analytical as well as the @xmath387 methods .
there are no exact analytical solutions of the @xmath2- , @xmath3- and @xmath392-wave system .
we thus expect a single effective solution for the @xmath2-,@xmath3-and @xmath392-wave system with some cosines of relative phases violating the cosine conditions . in the section ix .
we present evidence from the analysis of rybicki and sakrejda @xcite that the @xmath392-waves indeed violate the cosine condition .
below 1400 mev the process @xmath7 is described by @xmath2- , @xmath3- and @xmath392-wave transversity amplitudes . in the following
we shall use a new notation for these amplitudes .
the amplitudes @xmath396 and @xmath397 are the @xmath2-wave and @xmath3-wave unnatural exchange amplitudes with helicity @xmath274 .
the amplitudes @xmath398 and @xmath399 are the @xmath3-wave unnatural and natural exchange amplitudes with helicity @xmath400 .
there are three unnatural exchamge @xmath392-wave amplitudes @xmath401 corresponding to the helicities @xmath402 and two natural exchange @xmath392-wave amplitudes @xmath403 corresponding to helicities @xmath404 . with the notation @xmath405 in the spin mixing mechanism ( 6.17 ) and ( 6.18 ) the expressions for the @xmath2-,@xmath3- and @xmath392-wave kraus
amplitudes in terms of @xmath2-matrix amplitudes read as follows .
the three helicity zero amplitudes @xmath406 form a dephasing triplet @xmath407 the amplitudes @xmath408 and @xmath409 form two dephasing doublets . @xmath410
@xmath411 the amplitudes @xmath412 and @xmath413 form two dephasing singlets @xmath414 the matrices in ( 8.1)-(8.4 ) are unitary matrices @xmath415 and @xmath416 .
lorentz symmetry of kraus operators imposes a constraint @xmath5 .
this means that @xmath417 in ( 8.1 ) .
the unitarity of this matrix then implies @xmath418 and ( 8.1 ) takes the form @xmath419 @xmath420 is thus a dephasing singlet and the amplitudes @xmath421 and @xmath422 form a dephasing doublet .
the amplitude @xmath420 decouples from the @xmath2- and @xmath3- subsystem with the result that there can be no spin mixing of @xmath423 and @xmath424 amplitudes .
the matrix in ( 8.6 ) is a @xmath425 matrix . in the mass interval below 980 mev at low @xmath261 the @xmath2- and @xmath3-waves
dominate and @xmath392-waves are neglected .
there is no evidence for the presence of @xmath15 in the measured @xmath426 at high @xmath261 as predicted by the equations ( 8.5 ) .
the observed amplitudes @xmath398 and @xmath399 resonate at @xmath15 mass and at high @xmath261 @xmath399 are the largest amplitudes .
there is no evidence for the presence of @xmath15 in the amplitudes @xmath427 and @xmath428 at high @xmath261 .
this implies @xmath429 .
then ( 8.2 ) reads @xmath430 in this mass range there is a consistent body of evidence for @xmath6 mixing in the amplitudes @xmath431 and @xmath432 ( for a review see ref .
@xcite ) . to study the observable effects of the spin mixing mechanism we need to introduce the phases of the @xmath2-matrix transversity amplitudes .
in this subsection we omit the label @xmath2 of the @xmath2-matrix transversity amplitudes . in the part i. of this work @xcite
we have shown that the unitary evolution law imposes a constraint on the relative phases of the @xmath2-matrix amplitudes @xmath433 for all @xmath385 and corresponding @xmath434 .
it follows from the unitary conditions ( 8.7 ) that all unnatural exchange amplitudes @xmath30 as well as all natural exchange amplitudes @xmath31 share the same absolute phase up to an integer multiple of @xmath435 @xmath436 where @xmath437 and @xmath438 are integers .
the phases @xmath439 and @xmath440 depend on @xmath441 . the unitary phases ( 8.8 ) imply that if there is a resonance at the mass @xmath442 in the amplitudes @xmath443 ( @xmath444 ) with @xmath445 , then all contributing amplitudes @xmath446 ( @xmath447 ) with @xmath448 will have the same resonant phase near @xmath442 .
this does not imply that any of the amplitudes with @xmath448 must resonate and show a resonant peak or a dip at @xmath442 since their non - resonant moduli are mathematically allowed to have resonant phases .
although the unitary phases ( 8.8 ) appear to allow such resonance mixing , it is excluded by the the symmetries of the @xmath2-matrix and standard model . indeed , the production as well as the decay of a resonance is described entirely by the resonant @xmath2-matrix amplitudes @xmath449 while the mixing of resonances in the kraus amplitudes
@xmath450 arises entirely from the dephasing amplitudes .
with ( 8.8 ) the @xmath2-matrix transversity amplitudes then have the form latexmath:[\[\begin{aligned } u^j_{\lambda \tau } & = & e^{i\phi(s_\tau ) } e^{i\pi n(u^j_{\lambda \tau } ) } \nonumber n^j_{\lambda \tau } & = & e^{i\phi(n_\tau ) } e^{i\pi n(n^j_{\lambda \tau } ) } phases is presented in the table ii . in the part i. of this work @xcite . for the @xmath2- and @xmath3-wave subsystem below 980 mev
the relative phases read @xmath452 for masses @xmath453 mev there is a change of sign in the measured interference terms of the amplitudes prompting a change of relative phases for @xmath454 @xcite @xmath455 in ( 8.10 ) and ( 8.11 ) we have introduced the superscript @xmath456 to label the @xmath2- and @xmath3-wave @xmath2-matrix amplitudes to be used in the next subsection .
it follows that for @xmath457 mev @xmath458 while for the masses above 980 mev the signs of @xmath459 and @xmath460 are reversed .
the spin mixing matrix @xmath461 in ( 8.6 ) is a unitary @xmath425 matrix .
the most general @xmath425 matrix has the form @xcite @xmath462 with @xmath463 the spin mixing mechanism for @xmath2- and @xmath3-wave kraus amplitudes @xmath464 reads @xmath465 in the following we omit the label @xmath466 .
first we consider the @xmath15 mass region and @xmath457 mev . with the phases ( 8.12 ) the equations ( 8.14 ) read @xmath467
the moduli have a form @xmath468 . to fix the phases @xmath469 we require that the cosine @xmath470 of the relative phase @xmath471 is near its unitary value @xmath472 given by ( 8.10a ) @xmath473 from ( 8.15 ) and ( 8.16 ) we find @xmath474 where we have assumed @xmath475 .
the consistency with the unitary value requires that @xmath476 .
the unitarity of the resulting @xmath461 requires @xmath477 .
similarly we require that @xmath478 is near the unitary value @xmath479 .
this consistency implies that @xmath480 .
the final form of the spin mixing mechanism below 980 mev then reads @xmath481 with @xmath482 the equations ( 8.16 ) take the form latexmath:[\[\begin{aligned } \nonumber \nonumber of @xmath15 resonance in the amplitudes @xmath484 .
we now consider the @xmath18 mass region @xmath453 mev .
we shall focus on amplitudes with transversity @xmath454 . since there is no change in the amplitudes with @xmath485 we assume we still have @xmath486 and @xmath487 . with the reversed signs of @xmath432 and @xmath488 in ( 8.12 ) the final form of the spin mixing mechanism ( 8.14 ) for @xmath454 reads @xmath489 the moduli for @xmath454
have the form latexmath:[\[\begin{aligned } \nonumber \nonumber moduli retain the form ( 8.21 ) . the equations ( 8.21 ) and ( 8.23 ) both predict the presence of @xmath18 resonance in the amplitudes @xmath394 .
expressions for the relative phases can be solved exactly to read @xmath491 where @xmath492 . from the ratio @xmath493
we obtain a quadratic equation for the cotangent @xmath494 which has two solutions @xmath495 only @xmath496 is a physical solution for the assumed @xmath497 . the spin mixing parameters @xmath498 and @xmath499 depend on the c.m.s .
energy @xmath181 and the invariant dipion mass @xmath23 but do not depend on the transversity @xmath500 and momentum transfer @xmath261 .
for the expression on the r.h.s . of equation
( 8.28 ) to be independent of @xmath500 and @xmath261 for all @xmath501 we need @xmath502 and @xmath503 to be independent of @xmath500 and @xmath261 .
this independence for @xmath503 requires proportionality relations for the @xmath2-matrix amplitudes @xmath504 and @xmath505 with the same @xmath506 .
these relations ensure the required independence of @xmath502 .
the relations ( 8.29 ) and ( 8.30 ) do not apply to the helicity @xmath400 amplitudes @xmath507 and @xmath508 .
the independence of @xmath503 on @xmath500 and @xmath261 means that the relative phases do not depend on the transversity and , at a fixed @xmath181 and @xmath23 , on @xmath261 .
the spin mixing mechanism thus predicts that the solutions for the cosines of relative phases in amplitude analyses should be the same for both transversities and independent of @xmath261 at the same energy .
there are @xmath509 kraus amplitudes @xmath422 and @xmath421 .
the spin mixing parameters @xmath510 and @xmath511 are functions of energy @xmath181 and dipion mass @xmath23 but do not depend on the momentum transfer @xmath261 and the transversity @xmath500 .
we can identify the complex valued functions @xmath422 and @xmath421 with the complex valued analytical amplitudes obtained from the experimental data .
there are two solutions for the moduli of the @xmath2- and @xmath3-wave amplitudes and the cosines of their relative phases which implies @xmath512 .
it is also possible to cross combine the solution 1 ( 2 ) for the @xmath485 amplitudes with the solution 2 ( 1 ) for the @xmath454 amplitudes .
such four combinations of the solutions @xmath513 , @xmath514 , @xmath515 and @xmath516 imply @xmath517 . since the cosines do not determine the signs of the relative phases ,
there is a four - fold sign ambiguity @xmath518 in the relative signs of the relative phases in each solution in both @xmath512 and @xmath517 cases .
the amplitudes of the solutions @xmath519 and @xmath520 are simply complex conjugate amplitudes of the solutions @xmath521 and @xmath522 , respectively .
the cross solutions @xmath514 and @xmath515 are excluded by the spin mixing mechanism which requires @xmath523 and @xmath524 for each solution @xmath525 . in the following
we shall focus on the @xmath512 case with solutions 1 and 2 corresponding to combinations @xmath513 and @xmath516 . at this point
we do not know the @xmath2-matrix amplitudes @xmath526 and @xmath527 , i.e. the ratio @xmath503 , and the spin mixing parameters @xmath528 and @xmath529 so we can not make quantitative predictions for the moduli ( 8.23 ) and the cosines ( 8.24)-(8.26 ) .
however we can still make qualitative comparisons with the data and draw from them important conclusions .
figure 1 shows the @xmath2-wave moduli from the recent monte carlo analysis of the cern - cracow - munich data ( ccm ) at low momentum transfers @xmath530 ( gev / c)@xmath531 @xcite .
these results are nearly identical to the published results of the @xmath387 fit analysis of the same data by the ccm group @xcite shown in the figure 2 .
both analyses clearly indicate the presence of a @xmath15 peak in both solutions for the amplitude @xmath532 .
the solution 2 for @xmath532 exhibits a pronounced structure at 930 mev absent in the solution 1 .
the results of the recent monte carlo analysis @xcite are similar to the previously published monte carlo amplitude analyses of the ccm data at smaller monte carlo resolution @xcite .
the recent work includes analysis on unpolarized target and the determination of helicity amplitudes which both confirm the presence of the @xmath15 in the @xmath2-wave .
the 1997 @xmath387 fit analysis @xcite is in agreement with the 1979 @xmath387 analysis using ccm data from a different run @xcite .
independent evidence for @xmath15 in the @xmath2-wave comes from the analysis of itep data on @xmath7 at 1.78 gev / c at low @xmath261 @xcite , and from two analyses at high @xmath261 of ccm data @xcite and cern - saclay data on @xmath12 at 5.98 and 11.85 gev / c @xcite . for a full review with figures of all these results for the moduli and intensities
see ref .
this survey shows a remarkable consistency of the evidence for a rho - like state in the @xmath2-wave in @xmath533 production channel .
spin mixing mechanism explains the appearance of the @xmath15 peak in the measured amplitudes @xmath532 as due to the presence of a large resonating @xmath2-matrix amplitude @xmath534 in the expression for the @xmath2-wave modulus ( 8.23 ) for sufficiently large spin mixing parameter @xmath535 . in ccm data
this spin mixing appears suppressed in the amplitudes @xmath536 compared to @xmath532 .
the monte carlo analysis of ccm data @xcite shows a dip in the amplitude @xmath17 near the @xmath18 mass in both solutions while there is a sudden drop in @xmath532 at this mass .
these effects are due to the mixing of the decreasing @xmath2-matrix amplitude @xmath537 and increasing @xmath2-matrix amplitude @xmath538 above the @xmath390 threshold at 980 mev , indicating @xmath6 mixing in the @xmath3-wave amplitude .
[ htp ] -wave amplitudes @xmath536 and @xmath532 from the monte carlo analysis @xcite.,title="fig:",width=453,height=396 ] [ hp ] -wave amplitudes @xmath536 and @xmath532 from the @xmath387 fit analysis @xcite.,title="fig:",width=453,height=396 ] [ htp ] from the monte carlo amplitude analysis @xcite.,title="fig:",width=453,height=396 ] [ hp ] from the @xmath387 fit amplitude analysis @xcite.,title="fig:",width=453,height=396 ] [ htp ] from the monte carlo amplitude analysis @xcite.,title="fig:",width=453,height=396 ] [ hp ] from the monte carlo amplitude analysis @xcite.,title="fig:",width=453,height=396 ] figures 3 and 4 show the cosines for the relative phases @xmath539 for the monte carlo @xcite and for the @xmath387 fit @xcite analyses of the same ccm data , respectively
. the results are again nearly identical .
the most important but subtle difference is in the cosines for @xmath454 in the mass interval 700 - 800 mev . while in the monte carlo analysis
the values of this cosine are always less than 1.0 , in the @xmath387 fit analysis this cosine takes on the value of 1.0 at 749 mev bin .
this difference means that in the monte carlo analysis there can be no change of sign of the phase @xmath539 for @xmath454 below 1080 mev while this phase can change sign at the 749 mev bin in the @xmath387 analysis .
figures 5 and 6 show the cosines for the relative phases @xmath540 and @xmath541 , respectively , for the monte carlo analysis @xcite .
there are no corresponding figures available from the @xmath387 fit analysis @xcite . a careful inspection of the figures 3 , 5 and 6 shows that with the exception of solution 2 for @xmath470 the cosines of relative phases in monte carlo amplitude analysis are near the unitary values ( 8.10 ) and ( 8.11 ) as expected from the spin mixing relations ( 8.18 ) and ( 8.19 ) . also with the exception of solution 2 for @xmath470 ,
the cosines are approximately independent of the transversity @xmath500 .
this consistency is approximate because the amplitude analysis did not impose the spin mixing mechanism constraints @xmath542 and @xmath543 .
when these constraints are imposed in the amplitude analysis the consistency for @xmath470 becomes exact @xcite .
we now turn to the question of the dimension @xmath35 of the hilbert space @xmath61 . recall that @xmath544 in a non - unitary process @xmath545 @xcite . in the kraus representation ( 3.7 )
@xmath546 and @xmath547 .
the relevant hilbert space @xmath548 is the space of spin states for all final state particles forming the states @xmath1 and @xmath28 . the dimension @xmath35 must be common to all scattering processes . the simplest scattering process that involves interaction with the environment is @xmath0 .
at dipion mass @xmath23 the dimension @xmath549 is determined by the maximum dipion spin @xmath550 .
its simplest spin state describes the @xmath2-wave process @xmath551 . in the absence of @xmath3-waves in processes like @xmath552 or @xmath553
this is the only process near threshold .
the bnl high statistics meaurements of @xmath552 at 18.2 gev / c show that all @xmath392-wave amplitudes are consistent with zero for @xmath554 mev for all momentum transfers @xmath261 @xcite .
the evolution of @xmath555 due to its interaction with the environment then limits the dimension @xmath35 to @xmath556 . experimental evidence for @xmath36 can only come from the violation of the cosine condition ( 3.11 ) .
consider three observed amplitudes @xmath362 , @xmath363 and @xmath557 and their correlations @xmath558 , @xmath559 and @xmath560 . in general the phases @xmath561 , @xmath562 and @xmath563 violate the phase condition ( 3.10 ) but satisfy a phase gap condition @xmath564 which leads to a cosine gap condition @xmath565 where @xmath566 and @xmath567 are the phase gap and cosine gap , respectively .
they are related by the condition @xmath568 the cosine gap condition ( 9.2 ) is an unambigous signature of the non - unitary evolution law with @xmath36 . the first indication for the violation of the phase condition ( 9.1 ) and the cosine condition ( 9.2 ) came in 1985 from the amplitude analysis of cern measurement of @xmath7 at 17.2 gev / c at large momentum transfers @xmath569 ( gev / c@xmath570 by rybicki and sakrejda @xcite .
in that analysis they determined relative phases of @xmath3-wave amplitude @xmath571 ( @xmath572 in our notation ) and @xmath392-wave amplitudes @xmath573 and @xmath574 .
figure 10 shows the phase gap @xmath575 and the cosine gap @xmath576 @xmath577 these phase gaps and cosine gaps show large deviations from the unitary values 0 and 1 , respectively , as well as large fluctuations .
the analysis of rybicki and sakrejda supports the non - unitary evolution law with @xmath36 .
their analysis also finds evidence for a rho - like resonance in the @xmath2-wave , suggesting @xmath6 mixing at large @xmath261 ( see also ref .
@xcite ) .
[ hp ] and cosine gap @xmath578 for aplitudes @xmath571 , @xmath579 and @xmath574 .
data from ref .
@xcite.,title="fig:",width=453,height=396 ] in our new analysis @xcite the spin mixing mechanism allows to extract @xmath392-wave observables from the cern data .
analysis of the full @xmath392-wave subsystem for transversity @xmath485 reveals a violation of cosine conditions ( 3.11 ) by the amplitudes @xmath580 and @xmath581 .
detailed analysis of the amplitudes @xmath580 and @xmath581 determines that the number of interacting degrees of freedom of the environment is @xmath517 .
the evidence that @xmath582 brings up the issue of decoherence free and decohering amplitudes . this issue is relevant since the decoherence free subspaces are linked to observations of spin mixing .
kraus amplitudes forming decoherence free subspace do not depend on the interacting degrees of freedom @xmath466 since all kraus operators are equal @xmath583 , @xmath584 . in contrast
the decohering kraus amplitudes do depend on @xmath466 due to different kraus operators .
three or more kraus amplitudes are recognized as decoherence free amplitudes if all cosine conditions on the cosines derived from their measured bilinear terms are satisfied .
then we refer to the measured amplitudes also as decoherence free amplitudes .
if one or more cosine conditions are violated they evidence one or more decohering kraus amplitudes and we refer to the measured amplitudes as decohering amplitudes . in the absence of a theory of dephasing interactions the spin mixing matrices @xmath585
are only constrained by the experimental data from which we can deduce which amplitudes are decoherence free and which are decohering . in the sequel paper @xcite we report a new amplitude analysis of the cern data below 1080 mev using spin mixing mechanism ( 8.20 ) and ( 8.22 ) and the constraints ( 8.29 ) to determine anew the spin mixing @xmath2-and @xmath3-wave transversity amplitudes @xmath586 , and to determine the @xmath2-matrix transversity amplitudes @xmath587 and the spin mixing parameters @xmath125 .
the spin mixing mechanism also allows to extract full set of @xmath392-wave amplitudes for the transversity @xmath485 in a simultaneous analysis .
the spin mixing mechanism admits only a single physical solution for the @xmath2-and @xmath3-wave amplitudes and for the @xmath392-wave amplitudes @xmath588 corresponding to the solution 2 in figures 1 - 6 . interpreting this solution as kraus amplitudes implies that these amplitudes form a decoherence free subspace because @xmath36 .
the spin mixing parameters @xmath125 do not depend on @xmath466 .
[ hp ] and cosine gap @xmath578 for aplitudes @xmath571 , @xmath396 and @xmath579 .
data from ref .
@xcite.,title="fig:",width=453,height=396 ] there is no evidence for the presence of @xmath15 in the amplitude @xmath426 in our analysis @xcite and in the high @xmath261 analysis @xcite , as predicted by the equations ( 8.5 ) .
we do not expect @xmath589 to mix with @xmath397 because @xmath579 still decouples from the @xmath2- and @xmath24-wave amplitudes in agreement with the data .
figure 11 shows that in the analysis at low @xmath261 and @xmath590 mev @xcite the phase condition and cosine condition are satisfied by the relative phases @xmath539 , @xmath591 and @xmath592 for both @xmath500 .
the analysis of the full @xmath392-wave subsystem reveals @xmath15 mixing in the amplitudes @xmath593 and @xmath594 which is at variance with the analysis at high @xmath261 @xcite .
this mixing requires a small mixing of the @xmath2-matrix amplitudes @xmath595 and @xmath596 in the amplitudes @xmath597 and @xmath598 in ( 8.2 ) and ( 8.3 ) , respectively @xmath599 where the mixing matrix @xmath600 has a general form ( 8.13 ) .
the effective amplitudes @xmath601 and @xmath602 have the same @xmath2-matrix phases as @xmath603 and @xmath604 .
the amplitudes @xmath595 and @xmath596 have the same phases as @xmath603 and @xmath604 , respectively @xcite .
amplitude analysis determines the effective amplitudes . above 980 mev
there is no evidence for the presence of @xmath589 in the amplitudes @xmath398 and @xmath399 @xcite indicating @xmath605 at these masses .
our analysis of the amplitudes @xmath574 and @xmath606 in ref .
@xcite determines four phases @xmath607 for the two decohering amplitudes @xmath608 and @xmath609 in ( 8.4 ) @xmath610 which sets the dimension @xmath517 .
the results for the @xmath392-wave amplitudes mean that the amplitudes @xmath579 , @xmath611 and @xmath612 belong to the same decoherence free subspace with the @xmath2-and @xmath3-wave amplitudes .
this subspace includes three pairs of spin mixing amplitudes ( @xmath613 ) , ( @xmath408 ) and ( @xmath409 ) .
we conclude with the observation that the spin mixing occurs in decoherence free subspaces while there is no spin mixing in decohering subspaces .
we shall use the coleman - mandula no - go theorem @xcite to show that the non - unitary evolution law ( 3.7 ) is consistent with the @xmath2-matrix scattering dynamics and quantum field theory provided the dimension @xmath36 .
the theorem requires that in the decohering subspaces there can be no violation of lorentz symmetry while there must be a lorentz symmetry violating spin mixing in the decoherence free subspaces . a modern proof of the theorem is given by weinberg in ref .
assume that the generators @xmath614 and @xmath615 of the poincare group commute with the @xmath2-matrix .
assume that internal symmetry generators @xmath616 ( charges ) commute with the @xmath2-matrix and form an algebra given by commutation relations @xmath617=if^{abc}t^c\ ] ] coleman - mandula no - go theorem asserts that there is no physically meaningful @xmath2-matrix theory where the charges @xmath616 do not commute with the generators of the poincare group .
that means there is no mixing of space - time transformations and internal symmetry transformations .
supersymmetry evades this no - go theorem by replacing the commutation relations ( 11.1 ) with anticommutation relations @xcite .
the key assumption in the proof of the coleman - mandula theorem is the assumption that a two - body scattering always exists in the @xmath2-matrix theory as opposed to a situation where only a forward scattering is possible .
theories that allow only forward scattering evade the no - go theorem .
the matrix elements of the operators @xmath618 describing the pure dephasing interaction of the particle final states of @xmath1 all describe forward scattering of these states .
do they evade the no - go theorem ? to answer this question
let us first assume that there is only one kraus operator @xmath619
. this operator @xmath619 can not be the @xmath2-matrix for the @xmath2-matrix would allow a non - trivial scattering of the produced particle states .
thus @xmath619 is a genuine `` no - scattering '' operator .
the lorentz symmetry is conserved by @xmath619 and the coleman - mandula theorem is evaded
. then conserved internal charges associated with the kraus operator @xmath619 need not commute with the generators of the poincare group and in fact can carry a lorentz index .
these charges would include isospin of particles involved in the forward scattering in a complete contradiction with the @xmath2-matrix theory including supersymmetry . to resolve this contradiction
we note that the derivation of the no - go theorem assumes that the evolution of the scattering process is unitary . by assuming that there is more than one dephasing `` no - scattering '' operator in ( 3.7 ) , i.e. by assuming a non - unitary evolution with @xmath620
, we can `` evade the evasion '' and avoid any charges with lorentz index to resolve the contradiction .
the non - unitary evolution law does allow for unitary decoherence free channels in the hilbert space @xmath621 . in these subspaces
all kraus operators @xmath101 are equal to a unitary operator @xmath622 . in these spaces
we can not invoke the condition @xmath36 to `` evade the evasion '' of the no - go theorem .
neither can we invoke the breaking of the lorentz symmetry by the kraus operators . the only option
the dynamical system now has to `` evade the evasion '' in order to preserve its consistency with the @xmath2-matrix dynamics is to break lorentz symmetry by mixing some @xmath2-matrix amplitudes with different spins in the observed amplitudes .
this is a classical case of a spontaneously broken symmetry induced by the environment .
the spin mixing of the @xmath2-matrix amplitudes is a pure quantum mechanical effect due to the superposition principle .
it arises from the action of the kraus operators on the dipion - recoil nucleon states @xmath623 given by the superposition ( 4.14 ) .
since it affects only the observed amplitudes it co - exists with the fundamental lorentz symmetry of the particle dynamics and standard model .
quantum superposition principle co - exists with lorentz symmetry in another important quantum effect arising from this principle - quantum entanglement of spin states or other quantum states .
the apparent violation of lorentz symmetry by two entangled particles is well known since the famous epr paper . in this sense the spin mixing in particle scattering is analogous to quantum entanglement . in both cases
there is an apparent violation of the lorentz symmetry .
vector - scalar meson mixing is not a new idea . in 1977
chin studied @xmath624 mixing in nucleon - nucleon interactions in high density matter @xcite . in 2000 - 2002 gale and collaborators examined the effects of @xmath625 mixing on the dilepton production in relativistic heavy ion collisions @xcite measured at rhic .
the @xmath624 mixing was induced by the ground state of the system of the interacting nucleons , while the @xmath625 mixing originated in nucleon - nucleon excitations in the medium of nuclear matter . in both cases
there was no violation of fundamental symmetries because the interaction lagrangian conserved those symmetries including the lorentz symmetry .
the meson spin mixing and the spontaneous violation of lorentz symmetry was due to the interaction of the nucleon - nucleon scattering process with its environment .
it was this work on vector - scalar mixing that inspired the present author in 2003 to seek the solution to the puzzle of the appearance of a rho - like resonance in the @xmath2-wave in @xmath7 and its absence in @xmath13 in the idea of @xmath6 spin mixing .
the cern measurements of @xmath7 on polarized target at 17.2 gev / c are so far the only high statistics measurements of its kind , yet they revealed an interesting new physics beyond the standard model . in part
i. of this work we have shown that they reveal a nonunitary interaction of the produced final state with a quantum environment . to preserve the identity of the final state particles
this new interaction must be a pure dephasing interaction . in this part ii .
we required that the dephasing interaction be fully consistent with the standard model . from the complexity of the consistency requirements
emerges the simplicity of the spin mixing mechanism ( 6.17 ) and ( 6.18 ) , and its validation by the conservation of probability ( 7.21 ) . from this consistency alone
we have deduced that in @xmath0 processes the new interaction with the environment must be a dipion spin mixing interaction governed by the spin mixing mechanism . applied to the @xmath2- and @xmath3-wave subsystem in @xmath7
the theory predicts @xmath6 mixing in agreement with the evidence for such a mixing in all analyses of five experiments on polarized targets surveyed in ref .
@xcite . due to the selection rule @xmath5
the theory correctly predicts the absence of @xmath589 mixing in @xmath7 . since only @xmath190
are allowed in @xmath13 and in @xmath553 there is no spin mixing in these processes , only a change of the phase of the @xmath2-matrix partial wave amplitudes .
consider u - channel processes such as @xmath626 and @xmath627 . in these reactions
the two - pion systems @xmath628 and @xmath14 are accompanied by a spin 0 particle .
the conservation of the total angular momentum by the kraus operators in @xmath629 then requires @xmath630 .
therefore there is no spin mixing also in these processes . in a sequel paper
@xcite a new amplitude analysis of the cern data using spin mixing mechanism determines spin mixing parameters @xmath528 and @xmath529 , @xmath2-and @xmath3-wave @xmath2-matrix amplitudes and selects a single solution for the spin mixing amplitudes of a decoherence free @xmath2- and @xmath3-wave subsystem .
this analysis allows to extract @xmath392-wave observables from the cern data .
analysis of the full @xmath392-wave subsystem for transversity @xmath485 reveals a @xmath15 mixing in the dephasing doublets @xmath408 and @xmath409 .
detailed analysis of the amplitudes @xmath580 and @xmath581 determines that the number of interacting degrees of freedom of the environment is @xmath517 .
evidence from the analysis at high @xmath261 @xcite limits the dimension @xmath35 of the hilbert space of the environment to @xmath631 .
we use coleman - mandula no - go theorem to show that the consistency of the pure dephasing interactions with the standard model requires @xmath36 . as a result decoherence free subsystems
must spontaneously break lorentz symmetry while the lorentz symmetry is conserved by the decohering subspaces .
the breaking of lorentz symmetry or the violation of cosine conditions in the observed amplitudes are thus two unambigous signatures of dephasing interactions .
the obvious question arises - what is the physical nature of the quantum environment ?
the consistency of the quantum environment with the standard model suggests that it is a universal environment in the entire universe .
it can be viewed as a sea of quantum states @xmath96 across the universe .
the quantum states @xmath96 must be stable and can not directly interact with any of the particles of the standard model including the higgs boson .
the pure dephasing interaction involves only the exchange of quantum information between the two quantum states @xmath1 and @xmath68 .
as such it stands outside of the standard model . while the dephasing interaction can have strong observable effects , such as @xmath6 mixing , its effects are not observable in the usual studies of particle scattering .
since these experiments do not involve diparticle partial wave amplitudes in measurements on polarized targets , they can not observe the resonance mixing or the violations of the cosine conditions .
the measurements of relative phases alone can not provide evidence for the quantum environment since the needed unitary phases are known only for @xmath632 processes .
the dephasing interaction is largely a `` dark '' interaction .
the universal presence of the quantum environment and its pure dephasing interaction with the baryonic matter must manifest themselves in astrophysical observations .
such observations provide a convincing evidence for the existence of dark matter and dark energy which are both omnipresent environments in the universe with non - standard interactions with baryonic matter . in ref .
@xcite we present a physically motivated model of the quantum states @xmath96 which we propose to identify with the particles of a distinct component of the cold dark matter . in this picture
pure dephasing interactions are the interactions of baryonic matter with this form of dark matter .
these interactions are not rare events but they require high statistics measurements on polarized targets for their detection .
dedicated measurements of production processes such as @xmath0 and @xmath633 on polarized targets would provide new and accessible tools to explore the dark sector of the universe . | in part i. of this work we have presented evidence that the measured relative phases of transversity amplitudes in @xmath0 processes differ from those predicted by the unitary evolution law .
we ascribed this difference to a non - unitary interaction of the produced final state @xmath1 with a universal quantum environment in the physical universe .
this new kind of interaction must be a pure dephasing interaction in order to render the @xmath2-matrix dynamics of particle scattering processes accessible to experimental observation .
if the quantum environment is to be an integral part of the nature then its dephasing interactions must be fully consistent with the standard model .
in this work we impose on the dephasing interaction the requirements of the conservation of the identities of the final state particles including their four - momenta , lorentz symmetry , @xmath3-parity and the conservation of total angular momentum and isospin . from this consistency alone
we find that the dephasing interaction must be a dipion spin mixing interaction .
the observed amplitudes are a unitary transform of the corresponding @xmath2-matrix amplitudes .
the elements of the spin mixing matrix are forward scattering amplitudes of dipion spin states with recoil nucleon spin states with initial and final dipion spins @xmath4 and @xmath5 , respectively .
dipion helicities are conserved in this scattering .
these amplitudes are matrix elements of kraus operators describing the non - unitary interaction with the environment .
the theory predicts @xmath6 mixing in the @xmath2-and @xmath3-wave amplitudes in @xmath7 .
the predicted moduli and relative phases of the mixed amplitudes are in an excellent qualitative agreement with the experimental results .
the mixing of @xmath2-matrix partial wave amplitudes with different dipion spins to form the observable partial wave amplitudes is a new phenomenon beyond the standard model .
the spontaneous violation of rotational / lorentz symmetry observed in the measured @xmath2- and @xmath3-wave amplitudes is consistent with the conservation of that symmetry by the @xmath2-matrix and by the standard model lagrangian . |
enormous effort has gone into understanding the nature of sub - mm selected galaxies since they were first discovered with the submillimetre common user bolometer array ( scuba , holland et al .
1999 ) just under a decade ago ( e.g. smail et al . 1997 ; barger et al . 1998 ; hughes et al .
1998 ) . a general understanding of the sub - mm population and its role in galaxy evolution
is severely limited by the poor spatial resolution of current sub - mm telescopes and the faintness of these sources at other wavelengths ( e.g. pope et al .
the sub - mm population appears to consist of massive objects ( borys et al .
2005 ; greve et al . 2005 ) at @xmath1 ( chapman et al . 2005 ;
pope et al . 2005 ) .
these galaxies are rare and have very high star formation rates ( sfrs , lilly et al . 1999 ;
scott et al . 2002 ) , and hence may represent an early phase in the evolution of massive elliptical galaxies .
an outstanding question remains : are the incredible infrared luminosities of sub - mm galaxies powered by huge bursts of star formation or agn activity ? to assess in detail what powers the intense ir luminosity of sub - mm galaxies it is important to obtain a complete multi - wavelength picture .
the sample of sub - mm galaxies discussed in this paper is taken from the goods - n scuba super - map ( borys et al . 2003 ; pope et al . 2005 ) .
this is a purely sub - mm selected sample and it is almost completely identified at other wavelengths thanks to the deep multi - wavelength data available in the goods - n field ( see pope et al .
2006 for more details ) .
the addition of the mips photometry to the multi - wavelength dataset for sub - mm galaxies provides powerful constraints on the shape of the sed , the nature of the power source , and the source redshift . in the left panel of fig . 1
, we plot the @xmath2 flux ratio as a function of redshift as compared to several models . in general , this flux ratio as a function of redshift is lower than that of other ultra luminous infrared galaxies ( ulirgs ) of the same luminosity at low and high redshift .
this suggests that sub - mm galaxies either have higher levels of extinction or cooler dust temperatures .
this can be tested further by looking in detail at the mid - ir spectrum .
using deep multi - wavelength follow - up observations of sub - mm galaxies , it is possible to study their seds to derive fundamental properties such as dust properties and sfrs . in the right panel of fig . 1 , we plot a composite rest - frame sed for the sub - mm sources with spectroscopic redshift as compared to several models . we find that sub - mm galaxies have cooler average dust temperatures than those of local ulirgs of the same luminosity ( see pope et al .
2006 for more details ) .
this result is confirmed with the addition of deep @xmath3 m imaging of these sub - mm galaxies ( huynh et al .
this may indicate that their far - ir emission is more extended than that of local ulirgs , in which the majority of the ir emission comes from within the central kpc ( charmandaris et al .
the seds of sub - mm galaxies are also different from those of their high redshift neighbours selected at near - ir wavelengths ( e.g. bzk galaxies , daddi et al .
2005 ) , whose mid - ir to radio seds are more like those of local ulirgs .
we may understand this difference in terms of the evolutionary scenario advocated by vega et al .
( 2005 ) , in which enshrouded star forming galaxies undergo four distinct phases characterized by different mid- and far - ir colours . under their scheme , the sub - mm galaxies , with cooler temperatures , may be an earlier phase in the star formation than bzk galaxies .
many sub - mm galaxies show evidence of an agn , as determined either through x - ray observations ( alexander et al .
2005 ) , or with optical spectra ( swinbank et al
. 2004 ; chapman et al . 2005 ) , however it is unclear if the agn is a significant contributor to the bolometric luminosity of the galaxy .
mid - ir spectroscopy can help determine what powers sub - mm galaxies since these wavelengths contain a number of features which are sensitive to the nature of the energy source ( clavel et al .
2000 ; hudgins & allamandola 2004 ) .
we obtained _ spitzer _ irs observations of a sample of the sub - mm galaxies and agn in goods - n ( go2 pi : r. chary ) .
2 shows the rest - frame raw irs spectra for 10 sub - mm sources from this sample , offset for clarity .
all of these spectra show at least weak polycyclic aromatic hydrocarbon ( pah ) emission .
we have obtained several new redshifts from these irs spectra ( pope et al . in preparation ) .
3 shows the mean irs spectrum for these sub - mm galaxies compared to those of several local galaxies .
this composite spectrum was calculated by taking the mean of the individual spectra .
note that using the median instead of the mean does not make a significant difference to the shape of the composite spectrum .
we have excluded the top source in fig . 2 since its steeply rising mid - ir continuum is not seen in the other 9 spectra in our sample .
this source is not typical of our sample and is likely to have a stronger agn component . in the composite spectrum
, we see the 6.2 , 7.7 , 8.6 and 11.3@xmath4 m pah features much more clearly . while arp220 is often considered the typical local analog to high redshift sub - mm galaxies , it appears to have more silicate absorption and less 6.2@xmath4 m pah emission than the average sub - mm galaxy .
interestingly , the mid - ir spectra of the sub - mm galaxies seems to resemble that of m82 with the addition of a shallow power - law component ( @xmath5@xmath6 ) .
the total contribution of the power - law component to the total luminosity in this wavelength range is @xmath7 .
m82 is a typical local starburst galaxy but it is at least 2 orders of magnitude less luminous in the ir than the average sub - mm galaxy .
this composite suggests that the mid - ir emission in sub - mm galaxies is dominated by star formation activity and there is little contribution from an agn .
the average 7.7@xmath4 m line to continuum ratio for this sample is @xmath83.5 . comparing this with figure 4 from genzel et al .
( 1998 ) , we find that this ratio for sub - mm galaxies is consistent with the ratios for sbs and ulirgs and not as low as those found for agn .
our preliminary analysis suggests that sub - mm galaxies are powered primarily by sbs and not agn activity .
we are currently working on fitting templates to quantify the level of agn contributing to the ir luminosity and investigating various emission and absorption line strengths as diagnostics for the star formation activity .
combined with the other multi - wavelength data that are already tabulated on these sub - mm sources , we will create a complete picture of the energetics of sub - mm galaxies ( pope et al . in preparation ) .
these data will help determine if sub - mm galaxies and agn populations are connected via an evolutionary sequence .
thanks to the organizers of this conference for a very fruitful meeting .
we also thank our collaborators involved in the goods - n irs project .
this work was supported by the natural sciences and engineering research council of canada and the canadian space agency .
this work is based in part on observations made with the _
spitzer space telescope _ , which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
support for goods , part of the _ spitzer space telescope _
legacy science program , was provided by nasa through contract number 1224666 issued by jpl , caltech , under nasa contract 1407 .
alexander d.m . ,
et al . , 2005 , apj , 632 , 736 barger a.j . , et al . , 1998 , nat , 394 , 248 borys c. , et al . , 2003 , mnras , 344 , 385 borys c. , et al . , 2005 ,
apj , 635 , 853 chapman s.c . , et al . , 2005 , apj , 622 , 772 charmandaris , v. , et al . , 1999 , , 266 , 99 charmandaris v. , stacey g.j . , gull g. , 2002 , apj , 571 , 282 chary r. , elbaz d. , 2001 , apj , 556 , 562 clavel j. , et al . , 2000 , a&a , 357 , 839 daddi e. , et al .
, 2005 , apj , 631 , l13 draine b.t . , 2003 , ara&a , 41 , 241 fischer , j. , et al . 1999 , , 266 , 91 frster schreiber , n. m. , et al .
, 2003 , , 399 , 833 genzel , r. , et al .
1998 , , 498 , 579 greve t.r . , et al . , 2005 ,
mnras , 359 , 1165 holland w.s . , et al . , 1999 , mnras , 303 , 659 hudgins d. m. , allamandola l. j. , 2004 , aspc , 309 , 665 hughes d.h .
, 1998 , nature , 394 , 241 huynh m. et al .
, 2006 , apj submitted lilly s.j .
, et al . , 1999 , apj , 518 , 641 pope a. , et al . , 2005 , mnras , 358 , 149 pope a. , et al . , 2006 , mnras , 370 , 1185 scott s.e . , et al . , 2002 , mnras , 331 , 817 smail i. , ivison r.j . ,
blain a.w . , 1997 ,
apj , 490 , l5 swinbank a.m. , et al . , 2004 , apj , 617 , 64 vega o. , et al .
, 2005 , mnras , 364 , 1286 | submillimetre ( sub - mm ) galaxies have very high infrared ( ir ) luminosities and are thousands of times more numerous at @xmath0 than local ultra - luminous ir galaxies .
they therefore represent a key phase in galaxy evolution which can be missed in optical surveys .
determining their contribution to the global star formation rate requires dissecting their ir emission into contributions from starbursts ( sb ) and active galactic nuclei ( agn ) .
there are several examples of agn systems which masquerade as sbs in either the ir or x - ray , and sbs can often look like agn in some wavebands .
a combination of sb and agn emission is not unreasonable , given models of merger - driven evolution .
to assess in detail what powers the intense ir luminosity of sub - mm galaxies it is important to obtain a complete multi - wavelength picture .
mid - ir spectroscopy is a particularly good probe of where the intense ir luminosity is coming from .
we present the first results from a program to obtain _ spitzer _ irs spectroscopy of a sample of high redshift galaxies in the goods - n field , a large fraction of which are sub - mm galaxies .
this field is already home to the deepest x - ray , optical , ir and radio data .
we piece together the sub - mm data with the _ spitzer _ photometry and irs spectra to provide a well sampled ir spectral energy distribution ( sed ) of sub - mm galaxies and determine the contribution to the bolometric luminosity from the agn and sb components . |
it is wellknown that for relatively short polymer chains the standard rouse model can describe the dynamics of a melt reasonably well @xcite . on the contrary , for chain length
@xmath3 exceeding a critical length , the entanglement length @xmath4 , the behavior is usually described by the reptation model @xcite . here
we restrict ourselves to chain lengths @xmath5 , i.e. the entangled polymer dynamics will be beyond of our consideration .
the reason why in a dense melt the rouse model provides so well dynamical description for short chains is connected with a screening of the long - range hydrodynamic as well as the excluded volume interactions . as a result the fluctuations of the chain variables are gaussian .
but there are further essential questions : how does the bare monomeric friction coefficient @xmath6 and the entropic elastic modulus @xmath7 ( which are simple input parameters of the standard rouse model ) change due to the interactions of the test chain and the surrounding matix ?
why does such a simple model work so well for describing short chain melts ?
obviously , the corresponding answers can not be given by the rouse model , which describes only the dynamics of connected gaussian springs without further interactions .
on the other hand , at relatively low temperatures close to the glass transition of the surrounding matrix the deviations from the standard rouse behavior will be definitely more pronounced .
for example , monte carlo ( mc ) studies of the bond fluctuation model at low temperatures ( but still above the temperature region where possibly the glass transition mode coupling theory @xcite applies ) show that the rouse modes remain well - defined eigenmodes of the polymer chains and the chains retain their gaussian properties @xcite .
nevertheless , the relaxation of the rouse modes displays a stretched exponential behavior rather than a pure exponential .
it could even be expected that at temperatures below the glass transition temperature of the matrix @xmath0 the rouse modes are frozen out . in these temperature regimes
the interactions between monomers take a significant role and determine the physical picture of the dynamics as will be shown below .
the generalized rouse equation ( gre ) , which can be used for the investigation of the problems mentioned above , has been derived by using projection formalism methods and mode coupling approximations ( mca ) @xcite . as a result of projection operator formalism the time evolution of the test chain
is expressed in terms of a frequency matrix , which is local in time , and a memory function contribution due to the inter - chain forces exerted on the test chain segments . with the assumption that the frequency matrix term has the same form as in the standard rouse model ( linear elasticity with the entropic modulus @xmath8 ) all influence of the matrix chains reduce to the memory function contribution @xcite .
the projection operator methods appears to be exact but rather formal , and to derive explicit results further approximations have to be made , which can be hardly controlled often .
therefore it is instructive to use another alternative theoretical method to derive the gre .
recently , a non - pertubative variational method which is equivalent to a selfconsistent hartree approximation was used for the investigation of the dynamics of manifolds @xcite and sine - gordon model @xcite in a random media . as a starting point
the authors employed the standard martin - siggia - rose ( msr ) functional integral technique @xcite . here
we follow this approach to derive a gre and study the dynamics of a test polymer chain in a glass forming matrix .
+ the paper is organized as follows . in section 2
, we give a general msr - functional integral formulation for a test chain in a polymer ( or non - polymer ) matrix . under the assumption that the fluctuations of the test chain are gaussian
the hartree - type approximation is applied and a gre is finally derived .
the case when the fluctuation dissipation theorem ( fdt ) and the time homogenity are violated is also shortly considered . in section 3 on the basis of the gre some static and dynamical properties of the test chain are discussed .
in particular the theory of the test chain ergodicity breaking ( freezing ) in a glassy matrix is formulated .
section 4 gives some summary and general discussion .
the appendices are devoted to some technical details of the hartree - type approximation .
let us consider a polymer test chain with configurations characterized by the vector function @xmath9 with s numerating the segments of the chain , @xmath10 , and time @xmath11 .
the test polymer chain moves in the melt of the other polymers ( matrix ) which positions in space are specified by the vector functions @xmath12 , where the index @xmath13 numerates the different chains of the matrix .
the test chain is expected to have gaussian statistics due to the screening of the self - interactions in a melt @xcite .
we consider the simultaneous dynamical evolution of the @xmath9 and @xmath12 variables assuming that the interaction between matrix and test chain is weak .
+ the langevin equations for the full set of variables @xmath14 has the form @xmath15 @xmath16 where @xmath17 denotes the bare friction coefficient , @xmath18 the bare elastic modulus with the length of a kuhn segment denoted by @xmath19 , @xmath20 and @xmath21 are the interaction energies of test chain - matrix and matrix - matrix respectively , and @xmath22 , @xmath23 are the random forces with the correlator @xmath24 after using the standard msr - functional integral representation @xcite for the system ( [ r]-[f ] ) , the generating functional ( gf ) takes the form @xmath25-a_1\left[{\bf r}^{(p)}(s , t),{\bf { \hat r}}^{(p)}(s , t)\right]\nonumber\\&+&\sum_{p=1}^m\int_0^n ds\int_0^n ds'\int dt\:i{\hat r}_j(s , t)\frac{\delta}{\delta r_j(s , t)}v\left[{\bf r}(s , t)-{\bf r}^{(p)}(s',t)\right]\nonumber\\&+&\sum_{p=1}^m\int_0^n ds\int_0^n ds'\int dt\:i{\hat r}^{(p)}_j(s',t)\frac{\delta}{\delta r_j^{(p)}(s',t)}v\left[{\bf
r}^{(p)}(s',t)-{\bf r}(s , t)\right]\bigg\}\nonumber\end{aligned}\ ] ] where the dots represents some source fields which will be specified later and einstein s summation convention for repeated indices is used . in gf ( [ gf ] ) the msr - action of the free test chain
is given by @xmath26&=&-\int_0^n ds\int dt\bigg\{i{\hat r}_j(s , t)\left[\xi_0\frac{\partial}{\partial t}r_j(s , t)-\varepsilon\frac{\partial^2}{\partial s^2}r_j(s , t)\right]\nonumber\\&+&t\xi_0\left[i{\bf { \hat r}}(s , t)\right]^2\bigg\}\qquad.\end{aligned}\ ] ] as we will realize later the explicit form of the full action of the medium @xmath27 $ ] plays no role .
in principle it could have any form and in particular , for a polymer matrix , the following one @xmath28=-\sum_{p=1}^m\int_0^n ds\int dt\:i{\hat r}_j^{(p)}(s , t)\left[\xi_0\frac{\partial}{\partial t}r_j^{(p)}(s , t)-\varepsilon\frac{\partial^2}{\partial s^2}r_j^{(p)}(s , t)\right]\nonumber\\&-&\sum_{p=1}^m\int_0^n ds\int dt\:i{\hat r}_j^{(p)}(s , t)\frac{\delta}{\delta r_j^{(p)}(s , t)}\sum_{m=1}^m\int ds'{\tilde v}\left[{\bf { \hat r}}^{(p)}(s , t)-{\bf { \hat r}}^{(m)}(s',t)\right]\nonumber\\&+&\sum_{p=1}^m\int_0^n ds\int dt\:t\xi_0\left[i{\hat r}_j(s , t)\right]^2\end{aligned}\ ] ] in order to obtain an equation of motion for the test chain one should integrate over the matrix variables @xmath12 first . for this end
it is reasonable to represent gf ( [ gf ] ) as @xmath29-a_{0}\left[{\bf r}(s , t),{\bf { \hat r}}(s , t)\right]\right\}\label{gf1}\end{aligned}\ ] ] where the influence functional @xmath30 is given by @xmath31=-\ln\int\prod_{p=1}^{m}d{\bf r}^{(p)}(s , t)d{\bf { \hat r}}^{(p)}(s , t)\times\label{inf}\\&\times&\exp\bigg\{-a_{1}\left[{\bf r}^{(p)},{\bf { \hat r}}^{(p)}\right]\nonumber\\ & + & \sum_{p=1}^m\int_0^n ds\int_0^n ds'\int dt\:i{\hat r}_{j}(s , t)\frac{\delta}{\delta r_{j}(s , t)}v\left[{\bf r}(s , t)-{\bf
r}^{(p)}(s',t)\right]\nonumber\\&+&\sum_{p=1}^m\int_0^n ds\int_0^n ds'\int dt\:i{\hat r}_{j}^{(p)}(s',t)\frac{\delta}{\delta r_{j}(s',t)}v\left[{\bf r}^{(p)}(s',t)-{\bf r}(s , t)\right]\bigg\}\nonumber.\end{aligned}\ ] ] in the spirit of the mode coupling approximation ( mca ) @xcite the force between the test chain and the matrix should be expressed as a bilinear product of the two subsystems densities . in order to assure this we expand the influence functional ( [ inf ] with respect to the forces @xmath32 between the test chain and the matrix up to the second order .
this leads to @xmath33}&=-\ln\int\prod_{p=1}^{m}d{\bf r}^{(p)}(s , t)d{\bf { \hat r}}^{(p)}(s , t)\bigg\{\exp\left\{-a_{1}\left[{\bf r}^{p},{\bf { \hat r}}^{p}\right]\right\}\nonumber\\&+&\frac{1}{2!}\int d^{3}rd^{3}r'\int ds\int ds'\int dt\:i{\hat r}_{j}(s , t)\frac{\delta}{\delta r_{j}(s , t)}v\left[{\bf r}(s , t)-{\bf r}\right]\nonumber\\&\:&\qquad\times\int dt'\:i{\hat r}_{l}(s',t')\frac{\delta}{\delta r_{l}(s',t')}v\left[{\bf r}(s',t')-{\bf r}'\right]\left<\rho({\bf r},t)\rho({\bf r}',t')\right>_{1}\nonumber\\&+&\frac{1}{2!}\int d^{3}rd^{3}r'\int ds\int ds'\int dt\int dt'v\left[{\bf r}-{\bf r}(s , t)\right]v\left[{\bf r}'-{\bf r}(s',t')\right]\nonumber\\&\;&\qquad\times\nabla_{l}\nabla_{j}'\left<\pi_{l}({\bf r},t)\pi_{j}({\bf r}',t')\right>_{1}\nonumber\\&-&\frac{1}{2!}\int d^{3}rd^{3}r'\int ds\int ds'\int dt\int_{-\infty}^{t } dt'\:i{\hat r}_{j}(s , t)\frac{\delta}{\delta r_{j}(s , t)}v\left[{\bf r}(s , t)-{\bf r}\right]\nonumber\\ & \:&\qquad\times v\left[{\bf r}'-{\bf r}(s',t')\right]\nabla_{l}'\left<\rho({\bf r},t)\pi_{l}({\bf r}',t')\right>_{1}\nonumber\\&+&(t\leftrightarrow t')+{\cal o}(f^{3})\bigg\}\label{einf}\end{aligned}\ ] ] where the matrix density @xmath34 and the response field density @xmath35 were introduced and @xmath36 denotes cumulant averaging over the full msr - action @xmath37 $ ] of the matrix . in eq .
( [ einf ] ) the term @xmath38 is the same like the previous one but with permutated time arguments .
the terms which are linear with respect to @xmath39 vanishes because of the homogenity of the system . in the appendix
a we show that because of causality the correlator @xmath40 equals zero @xcite .
taking this into account and performing the spatial fourier transformation the expression for gf ( [ gf1 ] ) takes the form @xmath41\nonumber\\&+&\frac{1}{2}\int ds\:ds'\int dt\:dt'\:i{\hat r}_{j}(s , t)\int\frac{d^{3}k}{(2\pi)^{3}}k_{j}k_{l}\left|v(k)\right|^{2}s({\bf k},t - t')\nonumber\\&\:&\qquad\times\exp\left\{i{\bf k}\left[{\bf r}(s , t)-{\bf r}(s',t')\right]\right\}i{\hat r}_{l}(s',t')\nonumber\\&+&\int ds\:ds'\int dt\:dt'\:i{\hat r}_{j}(s , t)\int\frac{d^{3}k}{(2\pi)^{3}}k_{j}k_{l}\left|v(k)\right|^{2}p_{l}({\bf k},t - t')\nonumber\\&\:&\qquad\times\exp\left\{i{\bf k}\left[{\bf r}(s , t)-{\bf r}(s',t')\right]\right\}\bigg\}\label{gf2}\end{aligned}\ ] ] where the correlation function @xmath42 and the response function @xmath43 of the matrix are naturally defined .
going beyond the lrt - approximation would bring us multi - point correlation and response functions .
+ we should stress that in contrast to the matrix with a quenched disorder which was considered in @xcite in our case the matrix has its own intrinsic dynamical evolution which is considered as given . for example , for the glass forming matrix , which is our prime interest here , the correlation and response functions are assumed to be governed by the gtze mode - coupling equations @xcite .
the hartree approximation ( which is actually equivalent to the feynman variational principle ) was recently used for the replica field theory of random manifolds @xcite as well as for the dynamics of manifolds @xcite and sine - gordon model @xcite in a random media . in the hartree approximation the real msr - action is replaced by a gaussian action in such a way that all terms which include more than two fields @xmath44 or / and @xmath45 are written in all possible ways as products of pairs of @xmath44 or / and @xmath45 , coupled to selfconsistent averages of the remaining fields . as a result
the hartree - action is a gaussian functional with coefficients , which could be represented in terms of correlation and response functions . after these straightforward calculations ( details can be found in the appendix b )
the gf ( [ gf2 ] ) takes the form @xmath46\nonumber\\&+&\int_{0}^{n}ds\:ds'\int_{-\infty}^{\infty } dt\int_{-\infty}^{t}dt'\:i{\hat r}_{j}(s , t)r_{j}(s',t')\lambda(s , s';t , t')\nonumber\\&-&\int_{0}^{n}ds\:ds'\int_{-\infty}^{\infty } dt\:i{\hat r}_{j}(s , t)r_{j}(s , t)\int_{-\infty}^{t}dt'\lambda(s , s';t , t')\nonumber\\&+&\frac{1}{2}\int_{0}^{n}ds\:ds'\int_{-\infty}^{\infty } dt\int_{-\infty}^{t}dt'\:i{\hat r}_{j}(s , t)i{\hat r}_{j}(s',t')\chi(s , s';t , t')\bigg\}\label{hartree}\end{aligned}\ ] ] where @xmath47 and @xmath48 in eq .
( [ lambda],[chi ] ) the response function @xmath49 the density correlator @xmath50 with @xmath51 and the longitudinal part of the matrix response function @xmath52 are defined .
the pointed brackets denote the selfconsistent averaging with the hartree - type gf ( [ hartree ] ) .
+ up to now we considered the general off - equilibrium dynamics with the only restriction of causality @xcite .
we now assume that for very large time moments @xmath11 and @xmath53 , where the difference @xmath54 is finite so that @xmath55 , time homogenity and the fluctuation - dissipation theorem ( fdt ) holds .
this implies @xmath56 where @xmath57 . by using this in eq .
( [ hartree ] ) and after integration by parts in the integrals over @xmath53 the gf in hartree approximation takes the form @xmath58\frac{\partial}{\partial t'}r_{j}(s',t')\nonumber\\ & -&\int_{0}^{n}ds\:ds'\int_{-\infty}^{\infty}dt\:i{\hat r}_{j}(s , t)\bigg[\varepsilon\delta(s - s')\frac{\partial^{2}}{\partial s^{2}}+\beta\int\frac{d^{3}k}{(2\pi)^{3}}k^{2}|v(k)|^{2}s_{st}({\bf k})\nonumber\\ & \:&\:\times\left[f_{st}({\bf k};s , s')-\delta(s - s')\int_{0}^{n}ds^{''}f_{st}({\bf k};s , s^{''})\right]\bigg]r_{j}(s',t)\nonumber\\&+&t\int_{0}^{n}ds\:ds'\int_{-\infty}^{\infty } dtdt'\bigg[\xi_{0}\delta(t - t')\delta(s - s')+\theta(t - t')\beta\nonumber\\&\:&\:\int\frac{d^{3}k}{(2\pi)^{3}}k^{2}|v(k)|^{2}f({\bf k};s , s';t - t')s({\bf k};t - t')\bigg]i{\hat r}_{j}(s , t)i{\hat r}_{j}(s',t')\label{gfh}\end{aligned}\ ] ] where the subscript @xmath59 indicates the static correlation functions .
this generating functional immediately leads to the following generalized rouse equation ( gre ) @xmath60 where the memory function @xmath61 and the effective elastic susceptibility @xmath62\label{omega}\end{aligned}\ ] ] are defined .
the correlation function of the random force @xmath63 is given by @xmath64\label{force}\ ] ] as a result we have obtained basically the same gre as in the papers @xcite but with one additional elastic term .
this term ( see the 2-nd term in eq .
( [ omega ] ) ) is mainly inversely proportional to the temperature and is , in contrast to the first term , of an energetic nature .
the two factors of @xmath65 quantify the forces exerted by a pair of surrounding segments on the test chain segments @xmath66 and @xmath67 , whereas the @xmath68 and @xmath69 factors quantify the static correlations between the segments of surrounding and test chain segments , respectively . in @xcite only the entropic elastic part was taken into account . the memory function ( [ gamma ] ) has the same form as in @xcite and the relationship ( [ force ] ) is assured as soon as the fdt ( [ fdt1 ] ) and ( [ fdt ] ) is fullfilled . in this subsection
we give gre s for the more general case when the time homogenity ( stationarity ) and the fdt do not hold @xcite .
+ by employing the standard way @xcite one can derive two coupled equations of motion for correlators @xmath70 and response functions @xmath71 @xmath72g(s , s';t , t')\nonumber\\&+&\int_{0}^{n}ds^{''}\int_{-t'}^{t}dt^{''}\lambda(s , s^{''};t , t^{''})g(s^{''},s';t^{''},t')=\delta(s - s')\delta(t - t')\label{cor}\end{aligned}\ ] ] @xmath72c(s , s';t , t')\nonumber\\&+&\int_{0}^{n}ds^{''}\int_{-\infty}^{t}dt^{''}\lambda(s , s^{''};t , t^{''})c(s^{''},s';t^{''},t')\nonumber\\ & + & \int_{0}^{n}ds^{''}\int_{-\infty}^{t}dt^{''}\chi(s , s^{''};t , t^{''})g(s',s^{''};t',t^{''})=2t\xi_{0}g(s',s;t',t)\label{response}\end{aligned}\ ] ] with the initial conditions @xmath73 and @xmath74 in the stationary case all correlators and response functions in eq .
( [ cor]-[ini2 ] ) only depend from the differences of time moments , @xmath54 . if we assume again that fdt ( [ fdt1 ] ) and ( [ fdt ] ) holds , then from eq .
( [ response ] ) after performing the integrations by parts ( in the integrals over @xmath75 ) one arrive at the gre for @xmath76 @xmath77 of course , eq .
( [ gcor ] ) could be obtained immediately from eq .
( [ gre ] ) by multiplying both sides with @xmath78 , averaging and taking into account that because of causality @xmath79 at @xmath76 .
we will use the gre eq.([gcor ] ) , where the functions @xmath80 and @xmath81 are given by eqs .
( [ gamma ] , [ omega ] ) , in the next section for the investigation of the test chain ergodicity breaking ( freezing ) .
the new features of the gre ( [ gcor ] ) relative to the standard rouse equation are that it contains the integral convolution with respect to the @xmath82variable in the frictional term as well as in the elastic term .
the frictional term is also non - local in time .
all these things together should change the statical and dynamical behaviour of the gaussian test chain in comparison to the ideal chain .
+ we also should stress that the gre is substantially nonlinear because the memory function ( [ gamma ] ) depends from the test chain correlator @xmath83 in such a way that a positive feedback obviously exists .
that is the reason why one could expect that eq .
( [ gcor ] ) shows an ergodicity breaking in the spirit of gtze s glass transition theory @xcite . + as usual it is convenient to introduce the standard rouse mode variables @xcite : @xmath84 with the inverse transformation @xmath85 in general one also needs a 2-dimensional rouse transformation @xmath86 where functions like @xmath87 should be treated like @xmath88-matrices .
for example the density correlator ( [ dc ] ) should be considered as an exponential function from a @xmath89-matrix @xmath90 and the series expansion holds : @xmath91 we also assume that matrices in the rouse mode representation are nearby diagonal @xmath92 for any @xmath2 and @xmath93 not equal zero @xcite .
+ then as a result of rouse mode transformation the gre for the rouse mode time correlation function , @xmath94 , takes the form ( for @xmath95 ) @xmath96 where @xmath97 - 1\right\}s({\bf k},t)\label{gammap}\ ] ] and @xmath98\label{omegap}\ ] ] for @xmath99 the gre describes the dynamics of the centre of mass @xmath100 and has the following form @xmath101 with @xmath102 and @xmath103 as a result all rouse mode variables relax independently .
the conclusion that the rouse modes are still `` good eigenmodes '' even in the melt is supported by monte - carlo @xcite and molecular - dynamic @xcite simulations .
+ for cases where the assumption of diagonality [ 38,39,[diag ] ] can not be justified , the rouse modes do not decouple and one have to go back to eq .
( [ gcor ] ) . in the rouse mode representation
it reads as @xmath104 as we have already discussed in sec.ii.b the interaction with the surrounding segments renormalizes the elastic properties of the rouse chain so that the test chains elastic susceptibility is given by eq .
( [ omegap ] ) .
the additional elastic term in gre leads to the renormalized static normal modes correlator @xmath105 explicit evaluation of the @xmath106 can be done if we use for the static correlator @xmath107 the standard rouse expression @xmath108 then the calculation yields for the two limiting cases @xmath109+{\cal o}(p^{4 } ) & \frac{p\pi}{n}\ll 1\quad ( 50a ) \\
\left(\frac{p\pi}{n}\right)^{2}\varepsilon+\frac{\beta}{4\pi^{2}}\int_{0}^{l^{-1}}dk\:k^{4}|v(k)|^{2}s_{st}(k)\left(\frac{6}{l^{2}k^{2}}\right)\left(1-e^{-k^{2}r_{g}^{2}}\right ) & \frac{p\pi}{n}\simeq 1\quad\ : ( 50b)\end{array}\right.\nonumber\end{aligned}\ ] ] where @xmath110,\qquad r_{g}=\frac{nl^{2}}{6}\ ] ] and we have chosen @xmath111 as a cutting parameter .
it is evident from the previous eqs .
( 50@xmath112 , 50@xmath113 ) that at small @xmath2 * the elastic modulus gains an energetic component which , in contrast to the entropic part @xmath7 , increase with the cooling of the system , * initially absolutely flexible chains acquires a stiffness because of terms of order @xmath114 and higher . at large @xmath2
the elastic behaviour reduces to the standard rouse one , as it is expected . in fig.1 is shown the result of a numerical calculation of the static correlator ( [ cst ] ) .
the fourier component of the potential is taken , as it is customary e.g. in the theory of neutron scattering @xcite , in the form of a pseudo potential approximation , @xmath115 , where @xmath116 and @xmath117 have dimensions of molecular energy and distance , respectively .
the static structure factor @xmath68 is chosen in the form of the percus - yevick s simple liquid model @xcite .
one can see that for @xmath118 the small rouse mode index limit ( 50@xmath112 ) starts at @xmath119 whereas the opposit limit ( 50@xmath113 ) is fullfilled at @xmath120 . because the correlator @xmath121 depends mainly from @xmath122 , for relatively short test chains the high mode index limit ( 50@xmath113 )
is shifted into the window of calculations ( see fig.1 for n=20 ) .
+ at least qualitatively this deviation from the standard rouse behaviour have been seen by kremer and grest in their md - simulations ( see fig.3 in @xcite ) .
first we consider the case @xmath95 . in the nonergodic state
the rouse mode correlation functions can be represented as @xmath123 where the non - ergodicity parameter @xmath124 was introduced and @xmath125 . for the correlation function of the glassy matrix we can use the standard result of the glass transition theory @xcite @xmath126 where
the proximity parameter @xmath127 is defined and @xmath0 is the temperature of the matrix ergodicity breaking ( gtze temperature ) . in eq .
( [ gg ] ) @xmath128 is the non - ergodicity parameter of the matrix , @xmath129 is the characteristic time scale , @xmath112 is the characteristic exponent , @xmath130 and @xmath131 is some amplitude . + in order to derive the equation for @xmath132 let us take the limit @xmath133 in eq .
( [ grer ] ) keeping in mind the definitions ( [ psi ] ) and ( [ gg ] ) . very close to the test chain ergodicity breaking temperature @xmath1
, @xmath132 goes to zero ( a - type transition @xcite ) and we can expand the exponential function in eq .
( [ gammap ] ) up to the first order with respect to @xmath132 .
the solution of the resulting equation has the simple form @xmath134 the critical temperature @xmath1 is determined by the equation @xmath135 the numerical solution of eq .
( [ tc ] ) is given in fig.2 .
it is obviously that if the entropic part of @xmath106 dominates , the critical temperature is given by @xmath136 fig.2 really shows that this law ( [ pq ] ) is well satisfied due to the fact that the critical temperatures @xmath1 are quite high .
but for low temperatures the energetic contribution in @xmath106 is enhanced which leads to a deviation from this simple @xmath137-dependence .
+ now we consider the case for @xmath99 .
the equation ( [ c.m ] ) for the velocity of the center of mass @xmath138 leads to the equation for the velocity correlator @xmath139 where @xmath140 because of causality the correlator on the r.h.s .
( [ c.m.c ] ) has the form @xmath141 where , as it comes from eq .
( [ c.m ] ) @xmath142 taking into account the definition of @xmath143 and eq .
( [ force ] ) this yields to the correlator @xmath144 because of the causality property ( 61 ) only the @xmath145-functional term on the r.h.s .
( [ force ] ) contributes to the correlator ( [ lk ] ) .
therefore the resulting equation for the self - diffusion coefficient @xmath146 takes the form @xmath147}\label{d}\ ] ] which was obtained before in @xcite .
+ one can calculate the second term in the denominator of eq .
( [ d ] ) selfconsistently . because now the relevant times @xmath148 the approximation @xmath149 could be used in eq .
( [ c.m ] ) . then the density correlator ( [ 47 ] )
is given by @xmath150 with the use of eqs .
( [ de]),([gg ] ) and eq .
( [ gammat ] ) in the limit @xmath151 eq . ( [ d ] ) becomes @xmath152}\label{del}\ ] ] where the denominator is given by static properties only .
similar statements have been suggested already in @xcite the solution of eq .
( [ del ] ) has the simple form @xmath153 where @xmath154 finally , the temperature of the ergodicity breaking ( localization ) for the mode @xmath99 of the test chain is @xmath155 fig.3 shows the results of numerical calculations of @xmath156 and @xmath157 as functions of @xmath3 .
one can see that in the reasonable range of parameters @xmath158 . as a result
one can say that on cooling of a test chain in a glassy matrix the mode @xmath99 is the first to be freezed . on the subsequent cooling the modes @xmath159
are freezed successively , @xmath160 it is apparent that the system studied here is a nontrivial polymeric generalization of the model introduced by sjgren @xcite .
this model was used for the investigation of the @xmath161-peak in the spectrum of glass forming systems @xcite .
in this paper we have derived a gre for a test polymer chain in a polymer ( or non - polymer ) matrix which has its own intrinsic dynamics , e.g. the glassy dynamics @xcite .
we have used here the msr - functional integral technique which could be considered as an alternative to the projection operator formalism @xcite .
one of the difficulties in this formalism is the necessity of dealing with the projected dynamic , which is difficult to handle with explicitly . on the contrary in msr
- technique the dynamic of slow variables is well defined and several approximations which one have to employ could be justified . in the interaction of the test chain with the surrounding matrix only two - point correlation and response
functions are involved . in terms of mca
@xcite this obviously corresponds to the projection of the generalized forces only onto the bilinear variables : product of test chain density and matrix density . to handle with the action in the gf of the test chain we used the hartree - type approximation ( i.e. , equivalent to the feynman variational principle ) @xcite , which is reasonable when the fluctuations of the test chain are gaussian . in the case of a polymer melt ( high densitiy )
this is indeed the case due to the screening effects for the excluded volume @xcite .
the use of the hartree - type makes the problem that we deal with analytically amenable and results in the gre s for the case when the fdt holds as well as for the case when fdt does not hold . in this paper
we have restricted ourselves to the first case and have shown that the interaction with the matrix renormalizes not only the friction coefficient ( which makes the chain non - markovian ) but also the elastic modulus ( which changes the static correlator ) .
the form of the static correlator for the rouse mode variables is qualitatively supported by md - simulations @xcite .
as regards the dynamical behaviour , we have shown that the test chain in a glassy matrix ( with the matrix glass transition temperature @xmath0 ) undergoes the ergodicity breaking transition at a temperature @xmath162 .
the critical temperature @xmath1 could be parametrized with the rouse mode index @xmath2 and is a decreasing function of @xmath2 .
we have considered only the a - type transition which is assured by the bilinear term in the expansion of eq .
( [ gammap ] ) .
it seems reasonable that keeping the whole exponential function in eq .
( [ gammap ] ) might lead to a b - type transition also .
the results also essentially would change if the off - diagonal elements in the matrix ( [ 2rt ] ) can not be neglected ( see eq .
( [ grerq ] ) ) . in this case
only one ideal transition temperature @xmath163 would be possible .
the general theory of a a - type transition was discussed in @xcite .
this picture of freezing here should not be mixed with a different one , the underlying glass transition by itself ( e.g. the glass transition of the matrix at @xmath164 ) . according to the present view of this phenomenon @xcite , the spontaneous arrest of the density fluctuations
is driven by those of the microscopic lengthscale @xmath165 , where @xmath165 is the wave vector which corresponds to the structure factor s main maximum .
the freezing of these fluctuations then arrests the others through the mode coupling .
two of us gratefully acknowledge support from the deutsche forschungsgemeinschaft through the sonderforschungsbereich 262 ( v.g.rostiashvili ) and the bundesministerium fr bildung und forschung ( m.rehkopf ) for financial support .
we also greatly acknowledge helpful discussions with j.baschnagel , k.binder , k.kremer , w.gtze and r.schilling .
it is more convenient to handle with the spacial fourier transformation of this correlator @xmath166\right\}\right>_{1}\nonumber\\ & = & \sum_{a , b=0}^{\infty}\frac{1}{a!b!}\sum_{p , m=1}^{m}\int ds ds'\left < i{\hat r}_{l}^{(p)}(s , t)i{\hat r}_{j}^{(m)}(s',0)[i{\bf k}{\bf r}^{(p)}(s , t)]^{a}[-i{\bf k}{\bf r}^{(m)}(s',0)]^{b}\right>_{1}\nonumber\end{aligned}\ ] ] such multi - point cumulant response functions ( mrf ) were considered in @xcite .
the causality condition for these functions asserts that the time argument of at least one @xmath167-variable should be the latest one , otherwise this mrf equals zero . because of the same reason self - loops of response functions vanish @xcite .
mrf s which consists only of @xmath168-variables also vanish .
+ in the case ( [ ap ] ) all time arguments of the @xmath167-variables are equal to the corresponding time arguments of @xmath168-variables and as a result the mrf in eq .
( [ ap ] ) vanishes .
in order to calculate the bilinear hartree action , we follow the way mentioned in sec.ii.b . with these strategy in mind the 2-nd term in the exponent ( [ gf2 ] )
is evaluated as @xmath169 where @xmath170\right\}i{\hat r}_{l}(s',t')\right>\nonumber\\ i_{2}(s , s';t , t')&\equiv&\left<\frac{\delta}{\delta r_{j}(s , t)}\int \frac{d^{3}k}{(2\pi)^{3}}k_{j}k_{l}|v(k)|^{2}\exp\left\{i{\bf k}\left[{\bf r}(s , t)-{\bf r}(s',t')\right]\right\}i{\hat r}_{l}(s , t)\right>\nonumber\\ i_{3}(s , s';t , t')&\equiv&\left<\frac{\delta}{\delta r_{j}(s , t)}\int \frac{d^{3}k}{(2\pi)^{3}}k_{j}k_{l}|v(k)|^{2}\exp\left\{i{\bf k}\left[{\bf r}(s , t)-{\bf r}(s',t')\right]\right\}i{\hat r}_{l}(s',t')\right>\nonumber\\ i_{4}(s , s';t , t')&\equiv&\left<\frac{\delta}{\delta r_{j}(s',t')}\int \frac{d^{3}k}{(2\pi)^{3}}k_{j}k_{l}|v(k)|^{2}\exp\left\{i{\bf k}\left[{\bf r}(s , t)-{\bf r}(s',t')\right]\right\}i{\hat r}_{l}(s , t)\right>\nonumber\\ i_{5}(s , s';t , t')&\equiv&\left<\int\frac{d^{3}k}{(2\pi)^{3}}k_{j}k_{j}|v(k)|^{2}\exp\left\{i{\bf k}\left[{\bf r}(s , t)-{\bf r}(s',t')\right]\right\}\right>\nonumber\\ i_{6}(s , s';t , t')&\equiv&\bigg < i{\hat r}_{j}(s , t)i{\hat r}_{l}(s , t)\frac{\delta^{2}}{\delta r_{n}(s , t)\delta r_{n}(s',t')}\nonumber\\ & \:&\times\int \frac{d^{3}k}{(2\pi)^{3}}k_{j}k_{l}|v(k)|^{2}\exp\left\{i{\bf k}\left[{\bf r}(s , t)-{\bf r}(s',t')\right]\right\}\bigg>.\label{il}\end{aligned}\ ] ] the pointed brackets in eq .
( [ il ] ) represent the selfconsistent averaging with the gaussian hartree action .
taking this into account and using the generalized wick theorem @xcite , after straightforward algebra , we have @xmath171 where the last equation comes from the fact that the response function @xmath172 and @xmath173 . + the 3-rd term in the exponent eq .
( [ gf2 ] ) can be handled in the same way .
the response function for the isotropic matrix has the form @xmath174 where @xmath175 is the longitudinal part of the matrix response function .
then the hartree approximation of the 3-rd term in the exponent ( [ gf2 ] ) takes the form @xmath176 where @xmath177 taking into account eq .
( [ ap2 ] ) with eq .
( [ il ] ) and eq . (
[ j ] ) leads to the hartree - type approximation ( [ hartree ] ) .
99 m.doi and s.f.edwards , _ the theory of polymer dynamics _ , clarendon press , oxford , 1986 j.d.ferry , _ viscoelastic properties of polymers _ , wiley , n.y .
, 1980 w.gtze in , _ liquids , freezing and glass transition _ , ed . by j.p.hansen ,
d.levesque and j.zinn-justin , amsterdam , north - holland , 1991 k.okun , m.wolfgardt , j.baschnagel and k.binder , preprint w.hess , macromolecules , 21 , 2620 ( 1988 ) k.s.schweizer , j.chem.phys .
, 91 , 5802 ( 1989 ) k.s.schweizer , j.chem.phys . , 91 , 5822 ( 1989 ) h.kinzelbach and h.horner , j.phys.(france ) , 1,3 , 1329 ( 1993 ) d.cule and y.shapir , phys.rev .
e 53 , 1553 ( 1996 ) c.de dominicis and l.peliti , phys.rev .
b 18 , 353 ( 1978 ) r.bausch , h.k.janssen and h.wagner , z.phys .
b 24,113 ( 1976 ) v.g.rostiashvili and r.schilling , z.phys .
b ( 1996 ) m.mezard and g.parisi , j.phys .
a 29 , 6515 ( 1996 ) 3-rd edition , clarendon press , oxford , 1996 , sec . 4.2 .
l.f.cugliandolo and j. kurchan , phys.rev.lett .
, 71 , 173 ( 1993 ) k.kremer and g.s.grest , j.chem.phys .
92 , 5057 ( 1990 ) s.w.lovesey , _ theory of neutron scattering _ , clarendon press , oxford , 1984 j .- p.hansen and i.r.mcdonald , _ theory of simple liquids _ , academic press , london , 1976 s.f
. edwards , t.a .
vilgis , physica scripta , * t13 * , 7 , ( 1986 ) t.a .
vilgis in `` disorder effects on relaxational processes '' ed .
a. blumen , r. richert , springer verlag , heidelberg , 1994 l. sjgren , phys.rev .
a 33 , 1254 ( 1986 ) w. gtze and l. sjgren , j.phys .
condensed matter 1 , 4183 ( 1989 ) j.zinn-justin , _ quantum field theory and critical phenomena _ , 3-rd edition , clarendon press , oxford , 1996 , sec .
t.franosch and w.gtze , j.phys .
( cond.mat . ) 6 , 4807 ( 1994 ) | we consider the langevin dynamics of a gaussian test polymer chain coupled with a surrounding matrix which can undergo the glass transition . the martin - siggia - rose generating functional method and the nonpertubative hartree approximation
are used to derive the generalized rouse equation for the test chain .
it is shown that the interaction of the test chain with the surrounding matrix renormalizes the bare friction and the spring constants of the test chain in such a way that the memory function as well as the bending dependent elastic modulus appear .
we find that below the glass transition temperature @xmath0 of the matrix the rouse modes of the test chain can be frozen and moreover the freezing temperatures ( or the ergodicity - nonergodicity transition temperature ) @xmath1 depends from the rouse mode index @xmath2 .
-15 mm classification : physics abstracts , 05.20 - 36.20 |
the intrinsic geometrical properties of the bloch band is of great importance in solid state physics . to zeroth order in electromagnetic fields ,
the effective mass tensor reflects the curvature of band dispersions , which describes the low energy behaviour near band extrema and enters into the carrier density of states and various transport properties.@xcite to first order , an accurate description of bloch electron dynamics not only requires the knowledge of the band dispersion , but also the berry curvature and the orbital magnetic moment as functions of the crystal momentum.@xcite the berry curvature reflects the intrinsic geometry of the bloch state fiber bundle , and its importance has been exemplified in the study of anomalous hall effects of charge and heat and in the investigation of orbital magnetization.@xcite how does the intrinsic geometry of bloch bands affect the second order response to electromagnetic fields ?
are there additional geometrical quantities emerge in the orbital magnetic susceptibility ? in this work , we present an exhaustive analysis of the electron wave - packet energy and orbital magnetic susceptibility by making a gauge invariant classification .
we are able to identify a geometrical contributions from the fermi sea , which we call the geometrical susceptibility , in the sense that it involves geometrical quantities including the berry curvature and the quantum metric .
the geometrical susceptibility is a novel mechanism for orbital magnetic susceptibility , which provides the dominant diamagnetic response around the band gaps , and is especially important in strongly spin / pseudospin - orbit coupled systems such as topological insulators and 2d semimetals .
we propose that by introducing particle - hole symmetry breaking , it is generally possible to enhance the orbital paramagnetism .
moreover , we derive a novel fermi surface conribution , arising from the enegy polarization in the brillouin zone , and competing with pauli and peierls - landau magnetism . to our delight , these fermi surface contributions , together with a langevin - like magnetic susceptibility and the geometrical susceptibility , can be calculated based on bloch states inside a single bloch band , and the only interband contribution is the van vleck paramagnetic susceptibility .
the various terms can dominate over different energy range of the bloch band .
the above understanding of the orbital magnetic susceptibility is under the assumption of the minimal coupling , in which case the magnetic field modifies the hamiltonian only through the magnetic vector potential .
we will also show that our theory can be easily extended beyond the minimal coupling assumption , and be used to describe the effect such as the zeeman energy from spins .
there are many other systematic studies of the orbital magnetic susceptibility , mostly based on the direct perturbation technique@xcite or greens function formalism.@xcite however , our result has two advantages : ( 1 ) it consists of only gauge - invariant terms , which are independent of the phase choice of the bloch states ; ( 2 ) it allows simple physical interpretations of each term and hence the precise classification of geometrical and interband effects .
our paper is organized as follows . in section
ii , we demonstrate that the first order correction to the wave - packet has clear physical interpretations and can be understood by two new concepts , i.e. the vertical and horizontal mixings to the unperturbed bloch states . in section iii , with first order correction to the wave - packet , we derive the wave - packet energy up to second order in magnetic field ; in section iv , we derive the general formula of the orbital susceptibility . by comparing with the orbital susceptibility for atomic systems and free particles
we identify two new mechanisms : the geometrical susceptiblity and the energy polarization susceptibility ; in section v and vi we provide two examples to demonstrate the importance of the geometrical effect in orbital magnetic susceptibility and how various contributions compete with each other ; in section vii , we discuss how to extend our theory beyond the minimal coupling assumption .
in the first order semiclassical theory , the wave - packet is the superposition of unperturbed local bloch states , i.e. @xmath0 where @xmath1 is the center of mass position of the wave - packet .
the wave - packet is also assumed to be peaked around some momentum @xmath2 in the brillouin zone .
then the effective lagrangian can be calculated for this wave - packet and expressed only in terms of its center of mass position @xmath1 , the gauge - invariant crystal momentum @xmath3 , and their time derivatives . from the euler - lagrange equations of motion
, one finds that the dynamics of @xmath1 and @xmath4 contains two geometrical corrections : the orbital magnetic moment @xmath5 that contributes a zeeman energy , and the berry curvature @xmath6 that modifies the dynamical structure.@xcite however , in the second order theory , the local bloch states and hence the wave - packet should be corrected up to first order .
these corrections have two origins : ( 1 ) the non - adiabatic correction due to the fact that the local hamiltonian itself is implicitly time - dependent through the semiclassical evolution of the center of mass position @xmath1 ; ( 2 ) the first order gradient correction to the local hamiltonian .
they will modify the local bloch state in two ways ( see fig.([geometry ] ) ) : ( 1 ) they can mix the bloch states from other bands but at the same point in the brillouin zone , which we call the vertical mixing ; ( 2 ) they can also mix the bloch states inside the same band but from different points in the brillouin zone . to derive the first order corrections to the bloch states , note
that the wave - packet is assumed to be localized in real space near the center of mass position @xmath1 , we expand the exact hamiltonian in weak magnetic field with respect to the deviation @xmath7 from @xmath1 ( assume minimal coupling and set @xmath8 for simplicity):@xcite @xmath9 . here @xmath10 is the local hamiltonian by taking the magnetic vector potential at @xmath1 , where @xmath11 is the hamiltonian without external fields , and @xmath12 , @xmath13 are the momentum and position operators .
@xmath14 is the first order gradient correction to @xmath15 : @xmath16 $ ] , where @xmath17 $ ] is the velocity operator . @xmath18_i
[ \bm b\times ( \hat{\bm q}-\bm r_c)]_j$ ] is the second order perturbation to @xmath15 , where @xmath19 is the hessian matrix . for nonrelativistic pauli and schr@xmath20dinger hamiltonians , @xmath21 , where @xmath22 is the bare electron mass . for the relativistic dirac hamiltonian , @xmath23 . for calculating the orbital magnetic susceptibility , corrections to @xmath15 beyond second order will not contribute , and hence they are not included in the following discussion . by using the above gradient expansion of the hamiltonian in the time - dependent schr@xmath20dinger equation for the wave - packet
, we derive the modification to the periodic part of the bloch states @xmath24:@xcite @xmath25 is a normalization factor to ensure @xmath26 to first order , @xmath27 , @xmath28 , @xmath29 is the interband berry connection , @xmath30 is the velocity matrix element , and @xmath31 .
the subscripts @xmath32 , @xmath22 , and @xmath33 are band indices .
@xmath34 represents the band energy .
the partial derivative @xmath35 is with respect to the crystal momentum . from this modified bloch states in eq.([bloch ] )
, the influence of the magnetic field is quite heuristic and preludes the form of the wave - packet energy in eq.([energy ] ) .
first consider the second term in eq.([bloch ] ) .
it contains local bloch states at the same momentum from other bands ( @xmath36 ) and preserves the lattice translational symmetry . the essential quantity @xmath37 is gauge - invariant and its form is quite suggestive , which are couplings between the magnetic field and interband matrix elements of magnetic dipole moment @xmath38 .
note that the correct expression of @xmath38 includes the time dependence of @xmath1 in the bloch states , which is in fact the nonadiabatic effect mentioned previously.@xcite the remaining part in @xmath39 is from the interband part ( @xmath40 ) of the postion operator @xmath13 in @xmath14 in the bloch representation .
we call this correction with @xmath37 the vertical mixing since it mixes bloch states from different bands at the same @xmath41-point in the brillouin zone ( as illustrated in fig.([geometry ] ) ) . on the contrary , the first term in eq.([bloch ] ) only contains the bloch state inside the same band ( band 0 ) .
compared with the original bloch state @xmath24 , its crystal momentum is shifted to @xmath42 , hence breaking the lattice translational symmetry .
an interesting fact is that this term can be obtained if we initially take the position operator in the vector potential in the exact hamiltonian @xmath43 as a c - number and later recover it in @xmath44 as an operator .
this shift of momentum is due to the intraband part of the postion operator @xmath13 in @xmath14 in the bloch representation .
we emphasize that such type of correction without lattice trasnlational symmetry must appear , since the exact hamiltonian @xmath43 has an argument @xmath42 and does not generally respect such symmetry .
heuristically , the correct single band wave - packet should be constructed from the eigenstates of the exact hamiltonian which does not respect the lattice translational symmetry in general . however , since the susceptibility is a second order thermodynamic response , we can approximate @xmath43 by @xmath15 , with gradient corrections up to second order . unlike @xmath43 , @xmath15 is periodic with a well - defined brillouin zone .
the influcence of the actual lattice translational symetry breaking thus manifests as a first order correction to the local bloch states in the form of the momentum shift .
remarkably , this correction to @xmath44 can be rewritten simply in terms of the shift of momentum @xmath45 , and reads @xmath46 , where @xmath47 is the gauge - covariant derivative acting on the bloch states and @xmath48 is the intraband berry connection .
we will see that the momentum shift @xmath49 enters the description of electrons orbital motion through the band geometrical properties , yielding crucial geometrical corrections as discussed later .
since the correction from @xmath49 mixes bloch states at neighbouring @xmath41-points in the same band , we call it the horizontal mixing . to sum up , the magnetic field affects the local bloch states @xmath44 in two ways : ( i ) it vertically mixes @xmath44 with the bloch states @xmath50 from other bands ; ( ii ) it also horizontally shifts the bloch states along the path @xmath49 according to the affine connection ( berry connection ) @xmath51 in the brillouin zone .
we illustrate the two types of mixing of bloch states schematically in fig.[geometry ] .
with the modified bloch states derived in the previous section , we can construct the corrected wave - packet @xmath52 up to first order and derive the corresponding effective lagrangian up to second order.@xcite from this lagrangian , the wave - packet energy is obtained , which is the crucial quantity for evaluating the orbital magnetic susceptibility . by expressing the wave - packet energy in a gauge - invariant form , the physical meaning of various contributions is interpreted .
the concept of vertical and horizontal mixings is important to understand the wave - packet energy . from eq.([bloch ] ) , the corrected wave - packet up to first order reads : @xmath53 is a coefficient peaked at @xmath2 : @xmath54 , @xmath55 is connected to @xmath56 based on eq.([bloch ] ) : @xmath57\cdot \bm a_{n0}-{g_{n0}\over \varepsilon_0-\varepsilon_n } c_0\,.\ ] ] in eq.([cn ] ) , the first term is derived from the horizontal mixing by integration by parts , and the second term is from the vertical mixing in eq.([bloch ] ) .
the effective lagrangian for the wave - packet reads @xmath58 .
it can be transformed to a langrangian for the single band wave - packet : @xmath59 .
we then calculate the wave - packet energy up to second order as follows : @xmath60 , where @xmath61 is the contribution from the dynamical part of the lagrangian .
the wave - packet energy corrected up to second order reads ( see appendix a for details ) : @xmath62 here @xmath63 is the berry curvature , @xmath64 is the orbital magnetic moment , @xmath65 is the quantum metric of @xmath41-space,@xcite @xmath66 is the inverse of effective mass tensor , @xmath67+{1\over 4}\partial_i[(\bm b\times \bm a_{0n})_i\bm a_{n0}]+c.c.$ ] is the field - induced positional shift of the wave - packet center ( its second term is a geometrical quantity from the horizontal mixing and is proportional to the christoffel symbol in @xmath41-space@xcite ) , and @xmath68 $ ] is a single band quantity representing the energy polarization density in @xmath41-space .
indices @xmath69 , @xmath70 , @xmath41 , @xmath71 , @xmath72 and @xmath73 refer to cartesian coordinates and repeated indices are summed over .
@xmath74 is the totally antisymmetric tensor in three dimension . for the last term in eq.([energy ] ) , we choose @xmath75 ( for pauli and schr@xmath20dinger hamiltonians ) for simplicity and a more general formula is given in appendix a. all physical quantities in eq.([energy ] ) should be understood as functions of the gauge - invariant crystal momentum @xmath4 , and the partial derivatives are with respect to @xmath4 . in eq.([energy ] ) , various terms are gauge - invariant and can be characterized by their geometrical and physical meaning .
the two terms in the first line of eq.([energy ] ) are the band energy plus the magnetic dipolar energy , which is the result obtained in the first order semiclassical theory .
the two terms in the second line are the geometrical energies , in the sense that they consist of single band geometrical quantities , i.e. the berry curvature and the quantum metric . on the brillouin zone ,
the hilbert space with single band bloch states @xmath44 forms a fiber bundle , whose curvature is characterized by the berry curvature @xmath6,@xcite and the distance in which is captured by the quantum metric.@xcite for the remaining two quantities , @xmath76 depends only on the band dispersion and the magnetic moment @xmath5 is a single band quantity .
it is interesting to note that @xmath77 and @xmath5 actually form a conjugate pair : they are proportional to the real and the imaginary part of @xmath78 , respectively .
so are the quantum metric and the berry curvature , with respect to the quantity @xmath79 .
thus the less obvious meaning of @xmath80 and @xmath5 can be understood from their well studied conjugate partners @xmath6 and @xmath77 . the first term in the third line of eq.([energy ] )
is a real space polarization energy .
the magnetic field shifts the wave - packet center by @xmath81,@xcite hence modifying the magnetic dipole moment and the wave - packet energy .
the next term is a @xmath41-space polarization energy .
this can be understood by noticing that the the momentum shift @xmath49 gives rise to a second order energy polarization in @xmath41-space , @xmath82 .
similar to the relation between electric polarization and charge , the divergence of such energy polarization yields a local energy correction .
we find that this term is a single band quantity , and is related to the quadrupole moments of the velocity operator ( see appendix b for details ) . in the fourth line of eq.([energy ] ) , the first term is a standard second order perturbation energy through virtual interband transitions .
the last term in eq.([energy ] ) is from the perturbation of @xmath83 .
note that this term vanishes for the dirac hamiltonian , and for the nonrelativistic pauli and schr@xmath20dinger hamiltonians , it comes with the bare electron mass @xmath22 .
the above geometrical and physical meanings of the second order wave packet energy are also suggested by the vertical and horizontal mixings in the wave packet .
such two types of corrections in @xmath52 originates from the bloch represention of @xmath14 .
therefore , they enter the wave packet energy in eq.([energy ] ) through both @xmath14 and @xmath52 . if two horizontal mixings are combined to yield a second order energy , only the neighbourhood in the brillouin zone is involved , and we should obtain a purely geometrical contribution as in the second line in eq.([energy ] ) . on the contrary , if two vertical mixings are combined , then virtual interband transition is involved , and we obtain an interband effect as the first term in the fourth line of eq.([energy ] ) .
if the horizontal and vertical mixing are combined together , we obtain the @xmath41-space polarization energy , which is a single band but not necessarily geometryical quantity .
the various second order energy corrections in eq.([energy ] ) are indispensable for the evaluation of the orbital magnetic susceptibility .
the general approach is to evaluate the thermodynamic grand potential @xmath84 $ ] , where @xmath85)$ ] , @xmath86 is the temperature and @xmath87 is the chemical potential .
then the magnetic susceptiblity is simply the second order derivatives of this grand potential with respect to the magnetic field . by comparing with the susceptibility in atomic systems and free particles , the physical meaning of various terms in the orbital susceptibility
is interpreted .
various mechamisms are more clear when we rewrite the free energy in the wannier function representation and take the atomic insulator limit .
we find that in solids , there is a novel and purely geometrical contribution which depends solely on geometrical properties of bloch bands in the brillouin zone , and we call it the geometrical susceptibility . under external magnetic field , the semiclassical limit of the grand potential written in terms of physical variables is given by:@xcite @xmath88{d^3 k_c\over 8\pi^3}.\ ] ] here @xmath89 is the system volume and @xmath90 is the modified density of states.@xcite the first term in the bracket of eq.([fe ] ) is from the semiclassical grand potential density with second order energy correction , which yields the semiclassical free energy .
the second term @xmath91 is the peierls - landau magnetic energy : @xmath92 where @xmath93 is the energy derivative of the fermi distribution function @xmath94 .
for isotropic bands , the effective mass tensor @xmath95 takes a diagonal form , and @xmath91 reduces to its familiar form.@xcite this term originates from the discreteness of the landau levels , and appears when we transform the free energy from the quantum version to its semiclassical limit.@xcite we combine eq.([energy ] ) with eq.([fe ] ) and expand the free energy to second order : @xmath96 . at zeroth order , @xmath97 . at first order , @xmath98 , which yields the same magnetization as in ref ..
the second order @xmath99 is required for the magnetic susceptibility @xmath100 , and reads ( for simplicity we take @xmath75 ) @xmath101 magnetisms in the first line of eq.([sus ] ) are contributions from the fermi surface .
the first two contributions are the peierls - landau magnetism , and the pauli paramagnetism for the orbital moment @xmath5 . in solids ,
the peierls - landau magnetism can be paramagnetic , especially near the band saddle points.@xcite these two contributions are prominant around singular points where the density of states diverges .
for example , around the saddle point , peierls - landau magnetism dominates in general , @xcite and for the spin-1 continuum model , the pauli paramagnetism contributes to a paramagnetic peak at @xmath102 where the flat band emerges as explained in sect.vi .
the third term is due to the @xmath41-space energy polarization in eq.([energy ] ) , and is first identified here .
similar to the pauli and peierls - landau magnetism , it also involves only single band quantities.this term generally compete with the pauli orbital and peierls - landau magnetisms as illustrated in sect.v , except at the band maxima or minima where @xmath103 vanishes . the other terms in eq.([sus ] ) are fermi sea contributions .
the first term in the second line yields the van vleck susceptibility originated from the vertical mixing energy in eq.([energy ] ) .
it is the only interband contribution to the orbital magnetic susceptibility .
it is always paramagnetic after summing over all the occupied bands , similar to the van vleck paramagnetic susceptibility in atomic systems .
the second one yields a langevin - like magnetic susceptibility from the last term in eq.([energy ] ) .
it can be expressed in a compact form using the quantum metric @xmath80 , which describes the intrinsic fluctuation of position - position operators ( @xmath104 ) in the bloch representation : @xmath105 $ ] ( see appendix b for details ) . for pauli or schr@xmath20dinger hamiltonian with constant mass
, this term yields diamagnetic response along directions that diagonalize @xmath80 .
its expression will change for effective hamiltonians with a general hessian matrix @xmath106 as given in appendix a. for dirac hamiltonian , the lagnevin - like magnetic susceptibility vanishes .
magnetic free energies in the third line in eq.([sus ] ) have no analogs as in atomic physics or for free particles , and are first identified here .
we call the susceptibility from these two terms the geometrical magnetic susceptibility , because they are due to the geometrical energies in eq.([energy ] ) from the horizontal mixing of bloch states and the geometrical correction to the density of states in eq.([fe ] ) . notice that the first term consists of the berry curvature , it is important when the band structure contains monopole or other nontrivial topological structures .
for example , for two band systems with the particle - hole symmetry , geometrical magnetic susceptibility always yields a diamagnetic susceptibility and is a prominant or even dominant contribution in the band gap . in the limit of atomic insulators where the hopping between lattice sites is suppressed , these contributions in the second line in eq.([sus ] ) will reduce to the familiar van vleck paramagnetic susceptibility and langevin diamagnetic susceptibility in atomic physics . to demonstrate this
, we use the wannier function representation .
notice that the periodic part of the bloch function can be expressed in terms of the wannier function : latexmath:[\[\label{wannier } where @xmath32 is the band index and @xmath108 is the wannier function localized at @xmath109 for band @xmath32 . with the help of eq.([wannier ] ) , we can express the van vleck paramagnetic susceptibility , langevin magnetic susceptibility and geometrical magnetic susceptibility in terms of wannier functions . to simplify the discussion
, we consider an insulator ( with each band either completely filled or empty ) with a single atom at each lattice site and then suppress the inter - lattice - site hopping ( see appendix c for details ) .
the van vleck term reads : @xmath110 |w_0(\bm r)\rangle|^2\,,\end{aligned}\ ] ] where @xmath111 is the total number of atoms and @xmath89 is the sample volume .
notice that this is the same as the van vleck paramagnetic susceptibility in atomic systems for the energy level @xmath112 .
the langevin term reads : @xmath113 which is the familiar langevin diamagnetic susceptibilityin atomic systems .
furthermore , the remaining fermi sea contribution to the orbital susceptibility in eq.([sus ] ) , i.e. the geometrical magnetic susceptibility , vanishes in the atomic insulator limit by similar derivations .
therefore , our formula indeed reduces to the correct result in the atomic insulator limit . to conclude the discussion of the orbital susceptibility in the wannier funtion representation , we comment that the fermi sea contribution to the orbital susceptibility is due to two types of effects , i.e. intra - lattice - site transition similar to that in the atomic physics and inter - lattice - site hopping which is unique in crystalline solids . our classification in eq.([sus ] ) thus provide a reasonable extrapolation of the orbital susceptibility from atomic systems to crystalline solids :
on one hand , the van vleck paramagnetic susceptibility and langevin magnetic susceptibility reduce to their counterparts in atomic systems and the geometrical magnetic susceptibility vanishes in the limit of atomic insulators ; on the other hand , even though they all contain inter - lattice - site hopping contribution , the van vleck paramagnetic susceptibility and the langevin magnetic susceptibility in solids still preserve their essential properties established in the atomic systems , i.e. the van vleck paramagnetic susceptibility is always paramagnetic and depends on the energy interval between two electronic states and the langevin magnetic susceptibility is diamagnetic along directions that diagonalize the quantum metric .
furthermore , we emphasize that the geometrical magnetic susceptibility is indeed unique in crystalline solids and is a novel mechanism of orbital susceptibility that only depends on the geometrical quantities in @xmath41-space . in atomic systems , the van vleck paramagnetic susceptibility is generally small due to the large separation between electronic levels , although it has a notable exception in the atomic lanthanide series , where the energy interval between the ground state and the first excited states is small . in solids ,
different energy levels can be very close , e.g. near the topological transition points . in this case , the van vleck paramagnetic susceptibility become large .
however , we will show that in the circumstance where the linear band crossing occurs , it is another contribution , i.e. the geometrical contribution , that dominates over other orbital susceptibilities .
the langevin magnetic susceptibility in atomic systems is also small and only important for close shell atoms .
likewise , in solids the langevin magnetic susceptibility is usually discussed for insulators where electrons are localized . however , our theory suggests that it is connected to the intrinsic expansion of the localized wave - packet , i.e. the quantum metric ( see appendix b for details ) .
therefore , for both insulators and metals , the langevin mangetism is well defined and can be sizable as illustrated in sect.v .
it can contribute to the prominant paramagnetic plateau as discussed in ref .. it can even dominate over other susceptibilities as in the continuum model of the double layer graphene .
eq.([sus ] ) and eq.([energy ] ) are the main results of this work . by developing a gauge - invariant second order semiclassical theory , we successfully obtain the complete orbital magnetic susceptibility in a compact form , with each term gauge - invariant and having a clear physical meaning .
of all the terms in eq.([sus ] ) , only the van vleck contributions involve interband processes , while other terms are single band properties .
particularly , we are able to cast these single band terms ( except for energy polarization and the langevin - like term in general case ) in a form only involving the single band quantities @xmath77 and @xmath5 , and geometrical quantities @xmath6 and @xmath80 , clearly demonstrating their intrinsic geometrical identity .
as a concrete example to show how various terms contribute to the total magnetic susceptibility and the importance of the geometrical magnetic susceptibility , we first consider the following tight - binding model defined on a honeycomb lattice:@xcite @xmath114 where @xmath115 is the electron annihilation ( creation ) operator on site @xmath69 , the first and the second terms are the nearest - neighbor and second - neighbor hopping terms , the third term is a staggered potential with @xmath116 for the two sublattices , and @xmath73 , @xmath117 and @xmath118 are the strengths of the terms .
the staggered potential breaks the inversion symmetry and generates a gap of @xmath119 in the spectrum .
the second - neighbor hopping is introduced for breaking the particle - hole symmetry .
the various contributions to the orbital magnetic susceptibility are plotted in fig.[f_sus ] with ( a ) and without ( b ) particle - hole symmetry . in the presence of particle - hole symmetry ( @xmath120 ) , the energy polarization and
the van vleck contributions vanish identically . from fig.[f_sus](a ) , one observes that in the gap the fermi surface terms vanish and the geometrical magnetic susceptibility dominates , which leads to a large diamagnetic response .
the magnitude of the geometrical magnetic susceptibility decreases rapidly away from the gap and it ( along with peierls - landau term ) is compensated largely by the pauli orbital paramagnetic susceptibility which is peaked at the band edges where @xmath5 takes its largest value.@xcite two noticeable paramagnetic peaks are observed around the band saddle points due to the peierls - landau contribution , which is a general feature as discussed before.@xcite further away from the gap region , the susceptibility decreases gradually to zero .
our result of @xmath121 agrees with that from the exact quantum treatment.@xcite the physics around gap can be described by the gapped dirac model @xmath122 with @xmath123 where @xmath124 is the nearest - neighbor bond length.@xcite this model is widely used in the study of graphene , @xmath125 , topological insulator surfaces and thin films.@xcite here @xmath126 s are the pauli matrices . near the band edge ( @xmath127 ) , the three competing magnetic susceptibilities ( pauli @xmath128 , peierls - landau @xmath129 , and geometrical @xmath130 ) read explicitly @xmath131 we emphasize that for systems with two valleys connected by time reversal operation , such as graphene or @xmath125 , the geometrical magnetic susceptibilities from the two valleys have the same sign .
note that in this low energy model the total susceptibility vanishes identically outside gap , which seems contradicting to the result in fig.[f_sus](a ) where one sees a finite paramagnetic plateau .
the difference is due to two fermi sea contributions including the langevin and a term in geometrical magnetic susceptibility resulting from the nonzero hessian @xmath106 ( whereas @xmath106 vanishes in the low energy model ) which produce an overall shift of @xmath121 .
this was known as the lattice contribution " in previous studies.@xcite this agreement demonstrates the validity of our theory .
we point out that if one starts from the pauli or schr@xmath20dinger hamiltonian , the langevin magnetic susceptibility is diagmanetic as discussed previously .
but for an effective hamiltonian as given in eq.([honey ] ) , the langevin magnetic susceptibility can be paramagnetic as shown in fig.[f_sus ] .
therefore , this paramagnetic plateau is sizable as in fig.[f_sus ] only when these two bands are well separated from other bands such that the tight binding model is an appropriate approximation .
when particle - hole symmetry is broken by a finite @xmath117 , as shown in fig.[f_sus](b ) , the geometrical magnetic susceptibility still dominates in the gap and one notes that the paramagnetic plateau near gap is suppressed in the valence band while enhanced in the conduction band . now the energy polarization and the van vleck contributions are finite . with current parameters , the van vleck paramagnetic susceptibility takes a small value around the gap region , while the energy polarization contribution takes opposite signs between the two bands . the energy polarization term , along with the enhanced langevin and pauli terms are the main contributions to a large paramagnetic response between the conduction band edge and the saddle point .
this is different from the usual orbital paramagnetic susceptibility resulting from the peierls - landau contribution.@xcite in fact , the contribution from peierls - landau is less important even in the region near and above the saddle point in conduction band .
the total susceptibility there is more affected by the competition between geometrical , langevin , and energy polarization terms .
this is in contrast to valence band where the peierls - landau dominates while other contributions are suppressed .
this example illustrates that : ( 1 ) the geometrical magnetic susceptibility is an important contribution , especially around the band gap ; ( 2 ) different terms in our classification dominate over different energy ranges ; ( 3 ) it is possible to enhance the paramagnetic susceptibility by breaking the particle - hole symmetry .
in addition , we note that for a generic two band model @xmath132 , @xmath133 is finite when particle - hole symmetry is broken . from eq.([sus ] ) , we find that only the van vleck paramagnetic susceptibility depends quadratically on @xmath134 , while all other terms are either independent or only have linear dependence .
this implies that van vleck susceptibility can in principle dominate for large @xmath135 , which also leads to a strong paramagnetic response .
in the previous example , we show that the single band properties contribute greatly to the orbital susceptibility , especially the geometrical magnetic susceptibility . as a second example , we consider the gapped spin-1 continuum model which is a three band system and hence has an enhanced interband van vleck paramagnetic susceptibility .
we then compare the single band and interband contributions .
the model hamiltonian is @xmath136 , where @xmath137 are standard spin-1
matrices:@xcite @xmath138 and @xmath139 is diagonal with entries @xmath140 , @xmath33 and @xmath141 .
this model describes the low energy behaviour of the dice lattice.@xcite the band dispersions are @xmath142 , with @xmath143 .
if the fermi energy falls in the lowest band , compared with corresponding magnetic susceptibilities in eq.([dirac ] ) for the spin-@xmath144 model , pauli and peierls - landau magnetic susceptibilities are the same , and the geometrical magnetic susceptibility doubles : @xmath145 the van - vleck and @xmath41-space energy polarization magnetic susceptibilities are nontrivial and read : @xmath146 compared with the spin-@xmath147 continuum model , the total orbital susceptibility still vanishes if the fermi energy falls in the band , but in the band gap the van vleck paramagnetic susceptibility can compete with the geometrical magnetic susceptibility : it is half the geometrical magnetic susceptibility with an opposite sign .
there is another interesting feature about this model due to the flat band at zero energy which has a vanishing berry curvature but nonvanishing magnetic moment .
this interesting geometry yields a paramagnetic peak at zero temperature due to the pauli paramagnetism : @xmath148 . unlike the tight binding honeycomb model discussed previously , this pauli paramagnetism is not compensated by the geometrical magnetic susceptibility since @xmath6 vanishes for the flat band .
given that this paramagnetic peak has a bigger magnitude than the geometrical magnetic susceptibility , this system will exhibit a paramagnetic response at low temperature when fermi energy falls in the gap , in contrast to the diamagenetic response purely due to the geometrical magentism as discussed in the model in previous section . at the limit of vanishing band gap
, the total paramagnetic susceptibility is the same as obtained in ref ..
this example illustrates that : ( 1 ) generally if more bands are involved in the calculation of the orbital susceptibility , the interband effect ( van vleck magnetic susceptibility ) is enhanced , while the geometrical contribution is still prominant ; ( 2 ) the total susceptibility is greatly affected by band geometries .
up to now , we are working under the minimal coupling assumption , i.e. the magnetic field enters the hamiltonian only through the vector potential .
this assumption is exact for the dirac hamiltonian with a periodic potential , which yields the complete hilbert space for electrons in solids .
therefore , our expression of the orbital magnetic susceptibility is also complete in this situation .
however , for any effective hamiltonians defined in a subspace of the complete hilbert space , such as the schr@xmath20dinger hamiltonian , the pauli hamiltonian with the spin - orbit coupling , and the effective hamiltonian for some specific bands , the magnetic fields may have other effects .
our theory of the orbital magnetic susceptiblity can be easily extended to incorporate this situation .
based on the foldy - wouthuysen transformation in solid state context , the hamiltonian in a uniform magnetic field beyond minimal coupling assumption is : @xcite @xmath149 where @xmath150 and @xmath151 are appropriate matrix operators as functions of @xmath152 .
an example for @xmath150 is that if one derives the pauli hamiltonian from the dirac hamiltonian , one finds @xmath150 is the magnetic moment for spins : @xcite @xmath153 where @xmath112 is the electron band dispersion , and @xmath154 is the pauli matrices for electron spins .
the expression for @xmath155 is also given in ref ..
we can expand the new hamiltonian in eq.([fw ] ) around the center of mass position @xmath1 . compared with the same expansion for @xmath11 as given in sect.ii
, we find that the local hamiltonian @xmath156 is unchanged ; the first order correction is @xmath157-\bm b\cdot \hat{\bm \mu}(\bm p+{1\over 2}\bm b\times \bm r_c)\,,\ ] ] and it receives a new term from @xmath150 ; the second order correction is @xmath158_i [ \bm b\times ( \hat{\bm q}-\bm r_c)]_j\notag\\ & + \left\{\left[{1\over 2}\bm b\times ( \hat{\bm q}-\bm r_c)\right]\cdot \bm \partial\right\ } \hat{\bm \mu}(\bm p+{1\over 2}\bm b\times \bm
r_c ) \cdot \bm b\notag\\ & + b_ib_j\hat{h}_{ij}(\bm p+{1\over 2}\bm b\times \bm r_c)\,.\end{aligned}\ ] ] by following the same precedure as in sect ii - iv , we can derive the orbital magnetic susceptibility for the hamiltonian in eq.([fw ] ) .
the result is similar to eq.([sus ] ) , with the following modification : ( 1 ) the second term in eq.([newh1 ] ) is of first order with respect to the magnetic field , and its influence to the susceptibility is through @xmath5 and @xmath159 : @xmath160 and @xmath161 ; consequently , the pauli magnetism in eq.([sus ] ) will contain the contribution from @xmath150 ; the geometrical magnetic susceptibility and van vleck magnetic susceptibility are also affected ; ( 2 ) the second and third term in eq.([newh2 ] ) is of second order with respect to the magnetic field and its expectation value directly adds to the second order wave - packet energy and modifies the langevin magnetic susceptibility : @xmath162\,.\end{aligned}\ ] ]
in summary , we derive a compact gauge - invariant expression of the bloch wave - packet energy correct to second order in external magnetic field , which fully incorporates the first order correction of the wave - packet from vertical and horizontal mixings . based on this , we obtain a complete and compact formula for the orbital magnetic susceptibility , with important advantage that each term is gauge invariant and has clear physical meanings .
we demonstrate that other than the familiar pauli and peierls - landau magnetic susceptibilities , the orbital susceptibility in solids also consists of the van vleck paramagnetic susceptibility and langevin magnetic susceptibility , which reduce to their counterparts in atomic physics in the limit of atomic insulators .
more importantly , we identify two new contributions : the geometrical magnetic susceptibility derived from the berry curvature and the quantum metric which can dominate in a small energy gap , and the @xmath41-space energy polarization magnetic susceptibility , which competes with peierls - landau and pauli magnetic susceptibility on a fermi surface . we illustrate that the pauli , peierls - landau and geometrical magnetic susceptibility depend solely on band geometrical quantities and affect the orbital susceptibility greatly , as shown in two examples . _
acknowledgment._we acknowledge valuable discussions with d. l. deng , z. qiao , h. chen , j. zhou , x. li , r. cheng , and l. zhang .
we thank shan guan for helping with the figures .
we thank j .- n .
fuchs , a.raoux , f. pichon and g. montambaux for pointing out a numerical mistake in our previous manuscript .
qn is supported by nbrpc ( no.2012cb921300 and no.2013cb921900 ) , and nsfc ( no.91121004 ) .
say is supported by sutd - srg - epd-2013062 .
yg is supported by doe ( de - fg03 - 02er45958 , division of materials science and engineering ) and welch foundation ( f-1255 ) .
the wave - packet energy has several parts .
we first consider the contribution from @xmath15 first .
@xmath163 there is an important correction to eq.([lc ] ) : when the mixing of bloch states @xmath55 is taken into account , the normalization of the wave - packet is modified : @xmath164 to make the euler - lagrangian method valid , the wave - packet must be normalized , at least up to second order for our purpose .
therefore , @xmath56 must have a second order correction : @xmath165 , with @xmath166 .
this correction @xmath167 contributes to the energy in eq.([lc ] ) : @xmath168 here , the term @xmath169 arises due to the horizontal mixing in the coefficient @xmath55 .
compared with the renormalization condition in eq.([norm ] ) , the integration in eq.([lc ] ) has an additional energy factor @xmath112 in the integrand , which leads to the additional term @xmath169 due to the derivative of @xmath56 involved in the horizontal mixing .
however , this @xmath169 is cancelled by the contribution from the dynamical part of the langrangian : @xmath170 we only need to keep the term up to the second order in @xmath171 : @xmath172 plug eq.([dcn ] ) into eq.([dl ] ) , and we have @xmath173 where @xmath174 is the inverse effective mass tensor .
then we calculate the contribution to the wave - packet energy from the gradient correction @xmath14 : @xmath175\notag\\ & + \sum_{n\neq 0}{1\over 4 } \int d\bm p \ {
[ \bm b\times ( \hat{\bm d}-\bm r_{c})]_i^\star c_0^\star [ \bm b\times ( \hat{\bm d}-\bm r_{c})]_jc_0\notag\\ & \quad\quad\quad [ -i(v_i)_{0n}(a_j)_{n0}]+c.c.\}\,,\end{aligned}\ ] ] where @xmath5 is the orbital magnetic moment : @xmath64 , and @xmath81 is the positional shift : @xcite @xmath176+c.c.\,.\ ] ] the remaining contribution to the wave - packet energy is from the second order correction @xmath83 : @xmath177\notag\\ & + { 1\over 8 } \int d\bm p ( \bm b\times \bm \partial |c_0|)_i ( \bm b\times \bm \partial |c_0|)_j \langle 0| \gamma_{ij}|0\rangle \notag\\ & + { 1\over 8 } \langle 0|\gamma_{ij}|0\rangle ( \bm b\times \bm r_c)_i ( \bm b\times \bm r_c)_j\notag\\ & + \sum_{n\neq 0}{1\over 8 } [ ( \bm b\times \bm a_{0n})_i ( \gamma_{ij})_{n0 } ( \bm b\times \bm r_c)_j+c.c.]\notag\\ & + \sum_{n\neq 0}{1\over 16 } ( \bm b\times i\bm \partial)_i [ ( \gamma_{ij})_{0n } ( \bm b\times \bm a_{n0})_j]-(i\leftrightarrow j)\notag\\ & + { 1\over 8 } \sum_{(m , n)\neq 0 } ( \bm b\times \bm a_{0m})_i ( \gamma_{ij})_{mn } ( \bm b\times \bm a_{n0})_j\,.\end{aligned}\ ] ] after some cancellations , we have @xmath178 where @xmath179 . for relativistic dirac hamiltonian , @xmath180 simply vanishes . for nonrelativistic schr@xmath20dinger and pauli hamiltonian , @xmath181 is inverse mass times the identity matrix , and eq.([h2 ] ) reduces to a compact form : @xmath182 where @xmath183 is the quantum metric of @xmath41-space.@xcite by combining eq.([e2]),([e3]),([epsilon1]),([h2sim ] ) and the relation between effective mass tensor @xmath77 and the hessian matrix @xmath181 of @xmath15 : @xmath184\,,\ ] ] the wave - packet energy can be put in a compact form ( assume @xmath75 ) : @xmath185 various quantities in eq.([ap_energy ] ) have been explained in the main text .
note that originally , physical quantities such as @xmath44 and @xmath50 in eq.([bloch ] ) and ( [ wavepacket ] ) , @xmath186 , @xmath37 , @xmath112 and @xmath187 in eq.([bloch ] ) , ( [ cn ] ) , ( [ lc ] ) and ( [ dcn ] ) are all functions of @xmath188 .
however , since @xmath189 is a delta function localized around @xmath2 , after intergration with @xmath189 , those physical quantities in the final results in eq.([energy ] ) , ( [ epsilon1 ] ) , ( [ h2 ] ) and ( [ ap_energy ] ) are functions of the physical momentum @xmath3 .
the @xmath41-space polarization energy can be expressed in terms of the single band bloch states @xmath44 : @xmath190\,,\ ] ] where @xmath191 is the covariant derivative acting on the bloch state @xmath44 . in eq.([pol ] ) , the first term is from the gradient correction to @xmath15 , and the second term is from the nonadiabatic correction . to understand the geometric meaning of eq.([pol ] ) , we consider the quadrupole moment of the velocity operator under the wave - packet @xmath52 : @xmath192 [ ( i\bm \partial+\bm a_0-\bm r_c)_j c_0 ] ( \bm v_0)_k+c.c.\notag\\ & + \int d\bm p [ ( i\bm \partial+\bm a_0-\bm r_c)_i^\star c_0^\star ] c_0 \langle
u_0|\hat{v}_k |\hat{d}_j u_0\rangle+c.c.\notag\\ & + \int d\bm p c_0^\star [ ( i\bm \partial+\bm a_0-\bm r_c)_j c_0 ] \langle \hat{d}_i u_0|\hat{v}_k|u_0\rangle+c.c.\notag\\ & + \int d\bm p c_0^\star c_0 \langle \hat{d}_iu_0|\hat{v}_k|\hat{d}_j u_0\rangle+c.c.\,.\end{aligned}\ ] ] the @xmath41-space energy polarization is connected to the quadrupole moment of the velocity operator : @xmath193 the first term in eq.([quad ] ) vanishes . meanwhile ,
notice that @xmath49 in the definition of the energy polarization indicates that the horizontal mixing in the wave - packet is used . since the energy polarization combines the horizontal and vertical mixing , we should not take the intraband part of @xmath7 in @xmath14 in the bloch representation which represents the horizontal mixing in @xmath14 .
therefore , the remaining two contributions from eq.([quad ] ) yield : @xmath194\notag\\ & + { 1\over 4}\hat{e}_\ell \epsilon_{si\ell}\epsilon_{tjk}b_s b_t [ \langle \hat{d}_i u_0|\hat { v}_k |\hat{d}_j u_0\rangle+c.c.]\,,\end{aligned}\ ] ] the resulting @xmath41-space polarization energy is @xmath195\notag\\ & + { 1\over 4 } \epsilon_{si\ell}\epsilon_{tjk}b_s b_t \partial_\ell [ \langle \hat{d}_i u_0|\hat { v}_k |\hat{d}_j u_0\rangle+c.c.]\notag\\ & = { 1\over 4 } \epsilon_{si\ell}\epsilon_{tjk}b_s b_t \partial_\ell [ \langle \hat{d}_i u_0|\hat { v}_k |\hat{d}_j u_0\rangle+c.c.]\end{aligned}\ ] ] this yields the first term in eq.([pol ] ) . to obtain the second term ,
notice that the bloch state @xmath44 depends on time through @xmath1 , which yields a first order hamiltonian : @xmath196 through similar derivations for the polarization energy from @xmath14 , it is straightforward to derive the second term in eq.([pol ] ) .
finally , it is interesting that through similar derivations in eq.([quad ] ) , we find that the quantum metric @xmath80 is the intrinsic expansion of the wave - packet : @xmath197 in eq.([metric ] ) , the first term on the right hand side is the extrinsic expansion , which is determined by the shape of the wave - packet . from the semiclassical point of view , the particle simultaneously has the momentum and position as dynamical variables , and the extrinsic expansion of @xmath198 is ignored .
however , the second term in eq.([metric ] ) reflects the intrinsic expansion of the wave packet , due to the projection of the whole hilbert space onto a single band hilbert space .
we will discussion the van vleck paramagentism in the wannier function representation first . @xmath199 \big |^2\notag\\ & = \int { d^3k_c\over 8\pi^3 } \sum_n { f(\varepsilon_0(\bm k_c))\over \varepsilon_0-\varepsilon_n}{1\over n}\big |\sum_{\bm r_1}{1\over 2}\bm b\cdot \notag\\ & [ \langle w_n(\bm r_1)|\hat{\bm v}\times ( \bm r-\bm r_1 ) |w_0(\bm r_1)\rangle-{1\over n } \sum_{\bm r_3}\langle w_n(\bm r_1)|\hat{\bm v}|w_0(\bm r_1)\rangle \times\langle w_0(\bm r_3 ) |(\bm r-\bm r_3)|w_0(\bm r_3)\rangle]|^2\notag\\ & = { n\over v}\sum_n{f(\varepsilon_0)\over \varepsilon_0-\varepsilon_n}\big | \langle w_n(\bm r)|{1\over 2}\bm b\cdot[\hat{\bm v}\times ( \bm r-\bm r ) ] |w_0(\bm r)\rangle|^2\,.\end{aligned}\ ] ] in eq.([wanvv ] ) , the first equality is simply the expansion of the van vleck paramangetism for a single band @xmath112 in terms of the wannier function .
the second equality is derived by taking the atomic insulator limit and setting all inter - lattice - site hopping to zero : @xmath200 , @xmath201 and @xmath202 .
the last equality accounts for the fact that for atomic insulators , we only have flat band @xmath112 with no @xmath4 dependence , which coincides with the atomic energy level
. meanwhile , @xmath203 , since the electron is bound with each atom and its position expectation should coincide with the atom position .
we also use the fact that @xmath204 |w_0(\bm r)\rangle$ ] are identical for all lattice sites .
@xmath206\cdot \bm b|w_0(\bm r_1)\rangle \langle w_0(\bm r_4 ) |[\bm
b\times(\bm r-\bm r_4)]\cdot [ ( \varepsilon_0-\hat{h}_c)(\bm r-\bm r_3)]|w_0(\bm r_3)\rangle\right)\notag\\ & = { 3\over 8 } { f(\varepsilon_0)\over n^2 }
\sum_{\bm r_1,\bm r_2,\bm r_3,\bm r_4 } \delta_{(\bm r_1+\bm r_3-\bm r_2-\bm r_4),0}\notag\\ & \left ( \langle w_0(\bm r_2)|[(\bm r-\bm r_2)\times ( \bm r-\bm r_1)]\cdot \bm b|w_0(\bm r_1)\rangle \langle w_0(\bm r_4 ) |[\bm b\times(\bm r-\bm r_4)]\cdot [ ( \varepsilon_0-\hat{h}_c)(\bm r-\bm r_3)]|w_0(\bm r_3)\rangle\right)\notag\\\end{aligned}\ ] ] in the above equation , the second equality uses the fact that in the atomic insulator limit , the band dispersion becomes flat .
notice that in the last line in eq.([wangeo1 ] ) , @xmath207 if @xmath208 .
therefore , to obtain a nonzero contribution , we must have @xmath209 and @xmath210 due to the kronecker delta function .
this part of geometrical contribution contains only inter - lattice - hopping effect , and vanishes in the limit of atomic insulators . due to the prefactor @xmath212 ,
it is obvious that this remaining contribution to the geometrical magnetic free energy also contains only inter - lattice - hopping effect , and vanishes in the limit of atomic insulators .
100 n. w. ashcroft and n. d. mermin , solid state physics ( saunders , philadelphia , 1976 ) .
d. xiao , m .- c .
chang , and q. niu , rev .
82 , 1959 ( 2010 ) .
n. nagaosa , j. sinova , s. onoda , a. h. macdonald and n. p. ong , rev .
82 , 1539 ( 2010 ) .
y. kagan and l. a. maksimov , phys .
100 , 145902 ( 2008 ) .
l. zhang , j. ren , j .- s .
wang , and b. li , phys .
105 , 225901 ( 2010 ) .
t. qin , j. zhou , and j. shi , phys .
b 86 , 104305 ( 2012 ) .
s. onoda , n. sugimoto , and n. nagaosa , phys .
97 , 126602 ( 2006 ) .
j. shi , g. vignale , d. xiao and q. niu , phys .
99 , 197202 ( 2007 ) .
e. n. adams , phys .
rev . 89 , 633 ( 1953 ) .
j. e. hebborn and e. h. sondheimer , j. phys .
solids 13 , 105 ( 1960 ) .
j. m. luttinger and w. kohn , phys .
97 , 869 ( 1955 ) .
l. m. roth , j. phys .
solids 23 , 433 ( 1962 ) .
e. i. blount , phys .
126 , 1636 ( 1962 ) .
g. h. wannier and u. n. upadhyaya , phys .
136 , a803 ( 1964 ) .
p. k. misra and l. kleinman , phys .
b. 5 , 4581 ( 1972 ) . m. koshino and t. ando , phys .
b 76 , 085425 ( 2007 ) .
g. a. h. schober , h. murakawa , m. s. bahramy , r. arita , y. kaneko , y. tokura , and n. nagaosa , phys . rev .
108 , 247208 ( 2012 ) .
g. gmez - santos and t. stauber , phys .
106 , 045504 ( 2011 ) .
a. raoux , m. morigi , j .-
fuchs , f. pichon , and g. montambaux , phys .
112 , 026402 ( 2014 ) .
a. raoux , f. pichon , j .-
fuchs , and g. montambaux , arxiv:1411.5940 .
h. fukuyama , prog .
45 , 704 ( 1971 ) .
g. vignale , phys .
67 , 358 ( 1991 ) .
y. gao , s. a. yang and q. niu , phys .
112 , 166601 ( 2014 ) .
j. anandan and y. aharonov , phys .
65 , 1697 ( 1990 ) .
t. neupert , c. chamon , and c. mudry , phys .
b 87 , 245103 ( 2013 ) .
r. d. king - smith and d. vanderbilt , phys .
b 47 , 1651 ( 1993 ) .
t. cai , s. a. yang , x. li , f. zhang , j. shi , w. yao , and q. niu , phys .
b 88 , 115140 ( 2013 ) .
d. xiao , g .- b .
liu , w. feng , x. xu and w. yao , phys .
108 , 196802 ( 2012 ) .
a. h. castro neto , f. guinea , n. m. r. peres , k. s. novoselov and a. k. geim , rev .
81 , 109 ( 2009 ) .
qi and s .- c .
zhang , rev .
83 , 1057 ( 2011 ) .
m. z. hasan and c. l. kane , rev .
82 , 3045 ( 2010 ) .
d. xiao , w. yao , and q. niu , phys .
99 , 236809 ( 2007 ) .
d. bercious , d.f .
urban , h. grabert , and w. h@xmath213usler , phys .
a 80 , 063603 ( 2009 ) .
l. l. foldy and s. a. wouthuysen , phys .
78 , 29 ( 1950 ) .
e. i. blount , phys .
128 , 2454 ( 1962 ) .
chang and q. niu , j. phys .
: condens .
matter 20 , 193202 ( 2008 ) . | within the wave - packet semiclassical approach , the bloch electron energy is derived to second order in the magnetic field and classified into gauge - invariant terms with clear physical meaning , yielding a fresh understanding of the complex behavior of orbital magnetic susceptibility .
the berry curvature and quantum metric of the bloch states give rise to a geometrical magnetic susceptibility , which can be dominant when bands are filled up to a small energy gap .
there is also an energy polarization term , which can compete with the peierls - landau and pauli magnetism on a fermi surface .
all these , and an additional langevin susceptibility , can be calculated from each single band , leaving the van vleck susceptibility as the only term truly from interband coupling . |
although direct manifestations of new physics are searched for at the high energy frontier , new phenomena can also be looked for at low energies via precision measurements .
the new effects might be observed in rare decays of the positronium ( see e.g. @xcite ) .
positronium ( ps ) , the positron - electron bound state , is the lightest known atom , which at the current level of experimental and theoretical precision is bound and self - annihilates through the electromagnetic interaction @xcite .
this feature has made positronium an ideal system for testing the accuracy of quantum electrodynamics ( qed ) calculations for bound states , in particular for the triplet ( @xmath7 ) state of @xmath8 ( orthopositronium , @xmath9 ) @xcite .
due to the odd - parity under c - transformation @xmath9 decays predominantly into three photons with a lifetime in vacuum of @xmath10 ns @xcite-@xcite .
the singlet ( @xmath11 ) state ( parapositronium , @xmath12 ) decays predominantly into two photons with a lifetime in vacuum of @xmath13 ps @xcite . the longer lifetime of @xmath9 ( due to the phase - space and additional @xmath14 suppression factors ) gives an enhancement factor @xmath15 in the sensitivity to an admixture of potential new interactions not accommodated in the standard model ( sm ) @xcite .
this paper focuses on a new search for @xmath1 decays . by invisible we mean photonless decays , i.e. decays which are not accompanied by energy deposition in a hermetic calorimeter . in the sm the decay into a neutrino - antineutrino pair
has a branching ratio of @xmath16 @xcite .
evidence for invisible decays in the region @xmath17 would , therefore , unambiguously signal the presence of new physics .
new models that are relevant to the @xmath18 decay mode predict the existence either of i ) extra - dimensions @xcite , or ii ) fractionally charged particles @xcite , or iii ) a new light vector gauge boson @xcite , or iv ) mirror particles , which could be candidates for dark matter @xcite-@xcite .
the first experiment to search for invisible decay channels of o - ps was performed by atoyan et al .
their result on @xmath19 @xmath20 ( @xmath21 c.l . ) excluded this channel as a possible explanation of the @xmath9 lifetime anomaly ( for a recent review see e.g. @xcite ) .
this search was repeated by mitsui et al . who found a branching ratio @xmath19 @xmath22 ( @xmath21 c.l .
furthermore , they could place a limit on the existence of milli - charged particles and on the photon mirror - photon mixing .
this result was corrected in @xcite by taking into account the suppression factor for the mixing due to the presence of matter .
the rest of the paper is organized as follows .
the details of the experimental setup are reported in section [ sec : experiment ] .
the expected background is presented in section [ bkg ] . in sections
[ sec : datareduction ] and [ sec : results ] the data analysis and the results are described .
the interpretation and the conclusion are reported in sections [ sec : interpretation ] and [ sec : conclusion ] .
the experimental signature of @xmath1 decay is the apparent disappearence of the energy @xmath23 expected in ordinary decays in a hermetic calorimeter surrounding the @xmath9 formation target .
the readout trigger for the calorimeter is performed by tagging the stopping of a positron in the target with high efficiency . for the design of the experimental setup aiming at a sensitivity @xmath19 @xmath24 ,
the following criteria were considered : 1 .
the probability not to detect all direct @xmath25 annihilation photons was suppressed to @xmath26 using a thick hermetic crystal calorimeter , minimizing the dead material and with a good energy resolution .
the probability to loose all photons in 3@xmath27 decays is consequently even smaller .
the region around the target was designed with as little dead material as possible in order to reduce photon losses .
3 . by the appropriate choice of a porous target material and its dimensions a high fraction of @xmath9
was produced resulting in a suppression of the background from the the direct @xmath25 annihilation and @xmath12 decays and in high statistics .
the trigger rate and the daq speed were maximized for statistics .
an efficient positron tagging system was designed to provide a clean trigger for positronium formation .
a method was developed to suppress the background from the electrons emitted in the ec process ( shake - off electrons ) .
an efficient identification of the 1.27 mev photon emitted by the @xmath28na radioactive source was achieved with a method to veto the charged particles entering the trigger counter , thus , the backgrounds related to them could be reduced .
the schematic illustration of the detector setup is shown in figure [ detector ] ( the detailed description of the experimental technique and setup can be found in refs .
@xcite ) .
positrons are produced from a @xmath28na source with an activity of @xmath29 kbq .
the @xmath28na has a half life of 2.6 years and has a @xmath30-value for the nuclear transition to @xmath31ne of @xmath32 mev .
this is the maximum energy available for the particles involved in one of the three possible decay modes of @xmath28na : 1 .
decay mode a ( br @xmath33 ) : the @xmath34 decay with end - point energy 0.546 mev .
the positron is always followed by the prompt emission of a 1.27 mev photon ( @xmath35 3.7 ps ) from the @xmath31ne@xmath36 de - excitation to the ground state .
decay mode b ( br @xmath37 ) : the electron capture process ( ec ) , where an orbital electron is captured by the nucleus , and only a 1.27 mev photon and a neutrino are emitted from the source . in some rare cases ( with a probability @xmath38 ) , an orbital electron is ejected , due to the sudden change in the nucleus charge ( shake - off ) @xcite .
decay mode c ( br @xmath39 ) : there is no photon emission because the transition goes directly to the ground state .
the end point energy of the positron is 1.83 mev .
therefore , in most @xmath9 ( @xmath12 ) decays we expect 3(2 ) photons with a summed energy equal to @xmath23 and one photon with an energy of 1.27 mev ( see figure [ opsregion ] ) .
formation target of the setup .
_ _ ] photons from the direct @xmath25 annihilation in flight or from the positronium decays were detected in a hermetic , segmented bgo calorimeter ( the ecal ) .
two endcap counters called tbgo and fbgo ( see figures [ detector]-[opsregion ] ) surrounded the target on each side . at the analysis level
the 1.27 mev photon ( `` the triggering photon '' ) was required to be identified in the tbgo counter .
the calorimeter was instrumented with charge and time readout .
the activity of the source was chosen to maximize the trigger rate versus the inefficiency of signal detection ( mostly due to pileup events ) .
the source was prepared by evaporating drops of a @xmath28na solution directly on a 100 @xmath40 m thick and 2x8 mm@xmath41 wide plastic scintillator ( see figure [ fiber ] ) fabricated by squeezing a 500 @xmath40 m diameter scintillating fiber ( bicron bf-12 ) . in this way , no dead material was introduced for a source holder .
the s - shape of the fiber ( see figure [ detector ] ) was selected to avoid background from back - to - back 511 kev annihilation photons .
na source on the squeezed part in the center .
_ _ ] the scintillating fiber was read out at both ends by two photomultipliers ( philips xp2020 ) located outside the detector ( see figure [ detector ] ) .
the coincidence of the two pmt signals was used to tag the passage of a positron through the fiber and acted as a start signal for the data readout system .
the use of two pmts in coincidence , instead of a single one @xcite , lowered the ratio between fake and real positron triggers to @xmath42 .
opposite to the source , a 4x8x8 mm@xmath43 sio@xmath44 aerogel piece ( type sp30 , purchased from matsushita electric works ) was placed in contact with the squeezed fiber ( see figure [ opsregion ] and [ fbgofiber ] ) .
positrons stopping in the aerogel target may form positronium ( the formation probability is @xmath45 @xcite ) which can migrate into the aerogel pores .
the collisions with the walls of the pores did not appreciably quench the @xmath9 : when the aerogel was flushed with nitrogen , a fit to the distribution of the time difference between the start from the fiber and the stop from the calorimeter yielded @xmath46 ns , which is very close to the lifetime in vacuum @xmath47 ns @xcite .
the ecal was composed of 100 bgo crystals that surrounded the target region providing a nearly isotropic sphere of radius 200 - 220 mm ( see figure [ detector ] ) .
each crystal had a hexagonal cross - section of 61 mm diameter and a length of 200 mm .
the crystals of the inner most ring were wrapped in a 2 @xmath40 m thick aluminized mylar foil to minimize the photon energy absorption .
the other crystals were wrapped with teflon foils of 750 @xmath40 m thickness .
the crystals in the barrel and the fbgo were readout with etl 9954 photomultiplier tubes .
the energy deposited in the fiber is an important parameter in order to reject the background from the electrons emitted in the ec process ( shake - off electrons , see section [ bkg ] ) .
the mean number of photoelectrons detected in each xp2020 for a positron crossing the fiber was measured to be about 1.2 , thus , a cut on the energy deposited in the fiber using these signals would not be meaningful .
for this reason the fbgo was also used to measure the energy deposited in the fiber by the positron : the light emitted by the fiber could traverse the transparent aerogel and enter the fbgo through an aperture in the wrapping on its front face .
this light was then guided by the fbgo to the pmt attached on the back face of fbgo ( as illustrated in figure [ fbgofiber ] ) .
this method provided a mean number of photoelectrons equal to 13@xmath48 for a positron traversing the fiber @xcite .
the tbgo endcap was used in the off - line analysis to identify the 1.27 mev photon .
the energy resolution of the tbgo was an essential parameter to reduce backgrounds related to the misidentification of the 1.27 mev photon ( see also section [ bkg ] ) . to provide a better energy resolution ,
this crystal was coupled to an etl 9964 pmt with a more uniform light collection and a larger quantum efficiency than the etl 9954 .
for the same reason this crystal was of a better quality than the others and efforts were dedicated to maximize the light collection by keeping the amount of dead material introduced by the crystal wrapping as small as possible .
the best results were achieved with the crystal wrapped in the 3 m radiant mirror ( 64 @xmath40 m thickness ) : the resolution at 662 kev was measured to be about @xmath49 ( fwhm ) . to select the triggering photon , an energy window @xmath50 $ ] kev was used in the analysis . in addition , to veto charged particles ( positrons and electrons ) entering the tbgo , a 1 mm thick plastic scintillator ( bicron bc-400 )
was optically coupled to the tbgo front face ( for more details see @xcite ) , i.e. the same pmt was used to detect the light signals from the plastic scintillator and the tbgo crystal ( see section [ bkg ] for a discussion of backgrounds associated to charged particles entering the tbgo ) .
the signals from the plastic scintillator and the bgo could be distinguished because of the different decay time ( @xmath51 = 2.7 ns for the plastic scintillator and @xmath35 300 ns for the bgo ) . for this purpose ,
split signals from the pmt were fed into two adcs with different integration gates ( called short gate and long gate ) .
a vme system interfaced to a pc was used for data acquisition ( the daq rate was about 1800 events / s ) . for every trigger ,
five caen 32 channels qdc v792 modules recorded the charge of the crystal signals while a caen tdc v775 recorded the time information .
a trigger gate length of 2.9 @xmath40s was chosen to keep the probability of @xmath9 to decay after this time to @xmath52 .
a temperature stabilized light - tight box containing the calorimeter was built in order to keep the temperature of the bgo crystals in the range of @xmath53c .
water , whose temperature was controlled by two thermostats , circulated through copper tubes welded on two copper plates inside the box .
the experimental hall was air conditioned to keep temperature variations within @xmath54c .
the high - voltage dividers of all pmts were placed outside the box in order to avoid energy dissipation close to the crystals .
the bgo crystals were equipped with leds that could be pulsed periodically to monitor the response .
additionally , the gains of the pmts were also monitored to check their stability .
the detector was calibrated and monitored internally using the 511 kev annihilation photon and the 1.27 mev photons emitted by the @xmath28na source .
variations of the energy scale during the run period were within @xmath55 and corrected on the basis of an internal calibration procedure @xcite .
in order to reach the required sensitivity , the background must be reduced and controlled at the level of @xmath56 . to understand the different background sources and to cross - check the simulation , we perfomed an engineering run with a simplified version of our detector . during two months of data taking ,
the stability of the detector and its components was investigated .
the comparison between the background expected based on monte carlo ( mc ) simulations and the data of the engineering run is summarized in table [ bkgtbl ] .
the ecal thickness of 200 mm provides a probability of @xmath57 for two 511 kev photons to escape detection ( see background 1 in table [ bkgtbl ] ) . for three photons
decays , this probability is consequently even smaller .
if one ( or more ) annihilation photon ( e.g. backscattered from the target ) overlaps with the 1.27 mev in the tbgo , it can fall in the trigger energy window @xmath50 $ ] kev because of the finite energy resolution .
this introduces a background if the remaining annihilation photon gets absorbed in the dead material or escapes detection .
the separation between the upper bound of the trigger energy window and the sum of a triggering and a 511 kev photon was , thanks to the good energy resolution of the tbgo , 7@xmath58 of the 1786 kev peak ( 1275 + 511 kev ) .
thus , the level of this background is @xmath59 ( background 2 in table [ bkgtbl ] ) . in order to suppress the other sources of background related to the misidentification of the 1.27 mev photon ( backgrounds 3 , 4 and 5 listed in table [ bkgtbl ] ) , one had to veto charged particles entering the tbgo .
two processes are responsible for generating such triggers .
one is associated with decay mode b ( ec ) when the 1.27 mev photon interacts in the fiber , faking a positron signal .
if the scattered photon and the compton electron reach the tbgo , the sum of the energy of the two particles can be misidentified as the triggering photon without any energy in the rest of the calorimeter .
another background can occur in decay mode a if the 1.27 mev photon is not detected : a trigger can be produced by a positron that multiple scatters ( ms ) in the fiber and deposits enough energy to trigger the experiment .
if the positron reaches the tbgo with a kinetic energy of about 200 - 300 kev and the two 511 kev annihilation photons are completely absorbed in the tbgo an energy close to 1.27 mev will be reconstructed .
this will appear as an invisible decay since no energy is expected in the rest of the detector .
if the 1.27 mev photon is not present ( decay mode c ) a trigger can similarly be produced by the 1.83 mev positron . to veto these backgrounds the charged particle veto of the tbgo described in the previous section
was used such that the backgrounds 3 , 4 and 5 had a probability of @xmath60 in the final setup .
the ec photon may accidentally coincide with a trigger from the fiber generated either by the pmts noise or by some other particles emitted from possible unstable isotopes formed during the target activation .
the level of this background could be reduced with the selection of two xp 2020 with very low noise ( @xmath61 counts / s ) and the requirement of the coincidence between them .
in addition , a radioactive source with a controlled high purity was chosen . from the data of the engineering run ,
this background was estimated to be @xmath42 ( background 6 in table [ bkgtbl ] ) . during the decay mode b ( ec )
the fiber signal can be generated by shake - off electrons ( background 7 in table [ bkgtbl ] ) . since the probability of an electron ejection steeply drops with its emission kinetic energy ( more than 4 orders of magnitude in the first 100 kev ) the cut on the energy deposited in the fiber by the triggering particles was used to suppress this background .
the engineering run allowed to test these backgrounds as shown in table [ bkgtbl ] .
the expected fraction of zero energy events is 10@xmath62 smaller than what was measured .
this difference can be explained by the contribution from the shake - off electrons which was not included in the simulation .
indeed , in the engineering run there was no information about the energy deposited in the fiber so that no cut on the energy of the particles passing through the fiber could be applied .
the last column of the table lists the expectations for the final setup .
the total background is estimated to be at the level @xmath56 .
the threshold on the energy deposited in the fiber was set considering the uncertainty due to the number of photoelectrons and the subtraction method to be sure that no electron below 100 kev could trigger the fiber ; it was chosen , based on simulations , to be at 140 kev .
.__comparison between expected and measured background level for the different background sources in the engineering run and expected background level for the final setup ( see text for details ) .
_ _ [ cols="^,^,^,^,^ " , ]
after imposing the above requirements a final sample of 1.41@xmath63 events was obtained . for these events
the energies of all the 100 bgo crystals , except the tbgo , were summed .
figure [ ecalsum ] shows the spectrum of the total energy ( @xmath64 ) deposited in the ecal .
the peak at 1022 kev corresponds to the positronium mass ( @xmath65 ) .
the inset shows that no event is observed in the zero energy region . to define the upper energy cut on @xmath64 , below which an event is considered as photonless ( invisible ) , a dedicated run of @xmath66 triggers
was performed triggering the experiment only with the 1.27 mev photon and no requirement of the fiber @xcite .
the zero peak contained about 10.4@xmath67 of the events passing the selection cuts defined in section [ sec : datareduction ] .
this value was corrected by a factor 0.8 determined using the mc simulations to take into account the different detection efficiencies of the 1.27 mev photon in the case of the ec process and in the case of the transition with the positron .
the measured value was consistent with the expected fraction of electron capture events ( decay mode b ) .
the cut @xmath68 kev corresponding to the region containing 99@xmath62 of the events in the ec peak at zero energy was used to define the photonless events . to determine the signal inefficiency ,
mostly due to pileup , @xmath69 events have been collected using a random trigger formed by delaying the fiber coincidence by 16 @xmath40s ( half of the mean time interval between two events ) . for the 80 kev threshold defined above
, this gave an inefficiency of @xmath70 that was consistent with the prediction of the simulation .
this inefficiency was measured at the beginning of the data taking and , conservatively , was not corrected for the reduction of the source intensity during that period . for the fbgo , the broadening of the pedestal due to the contribution of energy deposited in the fiber had to be considered .
the energy of the two 511 kev annihilation photons was localized in two opposite - located crystals of the barrel and then the fbgo pedestal was built .
this resulted in a correction of 1@xmath62 in the efficiency of zero signal detection .
finally , the signal efficiency was estimated to be @xmath71 .
the mean fraction of @xmath9 in the data sample could be evaluated from the decay time curve by fitting the observed distribution to the function @xmath72 ( @xmath73 is the accidental background ) starting from the time @xmath74 ns when @xmath9 was completely thermalized in the target @xcite . after taking into account the estimated difference of efficiency for 2 and 3 gamma detection and the pick - off effect ( measured from the lifetime spectra with the same method as described in ref.@xcite ) ,
the fraction of @xmath9 in the data sample was @xmath75 @xcite .
since in the signal region no zero energy events were observed the upper limit for the branching ratio @xcite is : @xmath76 where @xmath77 is the number of @xmath9 in the selected sample
. figure [ ecalsum ] shows the extrapolation into the region of the zero signal with an exponential .
the integral from 0 to 80 kev of the function obtained from the fit gives an evaluation of the background contribution in this region .
the result is @xmath78 expected events , where the error was evaluated from the uncertainty related to the extrapolation procedure itself .
this experiment can also be used to obtain upper limits on @xmath79 and on @xmath80 . for this purpose
the different probabilities of positrons stopping in the fiber and in the aerogel were calculated with the help of simulations .
several papers were reviewed in ref.@xcite concluding that the probability to form positronium ( 75@xmath62 of o - ps and 25@xmath62 of p - ps ) per positron stopping is about 0.45 in the aerogel and between 0.2 and 0.4 in the fiber . in the aerogel a fraction of 0.05 to 0.1 of the p - ps atoms experience pick - off annihilation in about 2 ns before escaping out of the silica grains in the pores . for o - ps
this probability ranges from 0.28 to 0.45 . in a plastic scintillator ( the fiber )
the corresponding pick - off probabilities are 0.99 to 1 for @xmath9 and 0.05 to 0.1 for p - ps @xcite . from the simulations ,
the fraction of positrons stopping in the fiber is 0.43 and in the aerogel is 0.25 , therefore , the smallest possible fraction of p - ps decays ( including pick - off ) is @xmath81 .
thus , an upper bound for the invisible decay of p - ps can be calculated : @xmath82 where @xmath83 was used and the positrons that do not stop in the fiber or in the aerogel are assumed to annihilate directly .
+ the number of @xmath25 decays can be calculated subtracting from the total number of events the p - ps and o - ps decays , thus , one obtains @xmath84 and an upper limit for the branching ratio of : @xmath85
no event consistent with an invisible decay was found in the large sample of events . using eq . ( 32 ) of ref.@xcite , the bound for particles with a fraction @xmath86 of the electron charge can be plotted as a function of their mass @xmath87 for @xmath88 , as shown in fig .
[ milli]a .
thus , the region of the charge - mass parameter space , which was not excluded directly by the slac results @xcite and the previous search for @xmath1 @xcite , is covered by this experiment ( see fig . [ milli]b ) .
decay into milli - charged particles , excluded with this experiment , b ) comparison of our results ( the dashed region on the plot ) with other experimental ( slac @xcite and previous @xmath1 @xcite ) and astrophysical bounds ( the plot was taken from @xcite).__,title="fig : " ] decay into milli - charged particles , excluded with this experiment , b ) comparison of our results ( the dashed region on the plot ) with other experimental ( slac @xcite and previous @xmath1 @xcite ) and astrophysical bounds ( the plot was taken from @xcite).__,title="fig : " ] the strength of the photon mirror - photon mixing @xmath89 can be extracted from the limit on the @xmath19 with @xcite : @xmath90 by substituting a conservative value of @xmath91 for the collision rate of the @xmath9 against the walls of the aerogel pores @xcite .
@xmath92 is the decay rate of @xmath9 in vacuum and @xmath93 mhz is the contribution to the ortho - para splitting from the one - photon annihilation diagram involving @xmath9 @xcite .
using our result one can estimate the mixing strength to be @xmath3 .
this is close to the bbn limit of @xmath94 @xcite but does not cover all the region of interest suggested by the dama and cresst results @xcite and motivated by gut predictions @xcite and by string theory @xcite ( @xmath95 ) .
in this paper the results of a new search for an invisible decay of @xmath9 were reported .
since no event was found in the energy window [ 0,80 ] kev , an upper limit for the branching ratio was set : @xmath96 improving the best existing bound @xcite by a factor 7 . analyzed in the context of theoretical models
the negative result provides an upper limit on the photon mirror - photon mixing strength @xmath3 and rules out particles with a fraction @xmath4 ( for @xmath97 ) of the electron charge ( milli - charged particles ) .
furthermore , an upper limit on the branching ratios for the process @xmath5 and @xmath6 could be set .
* acknowledgements * + we thank the paul scherrer instititute ( villigen , switzerland ) for lending us the bgo crystals .
we gratefully acknowledge the help of b. eichler and j. neuhausen for the @xmath28na source preparation .
we wish to thank z. berezhiani , r. eichler , s. karshenboim , n.v .
krasnikov , v. matveev and v.a .
rubakov for useful discussions .
we are grateful to n.a .
golubev , l. knecht , g. natterer , j. p. peigneux and m. sauter for their essential help .
we wish to thank m. haguenauer for providing us with samples of scintillating fibers .
this work was supported by the swiss national science foundation and by the french ministery of foreign affairs through an econet program .
we thank the cern for its hospitality .
s.i.blinnikov and m.yu.khlopov on the possible signatures of mirror particles .
yadernaya fizika ( 1982 ) v. 36 , pp .
809 - 811 .
[ english translation : sov.j.nucl.phys .
( 1982 ) v.36 , pp.472 - 474 ] ; s.i.blinnikov and m.yu.khlopov on the possible astronomical effects of " mirror particles .
( 1983 ) , v. 50 , pp .
632 - 639 .
[ english translation : sov .
( 1983 ) v. 27 , pp .
371 - 355 ] .
h. primakoff and f. t. porter , phys . rev . * 89 * , 930 ( 1953 ) ; r. l. intermann , phys . rev . * 188 * , 1543 ( 1969 ) ; j. law and j. l. campbell , nucl . phys .
a * 199 * , 481 ( 1973 ) ; r. l. intermann , nucl .
a * 219 * , 20 ( 1974 ) | the results of a new search for positronium decays into invisible final states are reported .
convincing detection of this decay mode would be a strong evidence for new physics beyond the standard model ( sm ) : for example the existence of extra dimensions , of milli - charged particles , of new light gauge bosons or of mirror particles .
mirror matter could be a relevant dark matter candidate . in this paper the setup and
the results of a new experiment are presented . in a collected sample of about @xmath0
orthopositronium decays , no evidence for invisible decays in an energy window [ 0,80 ] kev was found and an upper limit on the branching ratio of orthopositronium @xmath1 could be set : @xmath2 our results provide a limit on the photon mirror - photon mixing strength @xmath3 and rule out particles lighter than the electron mass with a fraction @xmath4 of the electron charge .
furthermore , upper limits on the branching ratios for the decay of parapositronium @xmath5 and the direct annihilation @xmath6 could be set .
positronium decay , new physics , extradimension , hidden sector |
recent breathtaking advances in our understanding of the mathematical structure of scattering amplitudes in maximally supersymmetric yang - mills theory ( sym ) have so far had the most dramatic impact on tree - level amplitudes or quantities ( such as leading singularities or the planar integrand @xcite ) which are completely determined in terms of tree - level data .
the guiding principle behind these advances is the observation that amplitudes exhibit simplicity which is completely obscured by their traditional feynman diagram expansions .
in contrast the problem of actually evaluating the integrals which appear in multi - loop amplitudes remains extremely difficult in general ( see @xcite for an introduction to modern methods ) .
however data is starting to accumulate which suggests that the final results for amplitudes ( or even individual integrals ) can again be far simpler than one might have guessed on the basis of results obtained from more traditional integration approaches , such as mellin - barnes techniques .
in @xcite goncharov , vergu and the authors introduced to the sym literature a powerful mathematical tool first developed in @xcite , called the symbol of a transcendental function , for analyzing functions of the type which appear in all currently known sym loop amplitudes .
the symbol serves as a sort of motivic roadmap , encoding in a simple way all of the information about a function s discontinuities without the need to introduce any explicit representation of the function in terms of ( generalized ) polylogarithm functions .
indeed one might reasonably say that the symbol encapsulates all of the physically relevant information that one may want to know about any given integral or amplitude .
we view the symbol as stepping stone halfway between an integrand and its integral , and it is our hope that it may help to guide us through the dark jungle of multi - loop amplitudes . the simple formula given in @xcite for the two - loop six - point maximally helicity violating ( mhv ) remainder function was made possible by applying motivic technology to the 17-page analytic expression which had been heroically evaluated by del duca , duhr and smirnov @xcite .
it is clear that we are now desperately in need of some technology which would allow one to directly compute the symbol of any given integral without the need for such heroic effort . since the planar integrand of sym is a rational form , and since the symbol of an integral involves only rational functions , it is reasonable to suppose that there should exist an algebraic operation which would allow one to simply write down the symbol of any given integrand
this would provide a concrete realization of part of the ambitious program suggested in @xcite ( see in particular section 7 ) which aims for a `` correspondence principle '' unifying scattering amplitudes , their motivic avatars , and the combinatorics of feynman diagrams reflected in the coproduct formula .
there too the symbol is not the ultimate goal but serves as just the simplest incarnation of the amplitudes and their motivic structure .
work towards unlocking the structure of general multi - loop integrals is underway , but given the recent flurry of activity @xcite on one - loop hexagon integrals in six dimensions we feel compelled to focus in this brief note on the special case of the one - loop @xmath0-gon integral in @xmath0 dimensions .
the motivic structure of this integral is sufficiently simple that its symbol can be read off directly from its feynman parameterization by utilizing a 1996 theorem on mixed tate motives due to a. goncharov .
these higher dimensional integrals have several applications in physics , for example they appear in dimensionally - regulated one - loop mhv amplitudes @xcite and in certain cases they can be related to higher - loop four - dimensional integrals @xcite . in section
ii we set up some notation .
the main result is presented in section iii , followed by several sample applications in section iv , including a particular three - mass hexagon integral in six dimensions .
note that in this note we work entirely at the level of symbols and do not address the problem of choosing an explicit representative for integrals in terms of ( generalized ) polylogarithm functions .
it is our understanding that efforts to automate this final step are under way @xcite .
in this note we consider the one - loop @xmath0-gon integral in @xmath0 dimensions , specifically @xmath2 considered as a function of @xmath0 points @xmath3 ( with an implied cyclic ordering ) in @xmath0-dimensional minkowski space .
we have chosen to include in this definition the prefactor @xmath4 which serves three intimately related purposes : it ensures that @xmath5 is invariant under dual conformal transformations @xcite , it normalizes the leading singularities of the integrand to @xmath6 , and it ensures that @xmath7 evaluates to a function of well - defined transcendentality @xmath1 .
( we caution the reader that our @xmath8 is close , but not exactly equal , to the quantity called @xmath8 in @xcite . )
results for the @xmath9 integrals have of course been known since antiquity ( see for example @xcite for a comprehensive modern treatment ) , while only very recently the first two cases of @xmath10 have been studied : the massless hexagon in @xcite and the one - mass hexagon in @xcite .
we remind the reader that the integral ( [ eq : fzero ] ) is said to be massless if cyclically adjacent pairs are all null - separated , i.e. @xmath11 for all @xmath12 .
more generally we say that the integral has @xmath13 masses if @xmath13 pairs of cyclically adjacent points are null - separated .
for @xmath14 the integral @xmath5 always converges , while for @xmath9 ( the case of box integrals in four dimensions ) it only converges in the fully massive case .
we will return to the question of divergences in detail below .
the @xmath15 integral is easily performed after introducing feynman parameters to combine the @xmath0 propagators , which leads to @xmath16 one may choose any nontrivial linear combination of the @xmath17 s inside the delta - function @xcite but our simple choice of @xmath18 makes it immediately manifest that ( [ eq : fone ] ) is best thought of as a projective integral .
specifically ( [ eq : fone ] ) is an integral over @xmath19 with homogeneous coordinates @xmath20 $ ] which has been written in the patch @xmath21 .
therefore , the one - loop @xmath0-gon integral in @xmath0 dimensions is the @xmath19 period integral @xmath22 where @xmath23 is the standard holomorphic form @xmath24 and where we now use the notation @xmath25 to emphasize that @xmath7 should be thought of as a function of a quadratic form @xmath26 , whose matrix entries are @xmath27 in our application .
integrals of the type ( [ eq : ftwo ] ) have been studied in the context of mixed tate motives , and belong to a class of objects called tate iterated integrals in @xcite ( see especially section 3 ) .
in particular @xmath25 is a special case of the more general period integrals @xmath28 studied in @xcite .
these are integrals of the type ( [ eq : ftwo ] ) involving any non - degenerate quadric @xmath26 and integrated over a cycle representing a generator of @xmath29 for an arbitrary simplex @xmath30 in general position relative to @xmath26 .
the cohomology group @xmath31 has canonical mixed hodge - tate structure which corresponds to a certain mixed tate motive @xcite . in our particular application
we shall only be interested in quadratic forms @xmath26 with vanishing diagonal entries ( since obviously @xmath32 for all @xmath12 ) .
moreover , the cycle of integration of interest in ( [ eq : ftwo ] ) , inherited from the identification with ( [ eq : fone ] ) , involves the particularly simple simplex @xmath30 whose @xmath0 faces are just the coordinate hyperplanes @xmath33 .
these two simplifications of our particular @xmath25 s compared to the more general @xmath28 s studied in @xcite are not of essential mathematical importance but greatly streamline the bookkeeping involved in applying the results of that paper .
we will now write a simple recursive formula for the symbol @xmath34 of @xmath25 . the recursion expresses @xmath34 as a sum of @xmath35 terms involving @xmath36 , where the notation @xmath37 denotes the @xmath38 matrix obtained by deleting the rows and columns @xmath12 and @xmath39 from the @xmath40 matrix @xmath26 ( i.e. , the quadric @xmath37 is just the intersection @xmath41 ) . in terms of the feynman graph corresponding to the one - loop integral
, going from @xmath26 to @xmath37 amounts to turning the @xmath0-gon into a @xmath42-gon by deleting the two propagators @xmath43 and @xmath44 .
there are @xmath35 different ways of doing this , and each way leads to one term in the recursive sum . to obtain the recursion we apply theorem ( 4.10 ) of @xcite ( preprint version ) , which for us becomes @xmath45 where @xmath46 note that no summation of indices is implied on the right - hand side , which is expressed in terms of the @xmath47 , @xmath48 and @xmath49 entries of the inverse matrix @xmath50 .
let us comment briefly on the geometric content of ( [ eq : rdef ] ) .
for each choice of @xmath12 and @xmath39 , there are precisely two codimension-1 hyperplanes @xmath51 , @xmath52 in @xmath19 which contain the intersection @xmath53 and which are tangent to the quadric @xmath54 .
so for each @xmath12 and @xmath39 we have four natural codimension-1 hyperplanes @xmath55 all of which contain @xmath53 as a common codimension-2 subspace , and the quantity @xmath56 in ( [ eq : rdef ] ) is the cross - ratio naturally associated to this configuration .
it follows immediately from ( [ eq : df ] ) that the symbol @xmath34 satisfies the recursion @xmath57 in what follows we shall only make use of ( [ eq : symbolrecursion ] ) , though of course ( [ eq : df ] ) is a slightly stronger statement which realizes a precise set of differential relations between the infinite class of one - loop @xmath0-gon integrals .
various types of differential relations amongst multi - loop integrals have long been studied ; see @xcite for some recent examples which also relate integrals in different dimensions .
we begin with a couple of almost trivial examples demonstrating the application of the recursion ( [ eq : symbolrecursion ] ) . in general mathematical applications
one may seed the recursion with the result of the @xmath58 integral @xmath59 however in physics applications , where we are always interested in quadrics @xmath26 with zero entries down the diagonal , this ` bubble ' integral is too singular to be of use by itself .
instead it is more effective to seed the recursion with box integrals ( @xmath9 ) . as mentioned above these
are divergent except in the fully massive case .
nevertheless , as we discuss in detail below , it is relatively straightforward to apply this recursion to arbitrary integrals by introducing if necessary a regularization parameter @xmath60 which is ultimately taken to zero .
the finiteness of the integral @xmath25 for @xmath61 is reflected in the fact that the regularization parameter always drops completely out of the symbol @xmath34 after summing together all contributions to ( [ eq : symbolrecursion ] ) . before proceeding let us make an important comment regarding notation . in the previous section we defined the integral @xmath5 for an arbitrary collection of @xmath0 points @xmath62 in minkowski space , without regard to whether or not any cyclically adjacent pairs were null separated . in the remainder of this paper
we shall adopt notation more conventional in the physics literature , whereby we consider a collection of @xmath63 points which are all mutually null , i.e. @xmath64 for all @xmath12 , but only a subset of @xmath0 of these @xmath65 s will be chosen as the @xmath0 arguments to the function @xmath5 .
let us start with the four - mass box integral , which corresponds to the quadric @xmath66{$x_j$ } \falabel(10,16)[]{$x_i$ } \falabel(4,10)[]{$x_l$ } \falabel(10,4)[]{$x_k$ } \end{feynartspicture } } } } \qquad q = \left(\begin{matrix } 0 & x_{ij}^2 & x_{ik}^2 & x_{il}^2 \cr x_{ij}^2 & 0 & x_{jk}^2 & x_{jl}^2 \cr x_{ik}^2 & x_{jk}^2 & 0 & x_{kl}^2 \cr x_{il}^2 & x_{jl}^2 & x_{kl}^2 & 0 \cr \end{matrix } \right)\ ] ] in @xmath67 .
the recursion ( [ eq : symbolrecursion ] ) expresses the symbol of this integral as a sum over six @xmath58 integrals of the type [ eq : fdegreeone ] ) : @xmath68 using ( [ eq : fdegreeone ] ) we find for example @xmath69 where @xmath70 in this case .
evidently the symbol @xmath71 is singular , but this can be regulated by replacing all of the zero entries on the diagonal of @xmath26 with @xmath60 .
then , at the level of the symbol , we can replace @xmath72 here we have freely dropped an overall factor of @xmath73 since such numerical constants drop out of the symbol .
note that we could have introduced the @xmath60 into @xmath74 as well but there is no need as this quantity has a finite limit , shown above , as @xmath75 .
adding up the six contributions ( [ eq : sixterms ] ) we find first of all that @xmath60 completely drops out of the symbol ( an important consistency check ) , and that the result can be assembled into the form @xmath76 which of course agrees precisely with the symbol of the well - known four - mass box function @xmath77 in terms of @xmath78 from a mathematical point of view it is perhaps simplest to restrict one s attention to the generic case of the completely massive @xmath0-gon integral , whose symbol may now be written down directly using the recursion ( [ eq : symbolrecursion ] ) seeded by ( [ eq : fourmass ] ) at @xmath9 .
however in physics we are often interested in integrals with fewer masses .
the application of the recursion ( [ eq : symbolrecursion ] ) in these cases will call into play various divergent box integrals which can be regulated in a manner similar to ( [ eq : regularization ] ) . for example
consider the three - mass box integral with corresponding quadric @xmath66{$x_j$ } \falabel(10,16)[]{$x_i$ } \falabel(4,10)[]{$x_{k+1}$ } \falabel(10,4)[]{$x_k$ } \end{feynartspicture } } } } \qquad q = \left(\begin{matrix } 0 & x_{i , j}^2 & x_{i , k}^2 & x_{i , k+1}^2 \cr x_{i , j}^2 & 0 & x_{j , k}^2 & x_{j , k+1}^2 \cr x_{i , k}^2 & x_{j , k}^2 & 0 & 0 \cr x_{i , k+1}^2 & x_{j , k+1}^2 & 0 & 0 \cr \end{matrix } \right)\,.\ ] ] its symbol may be obtained by setting @xmath79 in ( [ eq : fourmass ] ) and keeping only the leading - order terms for small @xmath60 , which leads to @xmath80 results for all other box integrals may be obtained as further specializations of this formula .
for example the two - mass easy or two - mass hard box functions can be obtained by setting @xmath81 or @xmath82 respectively .
note that if we use ( [ eq : fourmass ] ) as our starting point then it is not necessary to introduce an @xmath60 for each of the diagonal entries of @xmath26 ; in writing ( [ eq : fourmass ] ) all of those divergences have already been properly eliminated .
we only need to introduce an @xmath60 for any zero entry immediately adjacent to the diagonal ( i.e. for any massless vertex of the corresponding diagram ) .
just like we saw in ( [ eq : fourmass ] ) , all dependence on the regularization parameter @xmath60 will necessarily drop out when individually divergent contributions are properly assembled to produce the symbol of any of the finite functions @xmath7 for @xmath61 . in practice
this cancellation can serve as a useful check against calculational errors .
as our first non - trivial ) with the help of mathematica s integrate command ; in our experience it generally gets stuck starting at degree 4 where the classical polylogarithms do not provide a sufficiently large basis of functions . ]
demonstration of the main result let us write the symbol for the three - mass hexagon integral in six dimensions corresponding to the quadric @xmath83{$x_2 $ } \falabel(10.0000 , 16.0000)[]{$x_1 $ } \falabel(4.80385 , 13.0000)[]{$x_8 $ } \falabel(4.80385 , 7.00000)[]{$x_7 $ } \falabel(10.0000 , 4.00000)[]{$x_5 $ } \falabel(15.1962 , 7.00000)[]{$x_4 $ } \end{feynartspicture } } } } \qquad q = \left(\begin{matrix } 0 & 0 & x_{14}^2 & x_{15}^2 & x_{17}^2 & x_{18}^2 \cr 0 & 0 & x_{24}^2 & x_{25}^2 & x_{27}^2 & x_{28}^2 \cr x_{14}^2 & x_{24}^2 & 0 & 0 & x_{47}^2 & x_{48}^2 \cr x_{15}^2 & x_{25}^2 & 0 & 0 & x_{57}^2 & x_{58}^2 \cr x_{17}^2 & x_{27}^2 & x_{47}^2 & x_{57}^2 & 0 & 0 \cr x_{18}^2 & x_{28}^2 & x_{48}^2 & x_{58}^2 & 0 & 0 \cr \end{matrix } \right)\ ] ] in @xmath84 .
here we have chosen for convenience an integral with @xmath85 particles , which manifestly exhibits a @xmath86 symmetry .
we shall for convenience restrict the external kinematics to four dimensions ( i.e. all of the @xmath65 lie within a common four - dimensional subspace of six - dimensional minkowski space ) .
this allows us to express everything in terms of momentum twistors @xcite ; in particular @xmath87 where the constant of proportionality is irrelevant as it cancels out in all conformally invariant quantities .
we make this choice for purposes of notational simplicity only but we emphasize that the result ( [ eq : symbolrecursion ] ) is of course valid for completely general kinematics ) for the octagon or higher since for @xmath88 the determinant @xmath89 is identically zero for any collection of @xmath0 points @xmath62 which lie in a common four - dimensional subspace . ] .
a special feature of the three - mass hexagon ( [ eq : threemass ] ) is that no square roots of momentum twistor invariants appear in its symbol .
this would not be true if any two adjacent vertices of the hexagon were massive .
the recursion ( [ eq : symbolrecursion ] ) expresses the symbol @xmath90 as a sum of 15 terms , but we can use the @xmath86 symmetry of this configuration to cast the result into the form @xmath91 \end{aligned}\ ] ] where @xmath92 is the shift - by - three operator and we use the notation @xmath93 appearing in the first two entries of the symbol are a two - mass easy box @xmath94 and some three - mass boxes @xmath95 , @xmath96 , @xmath97 and @xmath98 , corresponding to the degenerations of the hexagon : ( 465,93)(5,1 ) ( 13,7)(13,13)(0.,)/straight0 ( 13,13)(7,13)(0.,)/straight0 ( 7,13)(7,7)(0.,)/straight0 ( 7,7)(13,7)(0.,)/straight0 ( 13,13)(17.2426 , 17.2426)(0.,)/straight0 ( 13,7)(17.2426 , 2.75736)(0.,)/straight0 ( 7,7)(2.75736 , 2.75736)(0.,)/straight0 ( 7,13)(2.75736 , 17.2426)(0.,)/straight0 ( 16,10)[]@xmath99 ( 10,16)[]@xmath100 ( 4,10)[]@xmath101 ( 10,4)[]@xmath102 ( 13,7)(13,13)(0.,)/straight0 ( 13,13)(7,13)(0.,)/straight0 ( 7,13)(7,7)(0.,)/straight0 ( 7,7)(13,7)(0.,)/straight0 ( 13,13)(17.2426 , 17.2426)(0.,)/straight0 ( 13,7)(17.2426 , 2.75736)(0.,)/straight0 ( 7,7)(2.75736 , 2.75736)(0.,)/straight0 ( 7,13)(2.75736 , 17.2426)(0.,)/straight0 ( 16,10)[]@xmath99 ( 10,16)[]@xmath103 ( 4,10)[]@xmath101 ( 10,4)[]@xmath102 ( 13,7)(13,13)(0.,)/straight0 ( 13,13)(7,13)(0.,)/straight0 ( 7,13)(7,7)(0.,)/straight0 ( 7,7)(13,7)(0.,)/straight0 ( 13,13)(17.2426 , 17.2426)(0.,)/straight0 ( 13,7)(17.2426 , 2.75736)(0.,)/straight0 ( 7,7)(2.75736 , 2.75736)(0.,)/straight0 ( 7,13)(2.75736 , 17.2426)(0.,)/straight0 ( 16,10)[]@xmath100 ( 10,16)[]@xmath103 ( 4,10)[]@xmath101 ( 10,4)[]@xmath102 ( 13,7)(13,13)(0.,)/straight0 ( 13,13)(7,13)(0.,)/straight0 ( 7,13)(7,7)(0.,)/straight0 ( 7,7)(13,7)(0.,)/straight0 ( 13,13)(17.2426 , 17.2426)(0.,)/straight0 ( 13,7)(17.2426 , 2.75736)(0.,)/straight0 ( 7,7)(2.75736 , 2.75736)(0.,)/straight0 ( 7,13)(2.75736 , 17.2426)(0.,)/straight0 ( 16,10)[]@xmath100 ( 10,16)[]@xmath103 ( 4,10)[]@xmath99 ( 10,4)[]@xmath102 ( 13,7)(13,13)(0.,)/straight0 ( 13,13)(7,13)(0.,)/straight0 ( 7,13)(7,7)(0.,)/straight0 ( 7,7)(13,7)(0.,)/straight0 ( 13,13)(17.2426 , 17.2426)(0.,)/straight0 ( 13,7)(17.2426 , 2.75736)(0.,)/straight0 ( 7,7)(2.75736 , 2.75736)(0.,)/straight0 ( 7,13)(2.75736 , 17.2426)(0.,)/straight0 ( 16,10)[]@xmath100 ( 10,16)[]@xmath104 ( 4,10)[]@xmath101 ( 10,4)[]@xmath102 their @xmath60-dependent symbols may be read off from ( [ eq : eighteen ] ) , and it is a nontrivial check of the correctness of ( [ eq : result ] ) that all of the @xmath60 dependence completely drops out at the level of the symbol . it is also straightforward to verify that in the limit @xmath105 , @xmath106 and @xmath107 the result ( [ eq : result ] ) reproduces the symbol of the massless hexagon computed in @xcite . in conclusion
let us emphasize that in contrast to the old mb method , the new mb ( motivic beauty ) technique does not require the evaluation of any integrals .
the only even slightly nontrivial step in obtaining ( [ eq : result ] ) was inverting the matrix @xmath26 shown in ( [ eq : threemass ] ) and simplifying the @xmath56 s from ( [ eq : rdef ] ) to the final form presented .
generalizations of this method to projective integrals of the type which appear in higher loop integrals remain under active investigation . in this note
we have worked entirely at the level of symbols , without addressing the important problem of finding explicit representations for integrals in terms of ( generalized ) polylogarithm functions .
this remains an interesting open problem , especially since our experience with @xcite has shown that analytic expressions for amplitudes can turn out to be much simpler than one might have guessed from their symbols .
we are grateful to s. caron - huot , j. henn , d. skinner , c. vergu and especially n. arkani - hamed and a. goncharov for useful discussions , correspondence , and encouragement .
this work was supported in part by the department of energy under contract de - fg02 - 91er40688 task j oji ( ms ) and task a ( av ) , and the national science foundation under grant nos .
phy05 - 51164 , pecase phy-0643150 ( av ) and advance 0548311 ( av ) .
journal of physics a special issue on `` scattering amplitudes in gauge theories , '' to appear , eds .
r roiban , m. spradlin and a. volovich , with contributions by h. elvang , d. z. freedman and m. kiermaier , arxiv:1012.3401 ; r. britto , arxiv:1012.4493 ; j. m. henn , arxiv:1103.1016 ; z. bern and y. t. huang , arxiv:1103.1869 ; r. m. schabinger , arxiv:1103.2769 ; a. brandhuber , b. spence and g. travaglini , arxiv:1103.3477 ; t. bargheer , n. beisert and f. loebbert , arxiv:1104.0700 ; j. bartels , l. n. lipatov and a. prygarin , arxiv:1104.0816 ; t. adamo , m. bullimore , l. mason and d. skinner , arxiv:1104.2890 ; l. j. dixon , arxiv:1105.0771 ; h. ita and j. drummond . n. arkani - hamed , j. l. bourjaily , f. cachazo and j. trnka , `` local integrals for planar scattering amplitudes , '' arxiv:1012.6032 [ hep - th ] .
d. gaiotto , j. maldacena , a. sever and p. vieira , `` pulling the straps of polygons , '' arxiv:1102.0062 [ hep - th ] .
n. arkani - hamed , j. l. bourjaily , f. cachazo , s. caron - huot and j. trnka , `` the all - loop integrand for scattering amplitudes in planar @xmath108 sym , '' jhep * 1101 * , 041 ( 2011 ) [ arxiv:1008.2958 [ hep - th ] ] . v. smirnov , _ feynman integral calculus _ , springer verlag ( 2006 ) .
a. b. goncharov , m. spradlin , c. vergu and a. volovich , `` classical polylogarithms for amplitudes and wilson loops , '' phys
. rev .
lett . * 105 * , 151605 ( 2010 ) [ arxiv:1006.5703 [ hep - th ] ] .
a. b. goncharov , `` galois symmetries of fundamental groupoids and noncommutative geometry , '' duke math j. * 128 * , 209 ( 2005 ) , [ arxiv : math/0208144 ] .
l. f. alday , `` some analytic results for two - loop scattering amplitudes , '' jhep * 1107 * , 080 ( 2011 ) [ arxiv:1009.1110 [ hep - th ] ] . v. del duca , c. duhr , v. a. smirnov , c. duhr and v. a. smirnov , `` the massless hexagon integral in @xmath109 dimensions , '' phys .
b * 703 * , 363 ( 2011 ) [ arxiv:1104.2781 [ hep - th ] ] .
l. j. dixon , j. m. drummond and j. m. henn , `` the one - loop six - dimensional hexagon integral and its relation to mhv amplitudes in @xmath108 sym , '' jhep * 1106 * , 100 ( 2011 ) [ arxiv:1104.2787 [ hep - th ] ] . v. del duca , c. duhr and v. a. smirnov , `` the one - loop one - mass hexagon integral in @xmath110 dimensions , '' jhep * 1107 * , 064 ( 2011 ) [ arxiv:1105.1333 [ hep - th ] ] .
v. del duca , l. j. dixon , j. m. drummond , c. duhr , j. m. henn and v. a. smirnov , `` the one - loop six - dimensional hexagon integral with three massive corners , '' phys .
d * 84 * , 045017 ( 2011 ) [ arxiv:1105.2011 [ hep - th ] ]
. v. del duca , c. duhr and v. a. smirnov , `` an analytic result for the two - loop hexagon wilson loop in @xmath111 sym , '' jhep * 1003 * , 099 ( 2010 ) [ arxiv:0911.5332 [ hep - ph ] ]
. v. del duca , c. duhr and v. a. smirnov , `` the two - loop hexagon wilson loop in @xmath111 sym , '' jhep * 1005 * , 084 ( 2010 ) [ arxiv:1003.1702 [ hep - th ] ] .
z. bern , l. j. dixon , d. c. dunbar and d. a. kosower , `` one loop selfdual and @xmath108 superyang - mills , '' phys .
b * 394 * , 105 ( 1997 ) [ arxiv : hep - th/9611127 ] . v. del duca , c. duhr ,
e. w. nigel glover and v. a. smirnov , `` the one - loop pentagon to higher orders in epsilon , '' jhep * 1001 * , 042 ( 2010 ) [ arxiv:0905.0097 [ hep - th ] ] .
b. a. kniehl and o. v. tarasov , `` analytic result for the one - loop scalar pentagon integral with massless propagators , '' nucl . phys .
b * 833 * , 298 ( 2010 ) [ arxiv:1001.3848 [ hep - th ] ] .
j. m. drummond , j. m. henn and j. trnka , `` new differential equations for on - shell loop integrals , '' jhep * 1104 * , 083 ( 2011 ) [ arxiv:1010.3679 [ hep - th ] ] . c. anastasiou and a. banfi , `` loop lessons from wilson loops in @xmath108 supersymmetric yang - mills theory , '' jhep * 1102 * , 064 ( 2011 ) [ arxiv:1101.4118 [ hep - th ] ] . c. duhr , h. gangl and j. r. rhodes , `` from polygons and symbols to polylogarithmic functions , '' arxiv:1110.0458 [ math - ph ] .
d. j. broadhurst , `` summation of an infinite series of ladder diagrams , '' phys .
b * 307 * , 132 ( 1993 ) .
j. m. drummond , j. henn , v. a. smirnov and e. sokatchev , `` magic identities for conformal four - point integrals , '' jhep * 0701 * , 064 ( 2007 ) [ arxiv : hep - th/0607160 ] .
a. b. goncharov , `` a simple construction of grassmannian polylogarithms , '' arxiv:0908.2238 [ math.ag ] z. bern , l. j. dixon and d. a. kosower , `` dimensionally regulated pentagon integrals , '' nucl .
b * 412 * , 751 ( 1994 ) [ arxiv : hep - ph/9306240 ] . c. bogner and s. weinzierl , `` periods and feynman integrals , '' j. math .
phys . * 50 * , 042302 ( 2009 ) [ arxiv:0711.4863 [ hep - th ] ] .
a. hodges , `` eliminating spurious poles from gauge - theoretic amplitudes , '' arxiv:0905.1473 [ hep - th ] . | we use a result on mixed tate motives due to goncharov @xcite to show that the symbol of an arbitrary one - loop @xmath0-gon integral in @xmath0 dimensions may be read off directly from its feynman parameterization .
the algorithm proceeds via recursion in @xmath1 seeded by the well - known box integrals in four dimensions . as a simple application of this method we write down the symbol of a three - mass hexagon integral in six dimensions .
brown - het-1612 nsf - kitp-11 - 076 1 cm |
quantum superposition and its resulting interference are arguably the most fundamental effects in quantum mechanics , which leads to the concept of complementarity . in 1927 , bohr@xcite first reviewed this subject , claimed that the wave- and particle - like behaviors of a quantum mechanical object are mutually exclusive in a single experiment and expressed it as complementarity .
so bohr complementarity is often superficially identified with the ` wave - particle duality'@xcite which emphasizes two extreme cases that each experiment must be described in terms of either waves or particles , as explained in the textbook .
theoretical investigations about the intermediate situations in which particle and wave aspects coexist have led to some quantitative statements about wave - particle duality@xcite , some of which are expressed as inequalities @xmath8 or @xmath9 about the complementarity between ` which - way ' ( ww ) information of a particle : the predictability @xmath10 and the distinguishability @xmath11 of the path , and the visibility @xmath12 of interference fringes@xcite .
bohr complementarity is often illustrated by means of interferometers and has been experimentally investigated in one - particle interferometer with individual particles , including photons@xcite , electrons@xcite , neutrons@xcite , atoms@xcite and nuclear spins in a bulk ensemble with nuclear magnetic resonance ( nmr ) techniques@xcite .
complementarity , however , is a more general concept which states that quantum systems possess properties that are equally real but mutually exclusive .
naturally , the techniques of standard one - particle interferometry have been extended to two - particle interferometry by employing correlated two - particle systems .
the two - particle interference phenomenon was first observed by ghosh and mandel@xcite using photon pair produced by parametric down - conversion , where the nonclassical effects associated with einstein - podolsky - rosen ( epr ) states were exhibited . since that
, two - particle interferometry has been intensively studied theoretically and experimentally@xcite .
whereafter , a fact between one- and two - particle interference was noticed by horne and zeilinger@xcite , i.e. , when the two - particle visibility is unity , the one - particle visibility is zero , and vice versa .
the visibilities of the one- and two - particle interference are described as a pair of complementary quantities . a systematic investigation of intermediate cases
was carried out by jaeger et al.@xcite , who showed that in a large family of states @xmath13 an interferometric complementarity holds between one - particle visibility @xmath14 and two - particle visibility @xmath15 : @xmath16 where the upper bound is saturated for a special family of states while most of the cases are limited in the inequality ( 1 ) .
however , the condition of achieving the upper bound has not yet been pointed out .
the complementarity relation of one- and two - photon interference has been experimentally demonstrated in a young s double - slit experiment by abouraddy et al.@xcite . except this experiment
, we have not still found other experimental versions in the intermediate regime .
moreover , inequality ( 1 ) can be strengthened to an equality with a suitable extension of operations on the pair of particles@xcite , but this conclusion is based on the different parameters of experimental configuration , that is , one can not observe such interference fringes to obtain @xmath14 and @xmath15 that satisfy the equality in practice@xcite . then , what is the condition of this complementarity being of the form of an equality @xmath17 ? and what is the reason why the complementarity relation is restricted in the inequality ( 1 ) ? in this paper , we analyze the one- and two - particle interference complementarity in the general case and examine the relation among @xmath18 , @xmath4 and the predication of the path @xmath19 to obtain an equality : @xmath20 for any pure two - particle source , which gives the answers of the two questions above , and , in fact , embodies two interferometric complementarity relations whose relations are discussed .
furthermore , we also clarify the relationships of @xmath21 the distinguishability of the path @xmath22 and the entanglement @xmath7 of the composite system which show an important view that this complementarity can arise due to the correlations . using nmr techniques detailed in our works@xcite
, the one- and two - particle complementarity relation of @xmath23 with @xmath24 has been experimentally tested with the ensemble - averaged spin states of @xmath25c and @xmath26h nuclei in @xmath25c - labelled chloroform @xmath25chcl@xmath27 , including two extreme cases : perfect two - particle interference fringe and no one - particle interference fringe for an entangled state and the opposite for a product state , and an intermediate case of a special family of states , by measuring both single and joint probability densities of spin states .
we briefly review some dinfinitions and two - particle interferometric complementarity in two extreme cases . a schematic arrangement for a two - particle interferometer with variable phase shifters
is given in fig .
the source s emits a pair of particles 1 + 2 , one of which propagates along the path a and/or a@xmath28 , through a variable phase shifter @xmath29 impinging on an ideal symmetric beam splitter bs@xmath30 , and is then registered in either beam @xmath31 or @xmath32 , while on the other side there is the analogous process on the other ( see fig .
the locution `` and/or '' refers in a brief way to a quantum - mechanic superposition , the details of which are specified by the state of particles 1 + 2 .
assuming that the pair of particles 1 + 2 are ideally two - level or spin-1/2 quantum entities , the hilbert space h@xmath30 associated with particle 1 is spanned by vectors @xmath33 and @xmath34 representing states of propagation in the paths a and a@xmath28 , and the hilbert space h@xmath35 associated with particle 2 by analogous vectors @xmath36 and @xmath37 . in the same way , the space of the output states is the subspaces spanned by the vectors @xmath38 and @xmath39 for particle 1 and by @xmath40 and @xmath41 for particle 2 .
according to the standard definition of visibility @xmath42 , the usual definition for single - particle fringe visibility is @xmath43 _
{ \max } -\left [ p\left ( k_{i}\right ) \right ] _ { \min } } { \left [ p\left ( k_{i}\right ) \right ] _ { \max } + \left [ p\left ( k_{i}\right ) \right ]
_ { \min } } , \qquad ( i=1,2),\ ] ] whereas for two - particle fringe visibility the corrected
definition@xcite is used : @xmath44 _ { \max } -\left [ \overline{p}\left ( k_{1}k_{2}\right ) \right ] _ { \min } } { \left [ \overline{p}\left ( k_{1}k_{2}\right ) \right ] _ { \max } + \left [ \overline{p}% \left ( k_{1}k_{2}\right ) \right ]
_ { \min } } , \ ] ] where the `` corrected '' joint probability @xmath45 to assure the reasonableness of the visibility @xmath46@xcite , and @xmath47 and @xmath48 denote the probabilities of single and joint detections , respectively .
these visibilities can be obtained by observing the variations of both single and joint probabilities as functions of the phases @xmath29 and @xmath49 . for simplicity ,
our experiments take the case of @xmath50 and @xmath51 representing the single - particle states with spin - up and spin - down , respectively .
consider two extreme cases@xcite : when particles 1 + 2 is prepared in an entangled state @xmath52 or in a product state @xmath53 \otimes \left [ \frac{1}{\sqrt{2}}\left ( \left| \uparrow \right\rangle _ { 2}+\left| \downarrow \right\rangle _ { 2}\right ) \right ] .\ ] ] the joint action of variable phase shifters and symmetric beam splitters can be described by a unitary operation @xmath54 where @xmath55 and subscript @xmath56 represents particle @xmath56 .
@xmath57 is the unitary mapping from the space of input states of the pair of particles 1 + 2 into the space of output states .
it is easy to show that the joint and single measurement probabilities in the space of output states are : for the @xmath58 state of eq.(4 ) , @xmath59 \\ p\left ( \left| \uparrow \right\rangle _ { 1}\left| \downarrow \right\rangle _ { 2}\right ) = p\left ( \left| \downarrow \right\rangle _ { 1}\left| \uparrow \right\rangle _ { 2}\right ) = \frac{1}{4}\left [ 1-\cos \left ( \phi _ { 1}+\phi _ { 2}\right ) \right ] \\ p\left ( \left| \uparrow \right\rangle _ { 1}\right ) = p\left ( \left| \downarrow \right\rangle _ { 1}\right ) = p\left ( \left| \uparrow \right\rangle _ { 2}\right ) = p\left ( \left| \downarrow \right\rangle _ { 2}\right ) = \frac{1}{2 } \end{array}\ ] ] and for the @xmath60 state of eq.(5 ) , @xmath61 where @xmath62or@xmath63 . obviously , the probabilities of joint measurements in eqs.(8 ) exhibit cosinusoidal modulations via the variable phase @xmath64 which yields two - particle interference fringes for the entangled state @xmath58 , whereas there is no one - particle interference fringes .
conversely , a cosinusoidal dependence of the single probabilities on the variable phase @xmath64 in eqs.(9 ) yields one - particle interference fringes for the product state @xmath65 , but no genuine two - particle fringes , in that the variation of the joint probability @xmath66 results only from the variations of the single probabilities @xmath67 and @xmath68 .
this is the reason for using the `` corrected '' definition for two - particle fringe visibility .
further , eqs.(8 ) shows @xmath69 , a manifestation of quantum nonlocality associated with the entangled state @xmath58 . from the definitions of eqs.(2 ) and ( 3 ) , the theoretical outcomes of the visibilities are found to be @xmath70 for @xmath58 and @xmath71 for @xmath60 .
the phenomenon can also be partially explained via the spirit of the ` which - way ' ( ww ) interferometry experiment to test bohr complementarity : any attempt to gain ww information unavoidably destroys the interference pattern@xcite . for example , when the composite system of particles 1 + 2 is prepared in the entangled state @xmath58 , ww information of one particle is stored in the states of the other through their correlation .
consequently , as far as a single particle ( 1 or 2 ) is concerned , there are full ww information and no one - particle fringes .
however , for the pair of particles 1 + 2 , there is no way to determine whether 1 + 2 takes the composite path @xmath72 or the composite path @xmath73 , and hence perfect two - particle interference pattern displays .
in contrast , if the system is in the product state @xmath74 it is impossible to determine the path of one particle through the other , and thus there is one - particle interference .
now , we turn to the general case from two extreme @xmath75 and @xmath60 .
the most general state of pure two - particle source 1 + 2 that can be formed from the basis of @xmath76 is @xmath77 where @xmath78 where @xmath79 are complex numbers .
as the visibilities are independent of the phases @xmath80 , in our analysis , all of @xmath81 are taken to be real numbers@xcite .
likewise , the unitary operation @xmath82 in eq.(7 ) is applied to @xmath13 and straightforward calculations yield @xmath83 , \\
p\left ( \left| \downarrow \right\rangle _ { 1}\right ) = \frac{1}{2}\left [ 1 - 2\left ( \gamma _ { 1}\gamma _ { 3}+\gamma _ { 2}\gamma _ { 4}\right ) \cos \phi
_ { 1}\right ] , \\
p\left ( \left| \uparrow \right\rangle _ { 2}\right ) = \frac{1}{2}\left [ 1 + 2\left ( \gamma _ { 1}\gamma _ { 2}+\gamma _ { 3}\gamma _ { 4}\right ) \cos \phi
_ { 2}\right ] , \\
p\left ( \left| \downarrow \right\rangle _ { 2}\right ) = \frac{1}{2}\left [ 1 - 2\left ( \gamma _ { 1}\gamma _ { 2}+\gamma _ { 3}\gamma _ { 4}\right ) \cos \phi _
{ 2}\right ] , \end{array}\ ] ] and @xmath84\ ] ] where @xmath85 and the similar expressions for other @xmath86 .
it is readily verified that @xmath87 by using the normalization condition of eq.(11 ) . with the derivative method@xcite ,
the maximal and minimal values of @xmath88 are deduced as : @xmath89 , which can be achieved only when @xmath90 , i.e. , @xmath91 .
hence , on substituting for the maximal and minimal values of these probabilities in eqs.(2 ) and ( 3 ) , one gets @xmath92 and @xmath93 ^{2}\leq 1,\quad \left ( i=1,2\right ) .\ ] ] the complementarity relation is achieved for any pure state of a composite two - particle system . in this type of interference experiments ,
another interesting physical quantity is the ww information of one particle when the other serves as the ww maker .
because the two complementarities between path distinguishability and single - particle visibility and between one- and two - particle visibilities are intimately connected , the one - particle visibility @xmath18 enters in the same way in both of them@xcite . without any measurement ,
the available _ a priori _ ww knowledge of one particle is described by the predictability @xmath19 of the alternatives , which represents man s knowledge before measure ; while the distinguishability @xmath22 denotes the maximal information about the `` path '' that can be extracted from an appropriate measure , which is nature s information about the actual alternative ; usually @xmath94@xcite . for all pure states @xmath95 ,
the complementarity of single particle@xcite @xmath96 holds , while @xmath97 where , by definition , the predictability @xmath19 of the path for particle _ i _ reads@xcite @xmath98 where @xmath99 are the probabilities that particle @xmath56 takes one way @xmath100 or the other @xmath101 .
comparing inequality ( 16 ) with eq.(19 ) , an equality @xmath102 is achieved , which gives the relation among three physical quantities : @xmath103 , @xmath4 and @xmath19 .
in fact , eq.(20 ) covers both of two complementarity relations of inequalities ( 16 ) and ( 18 ) and implies the condition of saturating the upper bound of the two inequalities .
for example , if the state of particle _
i _ is prepared symmetrically ( @xmath104 ) so that @xmath24 , i.e. , there is nothing predictable about the paths in a symmetric one - particle counterpart , the one- and two - particle complementarity will be of the form of an equality . or otherwise , if particle _
i _ is an asymmetric state ( @xmath105 ) , that is , one path is more likely than the other to begin with @xmath106 , the complementarity is strictly limited in the inequality ( 16 ) , because partial ww information @xmath19 restrains full one - particle interference visibility @xmath18 .
as the complementarity is tightly associated with the property of the composite system , it would be natural to examine the relationships between these physical quantities and the entanglement @xmath7 of the system . by calculating the von neumann entropy@xcite
, the entanglement @xmath7 for the pure state @xmath95 in eq.(10 ) goes as @xmath107 as illustrated in fig .
2 , this monotonically increasing relation between the two - particle visibility @xmath4 and the entanglement @xmath7 indicates the visibility of two - particle interference fringes @xmath4 is entirely governed by the entanglement @xmath7 of pure two - particle source : the absence of entanglement will results in zero - visibility two - particle interference fringes ( e.g. , the @xmath60 state in eq.(5 ) , @xmath108 , @xmath109 ) , and the prefect entanglement , the full two - particle interference fringes ( e.g. , the @xmath58 state in eq.(4 ) , @xmath110 , @xmath111 ) . from eq.(21 ) one can not directly make out clear relations between @xmath18 or @xmath22 and @xmath7 . however , from eq.(20 ) and eq.(17 ) , @xmath112 is gained , which means that @xmath22 contains both of the _ a priori _ ww information @xmath19 and the additional information @xmath4 encoded in the entanglement between the object and the ww maker due to the entire determination of @xmath4 by @xmath7 .
since _ a priori _ knowledge @xmath19 only lies on the feature of initial single subsystem rather than that of the composite system , these quantities can be studied with a given _ a priori _ ww knowledge @xmath19 .
for instance , in a relevant symmetric one - particle interferometer @xmath113 , then @xmath114 and @xmath115 , which tie up with the entanglement @xmath7 ( see fig.2 ) : the entanglement @xmath7 is enhanced , more and more nature s ww information is stored in the states of the marker particle through their correlation so that @xmath22 rises whereas the one - particle visibility decreases , as discussed and experimentally testified in ref .
most of one - particle interference experiments@xcite are discussed in this kind of symmetric interferometer . if the interferometer begins with @xmath106 , @xmath18 and @xmath116 still bear the similar dependence on the entanglement @xmath7 , this monotonically decreasing relation for @xmath18 and this monotonically increasing relation for @xmath22 except for the wholly alterative amplitudes based on _ a priori _ knowledge @xmath19 , while @xmath4 does not alter the relation on @xmath7 at all .
consequently , this figure also illuminates an important view that quantum correlation is completely responsible for complementarity . in order to discuss these relations in greater detail , we study some special families of states .
when @xmath117 , the following two situations can be distinguished : \1 ) @xmath118 : @xmath13 corresponds to any product state ( @xmath108 ) , then @xmath119 . when @xmath24 ( i.e. , @xmath120 ) , we recover the @xmath60 state of eq.(5 )
( the sign makes no difference on these physical quantities ) , @xmath121 .
when @xmath106 , the one - particle interference visibility is limited by @xmath122 .
\2 ) @xmath123 or @xmath124 : @xmath125 or @xmath126 , which are entangled .
when @xmath24 , ( i.e. , @xmath127 or @xmath128 ) , we retrieve the @xmath58 state of eq.(4 ) or the similar state with @xmath129 and @xmath130 interchanged which are the maximal entangled states ( @xmath110 ) ; in either case @xmath131 and @xmath132 . when @xmath133 , ( @xmath134 or @xmath135 ) , the @xmath95 state is partially entangled ( @xmath136 ) , it then follows @xmath137 which is limited by @xmath138 in the intermediate regime , but @xmath131 and @xmath139 , the path can still be completely determined by a suitable measurement . when @xmath140 ( one of two @xmath81 in @xmath13 to be zero ) , the @xmath95 state again recovers a product state ( @xmath108 ) , @xmath119 and @xmath131 .
the cases discussed above are symmetric for particle 1 and particle 2 ( @xmath141 ) .
we continue on investigating a special asymmetric case ( @xmath142 ) : \3 ) when @xmath95 is of this form @xmath143 , \ ] ] in this case , @xmath144 , @xmath145 .
simple substitutions produce @xmath146 , @xmath147 and @xmath148 , which result in @xmath149 whereas @xmath150 . however , whether for particle 1 or for particle 2 , @xmath151 is satisfied .
if @xmath152 , we again turn to the state with @xmath153 : @xmath154 + \frac{1}{\sqrt{2}}\sin \alpha \left [ \left| \uparrow \right\rangle _ { 1}\left| \downarrow \right\rangle _ { 2}+\left| \downarrow \right\rangle _ { 1}\left| \uparrow \right\rangle _ { 2}\right ] .\ ] ] which was discussed in ref.@xcite : @xmath155 and @xmath156 which make @xmath157 .
in our experiments , the chosen quantum system is @xmath25c - labeled chloroform @xmath25chcl@xmath27 ( cambridge isotope laboratories , inc . )
molecules with the hydrogen nuclei ( @xmath26h ) for particle 1 and the carbon nuclei ( @xmath25c ) for particle 2 .
the spin - spin coupling constant @xmath158 between @xmath25c and @xmath159h is 214.95 hz . the relaxation times were measured to be @xmath160 @xmath161 and @xmath162 @xmath163 for the proton , and @xmath164 @xmath163 and @xmath165 @xmath163 for carbon nuclei .
experiments were performed on a brukerarx500 spectrometer with a probe tuned at 125.77mhz for @xmath25c and at 500.13mhz for @xmath26h by using a conventional liquid - state nmr techniques .
the quantum ensemble is firstly prepared in a pseudo - pure state @xmath166 from thermal equilibrium by applying line - selective pulses with appropriate frequencies and rotation angles as well as a consequent magnetic gradient pulse@xcite .
different two - particle source which corresponds to any @xmath95 state in eq.(10 ) can be prepared from the @xmath167 state by appropriate operatations , e.g. , the entangled state @xmath58 of eq.(4 ) , the product state @xmath168 of eq ( 5 ) and the special family of states @xmath169 of eq.(23 ) by applying , respectively , the following nmr pulse sequences : @xmath170 @xmath171 and @xmath172 to be read from left to right . here
@xmath173 denotes an @xmath174 rotation about @xmath175 axis on particle @xmath56 and so forth , and @xmath176 represents a time evolution of @xmath177 under the scalar coupling between spins 1 and 2 .
then the transformation @xmath178 to achieve the operations of phase shifters and beam splitters was realized by the nmr pulse sequence @xmath179 with @xmath180 and @xmath181 .
@xmath178 with a set of appropriate values @xmath182 and @xmath183 was repeatedly applied in experiments to vary the value of @xmath64 from @xmath184 to @xmath185 .
finally , the probabilities of single and joint measurements were obtained by quantum state tomography@xcite to reconstruct the diagonal elements of the output density matrices , which can be completed by employing reading - out pulses @xmath186 and @xmath187 after a gradient pulse field to record @xmath26h and @xmath188c spectra , respectively .
the quantitative measurements of the one- and two - particle interference visibilities were gained by monitoring the variations of these probabilities versus @xmath64 . in our experiments ,
the complementarity test is restricted in the cases of @xmath24 , that is , to verify @xmath157 . for two extreme cases
, we simultaneously varied the values of @xmath29 and @xmath49 with the respective increment of @xmath189 . applying the procedure stated above
, the experimental results are shown in fig .
as expected , for the entangled state @xmath58 , two - particle interference fringes are displayed ( fig .
3(b ) ) but almost no one - particle interference fringes ( fig .
3(a ) ) . and
the opposite situation for the product state @xmath60 ( fig .
3(c ) and ( d ) ) . from these experimental data points in fig .
3 , the measured values of the visibilities were extracted : @xmath190 @xmath191 , @xmath192 for the entangled state @xmath58 and @xmath193 @xmath194 , @xmath195 for the product state @xmath60 . by contrast with the theoretical expectations
( see sec ii ) , more experimental errors were introduced in the case of @xmath58 inasmuch as the preparation of the entangled state @xmath58 was more complicated than that of the product state @xmath60 .
the interferometric complementarity of the special family of states @xmath196 in the intermediate regime was testified with a similar procedure . as @xmath197 is a function of @xmath29 and @xmath49 whose extrema @xmath198 were found at @xmath199 in the theoretical calculations
, we scanned one of @xmath64 while fixing the other into @xmath200 for a given state @xmath201 , then repeated experiment with interchanging them , instead of simultaneously scanning @xmath29 and @xmath49 .
the procedure was repeated for different @xmath202 . for the convenience of experiments , we changed the values of @xmath202 from @xmath203 to @xmath204 with the increment of @xmath189 . from the desirable interference fringes shown from variations of the normalized populations versus @xmath64
, the visibilities were obtained .
the measured and theoretical visibilities of one- and two - particle interference @xmath205 @xmath206 and @xmath207 are plotted in fig .
4(a ) , together with entanglement @xmath208 of the two - particle system . these experimental results are in good agreement with the theoretical expectations .
this figure also shows the relationship between the visibilities and entanglement , i.e. , @xmath208 varies synchronously with @xmath207 and has the opposite variation trend with @xmath209 , consistent with the theoretical analyses in sec.iii . the experimental data of the complementarity relations for @xmath210 are also depicted in fig .
the initial state of two - particle source is a symmetric state with no _ a priori _ ww knowledge of particle 1 and 2 ( @xmath153 ) , so that all data in fig .
4(b ) should be unity . in these experiments ,
the estimated errors are less than @xmath211 due to the inhomogeneity of the rf field and static magnetic field , imperfect calibration of rf pulses and signal decay during the experiments .
if we take into account the imperfections of the experiments , the measured data in our nmr experiments agree well with theory .
for a general pure two - particle source , we derive the interferometric complementarity between one- and two - particle interference visibilities and obtain an equality among the one - particle interference visibility @xmath18 , two - particle interference visibility @xmath4 and the predication @xmath19 of the path of single particle : @xmath151 .
the equality not only embodies two complementarity relations of @xmath212 and @xmath213 , but also implies the conditions of their saturating the upper bound of them , e.g. , for the one- and two - particle interference , the duality of @xmath23 holds only when the interferometer involves a symmetric single - particle counterpart ( i.e. , @xmath24 ) .
in fact , the equality is also a demenstration of the relationships among one - paritcle wave - like ( @xmath18 ) , particle - like ( @xmath19 ) and two - particle wave - like ( @xmath46 ) attributes .
in addition , we also manifest the role of entanglement in this type of interference experiments , which reveals whether in a symmetric way or in an asymmetric one , the greater entanglement is the quantum composite system , the lower the visibility of two - particle interference fringes whereas the higher the visibility of two - particle interference fringes .
meanwhile , we also consider the relation between two interferometric complementarities .
the one- and two - particle interference complementarity has been testified for a kind of special symmetric interferometer ( @xmath153 ) in a spin nmr ensemble , including two extreme cases and a special family of states .
the experimental results showed the theoretical predictions of quantum mechanics within the experimental errors .
though the test was performed over interatomic distances , the experimental scheme originates from quantum version and their dynamical evolution is quantum mechanical .
therefore , these experiments also indicate that non - classical properties associated with entanglement can be displayed in nmr ensemble as long as the initial state is well prepared .
this work is supported by the national natural science foundation of china ( grant no .
10274093 ) and the national fundamental research program ( 2001cb309300 ) .
we also thank hanzheng yuan , zhi ren , and xu zhang for help in the course of experiments . c. o. alley and y. h. shih , in _ proceedings of the second international symposium on foundations of quantum mechanics in light of new technology _ ,
edited by m. namiki _
et al_. ( physical society of japan , tokyo , 1986 ) , p. 47 ; y. h. shih and c. o. alley , phys .
lett . * 61 * , 2921 ( 1988 ) ; c. k. hong , z. y. ou , and l. mandel , phys .
59 * , 2044 ( 1987 ) ; j. d. franson , _ ibid_. * 64 * , 2495 ( 1990 ) ; p. g. kwiat , w. a. vereka , c. k. hong , h. nathel , and r. y. chiao , phys .
a * 41 * , 2910 ( 1990 ) ; m. horne , a. shimony , and a. zeilinger , in _ quantum coherence _ , edited by j. anandan ( world scientific , singapore , 1990 ) , p. 356 .
2 the entanglement @xmath7 versus visibility @xmath4 for pure two - particle source ( denoted by the solid line ) , visibility @xmath18 and distinguishability @xmath22 of a single particle in different _ a priori _ ww knowledge @xmath19 : the dotted line of @xmath18 and the solid line of @xmath217 for @xmath24 ; the dashed line of @xmath18 and the dashdotted line of @xmath115 for @xmath218 . fig .
3 the single and `` corrected '' joint probabilities detected in the one- and two - particle interference , ( a ) and ( b ) for the entangled state @xmath219 , ( c ) and ( d ) for the product state @xmath65 . in (
a ) and ( c ) , data points @xmath220 @xmath221 @xmath222 and @xmath223 denote the single probabilities @xmath224 and @xmath225 , respectively ; in ( b ) and ( d ) , @xmath226 @xmath227 @xmath228 and @xmath229 denote the `` corrected '' joint probabilities @xmath230 @xmath231 @xmath232 and @xmath233 , respectively .
theoretical curves are depicted with the solid lines .
4 the visibilities and the complementarity relation of a special family of states @xmath234 as a function of the angle @xmath235 ( a ) data points @xmath226 @xmath229 and @xmath236 denote , respectively , the visibilities of one - particle interference @xmath237 @xmath238 and two - particle interference @xmath4 .
the solid lines are the theoretical expectations and the dotted line denotes the theoretical curve of entanglement @xmath7 .
( b ) experimental test of the complementarity relation @xmath239 ( denoted by data ponits @xmath228 ) and @xmath240 ( denoted by data ponits @xmath241 ) are plotted as a function of @xmath202 .
the solid line represents the theoretical expectations . | we analyze an interferometric complementarity between one- and two - particle interference in the general case : @xmath0 @xmath1 , @xmath2 , and further examine the relation among one - particle interference visibility @xmath3 , two - particle interference visibility @xmath4 and the predication @xmath5 of the path of a single particle .
an equality @xmath6 @xmath1 , @xmath2 is achieved for any pure two - particle source , which implies the condition of the complementarity relation to reach the upper bound and its relation to another interferometric complementarity between path information and interference visibility of a single particle .
meanwhile , the relationships of the complementarities and the entanglement @xmath7 of the composite system are also investigated .
using nuclear magnetic resonance techniques , the two - particle interferometric complementarity was experimentally tested with the ensemble - averaged spin states , including two extreme cases and an intermediate case . |
simple fluids are dominated by excluded volume effects , which , within lattice gas models , are accounted for by the single occupancy constraint , whereby each site on a lattice can be occupied by at most one molecule .
effective interactions between macromolecules or self - assembled aggregates in complex fluids , on the other hand , can be `` soft '' , i.e. lack an impenetrable core . a good example is the effective pair potential between the centres of mass ( cm ) of interacting polymer coils , obtained by taking statistical averages over monomer conformations for fixed distances @xmath0 between the cms @xcite .
recent extensive simulations of self - avoiding walk polymers carried out over a wide range of concentrations show that the repulsive state - dependent effective cm pair potential is of roughly gaussian shape , of width governed by the polymer radius of gyration , and of maximum amplitude @xmath1 @xcite ; this behaviour reflects the fractal nature of polymers in good solvent , leading to a low entropic cost at full overlap .
several other complex fluids have been shown to exhibit ultrasoft coarse - grained interactions , e. g. star polymers @xcite or effective particles considered within dissipative particle dynamics @xcite .
the penetrability of the corresponding `` gaussian core '' ( gc ) model , @xmath2 , leads to interesting phase behaviour at low temperatures ( @xmath3 ) , @xcite , but under conditions relevant for polymer solutions ( @xmath4 ) , the model behaves like a `` mean field '' fluid @xcite .
however binary gaussian core mixtures , characterised by different energy scales @xmath5 for the three types of pairs , lead to phase separation for moderate couplings @xcite .
this demixing , which occurs for purely repulsive , penetrable interactions , is of a very different nature than the usual phase separation of incompatible fluids , which is generally driven by the long - range attractive intermolecular forces . in this paper
we examine the simplest lattice gas representation of penetrable particles , in an effort to gain further insight into this novel class of phase transitions .
the penetrable nature of the effective interaction is reflected by allowing of each lattice site .
the simplest version of the model involves only on - site interactions of particles of the two species , with different energy penalties for the different types of pairs .
while a steepest descent estimate of the grand partition function predicts phase separation , it will be shown that a more accurate treatment reveals the spurious nature of this transition , as expected for an effectively zero - dimensional system .
the addition of nearest neighbour interactions leads to a genuine demixing transition , similar to that predicted for continuous versions of soft core mixtures examined earlier within the random phase approximation ( rpa ) @xcite .
we explore this behaviour by mean field theories and gibbs ensemble monte carlo simulations .
we also study the topology of the energy landscape .
consider a @xmath6-dimensional lattice of @xmath7 sites and coordination number @xmath8 , which can accommodate particles of two species .
the occupancy of each site is characterised by the two - component vector of integer occupation numbers @xmath9 .
particles interact only when they are on the same site or on nearest neighbour ( n.n . ) sites .
the on - site and off - site couplings are characterised by @xmath10 matrices of interaction energies @xmath5 and @xmath11 . in terms of the occupation numbers ,
the energy of the system reads : @xmath12 where the last summation is over n.n .
sites only . for given chemical potentials @xmath13 of the two species ,
the grand partition function is : @xmath14\right\}.\end{gathered}\ ] ] the first sum is over all possible distributions of occupation numbers of the @xmath7 sites , and the effective chemical potentials @xmath15 , @xmath16 have been introduced .
the main task of this paper is to approximately evaluate the grand partition function as a function of the temperature @xmath17 ( @xmath18 ) and of the chemical potentials @xmath19 and @xmath20 of the two species . using the standard rules of statistical mechanics
, this will lead directly to the pressure @xmath21 and to the mean occupation numbers @xmath22 and @xmath23 , and hence to the composition of the mixture .
phase separation will be signalled by a discontinuous change of the mean occupation number for well - defined values of @xmath24 , @xmath19 , and @xmath20 .
we shall first consider the simplest version of the model which only involves on - site interactions ( i.e. @xmath25 ) , before examining the case including nearest neighbour interactions .
in the absence of couplings between particles on different sites , the grand partition function , with @xmath25 , factorises into @xmath7 identical single site functions @xmath26 : @xmath27^l = \xi_{\text{site}}^l.\ ] ] the single site partition sum is not tractable analytically . for sufficiently large chemical potentials , implying mean occupation numbers @xmath28 ,
the latter may be treated as continuous variables , and the sums in eq . replaced by integrals : @xmath29 the integrand is sharply peaked around the most probable occupation numbers @xmath30 . adopting the standard steepest descent method
, we expand the logarithm of the integrand around @xmath31 to second order in @xmath32 , i.e. : @xmath33 where @xmath34 is the digamma function , and the covariance matrix @xmath35 is given by : @xmath36 with @xmath37 ( trigamma function ) .
the location of the maximum follows from the extremum condition @xmath38 which provides the relation between @xmath39 and @xmath31 . inserting the expansion into the integrand in eq . , and extending the integral to negative occupation numbers ( which is again justified provided @xmath40 ) , the resulting gaussian integral is easily evaluated and leads directly to the following equation of state : @xmath41 where @xmath42 is the volume per site . the virial pressure in eq . ,
derived on the basis of steepest descent , is thermodynamically inconsistent with the relation between chemical potentials and mean occupation numbers ( identical to their most probable values ) .
if the latter is integrated , an expression for the pressure follows which coincides with eq . , but without the logarithmic fluctuation term .
the situation is reminiscent of the rpa treatment of continuous penetrable ( e.g. gaussian ) core mixtures , where the virial and compressibility routes lead to equations of state similar in structure to eq . with ( virial ) and without ( compressibility ) the last term @xcite .
the two equations of state become identical in the high density limit pointing to the asymptotic `` mean - field fluid '' nature of systems of penetrable particles . in view of the above analogy
, it is not surprising that the equation of state leads to phase separation between sites with different mean occupation numbers when @xmath43 .
this predicted phase separation is obviously spurious , because of the effectively 0-dimensional nature of the model when restriction is made to on - site couplings .
the above steepest descent treatment has two obvious shortcomings .
the first deficiency is the replacement of the discrete sums in eq . by an integral , as well as the extension of the lower bounds in the integral to @xmath44 ;
these approximations are expected to break down for low mean occupation numbers @xmath45 , or equivalently for low chemical potentials @xmath46 .
this deficiency can be easily removed by reverting from the integrals to discrete sums over an interval of values of @xmath31 in the vicinity of the most probable values @xmath31 , and retaining the continuous integrals outside this interval . in the one - component case ,
the resulting single - site partition function reads : @xmath47.\end{gathered}\ ] ] where @xmath48 and @xmath49 , and the ceil function @xmath50 rounds to the nearest higher integer .
the steepest descent result is recovered for @xmath51 and sufficiently high @xmath52 , such that @xmath53 .
results for the equation of state are shown in fig .
[ fig:1 ] .
is seen to yield surprisingly accurate results , even at relatively low densities ( @xmath54 ) , and only breaks down in the @xmath55 limit .
a more fundamental shortcoming of the steepest descent method is that eq .
in fact predicts two maxima @xmath31 , separated by a saddle point . the situation is pictured in fig .
[ fig:2 ] .
the steepest descent method includes only the contribution of the highest ( global ) maximum of the integrand to the partition function .
this approximation leads to a discontinuous jump of the composition vector @xmath31 when the two maxima become equal , and hence to a first order phase transition . in reality ,
when the contributions of the two maxima of the configuration integral are properly included , the mean occupation numbers @xmath56 and @xmath57 vary continuously with the chemical potentials of the two species , as illustrated in fig .
[ fig:3 ] , according to expectation . as a function of the mean occupation numbers
@xmath58 for the lattice system with on site interaction @xmath59 , @xmath60 and no nearest neighbour interaction , at the chemical potentials @xmath61 , @xmath62 .
the appearance of a secondary maximum is clearly seen.,scaledwidth=90.0% ]
we now consider the model defined by eqns . and and including n.n .
interactions between particles of both species , i.e. @xmath63 .
the partition function no longer factorises into single site sums . proceeding as in section [ sec : case - site - inter ]
, we replace the discrete sums over occupation numbers by integrals and approximate the integrand by gaussians centred on each of the local maxima .
the latter are determined by a generalisation of eq .
, namely @xmath64 in the limit @xmath65 , the @xmath66 on different sites are independent and can each take two values , corresponding to the two maxima found in section [ sec : case - site - inter ] . when the n.n .
interactions are switched on , the @xmath66 are coupled so that a complex energy landscape emerges in @xmath67-dimensional occupation - number space , featuring a rapidly increasing number of maxima of the boltzmann factor . by continuity one can still expect @xmath68 solutions to eqns . in the most general case , at least for sufficiently weak coupling .
the approximate grand partition function now reads @xmath69 \\
\times \int d\delta{\boldsymbol{\rm n}}_1 \ldots \int d \delta { \boldsymbol{\rm n}}_l \exp\left(- \left[\frac{1}{2 } \sum_i \delta { \boldsymbol{\rm n}}_i \cdot { \boldsymbol{\rm \sigma}}_i \cdot \delta { \boldsymbol{\rm n}}_i + \sum_{\langle ij \rangle } \beta \delta { \boldsymbol{\rm n}}_i \cdot { \boldsymbol{\rm \eta } } \cdot \delta { \boldsymbol{\rm n}}_j \right ] \right)\end{gathered}\ ] ] where the covariance matrices @xmath70 are obvious generalisations of eq . , with @xmath45 replaced by @xmath71 .
the grand partition function is thus expressed as a sum over all `` configurations '' @xmath72 which maximise the integrand in the continuous representation generalising eq . ,
i.e.which satisfy the set of eqns . .
the gaussian fluctuation integrals in the second factor depend on the configuration @xmath72 through the covariance matrices @xmath70 . for given @xmath73
the coupled gaussian integrals can be calculated explicitly by a standard normal mode analysis .
the locations of the local maxima @xmath74 depend on the occupation numbers of the neighbouring cells @xmath75 via .
were that not the case , then the partition function would exactly factor into a product of a partition sum that is isomorphic to that of an ising model ( first factor in ) and the partition sum of a gaussian model .
the phase behaviour of a very similar system has been investigated in ref .
the model we examine can thus only be mapped onto an ising ( plus gaussian ) model in the limit of weak nearest neighbour interactions @xmath76 . in order to gain further insight
, we consider the `` ground state '' , corresponding to the low temperature or strong coupling limit . in that limit
all @xmath66 corresponding to the lowest minimum on the total energy surface are expected to be equal ( @xmath77 ) , and determined by a single extremum condition : @xmath78 where @xmath8 is the coordination number of the lattice . in the low temperature limit
only the contribution of this one configuration to the sum in equation is taken into account . the second factor in this equation
then becomes invariant with respect to translations , so that the corresponding integral is best evaluated in fourier space .
we therefore introduce the fourier components of the density fluctuations @xmath79 where @xmath80 denotes the position of the lattice site @xmath81 in units of the lattice constant . assuming born - von karman boundary conditions
the wave vector @xmath82 can only assume discrete values .
the total interaction energy can be written in fourier space as @xmath83 where the matrix @xmath84 is given by @xmath85 for a simple cubic lattice we obtain @xmath86 the fourier modes of the density fluctuations thus decouple , so that the multiple integral in equation can be evaluated in fourier space as a product of gaussian integrals @xmath87 \right)\\ = l^l \int d\delta\tilde{{\boldsymbol{\rm n}}}_{{\boldsymbol{\rm k}}_1 } \ldots \int d\delta\tilde{{\boldsymbol{\rm n}}}_{{\boldsymbol{\rm k}}_l } \exp \left(- \frac{l}{2 } \sum_{{\boldsymbol{\rm k } } } \delta \tilde{{\boldsymbol{\rm n}}}_{{\boldsymbol{\rm -k } } } \cdot { \boldsymbol{\rm d}}({\boldsymbol{\rm k } } ) \cdot \delta \tilde{{\boldsymbol{\rm n}}}_{{\boldsymbol{\rm k } } } , \right ) = ( 2 \pi)^l \prod_{{\boldsymbol{\rm k } } } \frac{1}{\sqrt{{\lvert { \boldsymbol{\rm d}}({\boldsymbol{\rm k } } ) \rvert}}}\end{gathered}\ ] ] the grand partition function then reduces to : @xmath88^l\\ \times ( 2 \pi)^l \exp\left\{- \frac{l}{2 } \int \frac{d{\boldsymbol{\rm k}}}{(2\pi)^3 } \ln\left({\lvert { \boldsymbol{\rm d}}({\boldsymbol{\rm k } } ) \rvert}\right)\right\},\end{gathered}\ ] ] leading to the equation of state @xmath89 + \frac{1}{2 } { \boldsymbol{\rm n}}^ * \cdot ( { \boldsymbol{\rm \epsilon}}+ q { \boldsymbol{\rm \eta } } ) \cdot { \boldsymbol{\rm n}}^ * + \ln ( 2 \pi ) - \frac{1}{2 } \int \frac{d{\boldsymbol{\rm k}}}{(2 \pi)^3 } \ln\left({\lvert { \boldsymbol{\rm d}}({\boldsymbol{\rm k } } ) \rvert}\right).\ ] ] in the limit of large occupation numbers ( @xmath40 ) and vanishing nearest neighbour interaction ( @xmath25 ) we recover equation . as in the case without nearest neighbour interaction , there is a thermodynamic inconsistency between the e.o.s . and the dependence of the chemical potentials on the occupation numbers .
integrating the latter would again lead to an e.o.s .
similar in structure to equation , albeit without the final two logarithmic terms .
the virial equation of state is shown in figure [ fig:4 ] .
for high nearest neighbour interaction ( @xmath90 no longer being positive definite ) the fluctuation term diverges at @xmath91 .
this instability indicates a transition of the ground state from the homogeneous phase to a microfluid phase where high and low occupation numbers alternate in a checkered fashion . to show that the microphase exists also for the one - component fluid ( for @xmath92 ) a canonical monte carlo simulation was run on a @xmath93 cubic lattice ( @xmath94 ) with the parameters @xmath95 , @xmath96 .
initially every lattice site was filled with 20 particles .
after equilibration a histogram of the occupation numbers occuring over the next 1 000 000 steps was recorded .
the result is shown in figure [ fig:4a ] .
it is clearly seen that the initial peak at @xmath97 has vanished and a bimodal distribiution around @xmath98 and @xmath99 has developed .
the development of the microphase is easy to understand when we estimate and compare the energetic and entropic costs of the microphase with ( mean ) occupation numbers @xmath100 and @xmath101 with those of a homogeneous phase with average density @xmath102 .
the entropy of the microphase per site is @xmath103 / 2 $ ] , compared to that of the homogeneous phase @xmath104 .
the difference between the two is simply the mixing entropy of the ideal gas , thereby favouring the homogeneous density distribution . on the other hand
the internal energy of the microphase per site is @xmath105 , compared to that of the homogeneous phase @xmath106 .
the difference between the two internal energies is @xmath107 . for strong nearest neighbour interaction
the microphase is therefore energetically favoured .
the nature of the transition from the homogeneous state to the micro state will be investigated elsewhere .
the explicit knowledge of the fourier components of the density fluctuations allows us to evaluate the structure factor of the fluid . for @xmath108
we obtain @xmath109 from the particle - particle structure factors @xmath110 the bhatia - thornton structure factors @xcite can be extracted .
these are compared with data from a canonical monte carlo simulation in figure [ fig:5 ] for the phase point @xmath111 , @xmath112 .
the agreement is reasonable , although there are still fairly large statistical errors , a problem commonly encountered in direct calculations of @xmath113 ( as opposed to fourier transforming @xmath114 ) .
the error bars on the simulation results ( not shown ) are rather large . in the calculations above we neglected all contributions from configurations in the partition sum other than the ( presumed ) ground state , where all occupation numbers are the same .
this is of course only justified when the other configurations have much higher energies . to check this hypothesis
, we explore the energy landscape of small lattice systems with @xmath115 . for a @xmath116 lattice systems
, one can explicitly enumerate all local energy minima . in the case of no nearest neighbour interactions
the boltzmann factor has @xmath117 degenerate local maxima . slowly switching on nearest neighbour interactions by setting @xmath118 and varying @xmath119
, one can numerically follow the local maxima of the boltzmann factor . due to the high symmetry , many local minima stay degenerate even at finite nearest neighbour interaction .
when increasing @xmath76 , some local minima turn into saddle points and vanish , as illustrated in figure [ fig:6 ] .
the remaining excited energy states depending on @xmath119 are shown in the same figure .
the increasing gap between the ground state and the excited states is clearly seen . for larger lattices
the explicit enumeration of all local minima is no longer feasible because of their large number .
we therefore resort to a simple metropolis algorithm employed in numerical optimization problems @xcite . instead of systematically searching the @xmath67 dimensional occupation number state , we generate trial configurations randomly from the previously found local minimum , analogous to the random particle displacements used in monte carlo simulations .
after each random step in the space of occupation numbers a local minimization of the energy is performed .
the random step is then accepted or rejected based on the energy difference of the two local minima .
effectively the very rugged energy landscape is therefore replaced with a locally constant function , thereby eliminating the numerical problems one usually encounters when solving global optimization problems . in figure [ fig:7 ] the resulting local energy minima
are shown together with the ( presumed ) homogeneous ground state .
all the minima lie on or slightly above the presumed ground state , thereby supporting our hypothesis that the ground state is the homogeneous state . in figure [ fig:8 ] the difference between the local minima and the ground state energy
is plotted .
even though some local minima are not found during the search , one can clearly see a trend of an increasing gap between the ground state and the first excited state when the nearest neighbour interaction increases . for a nearest neighour interaction that equals
10 percent of the on - site interaction one finds a gap of approximately @xmath120 .
even stronger nearest neighbour interaction should further widen the gap , leading to a higher accuracy of the ground state approximation .
to test the validity of the ground state approximation , gibbs ensemble monte carlo simulations @xcite of the model have been performed on a @xmath121 lattice . after an equilibration time of 1 000 000 steps two lattices were allowed to exchange particles and random moves within each lattice were attempted according to the standard metropolis algorithm .
the simulations were restricted to symmetric interactions with @xmath122 and @xmath123 . since in these systems
the equation of state must be invariant with respect to relabelling of the species ( or equivalently mirroring the concentration @xmath124 ) , volume changes of the lattice were not necessary , i. e. the two lattice systems were automatically under equal pressure .
two typical phase diagrams are shown in figure [ fig:9 ] and [ fig:10 ] .
it is seen that the ground state approximation predicts the correct shape of the phase diagram , albeit underestimates the densities of the coexistencing phases .
the discrepancy becomes larger for larger interaction energies , when the contributions of the excited states are no longer negligible , and the ground state approximation becomes less accurate .
the lattice model introduced in this paper , which allows for multiple occupancy of lattice sites by one or two species , with positive energy penalties , and includes repulsive nearest neighbour interactions , leads to intersting phase behaviour . in the one - component version ,
a steepest descent analysis allowing for gaussian fluctuations , predicts microphase separation onto two interpenetrating sub - lattices with different mean occupation numbers ( `` checkered '' phase ) when the ratio @xmath125 of nearest - neighbour to on - site couplings exceeds a threshold .
this prediction is confirmed by monte carlo simulations .
the two - component extension of the model leads to a macroscopic demixing transition into phases of different concentrations of the two species .
the steepest descent analysis predicts such a phase separation , even in the absence of nearest - neighbour interactions , but this is shown to be due to the expected break - down of the approximation . in the presence of nearest - neighbour interactions , the steepest - descent analysis corresponds to a `` ground state '' approximation . a numerical search of local minima of the energy surface points to a significant gap ( @xmath126 ) between the homogeneous `` ground - state '' and the first `` excited states '' , thus providing support for the validity of the approximation .
the predicted phase diagrams agree reasonably well with the results of gibbs ensemble mc simulations .
the present model is directly inspired by the continuous `` gaussian core '' model , first introduced by stillinger @xcite , and which has recently been shown to provide an adequate coarse - grained description of interacting , interpenetrating polymer coils @xcite .
the corresponding lattice model , considered here , appears to have some unorthodox features .
contrary to more familiar lattice gas models , it is not isomorphous to an ising spin model .
hence it is not yet clear whether its critical behaviour belongs to the ising universality class .
the binary version is expected to exhibit both the microphase separation found here in the one - component case , and the macroscopic phase separation anologuous to that of the continuous binary gaussian core model @xcite .
the interplay between these two phase transitions should lead to interesting and novel behaviour , which will be explored in future work . | a binary lattice gas model that allows for multiple occupancy of lattice sites , inspired by recent coarse - grained descriptions of solutions of interacting polymers , is investigated by combining the steepest descent approximation with an exploration of the multidimensional energy landscape , and by gibbs ensemble monte carlo simulations . the one - component version of the model , involving on site and nearest neighbour interactions , is shown to exhibit microphase separation into two sub - lattices with different mean occupation numbers . the symmetric two - component version of the multiple occupancy lattice gas is shown to exhibit a demixing transition into two phases above a critical mean occupation number . |
molecular simulation is a well - established approach for the analysis of fluid interfaces and their molecular structure .
much work has been dedicated to the interfacial tension.@xcitefor a fluid interface , there are ( at least ) three different aspects in which its size can be varied , each of which may affect the interfacial tension : * curvature effects , depending on the local characteristic radii of curvature * capillary wave effects , depending on the range of wavelengths permitted by the morphology and size of the interface * confinement effects , which arise due to spatial restrictions imposed on a fluid phase by one or several interfaces or walls according to the tolman @xcite approach , the interfacial tension of a nanodroplet deviates from that of a planar interface due to its extremely curved shape.@xcite however , it should be noted that all three phenomena are present when the size of a droplet is varied : smaller droplets have a higher curvature , a smaller range of capillary wavelengths , and a more significant deviation from bulk - like behaviour due to confinement .
in addition to curvature , the other effects might therefore also have a significant influence on the formation of droplets in a supersaturated vapor , where the size of the critical nucleus and the nucleation rate are of major interest.@xcite a similar case is cavitation , where bubbles emerge in a liquid phase.@xcite on the employed cutoff radius @xmath2 , for the truncated and shifted lj potential without long - range effects ( ljts ) and for the full lj potential with a long - range correction ( lrc ) according to janeek.@xcite the simulations were carried out in the canonical ensemble with @xmath3 = 2 048 particles at @xmath4 = 0.7 using a simulation box with an elongation of @xmath5 = 66 in the direction perpendicular to the vapor - liquid interfaces.,width=293 ] spherically curved interfaces of droplets were simulated for the first time by molecular simulation in the early 1970s.@xcite nonetheless , while curvature effects are relatively well - studied , there are only astonishingly few investigations related to the other size effects , which are also present in the case of planar interfaces .
several previous works address the influence of small simulation volumes,@xcite which is usually discussed in terms of capillary wave effects.@xcite the present study considers the influence of the liquid slab thickness , i.e. of confinement by two parallel planar vapor - liquid interfaces which are close to each other .
this effect was previously investigated by weng et al .
@xcite who , however , did not find a systematic correlation .
the computation of interfacial properties is always done in a single simulation volume containing both phases , the liquid and the vapor phase , separated by the interface . for the calculation of the bulk properties
there are many other methods , like grand equilibrium,@xcite @xmath6 plus test particle,@xcite or the gibbs ensemble method.@xcite the surface tension can be computed for example via the virial route or the surface free energy.@xcite the virial route is directly related to the common approach for calculating the pressure in a molecular simulation.@xcite it is known that the pressure in dependence of the density exhibits a van der waals loop in the two phase region.@xcite over the temperature for large liquid slabs .
comparison of results of different authors : ( red squares ) this work , ( blue triangles ) janeek,@xcite ( red diamonds ) holcomb et al.,@xcite ( brown circles ) potoff and panagiotopoulos,@xcite ( black circles ) int veld et al.,@xcite ( blue stars ) , as well as lpez lemus and alejandre,@xcite ; ( solid line ) regression from eq .
( [ eq : surface]).,width=293 ] at interfaces , the long range contribution to the interaction potential plays an important role for all thermodynamic properties.@xcite nonetheless , there are also simulations applying a truncated and shifted potential , which neglects the whole long range contribution.@xcite when dealing with a homogeneous system , long range corrections are only needed for energy and pressure,@xcite while in an inhomogeneous configuration , also the dynamics of the systems needs to be appropriately corrected.@xcite ( top ) and differential pressure @xmath7 ( bottom ) over the @xmath8 coordinate ( i.e. the direction perpendicular to the interface ) . the temperature is @xmath9 .
the blue dashed line corresponds to the minimum stable configuration which is @xmath10 for this temperature , while the red dash dotted one corresponds to @xmath1 = 7 and the black solid one to @xmath1 = 12 .
the dotted line in the upper picture represents the bulk liquid density and the difference between the vertical dashed lines in the upper picture represent the equimolar slab thickness.,width=293 ] in the present work , the influence of the liquid slab thickness on thermodynamic properties is discussed .
a suitable long range correction is used to obtain accurate and validated results .
over the density for the temperature @xmath9 and @xmath11 . the blue circles and brown triangles correspond to the minimum stable configuration for the corresponding temperature , while the red squares and green stars correspond to @xmath1 = 12 .
the dotted line represents the zero line.,width=293 ] in this study , the lennard - jones potential ( lj ) @xmath12\ ] ] is employed , where @xmath13 and @xmath14 are the energy and size parameters and @xmath15 is the distance between the two particles @xmath16 and @xmath17 .
the standard lj parameters @xmath18 and @xmath19 are used , yielding a reduced lj potential .
.surface tension @xmath20 in dependence of the slab thickness @xmath1 ( low temperatures ) .
the total elongation of the simulation box is indicated as @xmath5 in the direction perpendicular to the vapor - liquid interfaces and as @xmath21 ( @xmath22 ) in the other spatial directions . the statistical error in terms of the final digit
is shown in parentheses . [
cols="^,^,^,^,^,^",options="header " , ] in fig .
[ fig : vdw_07 ] , the plot of the differential pressure over the density is shown , which is obtained from the data displayed in fig .
[ fig : ddy ] . additionally results for @xmath11 are shown . the plot in fig .
[ fig : vdw_07 ] exhibits van der waals loops in all cases .
the red squares and green stars correspond to a large liquid slab , while the blue circles and brown triangles show the result of the smallest stable liquid slab .
like in fig .
[ fig : ddy ] , it is obvious that the differential pressure does not reach zero in the latter case .
the resulting surface tensions are shown in tabs .
[ tab : data_gamma1 ] and [ tab : data_gamma2 ] .
for all temperatures the surface tension decreases when the liquid slab thickness decreases . over the slab thickness @xmath1 for different temperatures .
the dashed lines represent the expression @xmath23 , where the temperature - dependent coefficients were adjusted to the simulation results , yielding @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 , and @xmath30.,width=293 ] the differential pressure in the center of the liquid slab , i.e. the minimum differential pressure , is shown in tabs . [
tab : data_pd1 ] and [ tab : data_pd2 ] . as already discussed , the differential pressure in the center of the liquid slab increases with decreasing slab thickness .
the density corresponding to the minimum differential pressure , i.e. the maximum density , in the liquid slab is shown in tabs .
[ tab : data_rho1 ] and [ tab : data_rho2 ] . at the largest slab thickness
the value agrees with the bulk properties . for lower slab thicknesses the density does not reach the bulk values .
the minimum stable slab thickness increases with increasing temperature . reducing the slab thickness below this point results in a rupture of the liquid phase and a transition from planar to cylindrical or spherical symmetry . in fig .
[ fig : surfacetension_rrho ] , the relative surface tension reduced by @xmath31 as obtained from the large slab simulations is plotted over the slab thickness for different temperatures .
confinement between two planar vapor - liquid interfaces reduces the surface tension , and the numerical data suggest that this effect is of the order @xmath32 . in fig .
[ fig : density_rrho ] , the reduced density @xmath33 is plotted over the slab thickness for different temperatures .
the relative density also decreases upon decreasing the slab thickness and , similar to the surface tension , this effect is approximately proportional to @xmath32 and becomes more significant at high temperatures . over the slab thickness @xmath1 for different temperatures .
the dashed lines represent the expression @xmath34 , where the temperature - dependent coefficients were adjusted to the simulation results , yielding @xmath35 , @xmath36 , @xmath37 , @xmath38 , @xmath39 , @xmath40 , and @xmath41.,width=293 ] for a slab thickness @xmath42 , the surface tension agrees with the value for a large liquid slab , within the simulation uncertainty . for a fluid described by the lj potential , e.g. methane,@xcite
this means that confinement effects are significant for slabs which are thinner than 4.5 nm . at high temperatures ,
the density in the center of the liquid slab deviates more significantly from the bulk value .
the present results suggest that the reduction of the density and the surface tension due to confinement are related effects . in fig .
[ fig : density_surface ] , the respective ratios are displayed together , which discloses an approximately linear relationship .
the regression @xmath43 is obtained from the simulation results . over the reduced surface tension @xmath44 for different temperatures .
the dash dotted line represents the regression from eq .
( [ eq : regression]).,width=293 ]
in the present work , molecular simulation was applied to study the influence of the slab thickness on the interfacial properties for planar vapor - liquid interfaces .
the present results prove that such an effect exists for thin slabs and quantifies it for the lj fluid .
the surface tension decreases with decreasing slab thickness , and so does the density in the middle of the slab .
the differential pressure does not reach zero for liquid slabs smaller than 12 , which proves that under such conditions , a bulk - like region is absent .
the confinement effects for the surface tension and the density were found to scale with @xmath32 in terms of the slab thickness @xmath1 , so that a linear relation between both effects could be obtained .
the present results depart from those obtained by weng et al .
@xcite in a previous study , where no systematic correlation between the slab thickness and the surface tension was found . for a lj system at @xmath4
= 0.818 , weng et al . detected minor fluctuations around a constant value ( @xmath20 = 0.78 @xmath45 0.02 ) , without a clear tendency , for a range of slab thicknesses between @xmath46 and @xmath47.@xcite a juxtaposition with the present numerical data , cf . tab .
[ tab : data_gamma1 ] and fig .
[ fig : surfacetension_rrho ] , according to which varying the slab thickness to such an extent has a significant influence on @xmath20 , clearly shows that there is a contradiction between present simulation results and the postulate of weng et al .
that `` with film thickness surface tension values and density profiles show little variation.''@xcite for the simulations of weng et al.,@xcite however , no long - range cutoff correction was employed at all , and the computations were only carried out over 120 000 time steps , as opposed to a million time steps for the present series of simulations . since systems with an interface relax more slowly than the homogeneous bulk fluid , the extremely short simulation time could consitute a serious limitation , affecting the accuracy of the results obtained by weng et al .
@xcite to a significant extent .
the comparison with results from a recent study of malijevsk and jackson @xcite suggests that the present results on confinement by two parallel planar vapor - liquid interfaces might also carry over qualitatively to confinement by the opposite sides of the single spherical interface that surrounds a small droplet .
therein , malijevsk and jackson come to the conclusion that for liquid drops , the size dependence of the surface tension is best described by two distinct , additive terms : the conventional tolman term , representing curvature , which increases the surface tension ( i.e. the tolman length is found to be negative ) , as well as `` an additional curvature dependence of the @xmath48 form '' which causes an eventual decrease of the surface tension `` for smaller drops.''@xcite furthermore , malijevsk and jackson observe that the characteristic droplet radius , below which this negative corrective term becomes dominant , `` increases with increasing @xmath2 '' and conjecture that `` such a crossover occurs when no ` bulk ' region can be assigned inside the drop . in this case
even particles in the centre of the drop ` feel ' the interface . ''
the present results lend further plausibility to this conjecture of malijevsk and jackson . there could be a relation between their @xmath48 term and the @xmath32 confinement effect from the present study . according to such a hypothesis
, these contributions would both represent the deviation from bulk - like behavior of the liquid phase due to confinement .
the authors gratefully acknowledge financial support from deutsche forschungsgemeinschaft ( dfg ) within the collaborative research center ( sfb ) 926 .
they thank stefan eckelsbach , philippe ungerer , and jadran vrabec for fruitful discussions as well as advait deshpande and srishti srivastava from iit bombay for performing some of the present simulations .
the present work was conducted under the auspices of the boltzmann - zuse society of computational molecular engineering ( bzs ) , and the simulations of the largest slab were carried out on the xc 4000 supercomputer at the steinbuch centre for computing , karlsruhe , under the grant mocos .
[ 2][]#2 [ 2]#2 biblabel#1#1 .
, , . _ _ ; * * ( ) : . , , , ,
. _ _ ; * * ( ) : . ,
. _ _ ; * * ( ) : .
, , , , , . _ _ ; * * : .
. _ _ ; * * ( ) : . , .
_ _ ; * * ( ) : . , , , , , , ,
. _ _ ; * * : . ,
. _ _ ; * * : .
_ _ ; * * : .
. _ _ ; * * ( ) : .
. _ _ ; * * ( ) : . , , .
_ _ ; * * : . .
_ _ ; * * : .
, , , , . _ _ ; * * : . , ,
. _ _ ; * * ( ) : .
, . _ _ ; * * ( ) : . , ,
. _ _ ; * * : . , ,
. _ _ ; * * ( ) : .
. _ _ ; * * ( ) : .
, , . _ _ ; * * ( ) : . , ,
. _ _ ; * * : .
, , , . _ _ ; * * : .
. _ _ ; * * : .
, , , . _ _ ; * * : . , , ,
. _ _ ; * * ( ) : .
, , , , , , , . _ _ ; * * ( ) : . , .
_ _ ; * * ( ) : . ,
. _ _ ; * * ( ) : .
, , , . _ _ ; * * ( ) : . , .
_ _ ; * * ( ) : . ,
. _ _ ; * * ( ) : . .
_ _ ; * * ( ) : . , .
_ _ ; * * ( ) : . ,
. _ _ ; * * : . , , .
_ _ ; * * : .
_ _ ; * * : .
. _ _ ; * * : . , .
_ _ ; * * : . , ,
. _ _ ; * * ( ) : .
. _ _ ; * * ( ) : . , , .
_ _ ; * * ( ) : . , . _ _ ; * * : . , . _ _ ; * * ( ) : .
, , , , . _ _ ; * * : . ,
. _ _ ; * * : .
_ _ ; * * ( ) : .
_ _ ; * * : . .
_ _ ; * * ( ) : .
. _ _ ; * * ( ) : . .
_ _ ; * * ( ) : .
_ _ ; * * ( ) : . , ,
. _ _ ; * * ( ) : .
, , . _ _ ; * * ( ) : .
. _ _ ; * * ( ) : . , . _ _ ; * * ( ) : . , ,
. _ _ ; * * ( ) : . .
_ _ ; * * ( ) : .
, , , . _ _ ; * * : . , , .
_ _ ; * * : . ,
. _ _ ; * * : . , ,
. _ _ ; * * ( ) : . ,
. _ _ ; * * ( ) : . , ,
. _ _ ; * * : . , .
_ _ ; * * ( ) : . | one of the long standing challenges in molecular simulation is the description of interfaces . on the molecular length scale
, finite size effects significantly influence the properties of the interface such as its interfacial tension , which can be reliably investigated by molecular dynamics simulation of planar vapor - liquid interfaces . for the lennard - jones fluid ,
finite size effects are examined here by varying the thickness of the liquid slab .
it is found that the surface tension and density in the center of the liquid region decreases significantly for thin slabs .
the influence of the slab thickness on both the liquid density and the surface tension is found to scale with 1/@xmath0 in terms of the slab thickness @xmath1 , and a linear correlation between both effects is obtained .
the results corroborate the analysis of malijevsk and jackson , j. phys .
: cond . mat .
24 : 464121 ( 2012 ) , who recently detected an analogous effect for the surface tension of liquid nanodroplets .
surface tension , long range correction , finite size effects |
there have recently been a number of suggestions for a quantum computer architecture that use quantum dots of varying kinds@xcite .
if these schemes are to be practical many important physical questions need to be answered , one of which is how to readout physical properties such as charge or spin at a single electron level@xcite . in this paper
we present a quantum trajectory analysis of a general scheme to readout a single electronic qubit using a single electron transistor ( set ) .
we adopt a general phenomenological description of the set in which the tunneling rate through the set is conditioned on the occupation or otherwise of a nearby quantum dot .
we consider two spatially separated quantum dots which are strongly coupled so that delocalized states of their relevant degrees of freedom can form . to be specific
, we imagine each dot to have a single electronic bound state that can be occupied .
thus the average occupation number of each dot must be less than unity .
this restriction can easily be removed to account for spin , or multiple electron states .
we label each dot with an index @xmath0 and let @xmath1 represent the fermi annihilation and creation operators for each single electron state ( see figure 1 ) .
the two dots are strongly coupled via the tunnel coupling hamilton @xmath2 thus the total hamiltonian of the two - dot system is @xmath3 in what follows we will work in an interaction picture and assume that the energies of each bound state are equal ( again this can be relaxed ) .
coulomb blockade effects have been ignored at this stage , but can easily be included without significantly changing the results of this paper .
the single particle eigenstates of this hamiltonian are even and odd superpositions of the bare states of each well .
such states are thus delocalized over the two - dot system and are sometimes called molecular states in the literature@xcite .
the localized states can then be represented as an even and odd superposition of the delocalized states .
the localized states are not stationary ; rather , the system will periodically oscillate between them .
that is to say the system will tunnel coherently between the two dots . to this coherent system
we add a measurement device which determines the presence of an electron on one of the dots , say dot @xmath4 , which we shall refer to as the target ( see figure [ fig1 ] ) .
the model is based on a set tunnel junction containing a single bound state on the island .
the interaction between the target and the set is via a coulomb blockade .
thus the interaction hamiltonian between the set and the target must commute with the target electron number operator , @xmath5 .
this makes it a qnd ( quantum nondemolition measurement ) of electron number@xcite .
the coulomb blockade changes the current flowing through the set . in simple terms
if there is no electron on the target the island state is biased so as to allow little or no current to flow though the set .
this is the quiescent state of the set .
however when there is an electron on the target , the coulomb blockade shifts the bound state on the island to allow a greater current to flow through the device ( see figure [ fig2 ] ) we derive a master equation to describe the behaviour of the target system .
this master equation describes the unconditional evolution of the measured system when the results of all measurement records ( that is current records ) are averaged over .
this will tell us the rate at which coherence in the target system is destroyed by the measurement .
however we also need to know how the system state depends on the actual current through the device in order to determine how quickly the conditional state of the electron becomes localized which is measure of the quality of the measurement .
one approach to this problem is to keep track of many different states of the system , corresponding to the different numbers of electrons which have tunneled through the set .
this is the approach used for example in ref .
@xcite . here
we adopt an alternate method which gives a more intuitive picture for the conditional dynamics .
we use a conditional _ stochastic _ master equation which gives the evolution of the measured system , conditioned on a particular realization of the measured current .
the instantaneous state of the target conditions the measured current while the measured current itself conditions the future evolution of the measured system in a self consistent manner .
this approach to measurements has been variously called the quantum trajectory method @xcite or quantum monte carlo method @xcite .
consider a two dot system with the coupling in eq .
( [ tunnel ] ) .
if the electron is in dot-2 the quiescent rate of current tunneling though the set is a constant which we will denote @xmath6 .
however if there is an electron on dot-1 , the rate of tunneling through the set changes to @xmath7 with @xmath8 . if @xmath6 does not equal zero then a current spike ( resulting from a tunnel event in the set ) does not necessarily imply that the electron in the measured system is in dot-1 . in an ideal device the quiescent tunneling rate ,
@xmath6 is zero . in reality
johnson noise on the circuit containing the set will give a non - zero quiescent tunneling current .
assuming that the set island state can be adiabatically eliminated , it is possible to derive a master equation for the state of the coupled dot system .
this is done in the appendix , and the result is @xmath9+\gamma_{{\rm dec}}{\cal d}[c_1^\dagger c_1]\rho = { \cal l}\rho\ ] ] where the irreversible part is defined for arbitrary operators @xmath10 and @xmath11 by @xmath12b={\cal j}[a]b - { \cal a}[a]b , \label{uncond}\ ] ] where @xmath13b & = & ab a^\dagger \label{defcalj}\\ { \cal a}[a]b & = & \frac{1}{2}(a^\dagger ab+b a^\dagger a ) . \label{defcala}\end{aligned}\ ] ] the decoherence rate is given by @xmath14 the fact that the irreversible term is a function of the number operator in the target qubit is an indication that this describes a qnd measurement of the occupation of the dot-1 .
it is easy to verify that the stationary solution of this master equation is an equal mixture of the two accessible electronic states .
the stochastic record of measurement ideally comprises a sequence of times , being the times at which electrons tunneled through the set . in practice of course these events are not seen due to a finite frequency response of the circuit ( including the set ) which averages each event over some time .
however for the purpose of this paper we will take the zero response time limit . in this limit
the current consists of a sequence of @xmath15 function spikes .
formally we can write @xmath16 , where @xmath17 is a classical point process which represents the number ( either zero or one ) of tunneling events seen in an infinitesimal time @xmath18 , and @xmath19 is the electronic charge .
we can think of @xmath17 as the increment in the number of electrons @xmath20 in the collector in time @xmath18 .
it is this variable , the accumulated electron number transmitted by the set , which is used in ref .
@xcite . the point process @xmath17 is formally defined by the conditions @xmath21^{2 } & = & dn(t ) \label{zo } \\
{ \rm e}[dn(t)]/dt & = & d_{0}{\rm tr}[(1-n_1)\rho_{{\rm c}}(t)(1-n_1 ) ] + ( d_{0}+d_{1}){\rm tr } [ n_1 \rho_{{\rm c}}(t ) n_1 ] \nonumber \\ & = & d_{0}+ d_{1}\left\langle{\ n_1 } \right\rangle_{{\rm c}}(t ) . \label{mean}\end{aligned}\ ] ] here @xmath22 $ ] denotes a classical average of a classical stochastic process @xmath23 , and @xmath24 is the occupation number operator for the first dot .
the first of these equations simply expresses the fact that @xmath17 equals zero or one .
the second says that the rate of events is equal to a background rate @xmath25 plus an additional rate @xmath26 if and only if the electron is in the first dot . in eq .
( [ mean ] ) , the system state matrix @xmath27 is _ not _ the solution of the master equation ( [ me1 ] ) .
that is because if one has a record of the current @xmath28 through the set then one knows more about the system than the master equation indicates .
that is to say , @xmath29 is actually conditioned by @xmath30 for @xmath31 , hence the subscript c. the first way of writing eq .
( [ mean ] ) hints at how @xmath32 conditions @xmath27 . from the appendix ,
the state at time @xmath33 given @xmath34 is @xmath35\ ] ] this is an unnormalized state whose norm is equal to the probability of that event ( @xmath34 ) occurring , as seen above in eq .
( [ mean ] ) .
the normalized state can be written more elegantly as @xmath36 + 2d_{0}{\cal d}[n_{1}])\rho ( t)}{{\rm tr}\{(d_{0}+d_{1}{\cal j}[n_{1}]+2d_{0}{\cal d}[n_{1}])\rho ( t)\}}=% \frac{(d_{0}{\cal j}[1-n_{1}]+(d_{0}+d_{1}){\cal j}[n_{1}])\rho ( t)}{{\rm e}% [ dn(t)/dt ] } , \label{rho1}\ ] ] where @xmath37 is as defined above in eq .
( [ defcalj ] ) . to write the full conditioned evolution we need to know the density operator @xmath38 given that @xmath39 .
this can be found from eq .
( [ rho1 ] ) plus the fact that when averaged over the observed classical point process @xmath40 , @xmath41 that is to say , _ on average _ the system still obeys the master equation ( [ me1 ] ) . from this equation ,
we obtain @xmath42\rho(t ) + ( d_{0}+ d_{1}){\cal a}[n_1]\rho(t)+i[h,\rho(t)]\ } , \\ & = & \rho(t)-dt\left\ { { d_{0}\rho(t)+d_{1}{\cal a}[n_1]\rho(t)-i[h,\rho(t ) ] } \right\},\end{aligned}\ ] ] where @xmath43 is as defined above in eq .
( [ defcala ] ) . once again this state
is unnormalized and its norm gives the probability that @xmath39 , that is @xmath44 = 1 - { \rm e}[dn(t)]\ ] ] using the variable @xmath17 explicitly , the conditioned state at time @xmath33 is @xmath45 } + [ 1-dn(t)]\frac{\tilde{\rho}_{0}(t+dt)}{{\rm tr}[\tilde{\rho}% _ { 0}(t+dt)]}.\ ] ] since @xmath17 is almost always zero we can set @xmath46 and expand this expression to finally obtain the stochastic master equation , conditioned on the observed event in time @xmath18
@xmath47 + 2d_{0}{\cal d}[n_1]}{d_0+d_1 { \rm tr}[\rho_{{\rm c } } n_1]}% -1\right ] \rho_{{\rm c } } \nonumber \\ & & \mbox{}+\ , dt\left\{-d_1{\cal a}[n_1]\rho_{{\rm c}}+d_1{\rm % tr}[\rho_{{\rm c } } n_1]\rho -i[h,\rho_{{\rm c}}]\right\ } \label{condev}\end{aligned}\ ] ] note that averaging this equation over the observed stochastic process ( by setting @xmath40 equal to its expected value ) gives the unconditional master equation ( [ me1 ] ) .
we now analyze in some detail the ensemble averaged properties of the system based on the unconditional master equation .
in particular we calculate the stationary noise power spectrum of the current through the set when there is the possibility of coherent tunneling between dot-2 and the measured dot-1 .
the details of how the quantum stochastic processes in the set determine the average current though the set are given in reference @xcite .
the link with the stochastic formalism of the preceding section is that the current @xmath48 through the set is given by @xmath49 first we calculate the steady state current @xmath50_\infty \nonumber \\ & = & e(d_0+d_1\langle c_1^\dagger c_1\rangle_{\infty } ) \\ & = & e(d_0+\frac{d_1}{2 } ) , \label{current}\end{aligned}\ ] ] where the @xmath51 subscript indicates that the system is at steady - state . the fluctuations in the observed current , @xmath52 are quantified by the two - time correlation function : @xmath53_\infty \\ & = & e i_{\infty } \delta ( \tau ) + \left\langle
i(t),i(t+\tau ) \right\rangle _ { \infty } ^{\tau \neq 0 } .
\label{eq - corr}\end{aligned}\ ] ] here @xmath54 .
the fact that the multiplier of the shot noise is @xmath55 rather than the usual @xmath56 is because of the approximation we have made in treating the set .
specifically , we have adiabatically eliminated the set by taking the limit where any electron which tunnels onto the set island from the emitter immediately tunnels off to the collector .
this means that , on the time scales we are interested in , there is a perfect correlation between the emitter current and collector current .
this leads to a doubling of the shot noise level .
of course , at very high frequencies , higher than we are interested in , the true shot noise level of @xmath56 could still be seen in principle . to relate these classical averages to the fundamental quantum processes occurring in the well
we apply the theory of open quantum system @xcite to the present system .
specifically , we can relate the correlation function for the current to the following quantum averages
@xmath57 + 2d_{0}{\cal d}[n_{1}])e^{{\cal l}% \tau}(d_{0}+d_{1}{\cal j}[n_{1}]+2d_{0}{\cal d}[n_{1}])\rho _ { \infty } } \right].\ ] ] because @xmath58 is an equal mixture of the two electron states , it satisfies @xmath59\rho _ { \infty } = 0.\ ] ] in addition , the following identies for arbitrary operators @xmath10 and @xmath11 are easy to prove : @xmath60b}\right ] \equiv 0 $ ] , @xmath61b}\right ] \equiv { \rm tr}[n_{1}b]$ ] , @xmath62={\rm tr}[b]$ ] , and @xmath63=% { \rm tr}[a\rho _ { \infty } ] $ ] .
using these simplfications we obtain @xmath64\rho _ { \infty } } \right ] -{\rm tr}[n_{1}\rho _ { \infty } ] ^{2}}\right\ } .\ ] ] evaluating this expression we find @xmath65 where @xmath66 and where the first term represents the shot noise component as discussed above .
the power spectrum of the noise is @xmath67 which evaluates to @xmath68 in the case that @xmath69 the spectrum will have a double peak structure indicating that coherent tunneling is taking place between the two coupled dots . for smaller @xmath70
only a single peak appears in the spectrum .
we can thus use the noise power spectrum of the current though the set as a means to measure the tunnel coupling between dots if the tunnel coupling is high enough .
we illustrate this in figure [ fig3 ] .
we now return to the stochastic master equation for the conditioned state , @xmath71 + 2d_{0}% { \cal d}[c_{1}^\dagger c_{1}]}{d_0+d_1 \mbox{tr}[\rho_{{\rm c } } c_1^\dagger c_1]}-1\right ] \rho_{{\rm c } } \nonumber \\ & & \mbox{}+dt\left\{-d_1\frac{1}{2}\{c_1^\dagger c_1,\rho_{{\rm c}}\}+d_1% \mbox{tr}[\rho_{{\rm c } } c_1^\dagger c_1]\rho -i[h,\rho_{{\rm c}}]\right\}. \label{condev2}\end{aligned}\ ] ] comparing this to the unconditional master equation @xmath72+\gamma_{{\rm dec}}{\cal d}[c_{1}^\dagger c_{1}]\rho\ ] ] we see that decoherence between the two coupled dots , 1 and 2 , takes place at the rate @xmath73 , but that the system decides between the two possibilities ( electron on dot-1 or on dot-2 ) on a time scale that depends on @xmath26 and @xmath6 in some more complicated way .
of course this measurement time scale is necessarily at least as large as the decoherence time scale because successfully distinguishing between the two dots would by definition destroy any coherence between them .
the different measurement time scales can be derived most easily by introducing the bloch representation of the state matrix : @xmath74 where the pauli matrices are defined using the fermi operators for the two dots @xmath75 in this representation the means of the pauli matrices @xmath76 are given by the respective coefficient @xmath77 , with @xmath78 . the stochastic master equation can now be written as a set of coupled stochastic differential equations for the bloch sphere variables as @xmath79 again the c - subscript is to emphasize that these variables refer to the conditional state .
if we average over the noise , the ensemble dynamics is then seen to be given by @xmath80 where @xmath81 $ ] denotes the averaging over the ensemble of conditional states .
these equations are exactly what would be obtained directly from the ensemble averaged master equation eq([me1 ] ) .
in particular we note that the average population difference @xmath82 between the dots is a constant of the motion in the absence of any free hamiltonian .
however the stochastic differential equations enable us to calculate important averages that are _ not _ obtainable from the master equation .
for example , if the model does indeed describe a measurement of @xmath83 , then , in the absence of tunneling , we would expect to see the conditional state become localized at either @xmath84 or @xmath85 .
indeed for @xmath86 we can see from the conditional equation for @xmath87 that @xmath88 is a fixed point .
we can take into account both fixed points by considering @xmath89 . in the absence of tunneling
this must must approach @xmath4 for all trajectories , since the system will eventually become localized due to the measurement in one dot or the other .
therefore it is sensible to take the ensemble average @xmath90 $ ] and find the rate at which this deterministic quantity approaches one . noting that for a stochastic variable we have @xmath91 , and that @xmath92={\rm e}% [ dn]=[d_{0}+d_{1}(1-z_{{\rm c}})/2]dt$ ] , we find that @xmath93}{dt}={\rm e}\left [ \frac{d_{1}^{2}(1-z_{{\rm % c}}^{2})^{2}}{4d_{0}+2d_{1}(1-z_{{\rm c}})}\right ] .\ ] ] if the system starts state which has an equal probability for single electron to be on each dot then @xmath94 and in the ensemble average this would remain the case .
however if we ensemble average @xmath95 over many quantum trajectories then for short times we find @xmath96=\frac{d_1 ^ 2}{4d_0 + 2d_1}\delta t \label{z2t}\ ] ] that is to say , the system tends towards a definite state ( with @xmath97 so @xmath98 ) at an initial rate of @xmath99 . for vanishing @xmath25 ,
this is the same as the decoherence rate , @xmath100 , as expected .
but for @xmath101 , the rate goes to @xmath102 .
that is , the rate at which the system becomes localized at one or the other dot is much less than the decoherence rate .
this result can not be obtained from the ensemble averaged master equation alone .
it is a direct reflection of the fact that for @xmath103 a tunneling event can not be unambiguously attributed to the location of an electron on the double dot system .
as the rate of localization is a direct indication of the quality of the measurement , we can use the localization rate defined as @xmath104 as an important parameter defining the quality of the measurement .
this parameter is related to the signal - to - noise ratio for this measurement as we now show . for a poisson process at rate @xmath105 , the probability for @xmath106 events to occur in time
@xmath107 is @xmath108 the mean and variance of this distribution are equal and given by @xmath109 .
now consider an electron which is , with equal likelihood , in either dot , so that @xmath110 . if the electron is in dot 1 then the rate of electrons passing through the set is @xmath111 ; if it is in dot two then it is just @xmath25 .
these two possibilities will begin to be distinguishable when the difference in the means of the two distributions @xmath112 is of order the square root of the sum of the variances .
that is , when @xmath113 solving this for @xmath107 gives a characteristic rate @xmath114 the right hand side of this expression is simply twice the @xmath115 defined above .
a similar conclusion is reached in reference @xcite . in the ideal limit of no quiescent current in the set @xmath116
, the stochastic master equation can be replaced by a stochastic schrdinger equation , and will collapse to a single possibility at a rate @xmath117 which is the same as the decoherence rate .
the effect of @xmath25 is most clearly seen in the other limit , @xmath118 , as noted above . in this limit
the single electron makes only a small relative change in the tunneling rate through the set .
as the rate of jumps also becomes large , the trajectories in this limit take on the appearance of diffusion rather than jumps .
the rate at which the electron localizes into one well or the other scales as @xmath119 , which is much longer than the decoherence time scale @xmath120 . [ earlier ]
we now turn to numerical simulations of the conditional evolution and to estimate the conditions for a good measurement . unlike traditional condensed matter measurements we wish to describe repeated measurements made on a single quantum system rather than a single measurement made upon an ensemble of systems .
to do this we use the conditional dynamics of the measured system given a particular measurement record as described by the above stochastic evolution equation ( [ condev2 ] ) we return to the bloch description defined in eq .
( [ bloch ] ) . in what follows we will assume that @xmath121 . for the form of tunneling used here the value of @xmath122 does not in fact change under either conditional or ensemble averaged dynamics .
if the conditional state of the system remains in a pure state then @xmath123 .
as noted previously this can only occur if the bare tunneling rate ( @xmath6 ) is zero , when a tunneling event can unambiguously be attributed to the occupation of the target dot and no information is lost about the state of the system . in the more realistic case in which @xmath124 , we can use the quantity @xmath125 as a measure of the purity of the sate , or equivalently as a measure of how much information the conditional record of measurements gives about the actual state of the two coupled dots .
if the conditional state is a maximally mixed state of a two state system then @xmath126 .
we now describe in detail the numerical simulation of the conditional dynamics .
first we consider the case @xmath127 , so the system is always in a pure state .
typical trajectories are shown in fig . 4 for various values of @xmath128 . for small @xmath129 we see little evidence for coherent tunneling .
most of the time the electron is localized almost entirely in one well or the other .
however , there is an asymmetry between the wells .
a transition from dot-2 into dot-1 ( the target ) is sudden , occurring whenever an electron tunnels through the set .
a transition the other way takes a time of order @xmath130 .
this time is still much smaller than the average time between state - changing transitions , which can be shown analytically to be @xmath131 .
thus over a long time , as shown in fig .
4(a ) , the system still has the appearance of a random telegraph .
this behaviour gives rise to a single - peaked noise spectrum , as shown in fig .
3(a ) . for moderate @xmath132
the system is no longer well - localized in one dot or the other .
rather , the dynamics is complicated with clearly non - sinusoidal oscillations from one dot to the other interspersed with jumps into the target dot .
the fact that oscillations are present gives peaks in the current noise spectrum , as shown in fig .
the position of these peaks is a frequency less than @xmath70 , as shown analytically in sec .
for @xmath133 the dynamics once again becomes simple , with nearly sinusoidal oscillations interspersed with jumps which occur with an average rate of @xmath100 .
this corresponds to a noise spectrum having a very sharp feature at @xmath134 , as shown in fig .
3(c ) . the change in behaviour as @xmath70 increases is summarized in fig . 5 . there
we plot e@xmath135 $ ] versus @xmath136 .
the quantity e@xmath137 $ ] measures how well localized the electron is at one well or the other , and would be @xmath4 if the electron were always localized and @xmath138 if it were never localized .
it is actually possible to calculate this quantity numerically without using a stochastic ensemble , as follows . with no background current , every time an electron tunnels through the set the electron on the dots is known to be on dot @xmath4 .
if there are no further set tunneling events for a time @xmath139 later then from eq .
( [ rho0 ] ) , the system evolves up to that time by the equation @xmath140\}.\ ] ] because there is no background current , and because the initial state is pure , it is possible to rewrite this in terms of a non - hermitian schrdinger equation @xmath141 here it must be remembered that the norm of this state represents the probability for the event that no electron has passed through the set since the last one a time @xmath139 ago : @xmath142 = \langle{\tilde\psi_{0}(t)}|{\tilde% \psi_{0}(t)}\rangle.\ ] ] it is not difficult to show that the solution to eq .
( [ sse ] ) satisfying the initial condition @xmath143 is @xmath144 where the occupation numbers refer to the dots one and two in order . here
@xmath145 and @xmath146 are real numbers defined by @xmath147 where @xmath148\ ] ] the conditioned quantum expectation value for @xmath149 is @xmath150 now in steady state the probability @xmath151 that there is no increment in @xmath20 for a time @xmath139 ago is related to the probability @xmath152 that the last increment was a time @xmath139 ago by @xmath153 since at steady state all conditioned states are uniquely identified by how long it has been since the last set event , the ensemble average for @xmath154 is simply given by @xmath155 & = & \int_{0}^{\infty } q_{0}(t ) [ z_{0}(t)]^{2}dt \\ & = & \left [ { \int_{0}^{\infty } \left ( { \beta^{2}+\alpha^{2 } } \right)dt } \right]^{-1 } { \int_{0}^{\infty}\frac{\left ( { \beta^{2}-\alpha^{2 } } \right)^{2 } } { \beta^{2}+\alpha^{2 } } dt } .\end{aligned}\ ] ] unfortunately it does not appear possible to evaluate the second integral here analytically .
however a numerical integration is easy .
the results , shown in fig .
5 , is in agreement with the ensemble averages obtained numerically using the stochastic master equation .
we next consider the case where @xmath156 .
we show two plots , both with @xmath157 , which is a regime in which coherent tunneling is clearly evident in the current noise spectrum . the first plot , in fig .
6 , is for @xmath158 . here
coherent oscillations are still evident in @xmath82 , but @xmath82 rarely attains its extreme values of @xmath159 .
the conditioned state is no longer pure , even immediately after a count .
also , the conditioned state following a count now depends on the state before the count . for this reason
an exact solution by the method of the preceding section is impossible . the second plot , in fig .
7 , is for @xmath160 . here
coherent oscillations are no longer obvious in the condition mean of @xmath149 , even thought they are present in the spectrum ( as small features above the shot noise ) , as calculated in sec . iii . in this regime @xmath161
so the diffusive limit discussed at the end of sec .
[ earlier ] applies .
the three parameters we need to compare our theoretical results with experiment are @xmath70 , @xmath6 and @xmath117 .
the two incoherent tunneling rates can be obtained by considering how they determine the steady state current through the device .
this is given in eq ( [ current ] ) .
our model implicitly assumes that the quiescent noise in the set is shot - noise limited , based as it is on elementary tunneling events .
however from the point of view of the macroscopic circuit in which the set is placed , the tunnel junctions appear as a capacitor in series with a resistor ( see note 14 in reference @xcite ) .
if the resistance of the junction is @xmath162 and the capacitance is @xmath163 the fundamental time constant for the junction is @xmath164 .
this sets an upper bound for the tunneling rate @xmath165 for a typical al / alo@xmath166/al junctions we have @xmath167f and @xmath168 and thus the time constant is @xmath169 . for a double tunnel junction device
, the maximum conductance is achieved for a symmetric pair of tunnel junctions .
the value of @xmath117 in this case is given by @xmath170 where @xmath171 is the tunneling rate of the set under conditions of maximum conductance ( see appendix ) .
we thus estimate that @xmath172 .
the background tunneling rate through the set depends on temperature as well as the bias conditions ( see appendix ) .
typical maximum and minimum conductance for the set at different temperatures have been measured by joyez _
_ @xcite . at a temperature of @xmath173
the minimum conductance is approximately zero , and thus at this temperature we can safely take @xmath116 , the zero temperature result .
however at a temperature of @xmath174 the ratio of the maximum to minimum conductance is 2.2 .
this indicates that at 400 mk , @xmath175 .
the value we choose for the tunneling rate depends strongly on the particular quantum dot system .
we will consider the value appropriate for the single electron measurement scheme of kane _
et al._@xcite . in this model
the two localized states correspond to an electron on a tellurium ion donor or at a nearby interface below the donor .
the tunneling rate is then between the donor state and the interface state .
_ estimate that for this system @xmath176 .
this value is too low to observe a peak in the noise power spectrum away from zero frequency .
however the tunneling rate could be increased by changing the donor interface bias voltage . in conclusion
we have presented a simple model to describe single electron measurements on a coupled double quantum dot system using an set .
we have given both the stationary ( ensemble averaged ) properties of the current through the set as well as the conditional dynamics of repeated measurements on a single system .
this illustrates how quantum trajectory methods may be naturally adapted to single electronics , and aid in the interpretation of ensemble averaged properties .
we believe these models will prove useful in current attempts to fabricate quantum logic gates in solid state devices .
it will suffice to consider a single quantum dot near the set .
this allows us to remove any reference to the electron field labeled by @xmath177 .
the set is modeled as a single biased double barrier ( single well ) device with a single bound state on the well described by the fermi operators @xmath178 .
the total hamiltonian for the system including the reservoirs is @xmath179 the term @xmath180 is the coulomb blockade term and is given by@xcite @xmath181 where @xmath182 is the coulomb blockade energy gap ( see figure [ fig2 ] ) .
note this term can only be nonzero if there is an electron on the island and on the dot , in which case the energy of the island electron is increased .
the terms @xmath183 described the tunneling between the many modes in the left and right ohmic contacts and the bound state on the set @xcite @xmath184 where @xmath185 are respectively the fermi field annihilation operators for the left and right reservoir states at momentum @xmath186 .
the tunneling matrix elements between respectively the left and right ohmic contacts and the island are @xmath187 .
the free hamiltonian for the the system is @xmath188 we now transform to an interaction picture to remove the terms @xmath189 .
the dynamics in the schrdinger picture is now described by the time dependent hamiltonian @xmath190 using the fact that @xmath191 we find @xmath192 where @xmath193 notice that if there is no electron on the dot and @xmath194 then the second term is zero and the first term is a standard tunneling interaction onto a bound state with energy @xmath195 . on the other hand
if there is an electron on the dot @xmath194 and the first term is zero and the second term is a standard tunneling interaction onto a bound state with energy @xmath196 as expected .
the derivation of the master equation for the state matrix @xmath105 for the system ( set and quantum dot ) can now proceed using standard techniques which we will sketch .
the objective is to obtain a semigroup evolution in lindblad form ( that is to say positivity - preserving irreversible dynamics ) for the state of the set island and the dot alone with no reference to the ohmic contacts .
the ohmic contacts are treated as perfect fermi thermal reservoirs with a very fast relaxation constants .
each ohmic contact ( left and right ) remains in thermal equilibrium with chemical potentials @xmath197 , but the total system is not in thermal equilibrium due to the external bias potential , @xmath198 with @xmath199 ( see references @xcite for further discussion ) . we first define a time interval @xmath200 which is slow compared to the dynamics of the island and the dot but very long compared to the time scale in which the ohmic contacts relax back to their steady state .
the change in the state matrix @xmath201 of the system ( set and dot ) and environment ( ohmic contacts ) from time @xmath139 to @xmath202 , to second order in the tunnel coupling energy , is given by @xmath203-\delta t\int_t^{t+\delta t}dt_1[h_i(t),[h_i(t_1),w(t_1)]]\ ] ] we now make the first markov approximation and assume that at any time the state of the total system may be approximated by @xmath204 , that is to say the left and right ohmic contacts instantaneously relax back to fermi distributions .
we now obtain an evolution equation for @xmath205 , the state of the island and the dot by tracing over the reservoirs .
the result is @xmath206{\cal d}[b(1-c_1^\dagger c_1)]r \\ & & + \left [ \gamma_l f_l(\omega_0)+\gamma_rf_r(\omega_0)\right ] { \cal d}% [ b^\dagger(1-c_1^\dagger c_1)]r \\ & & + \left [ \gamma^{\prime}_l(1-f_l(\omega_0+\chi))+\gamma^{\prime}_r(1-f_r(% \omega_0+\chi))\right ] { \cal d}[bc_1^\dagger c_1]r \\ & & + \left [ \gamma^{\prime}_lf_l(\omega_0+\chi)+\gamma^{\prime}_rf_r(% \omega_0+\chi)\right ] { \cal d}[b^\dagger c_1^\dagger c_1]r\end{aligned}\ ] ] where for arbitrary operators @xmath10 and @xmath11 , @xmath207b = ab a^\dagger -\frac{% 1}{2}(a^\dagger a b - ba^\dagger a)$ ] and where @xmath208 is the fermi filling probability for the left / right ohmic contact at the energy @xmath209 .
the rates @xmath210 and @xmath211 determine the rate of injection from the left ohmic contact into the island or emission from the island into the right ohmic contact under the conditions of no electron on the dot ( unprimed ) and with an electron on the dot ( primed ) .
these are evaluated using the second markov approximation as @xmath212 where @xmath213 and @xmath214 .
the ideal quiescent state of the set is defined as @xmath215 while @xmath216 . under these conditions
the master equation reduces to @xmath217r+\gamma_l{\cal d}% [ b^\dagger ( 1-c_1^\dagger c_1)]r+\gamma^{\prime}_r{\cal d}[b c_1^\dagger c_1]r + \gamma^{\prime}_l{\cal d}[b^\dagger c_1^\dagger c_1]r\ ] ] we now wish to derive a master equation for the state matrix @xmath218 for the dot alone .
this is easiest if we assume that @xmath219 are much larger than all other rates in the system . in this case
it is possible to adiabatically eliminate the set island using techniques similar to that in ref .
we expand the state matrix @xmath105 in powers of @xmath220 or @xmath221 as @xmath222 the equations of motion for @xmath223 and @xmath224 are @xmath225\rho_{1 } + \gamma_{l}{\cal j}[1-n_{1}]\rho_{0}-\gamma'_{r}{\cal a}[n_{1}]\rho_{1 } + \gamma'_{l}{\cal j}[n_{1}]\rho_{0 } \label{dotrho1 } \\ \dot{\rho}_{0 } & = & \gamma_{r}{\cal j}[1-n_{1}]\rho_{1 } - \gamma_{l}{\cal a}[1-n_{1}]\rho_{0 } + \gamma'_{r}{\cal j}[n_{1}]\rho_{1 } - \gamma'_{l}{\cal a}[n_{1}]\rho_{1 } \label{dotrho0}\end{aligned}\ ] ] here @xmath226 and @xmath37 and @xmath43 are as defined in eqs .
( [ defcalj ] ) , ( [ defcala ] ) . under the above conditions , we can slave @xmath223 to @xmath224 so that @xmath227 + \gamma'_{r}{\cal a}[n_{1}])\rho_{1 } = ( \gamma_{l}{\cal j}[1-n_{1}]+ \gamma'_{l}{\cal j}[n_{1}])\rho_{0}\ ] ] operating on both sides alternately by @xmath228 $ ] and @xmath229 $ ] it is easy to show that @xmath230\rho_{1 } & = & \gamma_{l}'{\cal j}[n_{1}]\rho_{0 } \\ \gamma_{r}{\cal j}[1-n_{1}]\rho_{1 } & = & \gamma_{l}{\cal j}[1-n_{1}]\rho_{0 } .\end{aligned}\ ] ] substituting these into eq .
( [ dotrho0 ] ) yields @xmath231\rho_{0}\ ] ] since the probability or their being an electron on the set is very small we can say @xmath232 . hence we have derived the master equation ( [ me1 ] ) ( without the hamiltonian term ) for the dot alone @xmath233\rho.\ ] ] here we have defined @xmath234 because the set collector reservoir in the two cases ( an electron on the dot and an electron not on the dot ) are independent ( due to the set energy shift ) , the state conditioned on an electron entering the collector is an incoherent mixture of the two possible paths . from quantum trajectory theory
@xcite , the unnormalized state conditioned on this event is @xmath235 + \gamma_{r}'{\cal j}[bn_{1}]\right)r .\ ] ] the norm of this state matrix gives the probability for this event , and is equal to the norm of @xmath236 + \gamma_{r}'{\cal j}[n_{1}]\right)\rho_{1 } .\ ] ] > from the adiabatic elimination procedure above , this is equal to @xmath237 + \gamma_{l}'{\cal j}[n_{1}]\right)\rho .\ ] ] this is the unnormalized state @xmath238 of the dot alone conditioned on an electron tunneling through the set . from this
it is easy to derive the rate of such tunnelings as @xmath239 | we describe the conditional and unconditional dynamics of two coupled quantum dots when one dot is subjected to a measurement of its occupation number using a single electron transistor ( set ) .
the measurement is made when the bare tunneling rate though the set is changed by the occupation number of one of the dots .
we show that there is a difference between the time scale for the measurement - induced decoherence between the localized states of the dots and the time scale on which the system becomes localized due to the measurement .
a comparison between theory and current experiments is made . |
in this paper we introduce a new approach for approximately counting in bounded degree systems with higher - order constraints .
for example , if we are given a cnf formula @xmath0 with @xmath3 variables and @xmath4 clauses with the property that each clause contains at least @xmath1 variables and each variable belongs to at most @xmath2 clauses we ask : how does @xmath1 need to relate to @xmath2 for there to be algorithms to estimate the number of satisfying assignments to @xmath0 within a @xmath5 multiplicative factor ? in the case of a monotone cnf formula where no variable appears negated , the problem is equivalent to the following : suppose we are given a hypergraph on @xmath3 nodes and @xmath4 hyperedges with the property that each hyperedge contains at least @xmath1 nodes and each node belongs to at most @xmath2 hyper edges . how does @xmath1 need to relate to @xmath2 in order to be able to approximately compute the number of independent sets . here
an independent set is a subset of nodes for which there is no induced hyperedge .
bordewich , dyer and karpinski @xcite gave an mcmc algorithm for approximating the number of hypergraph independent sets ( equivalently , the number of satisfying assignments in a monotone cnf formula ) that succeeds whenever @xmath6 .
bezakova et al .
@xcite gave a deterministic algorithm that succeeds whenever @xmath7 and proved that when @xmath8 it is @xmath9-hard to approximate the number of hypergraph independent sets even within an exponential factor .
more broadly , there is a rich literature on approximate counting problems . in a seminal work ,
weitz @xcite gave an algorithm to approximately count in the hardcore model with parameter @xmath10 in graphs of degree at most @xmath2 whenever @xmath11 and in another seminal work , sly @xcite showed a matching hardness result which was later improved in various respects by sly and sun @xcite and galanis , stefankovic and vigoda @xcite .
these results show that approximate counting is algorithmically possible if and only if there is spatial mixing .
moreover , weitz s result can be thought of as a comparison theorem that spatial mixing holds on a bounded degree graph if and only if it holds on an infinite tree with the same degree bound .
there have been a number of attempts to generalize these results to hypergraphs , many of which follow the approach of defining analogues of the self - avoiding walk trees used in weitz s algorithm @xcite .
however what makes hypergraph versions of these problems more challenging is that spatial mixing fails , even on trees .
and we can see that there are _ exponential _ gaps between the upper and lower bounds , since the algorithms above require @xmath1 to be linear in @xmath2 while the lower bounds only rule out @xmath12 .
we can take another vantage point to study these problems .
bounded degree cnf formulae are also one of the principal objects of study in the lovsz local lemma @xcite which is a celebrated result in combinatorics that guarantees when @xmath13 that @xmath0 has at least one satisfying assignment .
the original proof of the lovsz local lemma was non - constructive and did not yield a polynomial time algorithm for finding such an assignment , even though it was guaranteed to exist .
beck @xcite gave an algorithm followed by a parallel version due to alon @xcite that can find a satisfying assignment whenever @xmath14 .
and in a celebrated recent result , moser and tardos @xcite gave an algorithm matching exactly the existential result .
this was followed by a number of works giving constructive proofs of various other settings and generalizations of the lovsz local lemma @xcite .
however these works leave open the following question : under the conditions of the lovsz local lemma ( i.e. when @xmath1 is logarithmic in @xmath2 ) is it possible to approximately sample from the uniform distribution on satisfying assignments ?
approximate counting and approximate sampling problems are well - known to be related .
when the problem is self - reducible , they are in fact algorithmically equivalent @xcite .
however in our setting the problem is not self - reducible because as we fix variables we could violate the assumption that @xmath1 is at least logarithmic in @xmath2 .
it is natural to hope that under the conditions of the lovsz local lemma , that there is an algorithm for approximate sampling that matches the limits of the existential and now algorithmic results .
however the hardness results of bezkova et al .
@xcite imply that we need at least another factor of two , and that it is @xmath9-hard to approximately count when @xmath12 .
in fact , there is another connection between the lovsz local lemma and approximate counting .
scott and sokal @xcite showed that given the dependency graph of events in the local lemma , the best lower bound on the probability of an event guaranteed to exist by the lovsz local lemma ( i.e. the fraction of satisfying assignments ) is exactly the solution to some counting problem .
harvey , srivastava and vondrk @xcite recently adapted techniques of weitz to complex polydisks and gave an algorithm for approximately computing this lower bound .
this yields a lower bound on the fraction of satisfying assignments , however the actual number could be exponentially larger .
our main result is an algorithm to approximately count the number of solutions when @xmath1 is at least logarithmic in @xmath2 . in what follows ,
let @xmath15 , @xmath1 and @xmath2 be constants .
we prove : suppose @xmath0 is a cnf formula with at least @xmath1 variables per clause and at most @xmath2 clauses containing any one variable . for any @xmath16
there is a deterministic polynomial time algorithm for approximating the number of satisfying assignments to @xmath0 within a multiplicative @xmath17 factor .
moreover there is a randomized polynomial time algorithm to sample from a distribution that is @xmath18-close in total variation distance to the uniform distribution on satisfying assignments .
this algorithm closes an exponential gap between the known upper @xcite and lower @xcite bounds .
it also shows that under lovsz local lemma - like conditions not only is it possible to efficiently find a satisfying assignment , it is possible to find a random one .
moreover our approach is a significant departure from earlier techniques based either on path coupling @xcite or adapting weitz s approach to non - binary models and hypergraphs @xcite .
the results above appear in theorem [ thm : nonmonotonecounting ] and theorem [ thm : approximatesampling ] .
our approach starts from a thought experiment about what we could do if we had access to a very powerful oracle that could answer questions about the marginal distributions of individual variables under the uniform distribution on satisfying assignments .
we use this oracle and properties of the lovsz local lemma ( namely , bounds it gives on the marginal distribution of individual variables ) to construct a coupling between two random satisfying assignments so that both agree outside some logarithmic sized component . if we knew the distribution on what logarithmic sized component this coupling procedure produces , we could brute force and find the ratio of the number of satisfying assignments with @xmath19 to the number with @xmath20 to compute marginals at @xmath21
. however the distribution of what component the coupling produces depends intimately on the powerful oracle we have assumed that we have access to . instead
, we abstract the coupling procedure as a random root - to - leaf path in a tree that represents the state of the coupling .
we show that at the leaves of this tree , there is a way to fractionally charge assignments where @xmath19 against assignments where @xmath20 .
crucially , doing so requires only brute - force search on a logarithmic sized component .
finally , we show that there is a polynomial sized linear program to find a flow through the tree that produces an approximately valid way to fractionally charge assignments with @xmath19 against ones with @xmath20 , and that any such solution _ certifies _ the correct marginal distribution . from these steps
, we have thus bootstrapped an oracle for answering queries about the marginal distribution .
our main results then follow from utilizing this oracle . in settings where the problem is self - reducible @xcite
it is well - known how to go from knowing the marginal to approximate counting and sampling . in our setting ,
the problem is not self - reducible because setting variables could result in clauses becoming too small in which case @xmath1 would not be large enough as a function of @xmath2 .
we are able to get around this by using the lovsz local lemma once more to find a safe ordering in which to set the variables .
our algorithms have an interesting application in graphical models . directed graphical models are a rich language for describing distributions by the conditional relationships of their variables .
however very little is known algorithmically about learning them or performing basic tasks such as inference @xcite . in most settings , these problems are computationally hard . however we can study an interesting class of directed graphical models which we call _ cause networks_. see figure [ fig : causenetwork ] .
hidden variables .
a sample is generated by choosing each hidden variable to be @xmath22 independently with equal probability , and observing the truth values of each clause .
, width=226 ] in a cause network there is a collection of hidden variables @xmath23 that are chosen independently to be @xmath24 or @xmath25 with equal probability .
there is a collection of @xmath4 observed variables each of which is either an or or an and of several variables or their negations .
our goal is : given a random sample @xmath23 from the model where we observe the truth value of each of the @xmath4 clauses , to sample from the posterior distribution on the hidden variables .
this generalizes graphical models such as the symptom - disease network where the hidden variables represent diseases that a patient may have , and the clauses represent observed symptoms .
we will require the following regularity condition on our observations : a collection of observations is regular if for every observed variable , the corresponding clause is adjacent to ( i.e. shares a variable with ) at most @xmath26 or clauses that are false and at most @xmath26 and clauses that are true .
now , as an immediate corollary we have : [ corr : inference ] given a cause network where each observed variable depends on at least @xmath1 hidden variables , each hidden variable affects at most @xmath2 observed variables and @xmath27 , there is a polynomial time algorithm for sampling from the posterior distribution for any regular collection of observations .
this is a rare setting where there is an algorithm to solve an inference problem in graphical models but @xmath28 the underlying graph does not have bounded treewidth and @xmath29 correlation decay fails .
we believe that our techniques may eventually be applicable to settings where the observed variables are noisy functions of the hidden variables and where the hidden variables are not distributed uniformly .
in this paper , we will be interested in approximately counting the number of satisfying assignments to a cnf formula .
for example , we could be given : @xmath30 let s fix some parameters .
we will assume that there are @xmath3 variables and there are @xmath4 clauses each of which is an or of at least @xmath1 distinct variables .
finally , we will require a degree bound that each variable appears in at most @xmath2 clauses .
we will be interested in the relationships between @xmath1 and @xmath2 that allow us to approximately count the number of satisfying assignments in polynomial time .
the celebrated lovsz local lemma tells us conditions on @xmath1 and @xmath2 where we are guaranteed that there is at least one satisfying assignment : @xcite if @xmath31 then @xmath0 has at least one satisfying assignment .
moser and tardos @xcite gave an algorithm to find a satisfying assignment under these same conditions .
however the assignment that their randomized algorithm finds is not uniform from the set of all satisfying assignments .
our goal is to solve to be able to both approximately count and uniformly sample when @xmath1 is logarithmic in @xmath2 .
there are many more related results , but we will not review them all here .
instead we state a version of the asymmetric local lemma given in @xcite which gives us some control on the uniform distribution on assignments .
let @xmath32 be the collection of clauses in @xmath0 .
let @xmath33 $ ] denote the uniform distribution all assignments i.e. uniform on @xmath34 . finally , for a clause @xmath35 let @xmath36 denote all the clauses that intersect @xmath35 .
we can abuse notation and for any event @xmath37 that depends on some set of the variables , let @xmath38 denote all the clauses that contain any of the variables on which @xmath37 depends .
[ thm : distributionlll ] suppose there is an assignment @xmath39 such that for all @xmath40 we have @xmath41 \leq x(c ) \prod_{b \in \gamma(c ) } \big ( 1 - x(b ) \big ) \ ] ] then there is at least one satisfying assignment .
moreover the uniform distribution @xmath42 on satisfying assignments satisfies that for any event @xmath37
@xmath43 \leq \mbox{pr}[a ] \prod_{b \in \gamma(a ) } \big ( 1 - x(b ) \big ) ^{-1}\ ] ] notice that this inequality is one - sided , as it ought to be .
after all if we take @xmath35 to be some clause , and @xmath37 to be the event that @xmath35 is not satisfied then we know that @xmath44 = 0 $ ] even though @xmath45 $ ] is nonzero .
however what this theorem does tell us is that the marginal distribution of @xmath42 on any variable is close to uniform .
we will establish a quantitative version of this statement in the following corollary : [ corr : uniform ] suppose that @xmath46 .
then for every variable @xmath47 , we have @xmath48 \leq \frac{1}{2 } + \frac{2}{d^5}\ ] ] set @xmath49 for each clause @xmath15 , and consider the event @xmath37 that @xmath50 . now invoking theorem [ thm : distributionlll ] we calculate : @xmath51 & \leq & \mbox{pr}[x_i = t ] \prod_{b \in \gamma(a ) } \big ( 1 - x(b ) \big ) ^{-1 } \\ & \leq & \big ( \frac{1}{2 } \big ) \big ( 1 - \frac{1}{d^6 } \big ) ^{-1 } \leq \frac{1}{2 } + \frac{2}{d^5}\end{aligned}\ ] ] where the last inequality follows because @xmath52 .
an identical calculation works for the event @xmath53 .
all that remains is to check that the condition in theorem [ thm : distributionlll ] holds , which is a standard calculation : if @xmath15 is a clause then @xmath41 \leq \big ( \frac{1}{d^6 } \big ) \big ( 1 - \frac{1}{d^6 } \big ) ^d\ ] ] the left hand side is at most @xmath54 because each clause has at least @xmath1 distinct variables , and the right hand side is at least @xmath55 .
rearranging completes the proof .
notice that @xmath1 is still only logarithmic in @xmath2 but with a larger constant , and by increasing this constant we get some useful facts about the marginals of the uniform distribution on satisfying assignments .
now we are almost ready to define a coupling procedure .
the basic strategy that we will employ is to start from either @xmath19 and @xmath20 , and then sample from the corresponding marginal distribution on satisfying assignments .
if we sample a variable @xmath56 next , then corollary [ corr : uniform ] tells us that regardless of whether @xmath19 or @xmath20 , each clause has at least @xmath57 variables remaining and so the marginal distribution on @xmath56 is still close to uniform . thus we will try to couple the conditional distributions , when starting from @xmath19 or @xmath20 as well as we can , to show that the marginal distribution on variables that are all at least some distance @xmath58 away must converge in total variation distance .
there is , however , an important catch that motivates the need for a fix .
imagine that we continue in this fashion , sampling variables from the appropriate conditional distribution .
we can reach a situation where a clause @xmath15 has all of its variables except @xmath56 set and yet the clause is still unsatisfied .
the marginal distribution on @xmath56 is no longer close to uniform .
hence , reaching small clauses is problematic because then we can not say much about the marginal distribution on the remaining variables and it would be difficult to construct a good coupling . instead , our strategy is to use the lovsz local lemma once more , but to decide on a set of variables in advance which we call _
marked_. [ lemma : marked ] set @xmath59 .
suppose that @xmath60 .
then there is an assignment @xmath61 such that for every clause @xmath15 , it has at least @xmath62 marked and at least @xmath62 unmarked variables
. we will choose each variable to be marked or unmarked with equal probability , and independently .
consider the @xmath4 bad events , one for each clause @xmath15 , that @xmath15 does not have enough marked or enough unmarked variables .
then we have @xmath63 \leq 2 e^{-(\frac{1}{2})(\frac{1}{6})^2 k } = 2 c_0^{-k}\ ] ] which follows from the chernoff bound .
now we can appeal to the lovsz local lemma to get the desired conclusion .
only the variables that are marked will be allowed to be set to either @xmath24 or @xmath25 by the coupling procedure .
the above lemma guarantees that every clause @xmath15 always has enough remaining variables that can make it true that the marginal distribution on any marked variable always is close to uniform . now fix a variable @xmath21 .
we will build up two partial assignments , and will use the notation @xmath64 to indicate that the first partial assignment sets @xmath21 to @xmath24 , and the second one sets @xmath21 to @xmath25 . furthermore we will refer to the conditional distribution that is uniform on all satisfying assignments consistent with the decisions made so far in @xmath65 and @xmath66 . similarly we will refer to the other conditional distribution as @xmath67 . note that these distributions are updated as more variables are set .
we can now state our goal .
suppose we have partial assignments @xmath65 and @xmath68 .
then we will want to write @xmath69 where @xmath70 is the subformula we get after making the assignments in @xmath65 and simplifying i.e. removing variables that are @xmath25 , and deleting clauses that already have a variable set to @xmath24 .
similarly we will want to write @xmath71 finally , we want the following conditions to be met : 1 .
@xmath72 2 .
@xmath73 and @xmath74 share no variables , and similarly for @xmath75 and @xmath74 the crucial point is that if we can find partial assignments @xmath65 and @xmath68 where @xmath70 and @xmath76 meet the above conditions , then the conditional distribution on all variables in @xmath74 is exactly the same .
we will use the notation @xmath77 to denote the conditional distribution of @xmath66 projected onto just the variables in @xmath74 .
then we have : if the above factorization conditions are met , then @xmath78 from the assumption that @xmath79 and because @xmath73 and @xmath74 share no variables , it means that there are no clauses that contain variables from both the subformulas @xmath73 and @xmath74 .
any such clause would prevent us from writing the formula @xmath70 in such a factorized form .
thus the distribution @xmath66 is simply the cross product of the uniform distributions on satisfying assignments to @xmath73 and @xmath74 .
an identical statement holds for @xmath66 which completes the proof .
note that meeting the factorization conditions does _ not _ mean that the _ number _ of satisfying assignments to @xmath70 and @xmath76 are the same .
our goal in this subsection is to give a coupling procedure to generate partial assignments @xmath65 and @xmath68 starting from @xmath19 and @xmath20 respectively , that result in a factorized formula .
in fact , we will set exactly the same set @xmath80 of variables in both , although not all variables will be set to the same value in the two partial assignments and this set @xmath80 will also be random .
there are two important constraints that we will impose on how we construct the partial assignments , that will make it somewhat more tricky .
first , suppose we have only set the variable @xmath21 and next we choose to set the variable @xmath56 in both @xmath65 and @xmath68 .
we will want that the distribution on how we set @xmath56 in the coupling procedure in @xmath65 to match the conditional distribution @xmath66 and similarly for @xmath68 .
now suppose we terminate with some set @xmath80 having been set
. we can continue sampling the variables in @xmath81 from @xmath66 , and we are now guaranteed that the full assignment we generate is uniform from the set of assignments with @xmath19 .
an identical statement holds when starting with @xmath20 .
second , we will want that with very high probability , the coupling procedure terminates with not too many variables in the formula @xmath73 or @xmath75 .
finally , we will assume that we are given access to a powerful oracle : we will call the following a _ conditional distribution _ oracle : given a cnf formula @xmath0 , a partial assignment @xmath82 and a variable @xmath56 it can answer with the probability that @xmath83 in a uniformly random satisfying assignment that is also consistent with @xmath82 such an oracle is obviously very powerful , and it is well known that if we had access to it we could compute the number of satisfying assignments to @xmath0 exactly with a polynomial number of queries .
however one should think of the coupling procedure as a though experiment , which will be useful in an indirect way to build up towards our algorithm for approximate counting . 1 .
using lemma [ lemma : marked ] , label variables as marked or unmarked 2 .
initialize @xmath84 and @xmath85 3 .
initialize @xmath86 and @xmath87 4 . while there is a clause @xmath15 with variables in both @xmath88 and @xmath89 5 .
@xmath90 sequentially sample its marked variables ( if any ) from @xmath66 and @xmath67 , using @xmath25 to construct best coupling 6 .
@xmath90 case # 1 : @xmath15 is satisfied by variables already set in both @xmath65 and @xmath68 7 .
@xmath90 @xmath90 let @xmath80 be the variables in @xmath15 that have different truth values in @xmath65 and @xmath68 .
@xmath90 @xmath90 update @xmath91 , @xmath92 9 .
@xmath90 @xmath90 delete @xmath15 10 .
@xmath90 case # 2 : @xmath15 is not satisfied by variables already set in either @xmath65 or @xmath68 11 .
@xmath90 @xmath90 let @xmath80 be all variables in @xmath15 ( marked or unmarked ) 12 .
@xmath90 @xmath90 update @xmath91 , @xmath92 13 .
end notice that a clause @xmath15 can only trigger the while loop at most once .
if it ends up in case # 1 then it is deleted from the formula .
if it ends up in case # 2 then all its variables are included in @xmath88 and once a variable is included in @xmath88 it is never removed .
thus the procedure clearly terminates .
our first step is to show that when it does , the formula factorizes .
let @xmath93 be the set of remaining clauses which have all of their variables in @xmath88 .
similarly let @xmath94 be the set of remaining clauses which have all of their variables in @xmath89 . then set @xmath95 and let @xmath73 and @xmath75 be the simplification of @xmath96 with respect to the partial assignments @xmath65 and @xmath68 .
similarly set @xmath97 and let @xmath98 and @xmath99 be the simplification of @xmath100 with respect to the partial assignments @xmath65 and @xmath68 . [
claim : vi ] all variables with different truth assignments in @xmath65 and @xmath68 are in @xmath88 .
a variable is set in response to it being contained in some clause @xmath15 that triggers the while loop .
any such variable is moved into @xmath88 in both case # 1 and case # 2 .
now we have an immediate corollary that helps us towards proving that we have found partial assignments for which @xmath0 factorizes : @xmath101 recall that @xmath98 and @xmath99 come from simplifying @xmath100 ( which contains only variables in @xmath89 ) according to @xmath65 and @xmath68 . from claim [ claim : vi ] , we know that @xmath65 and @xmath68 are the same restricted to @xmath89 and thus we get the same formula in both cases .
now that we know they are equal , we can define @xmath102 .
what remains is to show that the subformulas we have are actually factorizations of the original formula @xmath0 : [ lemma : isfactored ] @xmath79 and @xmath103 when the while loop terminates , every clause @xmath15 in the original formula @xmath0 either has all of its variables in @xmath88 or in @xmath89 , or was deleted because it already contains at least one variable in both @xmath65 and @xmath68 that satisfies it ( although it need not be the same variable ) .
hence every clause in @xmath0 that is not already satisfied in both @xmath65 and @xmath68 shows up in @xmath104 .
some clauses that are already satisfied in both may show up as well . in any case , this completes the proof because the remaining operation just simplifies the formulas according to the partial assignments .
what remains is to bound the probability that the number of variables included in @xmath88 is at most @xmath105 .
first we need an elementary definition : when a variable @xmath47 is given different truth assignments in @xmath65 and @xmath68 , we call it a _ type @xmath106 _ error .
when a clause @xmath15 has all of its marked variables set in both @xmath65 and @xmath68 , but in at least one of them is not yet satisfied , we call it a _ type @xmath107 _ error . note that it is possible for a variable to participate in both a type @xmath106 and type @xmath107 error . in any case , these are the _ only _ reasons that a variable is included in @xmath88 in an execution of the coupling procedure : [ obs : errors ] all variables in @xmath88 are included either due to a type @xmath106 error or a type @xmath107 error , or both
. now our approach to showing that @xmath88 contains not too many variables with high probability is to show that if it did , there would be a large collection of disjoint errors .
first we construct a useful graph underlying the process : let @xmath108 be the graph on vertices @xmath88 where we connect variables if and only if they appear in the same clause together ( _ any _ clause from the original formula @xmath0 ) .
the crucial property is that it is connected : @xmath108 is connected this property holds by induction .
assume that at the start of the while loop , the property holds .
then at the end of the loop , any variable @xmath47 added to @xmath88 must have been contained in a clause @xmath15 that at the outset had one of its variables in @xmath88 .
this completes the proof .
now by observation [ obs : errors ] , for every variable in @xmath88 we can blame it on either a type @xmath106 or a type @xmath107 error . both of these types of errors are unlikely .
but for each variable , charging it to an error is problematic because of overlaps in the events . in particular , suppose we have two variables @xmath47 and @xmath109 that are both included in @xmath88 .
it could be that both variables are in the same clause @xmath15 which resulted in a type @xmath107 error , in which case we could only charge one of the variables to it .
this turns out not to be a major issue .
the more challenging type of overlap is when two clauses @xmath15 and @xmath110 both experience type @xmath107 errors and overlap . in isolation
, each clause would be unlikely to experience a type @xmath107 error .
but it could be that @xmath15 and @xmath110 share all but one of their marked variables , in which case once we know that @xmath15 experiences a type @xmath107 error , then @xmath110 has a reasonable chance of experiencing one as well .
we will get around this issue by building what we call a @xmath111-tree .
this approach is inspired by noga alon s parallel algorithmic local lemma @xcite where he uses a @xmath112-tree .
we call a graph @xmath24 on subset of @xmath88 a @xmath111-tree if each vertex is distance at least @xmath111 from all the others , and when we add edges between vertices at distance exactly @xmath111 the tree is connected . next
we show that @xmath108 contains a large @xmath111-tree : [ lemma:3tree ] any maximal @xmath111-tree contains at least @xmath113 vertices .
consider a maximal @xmath111-tree @xmath24 .
we claim that every vertex @xmath114 must be distance at most @xmath107 from some @xmath109 in @xmath24 .
if not , then we could take the shortest path from @xmath47 to @xmath24 and move along it , and at some point we would encounter a vertex that is also not in @xmath24 whose distance from @xmath24 is exactly @xmath111 , at which point we could add it , contradicting @xmath24 s maximality .
now for every @xmath47 in @xmath24 , we remove from consideration at most @xmath115 other variables ( all those at distance at most @xmath107 from @xmath47 in @xmath108 ) .
this completes the proof .
now we can indeed charge every variable in @xmath24 to a disjoint error : [ claim : disj ] if two variables @xmath47 and @xmath109 in @xmath24 are the result of type @xmath107 errors for @xmath15 and @xmath110 , then @xmath116 for the sake of contradiction , suppose that @xmath117 .
then since @xmath15 and @xmath110 experience type @xmath107 errors , all of their variables are included in @xmath88 .
this gives a length @xmath107 path from @xmath47 to @xmath109 in @xmath108 , which if they were both included in @xmath24 , would contradict the assumption that @xmath24 is a @xmath111-tree .
we are now ready to prove the main theorem of this section : [ thm : couplingterminates ] suppose that @xmath118
. then @xmath119 \leq \big ( \frac{3}{d } \big ) ^t\ ] ] suppose that @xmath120 .
then by lemma [ lemma:3tree ] we can find a @xmath111-tree @xmath24 with at least @xmath105 vertices .
the probability of any particular @xmath111-tree on @xmath105 vertices can be bounded by : @xmath121 this is because by observation [ obs : errors ] each vertex is caused by either a type @xmath106 or type @xmath107 error ( or both ) .
moreover by claim [ claim : disj ] the clauses that cause the type @xmath107 errors for each vertex in @xmath24 are disjoint . now the first term in the expression above is the probability of a type @xmath106 error , which follows from corollary [ corr : uniform ] .
the second term follows from the fact that each variable is contained in at most @xmath2 clauses each of which could cause a type @xmath107 error , from lemma [ lemma : marked ] which implies that each clause has at least @xmath122 marked variables , and again from corollary [ corr : uniform ] which implies that for any variable @xmath47 , the minimum of the probability it is set to @xmath24 or to @xmath24 in _ either _ @xmath65 or @xmath68 is conservatively at least @xmath123 ( because we chose the best coupling ) .
now it is well - known ( see @xcite ) that the number of trees of size @xmath105 in a graph of degree at most @xmath124 is at most @xmath125 .
moreover if we connect pairs of vertices in @xmath108 that are distance exactly @xmath111 from each other , then we get a new graph @xmath126 whose maximum degree is at most @xmath127 . thus putting it all together we have that the probability that @xmath128 can be bounded by @xmath129 where the last inequality follows from the constraint @xmath118 .
thus we can conclude that with high probability , the number of variables in @xmath88 is at most logarithmic .
we can now brute - force search over all assignments to count the number of satisfying assignments to either @xmath73 or @xmath75 .
the trouble is that we do not have access to the marginal probabilities , so we can not actually execute the coupling procedure .
we will need to circumvent this issue next .
in this section , we give an abstraction that allows us to think about the coupling procedure as a randomly chosen root - to - leaf path in a certain tree whose nodes represent states . first , we make an elementary observation that will be useful in discussing how this tree is constructed .
recall that the coupling procedure chooses _ any _ clause that contains variables in both @xmath88 and @xmath89 and then samples _ all _ marked variables in it .
we will assume without loss of generality that the choices it makes are done in lexicographic order .
so if the clauses in @xmath0 are ordered arbitrarily as @xmath130 and the variables are ordered as @xmath131 when executing the while loop , if it has a choice of more than one clause it chooses among them the clause @xmath132 with the lowest subscript @xmath133 .
similarly , given a choice of which marked variable to sample next , it chooses among them the @xmath109 with the lowest subscript @xmath134 .
the important point is that now we can think of a state associated with the coupling procedure , which we will denote by @xmath135 .
[ def : states ] the state @xmath135 of the coupling procedure specifies the following : 1 .
the set of remaining clauses @xmath136 i.e. that have not yet been deleted 2 . the partition of the variables into @xmath88 and @xmath89 3
. the set @xmath80 of variables whose values have been set , along with their values in both @xmath65 and @xmath68 4 . the current clause @xmath137 being operated on in the while loop , if any we will assume that the set @xmath138 of marked variables is fixed once and for all .
now the transition rules are that if @xmath137 has any marked variables that are unset , it chooses the lexicographically first and sets it . and when @xmath137 has no remaining marked variables to set , it updates @xmath136 , @xmath88 and @xmath89 according to whether it falls into case # 1 or case # 2 and sets the current clause to empty .
finally , if the current clause is empty then it chooses the lexicographically first clause from @xmath136 which has at least one variable in each of @xmath88 and @xmath89 to be @xmath137 .
finally , we can define the next variable operation : [ def : nextset ] let @xmath139 be the function that takes in a state @xmath135 , transitions to the next state @xmath140 that sets some variable @xmath56 and outputs @xmath141 .
note that some states @xmath135 do not immediately set a variable
e.g. if the next operation is to choose the next clause , or update @xmath136 , @xmath88 and @xmath89 .
these latter transitions are deterministic , so we let @xmath140 be the end resulting state and @xmath56 be the variable that it sets .
now we can define the stochastic decision tree underlying the coupling procedure : [ def : twosided ] given a conditional distribution oracle @xmath25 , the function @xmath142 and a stopping threshold @xmath143 , the associated _ stochastic decision tree _ is the following : 1 .
the root node corresponds to the state where only @xmath21 is set , @xmath84 , @xmath85 , @xmath86 and @xmath87 .
each node has either zero or four descendants .
if the current node corresponds to state @xmath135 , let @xmath144 . then if @xmath145 or if @xmath146 there are no descendants and the current node is a leaf corresponding to the termination of the coupling procedure or @xmath147 being to large .
otherwise the four descendants correspond to the four choices for how to set @xmath56 in @xmath65 and @xmath68 , and are marked with the state @xmath148 which incorporates their respective choices into @xmath140 .
3 . moreover the probability on an edge from a state @xmath140 to a state @xmath148 where @xmath56 has been set as @xmath149 and @xmath150 is equal to @xmath151 and the transition to the state where @xmath152 and @xmath153 has probability @xmath154 finally if @xmath155 then the transition to @xmath149 and @xmath153 is non - zero and is assigned all the remaining probability .
otherwise the transition to @xmath152 and @xmath150 is non - zero and is assigned all the remaining probability .
now we can use the stochastic decision tree to give an alternative procedure to sample a uniformly random satisfying assignment of @xmath0 .
we will refer to the process of starting from the root , and choosing a descendant with the corresponding transition probability , until a leaf node is reached as choosing a random root - to - leaf path " . 1 .
choose a random root - to - leaf path in @xmath80 2 .
choose a uniformly random assignment @xmath156 consistent with @xmath65 3 .
choose a uniformly random assignment @xmath157 consistent with @xmath68 4 .
output @xmath156 with probability @xmath158 $ ] , and otherwise output @xmath157 [ lemma : altgen ] the decision tree sampling procedure outputs a uniformly random satisfying assignment of @xmath0 .
we could alternatively think of the decision tree sampling procedure as deciding on @xmath156 or @xmath157 with probability @xmath159 vs. @xmath160 at the outset . then if we choose @xmath156 , and we only keep track of the choices made for @xmath65 , marginally these correspond to sequentially sampling the assignment of variables from @xmath66 . and
when we reach a leaf node in @xmath80 we can interpret the remaining choices to @xmath156 as sampling all unset variables from @xmath66 .
thus the output in this case is a uniformly random satisfying assignment with @xmath19 .
an identical statement holds for when we choose @xmath157 , and because we decided between them at the outset with the correct probability , this completes the proof of the claim .
now let @xmath135 be the state of a leaf node @xmath161 and let @xmath65 and @xmath68 be the resulting partial assignments .
let @xmath162 be the product of certain probabilities along the root - to - leaf path .
in particular , suppose along the path there is a transition with @xmath56 being set .
let @xmath163 be the probability of the transition to @xmath164
i.e. along the branch that it actually went down . and
let @xmath165 be the probability of the transition to @xmath166
i.e. where @xmath56 is set the same in @xmath65 but is set to the opposite value as it was in @xmath68 .
we let @xmath162 be the product of all @xmath167 over all such decision on the root - to - leaf path .
[ lemma : paths ] let @xmath168 be an assignment that agrees with @xmath65 .
then for the decision tree sampling procedure @xmath169 = p_1\ ] ] the idea behind this proof is to think of the random choice of which of the four descendants to transition to as being broken down into two separate random choices where we first choose @xmath170 and then we choose @xmath171 .
see figure [ fig : releveling ] .
now we can make the random choices in the decision tree sampling procedure in an entirely different order . instead of choosing the transition in the first layer , then the second layer and so on , we instead make _ all _ of the choices in the odd layers .
moreover at each leaf , we choose which assignment consistent with @xmath65 we would output .
this is the first phase .
next we choose whether to output the assignment consistent with @xmath65 or with @xmath68 . finally ,
we make all the choices in the even layers which fixes the root - to - leaf path and then we choose an assignment consistent with @xmath68 .
this is the second phase .
the key point is that once the output @xmath168 is fixed , all of the choices in the first phase are determined , because every time a variable @xmath56 is set it must agree with its setting in @xmath168 .
moreover each leaf node must choose @xmath168 for its assignment consistent with @xmath65 . and
finally , we know that the sampling procedure must output the assignment consistent with @xmath65 because @xmath168 agrees with @xmath65 and not @xmath68 ( because they differ on how they set @xmath21 ) . thus conditioned on outputting @xmath168 the only random choices left are those in the second phase . now the lemma follows because the probability of reaching leaf node @xmath161 is exactly the probability along the path of all of the even layer choices , which is how we defined @xmath162 . that is output by decision tree sampling .
, width=453 ] we can define @xmath172 in an analogous way to how we defined @xmath162 ( i.e. as the product of certain probabilities along the root - to - leaf path ) , and the lemma above shows that @xmath172 is exactly the probability of all the decisions made along the root - to - leaf path conditioned on the output being @xmath168 where @xmath168 agrees with @xmath68 .
the key lemma is the following : [ lemma : balance ] let @xmath173 be the number of satisfying assignments consistent with @xmath65 and let @xmath174 be the number of satisfying assignments consistent with @xmath68 .
then @xmath175 let @xmath161 be a leaf node .
consider a random variable @xmath176 that when we run the decision tree sampling procedure is non - zero if and only if we end at @xmath161 .
moreover let @xmath177 if an assignment with @xmath19 is output , and @xmath178 if an assignment with @xmath20 is output .
then clearly @xmath179 = 0 $ ] .
now alternatively we can write : @xmath180 = { \mathop{\bf e\/}}_a [ { \mathop{\bf e\/}}[z_u | a \mbox { is output}]]\ ] ] where @xmath168 is a uniformly random satisfying assignment of @xmath0 , precisely because of lemma [ lemma : altgen ] .
let @xmath181 be the total number of such assignments
. then @xmath182 = \big ( \frac{n_1}{n } \big ) ( p_1 ) ( 1-q ) + \big ( \frac{n_2}{n } \big ) ( p_2 ) ( -q)\ ] ] this follows because the only assignments @xmath168 that can be output at @xmath161 must be consistent with either @xmath65 or @xmath68 .
note that these are disjoint events because in one of them @xmath19 while in the other @xmath20 .
then once we know that @xmath168 is consistent with @xmath65 ( which happens with probability @xmath183 ) the probability for the decisions made in @xmath68 being such that we reach @xmath161 is exactly @xmath162 , as this was how it was defined .
the final term in the product of three terms is just the value of @xmath176 .
an identical argument justifies the second term . now using the fact that the above expression evaluates to zero and rearranging completes the proof .
the stochastic decision tree that we defined in the previous section is a natural representation of the trajectory of the coupling procedure .
however it has an important drawback that we will remedy here .
its crucial property is captured in lemma [ lemma : balance ] which gives a relation between 1 .
@xmath184 the conditional probability of an assignment consistent with @xmath185 reaching @xmath161 and 2 .
@xmath186 the number of assignments consistent with @xmath185 for @xmath187 .
however @xmath162 is the product of various ratios of probabilities along the root - to - leaf path .
this means that if we think of the transition probabilities as variables , the constraint imposed by lemma [ lemma : balance ] is far from linear and @xmath172 is a multiplication by one of two ratios of probabilities , _ which have different denominators_. for reasons that we will not digress into , this makes it challenging to encode the total probability @xmath162 as a flow in a tree . ] . in this section
, we will transform a stochastic decision tree into two separate trees , that we call _ one - sided stochastic decision trees_. these will have the property that the constraint imposed by lemma [ lemma : balance ] will be linear in the unknown probabilities that we think of as variables . ultimately we will show that any such pair can @xmath188 certify that a given value @xmath159 is within an additive inverse polynomial factor of @xmath189 $ ] and @xmath190 can be constructed in polynomial time through linear programming .
first we explain the transformation from a stochastic decision tree to a one - sided stochastic decision tree .
we will then formally define its properties and what we require of it .
now suppose we are given a stochastic decision tree @xmath80 .
let s construct the one - sided stochastic decision tree @xmath191 that represents the trajectory of the partial assignment @xmath65 .
when we start from the starting state @xmath135 ( see definition [ def : states ] ) , the four descendants of it in @xmath80 will now be four grand - children
. it s immediate descendants will be two nodes @xmath161 and @xmath192 , one representing the choice @xmath149 and one representing @xmath152 , where @xmath56 is the next variable set ( see definition [ def : nextset ] ) .
the two children of @xmath135 in @xmath80 that correspond to @xmath149 will now be the children of @xmath161 and the other two children will now be the children of @xmath192
. we will continue in this way so that alternate layers represent nodes present in @xmath80 and new nodes .
this alone does not change much the semantics of the trajectory .
all we are doing is breaking up the decision of which of the four children to proceed to , into two separate decisions .
the first decision is based on just @xmath65 and the second is based on @xmath68 .
however we will change the semantics of what probabilities we associate with different transitions . for starters , we will work with total probabilities .
so the total probability incoming into the starting node is @xmath106 .
let s see how this works inductively .
let s now suppose that @xmath135 represents the state of some node in @xmath80 ( not necessarily the starting state ) and @xmath161 and @xmath192 are its descendants in @xmath191
. then if the total probability into @xmath135 in @xmath191 is @xmath193 , we place @xmath193 along both the edges to @xmath161 and to @xmath192 .
this is because the decision tree is now from the perspective of @xmath65 , who perhaps has already chosen his assignment uniformly at random from those with @xmath19 but has not set all of those values in @xmath65 .
hence his decision is not a random variable , since given the option of transition to @xmath161 or @xmath192 he must go to whichever one is consistent with his hidden values .
however from this perspective , the choices corresponding to @xmath68 are random because he has no knowledge of the assignment that the other player is working with .
if we have @xmath193 total probability coming into @xmath161 , then the total probability into its two descendants will be @xmath194 and @xmath195 respectively , where @xmath163 and @xmath165 were the probabilities on the transitions in @xmath80 into the two corresponding descendants .
in particular , if @xmath163 is the probability of setting @xmath149 and @xmath150 and @xmath165 is the probability of setting @xmath149 and @xmath153 then @xmath194 is the total probability on the transition from @xmath161 to the descendant where @xmath150 and @xmath195 is the total probability on the transition from @xmath161 to the descendant where @xmath153 .
note that from corollary [ corr : uniform ] we have that @xmath196 this is an important property that we will make crucial use of later .
notice that it is a linear constraint in the total probability .
now we are ready to define a one - sided stochastic decision tree , which closely mirrors definition [ def : twosided ] .
[ def : onesided ] given the function @xmath142 and a stopping threshold @xmath143 , the associated _ one - sided stochastic decision tree _
for @xmath65 is the following : 1 .
the root node corresponds to the state where only @xmath21 is set , @xmath84 , @xmath85 , @xmath86 and @xmath87 .
each node has either two descendants and four _ grand_-descendants or zero descendants .
if the current node @xmath37 corresponds to state @xmath135 , let @xmath144 . then if @xmath145 or if @xmath146 there are no descendants and the current node is a leaf corresponding to the termination of the coupling procedure or @xmath147 being to large .
otherwise the two descendants correspond to the two choices for how to set @xmath56 in @xmath65 .
each of their two descendants correspond to the two choices for how to set @xmath56 in @xmath68 .
each grand - descendant is marked with the state @xmath140 which incorporates their respective choices .
let @xmath193 be the total probability into @xmath37 .
then the total probability into each descendant is @xmath193 .
moreover let the total probability into the grand - descendants with states @xmath149 and @xmath150 and @xmath149 and @xmath153 be @xmath197 and @xmath198 respectively .
then @xmath197 and @xmath198 are nonnegative , sum to @xmath193 and satisfy @xmath199 .
similarly , let the total probability into the grand - descendants with states @xmath152 and @xmath153 and @xmath152 and @xmath150 be @xmath200 and @xmath201 respectively .
then @xmath200 and @xmath201 are nonnegative , sum to @xmath193 and satisfy @xmath202 .
the one - sided stochastic decision tree for @xmath68 is defined analogously , in the obvious way .
finally we record an elementary fact : [ claim : match ] there is a perfect matching between the root - to - leaf paths in @xmath191 and @xmath203 , so that any pair of assignments @xmath156 and @xmath157 that takes a root - to - leaf path @xmath204 in @xmath191 , must also take the root - to - leaf path in @xmath203 to which @xmath204 is matched .
recall that the odd levels in @xmath191 and @xmath203 correspond to the nodes in @xmath80 .
therefore from a root - to - leaf path @xmath204 in @xmath191 we can construct the root - to - leaf path in @xmath80 , which in turn uniquely defines a root - to - leaf path in @xmath203 ( because it specifies which nodes are visited in odd layers , and all paths end on a node in an odd layer ) .
we are now ready to prove one of the two main theorems of this section : [ thm : constructive ] let @xmath158 $ ] and @xmath205 .
then there are two one - sided stochastic decision trees @xmath191 and @xmath203 that for any pair of matched root - to - leaf paths terminating in @xmath161 and @xmath192 respectively satisfy @xmath206 where @xmath173 and @xmath174 are number of satisfying assignments consistent with @xmath65 and @xmath68 respectively , and @xmath162 and @xmath172 are the total probability into @xmath161 and @xmath192 respectively .
moreover given @xmath207 and @xmath208 that satisfy @xmath205 there is an algorithm to construct two one - sided stochastic decision trees @xmath191 and @xmath203 that satisfy the above condition on all matched leaf nodes corresponding to a termination of the coupling procedure , which runs in time polynomial in @xmath4 and @xmath209 where @xmath143 is the stopping size .
the first part of the theorem follows from the transformation we gave from a stochastic decision tree to two one - sided stochastic decision trees .
then claim [ claim : match ] combined with lemma [ lemma : balance ] implies @xmath210 , which then necessarily satisfies @xmath211 .
rearranging completes the proof of the first part . to prove the second part of the theorem , notice that if @xmath143 is the stopping size , then the number of leaf nodes in @xmath191 and in @xmath203 is bounded by @xmath209
at each leaf node that corresponds to a termination of the coupling procedure , from lemma [ lemma : isfactored ] we can compute the ratio of @xmath173 to @xmath174 as the ratio of the number of satisfying assignments to @xmath73 to the number of satisfying assignments to @xmath75 .
this can be done in polynomial in @xmath4 and @xmath212 time by brute - force . finally , the constraints in definition [ def : onesided ] are all linear in the variables that represent total probability ( if we treat @xmath213 , @xmath214 , @xmath215 and all ratios @xmath216 as given constants ) .
thus we can find a valid choice of the total probability variables by linear programming .
this completes the proof of the second part .
recall that we will be able to choose @xmath217 and theorem [ thm : couplingterminates ] will imply that at most an inverse polynomial fraction of the distribution fails to couple .
thus the algorithm above runs in polynomial time for any constants @xmath2 and @xmath1 .
what remains is to show that any valid choice of total probabilities _ certifies _ that @xmath218 \leq q''$ ] .
we are now ready to prove the second main theorem of this section .
we will show that having any two one - sided stochastic decision trees that meet the constraints on the leaves imposed by theorem [ thm : constructive ] is enough to certify that @xmath189 $ ] is approximately between @xmath207 and @xmath208 .
this result will rest on two facts .
fix any assignment @xmath168
. then either 1 .
the assignment has too many clauses that restricted to marked variables are all @xmath25 or 2 .
the total probability of @xmath168 reaching a leaf node @xmath161 where the coupling procedure failed to terminate before reaching size @xmath143 is at most @xmath219 .
[ thm : certify ] suppose that @xmath118 . then any two one - sided stochastic decision trees @xmath191 and @xmath203 that meet the constraints on the leaves imposed by theorem [ thm : constructive ] and satisfy @xmath220 imply that @xmath221 \leq q '' + o\big(\frac{1}{n^c}\big)\ ] ] the proof of this theorem will use many of the same tools that appeared in the proof of theorem [ thm : couplingterminates ] , since in essence we are performing a one - sided charging argument . the proof will proceed by constructing a complete bipartite graph @xmath222 and finding a fractional approximate matching as follows .
the nodes in @xmath223 represent the satisfying assignments of @xmath0 with @xmath19 .
the nodes in @xmath224 represent the satisfying assignments of @xmath0 with @xmath20 .
moreover all but a @xmath219 fraction of the nodes on the left will send between @xmath225 and @xmath226 flow along their outgoing edges .
finally all but a @xmath219 fraction of the nodes on the right will receive between @xmath227 and @xmath228 flow along their incoming edges .
first notice that any assignment @xmath168 ( say with @xmath19 ) is mapped by @xmath191 to a distribution over leaf nodes , some of which correspond to a coupling and some of which correspond to a failure to couple before reaching size @xmath143 .
now consider matched pairs of leaf nodes ( according to claim [ claim : match ] ) that correspond to a coupling .
let @xmath162 and @xmath172 be the total probability of the leaf nodes in @xmath191 and @xmath203 respectively .
let @xmath173 and @xmath174 be the total number of assignments that are consistent with @xmath65 and @xmath68 , and let @xmath229 and @xmath230 be the corresponding sets of assignments . from the assumption that @xmath206 and the intermediate value theorem it follows that there is a @xmath231 which satisfies @xmath232 hence there is a flow that sends exactly @xmath233 units of flow out of each node in @xmath229 and which each node in @xmath230 receives exactly @xmath234 units of flow .
if every leaf node corresponding to a coupling , we would indeed have the fractional matching we are looking for , just by summing these flows over all leaf nodes .
what remains is to handle the leaf nodes that do not correspond to the coupling terminating before size @xmath143 .
consider any such leaf node @xmath161 in @xmath191 and the corresponding leaf node @xmath235 in @xmath203 . from lemma [ lemma:3tree ]
we have that there is a @xmath111-tree @xmath24 of size at least @xmath236 . for each node in @xmath24 , from claim [ claim : disj ] we have there are at least @xmath236 disjoint type @xmath106 or type @xmath107 errors .
* case # 1 : * suppose that there are at least @xmath237 disjoint type @xmath106 errors .
fix the @xmath111-tree @xmath24 , and look at all root - to - leaf paths that are consistent with just the type @xmath106 errors .
then the sum of their total probabilities is at most @xmath238 this follows because the constraint that @xmath239 ( and similarly for @xmath201 ) in definition [ def : onesided ] implies that for each path we can factor out the above term corresponding to just the decisions where there are type @xmath106 errors .
the remaining probabilities are conditional distributions on the paths ( after having taken into account the type @xmath106 errors ) and sum to at most one .
finally the total number of @xmath111-trees of size @xmath236 is at most @xmath240 .
thus for any assignment @xmath168 , if we ignore what happens to it when it ends up at a leaf node which did not couple and which has at least @xmath237 disjoint type @xmath106 errors , in total we have ignored at most @xmath241 of its probability .
* case # 2 : * suppose that there are at least @xmath237 disjoint type @xmath107 errors .
each type @xmath107 error can be blamed on either @xmath65 or @xmath68 or both ( e.g. it could be that the clause @xmath15 might only have all of its marked variables set to @xmath25 in @xmath65 ) . let s suppose that the assignment @xmath168 contributes at least @xmath242 disjoint type @xmath107 errors . in this case we will completely ignore @xmath168 in the constraints imposed by our flow .
how many such assignments can there be ?
the probability of getting any such assignment is bounded by @xmath243 thus if we ignore the flow constraints for all such assignments , we will be ignoring at most a @xmath18 fraction of the nodes in @xmath223 and the nodes in @xmath224 . the only remaining case is when the assignment @xmath168 ends up at a leaf node @xmath161 that has at least @xmath237 disjoint type @xmath107 errors , but it contributes less than @xmath242 itself . for each type @xmath107 error that it does not contribute to , it contributes to another type @xmath106 error .
the only minor complication is that the node responsible might not be in the @xmath111-tree @xmath24 .
however it is distance at most @xmath106 from the @xmath111-tree because it is contained in a clause that results in type @xmath107 error that does contain a node in @xmath24 .
now by an analogous reasoning as in case # 1 above , if we fix the pattern of these type @xmath106 errors
i.e. we fix the @xmath111-tree and the extra nodes at distance @xmath106 from it that contribute the missing type @xmath106 errors the sum of the total probability of all consistent root - to - leaf paths is at most @xmath238 now the number of patterns can be bounded by @xmath244 , which accounts for the inclusion of extra nodes that are not in @xmath24 . once again , for such an assignment @xmath168 if we ignore what happens to it when it ends up at a leaf node which did not couple and which has at least @xmath237 disjoint type @xmath107 but it contributes less than @xmath242 itself , in total we have ignored at most @xmath241 of its probability . now returning to the beginning of the proof and letting @xmath173 and @xmath174 be the total number of satisfying assignments with @xmath19 and @xmath20 respectively .
we have that the flow in the bipartite graph implies @xmath245 and the further condition @xmath246 which gives @xmath247 which completes the proof of the theorem .
here we show how to use our algorithm for computing marginal probabilities when @xmath1 is logarithmic in @xmath2 for approximate counting and sampling from the uniform distribution on satisfying assignments .
first , we show how to use an algorithm for computing marginal probabilities to do approximate counting in a monotone cnf , where no variable is negated .
this approach is standard , and appears in @xcite .
[ corr : monotonecount ] suppose we are given a monotone cnf formula @xmath0 on @xmath3 variables with at least @xmath1 variables per clause and at most @xmath2 clauses containing any one variable with @xmath118 .
let @xmath248 be the number of satisfying assignments .
then there is an algorithm that outputs a quantity @xmath249 that satisfies @xmath250 and runs in time polynomial in @xmath4 and @xmath251 .
first , we fix an ordering of the variables @xmath23 and a sequence of formulas @xmath252 .
let @xmath253 and let @xmath254 be the subformula we get when substituting @xmath255 into @xmath0 and simplifying .
notice that each such formula is a monotone cnf and inherits the properties we need from @xmath0 .
in particular , each clause has at least @xmath1 variables because the only clauses left in @xmath256 ( i.e. not already satisfied ) are the ones which have all of their variables unset .
also , each variable belongs to at most @xmath2 clauses because we have only removed variables and clauses .
thus we can appeal to theorem [ thm : constructive ] and theorem [ thm : certify ] to compute for each variable @xmath47 the quantity @xmath257 $ ] to within an additive @xmath258 where @xmath259 is the uniform distribution on satisfying assignments to @xmath256 .
let our estimate be @xmath260 .
since @xmath261 we have that @xmath260 and @xmath184 are also multiplicatively close , with @xmath262 . now
if we take the product of the @xmath184 s we get a telescoping product which computes the ratio of the number of satisfying assignments with all variables set to @xmath24 divided by the number of satisfying assignments to @xmath0 .
moreover each @xmath261 .
thus we conclude @xmath263 which completes the proof .
the above approach heavily used monotonicity to ensure that no clause becomes too small ( i.e. contains few variables , but is still unsatisfied ) .
this is a similarly issue to what happened with the coupling procedure , which necessitating using _ marked _ and _ unmarked _ variables , the latter being variables that are never set and are used to make sure no clause becomes too small .
we can take a similar approach here . in what follows
we will no longer assume @xmath0 is monotone .
[ lemma : nonmonotonehelper ] set @xmath264 .
suppose that @xmath265 .
then there is a partial assignment @xmath82 so that every clause is satisfied and each clause has at least @xmath122 unset variables .
moreover there is a randomized algorithm to find such a partial assignment that runs in time polynomial in @xmath4 , @xmath3 , @xmath1 and @xmath2 .
alternatively there is a deterministic algorithm that runs in time polynomial in @xmath4 and @xmath266 .
we will choose independent for each variable to set it to @xmath24 with probability @xmath267 , to set it to @xmath25 with probability @xmath267 and to leave it unset with probability @xmath268 .
now consider the @xmath4 bad events , one for each clause @xmath15 , that @xmath15 is either unsatisfied or has not enough unset variables ( or both ) .
then we have @xmath63 \leq e^{-(\frac{1}{2})(\frac{1}{6})^2 k } + \big ( \frac{3}{4 } \big ) ^k \leq 2 c_0^{-k}\ ] ] where the first term follows from the chernoff bound and represents the probability that there are not enough unset variables , and the second term is the probability that the clause is unsatisfied . once again we can appeal to the lovsz local lemma to show the existence . finally we can use the algorithm of moser and tardos @xcite to find such a partial assignment in randomized polynomial time .
moreover moser and tardos @xcite also give a deterministic algorithm that runs in time polynomial in @xmath4 and @xmath266 .
[ thm : nonmonotonecounting ] suppose we are given a cnf formula @xmath0 on @xmath3 variables with at least @xmath1 variables per clause and at most @xmath2 clauses containing any one variable with @xmath118 .
let @xmath248 be the number of satisfying assignments .
then there is a deterministic algorithm that outputs a quantity @xmath249 that satisfies @xmath250 and runs in time polynomial in @xmath4 and @xmath251 .
our proof follows the same basic outline as in corollary [ corr : monotonecount ] .
first we ( deterministically ) find a partial assignment that meets lemma [ lemma : nonmonotonehelper ] and let @xmath269 be an ordering of the set variables .
we define @xmath270 in the same way as the subformula we get by substituting in the assignments for @xmath271 and simplifying to get @xmath256 .
again let @xmath260 be our estimate for the marginal probabilities .
the key point is that @xmath272 would be empty , because all clauses are satisfied .
moreover each clause that appears in any formula @xmath256 for @xmath273 has at least @xmath122 variables because it has at least that many unset variables in the partial assignment .
moreover we can now output @xmath274 because @xmath272 has exactly @xmath275 satisfying assignments ( every choice of the unset variables ) and we have used the same telescoping product , but now to compute the ratio of the number of satisfying assignments to @xmath272 divided by the number of satisfying assignments to @xmath0 . here
we give an algorithm to generate an assignment approximately uniformly from the set of all satisfying assignments .
again , the complication is that our oracle for approximating the marginals works only if @xmath1 is at least logarithmic in @xmath2 so we need some care in the order we choose to sample variables .
first we give the algorithm : 1 .
using lemma [ lemma : marked ] , label variables as marked or unmarked 2 . while there is a marked variable @xmath21 that is unset 3 .
@xmath90 sample @xmath21 using @xmath25 4 .
@xmath90 initialize @xmath86 and @xmath89 to be all unset variables ( @xmath21 is already set ) 5 .
@xmath90 while there is a clause @xmath15 with variables in both @xmath88 and @xmath89 6 .
@xmath90 @xmath90 sequentially sample its marked variables ( if any ) using @xmath25 7 .
@xmath90 @xmath90 case # 1 : @xmath15 is satisfied 8 .
@xmath90 @xmath90 @xmath90 delete @xmath15 9 .
@xmath90 @xmath90 case # 2 : @xmath15 is unsatisfied 10 .
@xmath90 @xmath90 @xmath90 let @xmath80 be all variables in @xmath15 ( marked or unmarked ) 11 .
@xmath90 @xmath90 @xmath90 update @xmath91 , @xmath92 12 .
@xmath90 end 13 .
end 14 . for each connected component of the remaining clauses 15 .
@xmath90 enumerate and uniformly choose a satisfying assignment of the unset variables 16 .
end [ lemma : approxunif ] if the oracle @xmath25 outputs a marginal probability that is @xmath276 close to the true marginal distribution for each variable queried , then the output of the sampling procedure is a random assignment whose distribution is @xmath18-close in total variation distance to the uniform distribution on all satisfying assignments .
the proof of this lemma is in two parts .
first , imagine we were instead given access to an oracle @xmath108 that answered each query for a marginal distribution with the exact value .
then each variable set using the oracle is chosen from the correct marginal distribution .
and in the last step , the set of satisfying assignments is a cross - product of the satisfying assignments for each component .
thus the procedure would output a uniformly random assignment from the set of all satisfying assignments .
second , since at most @xmath3 variables are queried , we have that with probability at least @xmath277 all of the random decision of the procedure would be the same if we had given it answers from @xmath108 instead of from @xmath25 .
this now completes the proof .
[ thm : approximatesampling ] suppose we are given a cnf formula @xmath0 on @xmath3 variables with at least @xmath1 variables per clause and at most @xmath2 clauses containing any one variable with @xmath278 .
there is an algorithm that outputs a random assignment whose distribution is @xmath18-close in total variation distance to the uniform distribution on all satisfying assignments .
moreover the algorithm runs in time polynomial in @xmath4 and @xmath251 .
the proof of this theorem uses many ideas from the coupling procedure as analyzed in section [ sec : couple ] .
let @xmath279 be the formula at the start of some iteration of the inner while loop .
then at the end of the inner while loop , using lemma [ lemma : isfactored ] we can write : @xmath280 where @xmath96 is a formula on the variables in @xmath88 and @xmath100 is a formula on the variables in @xmath89 .
in particular , no clause has variables in both because the inner while loop terminated .
now we can appeal to the analysis in theorem [ thm : couplingterminates ] which gives a with high probability bound on the size of @xmath88 .
the analysis presented in its proof is nominally for a different procedure , the coupling procedure , but the inner while loop of the sampling procedure is identical except for the fact that there are no type 1 errors because we are building up just one assignment .
thus @xmath119 \leq \big ( \frac{3}{d } \big ) ^t\ ] ] the inner while loop is run at most @xmath3 times and so if we choose @xmath281 we get that with probability at least @xmath282 no component has size larger than @xmath283 . now
the brute force search in the last step can be implemented in time polynomial in @xmath4 and @xmath251 , which combined with lemma [ lemma : approxunif ] completes the proof .
recall , we are given a cause network and the truth assignment of each observed variable .
first we do some preprocessing . if an observed variable is an or of several hidden variables or their negation , and the observed variable is set to @xmath25 we know the assignment of each hidden variable on which it depends .
similarly , if an observed variable is an and and it is set to @xmath24 again we know the assignment of each of its variables . for all the remaining observed variables ,
we know there is exactly one configuration of its variables that is prohibited so each yields a clause in a cnf formula @xmath0 .
moreover each clause depends on at least @xmath284 variables whose truth value has not been set because the collection of observations is regular .
finally each variable is contained in at most @xmath2 clauses .
the posterior distribution on the remaining hidden variables ( whose value has not already been set ) is uniform on the set of satisfying assignments to @xmath0 and thus we can appeal to theorem [ thm : approximatesampling ] to complete the proof . | in this paper we introduce a new approach for approximately counting in bounded degree systems with higher - order constraints .
our main result is an algorithm to approximately count the number of solutions to a cnf formula @xmath0 with at least @xmath1 variables per clause and degree at most @xmath2 when @xmath1 is logarithmic in @xmath2 .
this closes an exponential gap between the known upper and lower bounds .
moreover our algorithm extends straightforwardly to approximate sampling , which shows that under lovsz local lemma - like conditions it is not only possible to find a satisfying assignment , it is also possible to generate one approximately uniformly at random from the set of all satisfying assignments .
our approach is a significant departure from earlier techniques in approximate counting , and is based on a framework to bootstrap an oracle for computing marginal probabilities on individual variables .
finally , we give an application of our results to show that it is algorithmically possible to sample from the posterior distribution in an interesting class of graphical models . |
the lightest nuclei can be accurately modeled as systems of nucleons interacting via effective two- and three - body forces motivated e.g. by meson exchange .
this picture is expected to break down at a higher energy scale , where the physics is more efficiently described in terms of subnuclear degrees of freedom .
few - body methods have been an essential tool for determining model hamiltonians that describe low - energy nuclear physics .
they also have the potential to be a useful framework for testing the limitations of viewing the nucleus as a few nucleon system .
the latter requires extending few - body models and calculations to higher energies . in order to successfully do this ,
a number of challenges need to be addressed .
these include replacing the nonrelativistic theory with a relativistic one , overcoming limitations imposed by interactions fit to elastic nucleon - nucleon ( nn ) scattering data , including new degrees of freedom that appear above the pion production threshold , as well as numerical problems related to the proliferation of partial waves characteristic for scattering calculations at higher energies .
thus the intermediate energy regime is a new territory for few - body calculations which waits to be explored . in this paper
we address two of the challenges .
we demonstrate that it is now possible to perform converged three - body scattering calculations at energies up to 2 gev laboratory kinetic energy .
key elements are a consistent implementation of a poincar symmetric quantum theory @xcite , and the use of direct integration methods that avoid the partial wave decomposition , successfully applied below the pion - production threshold @xcite . in a series of publications @xcite
the technique of solving the nonrelativistic momentum - space faddeev equation without partial waves has been mastered , for bound as well as scattering states .
the relativistic faddeev equation , based on a poincar invariant mass operator , has been formulated in detail in @xcite , showing that it has both kinematical and dynamical differences with respect to the corresponding nonrelativistic equation .
the formulation of the theory is given in a representation of poincar invariant quantum mechanics where the interactions are invariant with respect to kinematic translations and rotations @xcite .
the model hilbert space is a three - nucleon hilbert space ( thus not allowing for absorptive processes ) .
the method used to introduce the nn interactions in the unitary representation of the poincar group allows input of high - precision nn interactions @xcite in a way that reproduces the measured two - body observables . however in this study we use a simpler , spin - independent interaction consisting of a superposition of an attractive and a repulsive yukawa interaction that supports a bound state with the deuteron binding energy .
this is mathematically equivalent to three - boson scattering .
poincar invariance and @xmath0-matrix cluster properties dictate how the two - body interactions must be embedded in the three - body dynamical generators .
scattering observables are calculated using the faddeev equation formulated with the mass casimir operator ( rest hamiltonian ) constructed from these generators .
we want to point out that the relativistic faddeev equation with realistic spin - dependent interactions has been solved below the pion - production threshold in a partial wave basis @xcite . in order to estimate the size of relativistic effects the interactions employed in the nonrelativistic and relativistic calculations presented here
are chosen to be two - body phase shift equivalent .
this is achieved in this article by adding the interaction to the square of the mass operator @xcite .
thus differences in relativistic and nonrelativistic calculations first appear in the three - body calculations .
those differences are in the choice of kinematic variables ( jacobi momenta are constructed using lorentz boosts rather than galilean boosts ) and in the embedding of the two - body interactions in the three - body problem , which is a consequence of the non - linear relation between the two and three - body mass operators .
these differences modify the permutation operators and the off - shell properties of the kernel of the faddeev equation .
this article is organized as follows . in section
ii the formulation of the poincar invariant faddeev equation is given and numerical aspects for computing the faddeev kernel are discussed . in sections
iii and iv we present calculations for elastic and breakup processes in the intermediate energy regime from 0.2 to 1.5 gev .
our focus here is the investigation of the convergence of the multiple scattering series as a function of projectile kinetic energy .
we compare our calculations to selected breakup observables and investigate a simple approximation of the embedding of the two - body interaction into the three - body problem .
a detailed formulation of poincar invariant three - body scattering has been given in @xcite , where the driving term in the relativistic faddeev equation ( first order in the two - body transition operator ) has been used to evaluate cross sections for elastic as well as break - up scattering .
this is now being complemented by fully solving the relativistic faddeev equation based on the numerical techniques previously used to solve the non - relativistic faddeev equation @xcite . for the convenience of the reader essential equations
are repeated , but for the detailed derivation of the expressions we refer to ref .
@xcite .
the symmetrized transition operators @xmath1 for elastic scattering and @xmath2 for breakup reactions can be expressed in terms of the solution @xmath3 of the symmetrized faddeev equations @xmath4 where @xmath5 is the invariant mass operator for three non - interacting particles and the permutation operator @xmath6 is given by @xmath7 .
the operator @xmath3 is the solution to the symmetrized faddeev equation @xmath8 where the operator @xmath9 is the two - body transition operator embedded in the three - particle hilbert space and defined as the solution to @xmath10 here @xmath11 is the two - body interaction embedded in the three - body hilbert space and @xmath12 is the invariant mass operator for two interacting particles and a spectator . for calculating transition matrix elements ,
explicit basis states need to be introduced .
the momenta of the three particles can be labeled either by single - particle momenta @xmath13 , @xmath14 , and @xmath15 , or the total momentum @xmath16 and the relativistic poincar - jacobi momenta @xcite @xmath17 and @xmath18 .
the explicit relations between the three - body poincar jacobi momenta and the single particle momenta are @xmath19 then the poincar - jacobi momenta @xmath17 and @xmath18 are given as @xmath20 where @xmath21 .
in addition , the transformation from the single particle momenta @xmath22 to the poincar - jacobi momenta has a jacobian given by @xmath23 where for @xmath24 the jacobian becomes @xmath25 in the above expression we chose , without loss of generality , particle 1 as the spectator .
the poincar - jacobi momenta are relevant for the calculation of the permutation operator @xmath6 in eqs .
( [ eq:2.1 ] ) and ( [ eq:2.2 ] ) .
the matrix elements of the permutation operator are then explicitly calculated as @xmath26 , \label{eq:2.7}\end{aligned}\ ] ] where the function @xmath27 contains the product of two jacobians and reads @xmath28{(e(\mathbf{q } ) + e(\mathbf{q } + \mathbf{q}'))^2 - \mathbf{q}'^2 } \;\ ; \sqrt[4]{(e(\mathbf{q } ' ) + e(\mathbf{q } + \mathbf{q}'))^2 - \mathbf{q}^2 } } { \sqrt{e(\mathbf{q})e(\mathbf{q } ' ) } } .
\label{eq:2.8}\end{aligned}\ ] ] with @xmath29 .
the function @xmath30 is calculated as @xmath31 these permutation operators , which change the order of coupling , are essentially racah coefficients for the poincar group . in the nonrelativistic case the functions @xmath27 and @xmath32 both reduce to the constant 1 and have the relatively compact form of the matrix elements of @xmath6 given in e.g. @xcite . in matrix form the faddeev equation , eq .
( [ eq:2.2 ] ) , reads @xmath33 where we have factored out a delta function in the total momentum and set @xmath34 . inserting complete sets of states and explicitly evaluating
the permutation operator leads to @xmath35 the quantities @xmath36 and @xmath37 are determined by the laboratory kinetic energy @xmath38 of the incident nucleon , @xmath39 the nucleon rest mass is given by @xmath40 , the rest mass of the deuteron is @xmath41 , where @xmath42 is the deuteron binding energy .
the poincar jacobi momentum between projectile and target , @xmath43 , is related to @xmath38 by @xmath44 the invariant parametric energy @xmath45 which enters the two - body t - matrix is given by @xmath46 . since we consider bosons , we introduce the symmetrized two - body transition matrix @xmath47 @xmath48 this two - body t - matrix has a simple pole at @xmath49 .
thus , for the practical calculation we need to take this pole explicitly into consideration by defining @xmath50 and solving eq .
( [ eq:2.11 ] ) for @xmath51 @xmath52 for the explicit calculation we introduce the independent variables @xcite @xmath53 so that @xmath54 , is a function of 5 variables . in the variables of eq .
( [ eq:2.17 ] ) and defining @xmath55 , the final expression for eq .
( [ eq:2.11 ] ) reads @xmath56 while the deuteron pole can be numerically taken care of with a single subtraction in the @xmath57-integration , the free three - nucleon propagator in the 2nd term under the integral of eq .
( [ eq:2.19 ] ) contains singularities depending on @xmath57 as well as @xmath58 leading to a singular region in the @xmath59 plane . in order to simplify the calculation
, we carry out the integration of the kernel in a frame in which the @xmath60-axis is along the direction of @xmath17 . in this frame @xmath61 and @xmath62
is the azimuthal angle of @xmath63 . with these definitions
one has @xmath64 where @xmath65 and @xmath66 are the azimuthal angles of @xmath18 and @xmath37 in the frame described above .
since there is a freedom in choosing the @xmath67-axis , we may place @xmath37 in the @xmath68-plane , this gives @xmath69 . with this choice @xmath65
is evaluated as @xmath70 the remaining variables in eq .
( [ eq:2.19 ] ) are explicitly evaluated as @xmath71 and @xmath72 and @xmath73 with @xmath74 for the integration of the 3n propagator , each singularity in the @xmath58-integration ( for fixed @xmath57 ) is explicitly taken into account by a subtraction .
however , this leads to logarithmic singularities in @xmath57 at the boundaries @xmath75 .
these we integrate in the semi - analytic fashion introduced in ref .
@xcite by using cubic hermite splines .
while using cubic hermite splines is advantageous in dealing with the logarithmic singularities , this method is not as effective as gauss - legendre quadrature when integrating over large , non - singular regions .
thus , in order to make the most efficient use of both methods , we divide the interval of the @xmath57-integration into several integration regions , and use gauss - legendre quadrature in the non - singular integrals , while keeping the cubic hermite splines in the small regions around the singularities . with this procedure
we are able to successfully integrate over the faddeev kernel with sufficient accuracy .
for the final solution of eq .
( [ eq:2.19 ] ) the kernel is successively applied and the resulting terms are summed up as pad sums . at the higher energies we will also carry out the neumann sum .
our explicit calculations are based on a simple interaction of malfliet - tjon type consisting of a superposition of an attractive and repulsive yukawa interaction that supports a bound state with the deuteron binding energy .
the parameters of this nonrelativistic interaction are given in ref .
@xcite . in order to obtain a relativistic interaction which is phase shift equivalent with the nonrelativistic one
, we employ a scheme in which @xmath76 multiplied with the interaction is added to the square of the non - interacting two - body mass operator .
this procedure was introduced by coester , pieper and serduke @xcite and used here in the form given in @xcite .
it guarantees that differences in the relativistic and nonrelativistic calculations first appear in the three - body calculations . before entering a detailed study on relativistic effects ,
we want to present further details on the numerical quality of our solution of the relativistic faddeev equation .
one internal consistency check of the solution is provided by the optical theorem , which states that the total cross section , being the sum of the total elastic cross section , @xmath77 , and the total breakup cross section , @xmath78 , must be equal to the imaginary part of the transition operator for elastic scattering @xmath79 in forward direction .
in the center - of - momentum ( c.m . ) frame this relation reads @xmath80 listed in table [ table-1 ] are our fully relativistic calculations of the total cross sections for elastic scattering and breakup reaction for projectile energies from 0.1 to 2.0 gev , together with the total cross sections .
the total cross sections are calculated as sum of the elastic and breakup cross sections , @xmath81 , and via the optical theorem , @xmath82 , from the imaginary part of the operator @xmath79 in forward direction , @xmath83 . a comparison of those two numbers for the total cross section shows , that our calculations fulfill the optical theorem to about 1% or better up to 1 gev .
this error increases to about 3% at 2 gev .
here we did not push the calculations any further , since our model potential is too simple to take it to much higher energies anyway . for the sake of showing the numerical quality of our calculations , we included 2 gev in table [ table-1 ] , but will not show any further observables at this energy .
the transition amplitude of eq .
( [ eq:2.19 ] ) is a function of 5 variables , and is the solution of an integral equation in three dimensions .
thus , in the calculation the dependence of the result on the various choices of grids has to be considered .
as far as the momentum grids are concerned , the accuracy of the calculation is most sensitive to the @xmath84-grid , as already found in ref . @xcite . in fig .
[ fig1 ] we show the dependence of the relative error @xmath85 in the optical theorem as function of the size of the @xmath84-grid , @xmath86 , for a calculation at 1 gev projectile laboratory kinetic energy .
the slope of @xmath87 shows that indeed the accuracy of the calculation is strongly influenced by the size of this grid . for our calculation , @xmath86 = 50 is sufficient at 1 gev . next , we consider the sensitivity of the calculation to the size of different angle grids . in table
[ table-2 ] we give the cross sections for elastic scattering and breakup reactions together with the total cross section @xmath82 extracted from the optical theorem when varying the size of the different angle grids .
we can see , that the results are most sensitive with respect to the grids in @xmath88 and @xmath58 .
it is common wisdom in calculations using an angular momentum basis that as the energy of the projectile increases , the number of partial waves needed to obtain a converged result increases rather quickly . in our 3d calculations ,
all partial waves are included .
the increase in energy manifests itself in a two - body t - matrix acquiring a more pronounced peak structure in the forward and backward directions with respect to the angle between the two momentum vectors @xcite .
this peak structure at @xmath89 must be adequately covered in calculations at higher energies to ensure converged results . in fig .
[ fig2 ] we show the relative error @xmath90 in the optical theorem as function of the size of the @xmath88-grid for three different projectile laboratory kinetic energies .
the necessity of increasing the @xmath88-grid with increasing projectile energy is clearly seen .
whereas for 0.2 gev @xmath91 = 24 is clearly sufficient , at 0.5 gev one needs already at least 28 points , whereas at 1 gev a minimum of 36 points is required .
this conclusion is also reached in our table [ table-3 ] , where we show the relativistic differential cross section at selected angles while varying the @xmath88-grid .
it should be noted that the angle @xmath88 is related to the angular momentum of the relative motion between the spectator and the interacting pair .
the angle @xmath92 , which is related to the angular momentum of the interacting pair is not nearly as sensitive as @xmath88 . in table
[ table-2 ] we vary the @xmath92 grid from 20 to 24 points , and see hardly any difference . it is illustrative to contrast the computational algorithm for direct integration with the experience gained when using a partial wave basis in the 3n system .
our experience tells us that at @xmath38 = 200 mev the total angular momentum of the 2n subsystem @xmath93 needs to be @xmath94 = 5 to reach convergence .
furthermore , the maximum total angular momentum @xmath95 of the 3n system required to reach convergence is @xmath96 = 25/2 .
let us assume that @xmath97 , where @xmath98 is the maximal angular momentum of the projectile nucleon with respect to the target pair , @xmath99 is the spin of the projectile and @xmath100 is the relative orbital angular momentum between the projectile and target pair .
this leads to @xmath101 = 15/2 for a 3n scattering calculation at @xmath38 = 200 mev . disregarding the spin degree of freedom for the three nucleons , leading to the three - boson model under consideration here , we find that @xmath96 = 12 with @xmath102 = 5 and @xmath103 = 7 are necessary for a converged calculation at @xmath38 = 200 mev . in the three - boson case @xmath104 , the orbital momentum of the interacting pair , and @xmath100
take the role of @xmath93 and @xmath105 . in order to estimate the corresponding maximal number of angular momenta needed for @xmath38 = 1 gev
, one needs the effective deuteron radius @xmath106 , which leads to @xmath103 = 7 at @xmath38 = 200 mev .
nonrelativistically the 3n c.m .
energy is given as @xmath107 , leading to @xmath108 400 mev / c at @xmath38 = 200 mev and @xmath108 900 mev / c at @xmath38 = 1 gev .
if we roughly set @xmath109 , then we find at @xmath38 = 200 mev a value @xmath110 3.5 fm , which appears reasonable .
applying the same value at @xmath38 = 1 gev then leads to @xmath103 = 15 . using our experience in calculating the nn system in the gev regime @xcite , where one needs for converged nn observables at least @xmath94 = 14 at 1 gev
, we estimate that a converged partial wave 3n calculation of the three - boson system would need @xmath111 = 14 + 15 = 29 .
let us now regard the two cases , ( a ) @xmath38 = 200 mev , @xmath102 = 5 , @xmath103 = 7 , @xmath96 = 12 , and ( b ) @xmath38 = 1 gev , @xmath102 = 14 , @xmath103 = 15 , @xmath96 = 29 in a partial wave decomposition . to illustrate the tremendous number of partial waves needed in ( b ) compared to the feasible case ( a ) it is sufficient to consider simple algebra for different values of @xmath95 .
take for example
@xmath95 = 5 . then simple counting yields 30 different @xmath112 combinations in case ( a ) and 125 in case ( b ) . for @xmath95
= 10 this number increases in case ( b ) to 160 . moreover , since the number of total @xmath95 s at 1 gev is more than twice the number of @xmath95 s at 200 mev , it appears quite unreasonable to enforce a partial wave decomposition at energies far above 200 mev in the 3-boson ( nucleon ) system .
in addition , it would also be numerically very demanding to evaluate the various ingredients in the faddeev equation reliably for the very high angular momenta .
another advantage of using direct integration of vector variables in the faddeev equation is the simplicity of the permuation operators .
in the following we present our results for elastic and breakup scattering in the energy regime from about 200 to 1500 mev laboratory projectile kinetic energy . we start with a comparison of our model calculation to calculations based on a realistic nn force at lower energies in order to show that even though our model is very simple , we see similar features in the cross sections
. then we study relativistic effects at higher energies .
there are several questions we want to address .
first , we want to identify scattering observables that are sensitive to the difference between the relativistic and non - relativistic formulations of the three - body problems and to study the size of those relativistic effects as function of increasing energy .
this can at present only be done with our model interaction .
second , we want to study the convergence properties of the faddeev multiple scattering series as function of the projectile kinetic energy . here
the question of interest is , if , once the energy is high enough , it is sufficient to only consider the first few terms in the multiple scattering series .
in addition , we also want to study some approximations to our relativistic scheme .
the laboratory kinetic energy of 200 mev is a perfect energy to study if the features of the 3n system we find based on our model interaction are also present in calculations based on a realistic model of the nn interaction , which describes the nn observables with high accuracy .
the so - called high - precision interactions are fitted up to about 350 mev , but strictly speaking only valid below the pion - production threshold .
we also know @xcite that relativistic effects are already visible at 200 mev .
we choose the cd - bonn interaction @xcite for this comparison . in fig .
[ fig3 ] we show the @xmath113 total cross section extracted from the said database @xcite together with the total cross section obtained from the mt - iii interaction assuming bosonic symmetry .
the parameters of the mt - iii interaction @xcite are adjusted such that a two - body bound state at @xmath114 = 2.23 mev is supported .
[ fig3 ] shows that the experimental @xmath113 total cross section falls slightly below the two - body cross section predicted by our model at energies smaller than @xmath115 300 mev , is about equal between 300 and 400 mev , and then reaches a constant value from about 600 mev on , while our model prediction continues to decrease .
the slight rise of the experimental value around 600 mev is a manifestation of the influence of the @xmath116(1232 ) resonance in the nn system .
the cd - bonn interaction is fitted to nn observables to about 350 mev laboratory projectile energy and thus coincides with the said result up to that energy . in fig .
[ fig4 ] we show a comparison of elastic and breakup cross sections at 200 mev projectile laboratory energy for the 3d calculations based on our mt - iii model interaction and calculations based on a partial wave decomposition employing the cd - bonn potential .
the top row displays the differential cross section for elastic scattering .
we see that in both cases the difference between the fully relativistic calculation and the nonrelativistic one is overall quite small , and mostly visible at the backward angles , an observation already made in @xcite .
the differential cross section in the forward direction is much larger for our model interaction , which is consistent with the larger two - body total cross section .
in addition , there are more diffraction minima in the bosonic case than in the fermionic case , however , the minimum at around 130@xmath117 is present in both calculations . in the middle row
we display the cross section for inclusive breakup scattering as function of the laboratory kinetic energy of the ejected particle at fixed laboratory angle 18@xmath117 .
both cross sections are qualitatively similar , the fully converged faddeev calculation gives a lower cross section than the first order calculation , indicating the importance of rescattering contributions at this low energy .
the difference between the relativistic and nonrelativistic calculations is quite small in both cases . in the calculation based on the cd - bonn interaction
the fsi peak is more pronounced due to the virtual bound state in the @xmath118s@xmath119 state .
the latter is absent in the mt - iii model .
the bottom row shows the five - fold differential cross section as function of the arc - length @xmath0 for a configuration in which the laboratory angles @xmath120 are given in the scattering plane ( @xmath121 ) .
the position of the peaks is identical for both calculations , which is a manifestation that peak structures are given by the kinematics of the problem . in both cases
the relativistic calculation gives a significantly larger cross section for the central peak at @xmath122 mev compared to the nonrelativistic result , an increase by a factor of @xmath1231.5 for the full partial wave calculation and by a factor @xmath123 2 for the full 3d calculation .
this increase is already present in both first order calculations . for the mt - iii model
this trend is the same for all other peaks , whereas for the cd - bonn model the nonrelativistic calculations give a slightly larger cross section compared to the relativistic one in the peaks at small and large values of arc - length s. summarizing , the comparison of cross section obtained from our model interaction mt - iii with those given by a realistic nn interaction like cd - bonn at 200 mev indicates that , despite our model being quite simple , the qualitative features of especially the breakup cross sections are very similar .
the differences between the fully relativistic calculations and their nonrelativistic counterparts are still quite small at this low energy for elastic scattering and inclusive breakup . for the exclusive breakup , however , even at this energy complete configurations with large changes of the nonrelativistic cross section due to relativity can be found .
this sensitivity of the complete breakup to relativistic effects has already been observed in @xcite .
starting from our model interaction we now consider three - body scattering in the energy regime up to 1.5 gev . the total cross section for elastic scattering is related to the symmetrized transition operator @xmath79 of eq .
( [ eq:2.1 ] ) via @xmath124 in fig .
[ fig5 ] we display the total cross section for elastic scattering as a function of the projectile kinetic energy up to 1.5 gev obtained from our fully converged relativistic faddeev calculation as well as the one obtained from the first order term .
it is obvious that , especially for the energies below 300 mev , the contribution of rescattering terms is huge . since the logarithmic scale from the top panel
is unsuited to extract detailed information about the size of relativistic effects , we show in the lower two panels the relative difference of the relativistic calculations with respect to their nonrelativistic counterparts .
the bottom panel displays the relative difference between the relativistic first order term and its nonrelativistic counterpart as dotted line .
essentially the first order calculation does not show any effect .
this is theoretically consistent when having in mind that in first order ( @xmath125 ) only the two - body t - matrix enters into the cross section .
the relativistic two - body t - matrix is constructed to be phase - shift equivalent to the non - relativistic one via the cps method @xcite .
thus , seeing no difference between the fully relativistic and the corresponding nonrelativistic calculations indicates that relativistic effects are taken into account consistently at the two - body level .
doing the same comparison with fully converged faddeev calculations ( solid line in the middle panel ) indicates that relativistic effects in the three - body problem increase the elastic scattering total cross section with increasing energy . at our highest energy , 1.5 gev ,
this increase is about 8.3% .
often only effects due to relativistic kinematics are taken into account .
here we have the opportunity to study consequences of such a simple approximation . for the calculations labeled @xmath126
we only consider the lorentz transformations from laboratory to center - of - momentum ( c.m . )
frame and the relativistic phase space factor of eq .
( [ eq:3.1 ] ) , whereas the matrix elements of the operator @xmath79 are calculated from the solution of the nonrelativistic faddeev equation .
the relative difference between this calculation and a completely nonrelativistic calculation is indicated by the short - dashed line in the bottom panel of fig .
[ fig5 ] , where only the first order term is considered .
the triple - dotted curve in the middle panel is the same comparison , but now between fully converged faddeev calculations . for both ,
full and first order calculation , the effect is huge . to understand better
which piece of the kinematics included is responsible for this large enhancement of the cross section , we also plot in fig .
[ fig5 ] calculations ( labeled nr@xmath127 ) , which only contain the lorentz transformation between the laboratory and c.m .
frame , but carry the nonrelativistic phase space factor in eq .
( [ eq:3.1 ] ) .
the dash - dotted line in the lower panel show the 1st order calculation and the dotted line in the middle panel the full faddeev calculation . using a lorentz transformation in the change of frames
has the effect that the two - body t - matrix is calculated at a slightly different c.m .
momentum @xmath43 , and thus there is a small effect , about a 5% underestimation of the total cross section . the huge effect is entirely due to the relativistic phase space factor , and relativistic dynamics then has an equally large effect of the opposite sign .
this interplay of increasing effects due to the relativistic phase - space factor and decreasing effects due to relativistic dynamics has been already observed in the partial - wave based faddeev calculations with realistic interactions @xcite .
it led for elastic scattering cross section at 250 mev to relativistic effects which are relatively small and restricted to backward angles .
recent measurements of the neutron - deuteron ( @xmath128 ) differential cross section at 248 mev @xcite indicate that for discrepancies of theoretical predictions in this observable , short range components of a three - nucleon force are equally important .
the problem with approximating relativistic effects only through kinematics and phase space factors can be easily understood in the 2 + 1 body problem , where the phase equivalence is achieved by choosing the invariant mass as a function of the non - relativistic two - body hamiltonian , @xmath129 .
the eigenvalues equation for the scattering problems @xmath130 are equivalent , but the replacement of @xmath131 by @xmath132 must be compensated by replacing the interaction @xmath133 by @xmath134 . including only kinematic relativistic effects
is equivalent to making the replacement @xmath135 without making the compensating replacement @xmath136 .
we also study a more sophisticated approximation to the relativistic dynamics . in ref .
@xcite we described in detail how we obtain the transition amplitude of the 2n subsystem , @xmath137 , embedded in the three - particle hilbert space which enters the faddeev equation , eq .
( [ eq:2.10 ] ) .
the fully off - shell amplitude is the solution of a first resolvent type equation @xcite given by @xmath138 here @xmath139 is taken to be right half - shell with @xmath140 .
note that in this equation the unknown matrix element is to the _ left _ of the kernel .
it was suggested in ref .
@xcite that a reasonable approximation to this embedded 2n transition amplitude might by the born term of the above integral equation , which is @xmath141 in this approximation , the fully off - shell 2n transition amplitude is replaced by a half - shell amplitude .
the effect of this approximation is not large in elastic scattering , as shown in fig .
[ fig6 ] , where we plot the differential cross section in forward direction for the fully relativistic calculations and the ones containing the approximation of eq .
( [ eq:3.3 ] ) to the boost ( curves labeled h ) . consistently and independent of projectile energy , approximating the embedded two - body t - matrix by the half - shell t - matrix leads to an underprediction of the differential cross section in forward direction .
though not plotted , this also leads to a smaller total cross section for elastic scattering .
finally , we want to investigate the convergence of the multiple scattering series as a function of the projectile laboratory kinetic energy .
one might expect that with increasing energy only a few terms in the multiple scattering series are sufficient for a converged result .
our converged relativistic faddeev calculations now allow a detailed study .
this is of particular interest , since recently relativistic calculations in the energy regime around 1 gev have been published @xcite , which are carried out in a multiple scattering expansion of the faddeev equations up to 2nd order , and which use the off - shell continuation of the experimental nn amplitudes as two - body input . first we want to consider the convergence of the faddeev multiple scattering series in the total cross sections for elastic scattering as well as breakup reactions as a function of projectile kinetic energy . in the bottom row of fig .
[ fig7 ] the different orders ( successively summed up as neumann sum to the order indicated in the legend ) are shown as functions of the projectile laboratory energy . we see a distinct difference in the behavior of the elastic total cross section in comparison with the breakup total cross section .
while the elastic total cross section converges very rapidly , the total breakup cross section does not .
the left upper panel of fig .
[ fig7 ] shows the elastic total cross section as a function of the order in the multiple scattering series ( the orders are successively summed up the one indicated on the x - axis ) . even at 200 mev there is very little change due to contributions from the 2nd or higher order rescattering terms . for the higher energies , the 1st order term already captures the essential physics
this is very different for the total breakup cross section , where for 200 mev projectile energy the full solution of the faddeev equation is clearly necessary . for energies of 1 gev and higher ,
at least one rescattering contribution ( 2nd order in the multiple scattering series ) is necessary to come close to the full solution . since the total cross section for elastic scattering might be insensitive to higher orders in the faddeev multiple scattering series , we plot in fig . [ fig8 ] the differential cross section at forward and backward angles as a function of the order in the multiple scattering series for the same laboratory projectile energies .
here we see that at the lowest energy , 0.2 gev , the convergence is not as fast as the total cross section suggests .
in fact , at least 5 orders are necessary , which is consistent with the experience from nonrelativistic calculations at low energies @xcite . for energies of 1 gev and higher ,
the forward direction is converged at the 3rd order in @xmath142 , whereas the backward angle is not as sensitive ( it should be pointed out that the cross section in backward direction is about five orders of magnitude smaller than the one in forward direction ) .
it seems accidental that the multiple scattering series converges faster at 0.5 gev compared to 1 gev .
however , a similar finding was presented in ref .
@xcite , where it was observed that polarization observables for elastic proton - deuteron ( @xmath143 ) scattering at 395 mev were described better than those at 1.2 gev , when calculating the faddeev multiple scattering series up to the 2nd order .
the calculation of breakup cross sections requires knowledge of the matrix element @xmath144 in eq .
( [ eq:2.1 ] ) . for details of the derivation
we refer to ref .
@xcite and only give the final expressions here .
the five - fold differential cross section for exclusive breakup is given in the laboratory frame as @xcite @xmath145 here @xmath146 is the total energy of the system and @xmath16 its total momentum .
the subscripts @xmath147 and @xmath148 indicate the two outgoing particles . in inclusive breakup
only one of the particles is detected , and thus one of the angles in eq .
( [ eq:3.4 ] ) is integrated out .
this leads to the inclusive breakup cross section in the laboratory frame @xcite @xmath149 where @xmath150 is determined by the condition @xmath151 . in refs .
@xcite we already pointed out and demonstrated that relativistic kinematics is essential to obtain the correct position of e.g. the peak for quasi - free scattering ( qfs ) , especially at higher energies .
the difference between a nonrelativistic calculation of the breakup cross section and a relativistic one is quite large at higher energies .
however one may argue that this difference is artificially large , since it is natural to use relativistic kinematics at higher energies .
therefore , here we will _ not _ compare to entirely nonrelativistic calculations , but rather calculations where the three - body transition amplitude has been obtained from the solution of a nonrelativistic faddeev equation , but the transformations between the laboratory frame and the c.m .
frame are lorentz transformations .
this is equivalent to comparing the relativistic and non - relativistic calculations in the center - of - momentum frame .
in addition we use the relativistic phase space factor for the cross sections . in fig .
[ fig9 ] we show the inclusive breakup cross section as a function of the laboratory kinetic energy of the ejected particle at fixed angle @xmath152 for different projectile kinetic energies calculated from the full solution of the relativistic faddeev equations together with ` nonrelativistic ' calculations using the above defined relativistic kinematics .
there is still a shift of the position of the qfs peak towards lower ejectile energies , which increases with increasing projectile energy .
there is also a very visible effect of the relativistic phase space factor used together with the nonrelativistic three - body transition amplitude . at 1000
mev the size of the qfs peak is a factor of two larger compared to exact relativistic calculation . for the lower energies the 1st order calculation yields a significantly higher qfs peak compared to the full calculation , whereas for the higher energies
, the peak height is almost the same for the 1st order and the full calculation .
next we investigate in detail the convergence of the faddeev multiple scattering series in the region of the qfs peak as a function of the projectile energy . in fig .
[ fig10 ] we display calculations at selected energies from 200 to 1000 mev .
the solid line represents the solution of the relativistic faddeev equation , whereas the other curves show the neumann sum of the multiple scattering series containing the sum up to the order in the two - body t - matrix as indicated in the legend .
for the lowest energy , 200 mev , it is obvious that the multiple scattering series does not converge rapidly .
this changes considerably as the projectile kinetic energy grows . though the variation of the different orders is not as large anymore at 500 mev
, the multiple scattering series must still be summed up to 4th order in the qfs peak to coincide with the full result , whereas at 800 mev already the 2nd order is almost identical with the full result , and even a 1st order calculation can be considered quite good .
this trend continues as the energy grows .
of course , 1st order calculations are never able to capture the fsi peak at the maximum energy of the ejectile , nor do they describe the high energy shoulder of the qfs peak .
however , our study indicates that for energies in the gev regime it is very likely sufficient to consider only one rescattering term when studying inclusive breakup reactions in the vicinity of the qfs peak .
finally , we also want to study the approximation suggested in eq .
( [ eq:3.3 ] ) , namely replacing the off - shell two - body transition amplitude embedded in the three - body hilbert space by the half - shell one .
the calculations based on the approximation of eq .
( [ eq:3.3 ] ) and labeled ( h ) are plotted in fig .
[ fig11 ] together with the exact solution . considering only the 1st order calculation we observe a similar trend as in the differential cross section for elastic scattering , the approximation slightly underpredicts the exact result , independent of the energy under consideration .
however , when this approximate two - body transition amplitude is iterated to all orders in the faddeev equation , the deviations from the exact calculations become larger . at 800 and 1000
mev the iteration of the exact amplitude increases the cross section in the qfs peak , whereas it decreases for the approximation with respect to the 1st order term . at 200 and 500
mev the approximation does not only give a smaller cross section in the qfs peak but also fails to develop an fsi peak towards the maximum allowed ejectile energies .
from this we conclude that eq .
( [ eq:3.3 ] ) does not provide a good approximation for inclusive breakup cross sections .
our calculations indicate that at energies 1 gev or higher , it is important to carry out the poincar invariant aspects of the calculation exactly . they also indicate that it is sufficient to consider only one rescattering term to capture most features of the cross section .
although these conclusions are based on the use of a simple model two - body interaction , we conjecture that calculations based on realistic interactions will have similar characteristics . for our study of exclusive breakup scattering in the intermediate energy regime
we choose two different experimental situations where there are data available .
first we consider the @xmath153h(p,2p)n reaction at 508 mev , where the two outgoing protons are measured for a given angle pair @xmath154 in the scattering plane @xcite .
since the convergence of the multiple scattering series is already discussed in @xcite , we only want to investigate the effect of the approximations previously given in this reaction . in fig .
[ fig12 ] selected angle configurations are shown .
the left column of the figure shows the first order calculations and the right column the full solution of the faddeev equation .
the exact 1st order calculation is given by the dotted line in the left column and the exact full solution by the solid line in the right column .
the angle combination @xmath155 is a qfs configuration .
first , we see that in a qfs configuration , the 1st order calculation is already almost identical to the full faddeev calculation @xcite , whereas this is not the case for the other configurations shown . if only relativistic kinematics is considered , namely the lorentz transformations between laboratory and c.m .
frame together with the relativistic phase space factor , and a nonrelativistic three - body transition amplitude is employed , we obtain the double - dotted curve for the 1st order calculations and the dashed line for the full solution of the faddeev calculation .
again , the qfs configuration is quite insensitive to this approximation .
however , the deviation from the exact calculation is quite visible in the other two configurations shown .
finally , we also consider the approximation suggested by eq .
( [ eq:3.3 ] ) , which is indicated by the dash - dotted line , labeled ` h ' in the left column ( 1st order calculation ) and the dotted line in the right column ( full solution of the faddeev equation ) . here
we see that even in the qfs configuration there are already deviations of this approximation for the high energy shoulder .
the approximation underpredicts the full solution . this tendency becomes stronger for the other two configurations .
the interesting property of this approximation is that while it appears to be a reasonable approximation to the faddeev kernel , the errors in the approximation increase when the equation is iterated .
thus we conclude that this approximation , though simplifying the calculation of the two - body t - matrix embedded in the three - body hilbert space , does not seem to capture essential structures of the two - body t - matrix .
the failure of this approximation , which approximates the off shell two body transition operator in the faddeev equation with the half - shell transition operator , suggests that some care is necessary in modeling the off - shell behavior of the transition operators in more phenomenological schemes . for breakup reaction at a slightly higher energy we consider the @xmath118h(d,2p)n reaction at 2 gev deuteron kinetic energy @xcite . here
the two outgoing protons are measured .
energetically , this reaction would correspond to @xmath143 scattering at roughly 1 gev and thus is within the range of the calculations presented here . in fig .
[ fig13 ] we show the five - fold differential cross section as a function of the angle of the second detected proton for four different momenta of the first detected proton .
the full relativistic faddeev calculation is represented by the solid line . in order to investigate the convergence of the multiple scattering series we show the 1st order calculation as a dotted line ,
then successively add one ( 2nd order ) and two ( 3rd order ) rescattering terms to the leading order . in this reaction ,
the first two rescattering terms are about the same size , but have opposite sign , so that the 3rd order calculations are very close to the 1st order one .
we also observe that the 3rd order calculation is already so close to the full faddeev calculation that the neumann series can be considered converged with three terms .
in this work we demonstrated the feasibility of applying poincar invariant quantum mechanics to model three - nucleon reactions at energies up to 2 gev . this is an important first step for studying dynamical models of strongly interacting particles in the energy range where sub - nuclear degrees of freedom are thought to be relevant . at these energies
the poincar invariance of the theory is an essential symmetry . at lower energies non - relativistic quantum mechanical models are powerful tools for understanding the dynamics of strongly interacting nucleons . at higher energies the physics is more complicated , but one can expect that it is still dominated by a manageable number of degrees of freedom .
poincar invariant quantum mechanics is the only alternative to quantum field theory where it is possible to realize the essential requirements of poincar invariance , spectral condition , and cluster properties @xcite .
it has the advantage that the faddeev equation provides a mathematically well - defined method for exactly solving the strong interaction dynamics .
the faddeev equation in this framework is more complicated than the corresponding non - relativistic equation , due to the non - linear relation between the mass and energy in relativistic theories , but these difficulties can be overcome @xcite
. an important advance that allows these calculations to be extended to energies in the gev range is the use of numerical methods based on direct integrations , rather than partial wave expansions @xcite .
these have been successfully applied to the non - relativistic three - nucleon problem .
this paper demonstrates that they can also be successfully applied to the relativistic problem , even with its additional complications .
the model presented here involves three nucleons interacting with a spin - independent malfliet - tjon @xcite type of interaction .
it differs from more realistic interactions @xcite in that it is spin independent and it does not give a high - precision fit to the two - body scattering data .
in addition , the model is for fixed numbers of particles , not allowing pion production , which is an open channel at these energies .
while these limitations must be addressed in realistic applications , the three - body faddeev calculations presented in this paper provide a powerful framework for both testing approximations and for examining the sensitivity of scattering observables to relativistic effects . in order to investigate relativistic effects
, we treat the interaction as if it was determined by fitting the cross section obtained by solving the non - relativistic lippmann - schwinger equation to scattering data . when this is done with a realistic interaction the experimental differential cross section is properly transformed from the lab frame to the center of momentum frame before the fit is done .
the result of this process is that the computed differential cross section agrees with the fully - relativistic experimental differential cross section in the center of momentum frame as a function of the relative momentum .
thus , even though the two - body scattering observables are computed with a non - relativistic equation , there is nothing non - relativistic about the result . at the two - body level
the corresponding relativistic lippmann - schwinger equation must be designed to give the same scattering observables .
this can be achieved by expressing the relativistic mass operator as a simple function of the non - relativistic center of momentum hamiltonian @xcite .
the important consequence of this is that it does not make sense to relate the relativistic and non - relativistic two - body models using @xmath156 expansions ; the prediction of the relativistic and non - relativistic two - body models are identical .
real differences in the dynamics appear when the two - body dynamical operators are used to formulate the three - body dynamics . how this must be done in the two and three - body cases is dictated , up to three - body interactions , by cluster properties .
the faddeev equation for the relativistic and non - relativistic system have identical operators forms .
the permutation operators , two - body transition operators and free resolvents that are input to the faddeev equation have different forms in the relativistic and non - relativistic equations .
these differences are responsible for differences in the relativistic and non - relativistic three - body calculations .
the calculations presented in this paper have a number of consequences .
the most important result is a demonstration that direct integration methods can be successfully applied to extend the energy range for converged solutions to faddeev equations to intermediate energies .
our estimates of the number of partial waves needed for calculations at different energies suggest that it is not currently practical to extend existing partial wave calculations beyond a few hundred mev , while in this paper we have demonstrated convergence of the direct integration methods for laboratory energies up to 2 gev . while our model interaction is not realistic , when we compared the results of our calculations to relativistic calculations at 200 mev that have been performed with realistic interactions @xcite in a partial wave basis , we found that the qualitative features of the realistic model are reproduced in our simple model , suggesting that some of the conclusions derived from our model should be applicable to models with realistic interactions . having a model where it is possible to perform numerically exact solutions of scattering observables in the intermediate energy range provides us with a tool to test approximations that have been used in other calculations as well as to look for observables that are sensitive to the differences between the relativistic and non - relativistic models .
one common approximation that we tested is the replacement of non - relativistic kinematic factors by the corresponding relativistic kinematic factors in a non - relativistic model .
our tests clearly illustrated a big effect , but most of it is canceled by the associated dynamical corrections .
this suggests that including only kinematic corrections can actually provide large relativistic effects .
such an approach should never be used in the absence of a complete theory where relativistic effects can be rigorously estimated .
a second important set of approximations are multiple scattering approximations .
these are expected to improve at higher energies , but it is important to understand in the context of models based on realistic interactions how high these energies have to be for convergence .
our conclusion is that the convergence of the multiple scattering series is non - uniform .
even at 200 mev our calculations show that the first - order term reproduces the total elastic cross section ; for the total breakup cross section at least one more iteration is needed up to about 600 mev .
both of these observations turn out to be misleading when one investigates the differential cross sections .
while the total elastic cross section is reproduced at 200 mev by the first order term , the correct angular distribution requires at least five orders in the multiple scattering series .
even at 1 gev the first order approximation is not accurate enough at forward angles . for inclusive breakup reactions our computations show that the first order calculation does not give the right size of the quasifree peak even at 1 gev , however for 800 mev and above the second order term is a good approximation . for exclusive breakup the convergence of the multiple scattering series
even at 1 gev energy depends on a specific configuration .
another type of approximation that is employed is to use on - shell transition operators with a phenomenological representation of the off - shell dependence . in our formulation of the three - body problem that off shell behavior needs to be computed by solving a singular integral equation .
it was suggested in @xcite that simply replacing the off - shell two - body @xmath157 by its on - shell value might be a good approximation .
this was based on the observation that the difference between on and off shell faddeev kernel is small .
our calculations show that while this does not lead to a large effect in the elastic cross section , the off - shell effects lead to non - trivial modifications when one considers the breakup cross sections .
this shows that such an approximation should not be used and also suggests that phenomenological parameterizations of the off - shell behavior of the two - body amplitudes need to be carefully tested , especially for breakup reactions . while a number of calculations have shown small relativistic effects for the three - body binding energy , non - trivial effects
have already been observed in scattering observables at 200 mev @xcite .
our model confirms these previously observed effects and indicates that they continue into the intermediate energy region . our calculations exhibit a number of sensitivities to relativistic effects in the breakup observables
both the shape and size of the quasielastic peak differ from the non - relativistic quantities .
this paper demonstrates both the need for a relativistic description of few - nucleon dynamics in the intermediate energy range and shows that the problem is amenable to a numerically exact solution , using direct integration , for laboratory energies up to 2 gev
. in the future relativistic few - body calculations will be important tools for testing the validity of approximations , such as the eikonal approximation .
obviously extensions to include spin - dependent interactions , meson channels , and interactions that are fit to higher energy data will be needed for realistic applications .
the success of the calculations in this paper provide a strong motivation for continuing this program .
this work was performed in part under the auspices of the u. s. department of energy , office of nuclear physics , under contract no .
de - fg02 - 93er40756 with ohio university , contract no .
de - fg02 - 86er40286 with the university of iowa , and contract no .
de - ac02 - 06ch11357 with argonne national laboratory .
it was also partially supported by the helmholtz association through funds provided to the virtual institute `` spin and strong qcd '' ( vh - vi-231 ) .
we thank the ohio supercomputer center ( osc ) for the use of their facilities under grant phs206 .
part of the numerical calculations were performed on the ibm regatta p690 + of the nic in jlich , germany .
h. liu , ch .
elster , and w. glckle , phys.rev .
c*72 * , 054003 ( 2005 ) .
h. liu , ch .
elster , and w. glckle , few - body systems 33 , 241 ( 2003 ) .
elster , w. schadow , a. nogga , and w. glckle , few - body systems * 27 * , 83 ( 1999 ) .
t. lin , ch .
elster , w. n. polyzou and w. glckle , phys . rev .
c*76 * , 014010 ( 2007 ) .
wiringa , v.g.j .
stoks , r. schiavilla , phys .
c*51 * , 38 ( 1995 ) .
r. machleidt , phys .
c*63 * , 024001 ( 2001 ) .
stoks , r.a.m .
klomp , c.p.f .
terheggen , and j.j.de swart , phys .
c*49 * , 2950 ( 1994 ) . .
the total elastic and break - up cross sections together with the total cross section extracted via the optical theorem calculated from a malfliet - tjon type potential as a function of the projectile laboratory kinetic energy . [ cols="^,^,^",options="header " , ] in the optical theorem as a function of the grid points in the angle grid @xmath88 , when this grid is increased successively by 4 gauss - legendre points .
the different curves correspond to the three different laboratory projectile energies in gev , indicated in the legend .
[ fig2],width=302 ] = 200 mev .
the left column shows results obtained from the malfliet - tjon - iii potential assuming boson symmetry and no partial wave decomposition , the right column shows the corresponding realistic calculations obtained with the cd - bonn @xcite potential where partial wave decomposition is applied .
the top row displays the differential cross section for elastic scattering , the middle row shows the breakup cross section for inclusive scattering for the laboratory angle @xmath158 of the outgoing particle . the bottom row shows the five - fold differential for exclusive breakup reaction as a function of the arc - length s. the laboratory angles of the outgoing particles are @xmath159 , and @xmath160 .
the fully relativistic converged faddeev calculations are given by the solid lines ( r ) , the corresponding nonrelativistic calculations by the long - dashed lines ( nr ) .
in addition the relativistic ( dash - dot ) and nonrelativistic ( dotted ) first order calculations are shown .
[ fig4],width=566 ] ) are given as dotted line for a full faddeev calculation and as short - dashed line for the first order term .
calculations which only take into account the lorentz transformations between the laboratory and c.m .
frame ( labeled nr@xmath127 ) are shown as dotted line for the full faddeev calculation and as dash - dotted line for the 1st order one .
the middle panel shows the relative difference between the fully relativistic faddeev calculation and the nonrelativistic one ( solid line ) together with the difference to the nonrelativistic calculation if only relativistic kinematics is considered .
the bottom panel shows the corresponding relative differences when only the first order term is taken into account .
[ fig5],width=340 ] for selected laboratory kinetic energies .
the converged solution of the relativistic faddeev equation is given as solid line .
the dotted line shows the converged solution of the relativistic faddeev equation in which the fully off - shell 2n t - matrix is replaced by the half - shell t - matrix .
the corresponding first order calculation are given by the short - dashed line and the dash - dotted line .
for details see text .
[ fig6],width=566 ] of the emitted particle at an emission angle @xmath152 .
the incident laboratory kinetic energy for each cross section is indicated in each panel . the solid lines ( r full ) represent the converged relativistic faddeev calculation and the dotted line the corresponding first order calculations ( r 1st ) .
the lines labeled r@xmath161 correspond to calculations in which only relativistic kinematics is taken into account .
[ fig9],width=566 ] of the emitted particle at an emission angle @xmath152 .
the incident laboratory kinetic energy for each cross section is indicated in each panel .
the solid lines ( r ) represent the converged relativistic faddeev calculation .
the triple - dotted line shows the 1st order calculation , for the short dashed line the 2nd order contribution is added to the previous , for the dash - dotted line the 3rd order is added , and for the dotted line the 4th order .
[ fig10],width=566 ] of the emitted particle at an emission angle @xmath152 .
the incident laboratory kinetic energy for each cross section is indicated in each panel .
the solid lines ( r ) represent the converged relativistic faddeev calculation .
the dotted line ( h ) displays the calculation in which the fully off - shell two - body t - matrix is replaced by the half - shell one .
the calculations labeled 1st stand for the corresponding 1st order calculations .
[ fig11],width=566 ] h(p,2p)n reaction at 508 mev laboratory projectile energy for different proton angle pairs @xmath162-@xmath163 with respect to the beam axis as a function of the laboratory kinetic energy of the first detected proton .
the left column represents 1st order calculation , whereas the right column gives full solution of the faddeev equation .
the curves labeled r ( solid for the full faddeev calculation and dotted for the 1st order one ) represent the full relativistic calculations , whereas for the curves labeled r@xmath161 ( dashed in the right column and double - dotted in the left ) only relativistic kinematics is taken into account ( see text ) , and for the curves labeled h ( dotted in the right column and dash - dotted in the left ) the fully off - shell t - matrix is replaced by the half - shell one .
the data are taken from ref . @xcite .
[ fig12],width=566 ] h(d,2p)n at 2 gev deuteron energy as a function of the angle @xmath163 of the second of the outgoing protons for a fixed first proton momentum indicated in the figure .
the solid line represents the solution of the full relativistic faddeev equation .
the dotted line gives the result of the first order calculation .
for the dashed line the 2nd order term is added and for the dash - dotted line the 2nd and 3rd order terms are added .
the data are taken from ref .
[ fig13],width=566 ] | the relativistic faddeev equation for three - nucleon scattering is formulated in momentum space and directly solved in terms of momentum vectors without employing a partial wave decomposition .
the equation is solved through pad summation , and the numerical feasibility and stability of the solution is demonstrated .
relativistic invariance is achieved by constructing a dynamical unitary representation of the poincar group on the three - nucleon hilbert space .
based on a malfliet - tjon type interaction , observables for elastic and break - up scattering are calculated for projectile energies in the intermediate energy range up to 2 gev , and compared to their nonrelativistic counterparts .
the convergence of the multiple scattering series is investigated as a function of the projectile energy in different scattering observables and configurations .
approximations to the two - body interaction embedded in the three - particle space are compared to the exact treatment . |
edge ideals of graphs have been extensively studied by several authors ( see for instance @xcite , @xcite , @xcite , @xcite , @xcite ) . in recent years the interest focused on a generalization of the notion of graph , the so - called _ clutter _ ( see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
edge ideals have been introduced by villarreal in @xcite .
let @xmath5 be a field and @xmath6 a finite simple graph on @xmath7 vertices .
the edge ideal of @xmath6 is @xmath8.\ ] ] a graph @xmath6 is called cohen - macaulay if @xmath9 is cohen - macaulay .
similarly it is possible to define the edge ideal @xmath10 of a clutter @xmath11 .
a relevant class of graphs consists of bipartite graphs .
a finite simple graph @xmath6 is bipartite if there exists a partition of its vertex set @xmath12 , such that every edge of @xmath6 is of the form @xmath13 , with @xmath14 and @xmath15 . for every @xmath1 , @xmath0-partite
clutters are a natural extension of bipartite graphs .
they have been introduced by stanley in @xcite .
there are also other possible extensions , see e.g. @xcite . in @xcite ,
herzog and hibi characterized all cohen - macaulay bipartite graphs @xmath6 , showing that a bipartite graph is cohen - macaulay if and only if it can be associated to a poset @xmath2 under a suitable construction . in @xcite , ene , herzog and mohammadi extended this construction introducing the clutter @xmath3 , for @xmath1 , and describing the maps of the resolution of the alexander dual ideal @xmath16 of @xmath17 , the so - called _ cover ideal_. by using betti splitting techniques introduced in @xcite , in theorem [ cmdip ] we prove an explicit recursive formula for the graded betti numbers of @xmath16 , extending a result by francisco , h and van tuyl ( * ? ? ? * theorem 3.8 ) .
there are several preliminary results to the proof of the main theorem , but we think that proposition [ splitinside ] and proposition [ final1 ] could be results of independent interest . in the case of a poset @xmath2 with a unique maximal element , the formula can be considerably simplified ( see corollary [ simpler ] ) . in corollary [ cmbip ]
we give another formulation of ( * ? ? ?
* theorem 3.8 ) , on betti numbers of the cover ideal of a cohen - macaulay bipartite graph .
in example [ dunce ] we show that betti splittings could _ not exist _ for the alexander dual ideal @xmath18 of a simplicial complex @xmath19
. moreover we point out that there is no relation between the existence of a betti splitting of the stanley - reisner ideal @xmath20 and the existence of a betti splitting of the alexander dual ideal @xmath18 .
finally we give a sufficient condition that ensures the existence of an @xmath4-splitting for @xmath18 ( see proposition [ weakvd ] ) .
+ _ acknowledgements . _
the last section of this paper was inspired by the summer school _ discrete morse theory and commutative algebra _
, held at the institut mittag - leffler in stockholm in 2012 .
the author is grateful to the staff of the institut , to the organizers bruno benedetti and alexander engstrm , for their suggestions on discrete morse theory and properties of simplicial complexes .
thanks to paolo sentinelli for helpful discussions .
let @xmath5 be a field . throughout this paper , @xmath21 denotes the polynomial ring @xmath22 $ ] . for a monomial ideal @xmath23 ,
let @xmath24 be the _ graded betti numbers of @xmath25 _ and @xmath26 be the _
i - th total betti numbers of @xmath25_. if @xmath25 is generated in degree @xmath0 , we say that @xmath25 has a _ @xmath0-linear resolution _ provided @xmath27 for every @xmath28 and @xmath29 when the context is clear , we simply say that @xmath25 has a linear resolution .
denote by @xmath30 the set of variables dividing a monomial @xmath31 and by @xmath32 the set of minimal monomial generators of @xmath33 if @xmath25 is a square - free monomial ideal , let @xmath34 be its alexander dual ideal ( see for instance @xcite ) .
+ the idea of betti splitting has been introduced in @xcite , but in this paper we follow the more general setting of @xcite . _
( ( * ? ? ?
* definition 1.1 ) ) _ _ let @xmath25 , @xmath35 and @xmath36 be monomial ideals such that @xmath37 and @xmath32 be the disjoint union of @xmath38 and @xmath39 .
then @xmath40 is a _
betti splitting _ of @xmath25
if @xmath41 if this is the case @xmath42 _ in some cases we focus on a special splitting of a monomial ideal @xmath25 .
let @xmath4 a fixed variable of @xmath21 .
define @xmath43 , @xmath44 and let @xmath36 be the ideal generated by the remaining monomials of @xmath32 .
we call @xmath45 an _ @xmath4-partition_. if this is a betti splitting , we say that @xmath45 is an _ @xmath4-splitting _ of @xmath25 .
francisco , h and van tuyl , in @xcite , proved the following splitting results that we use extensively in this paper .
[ urur ] _ ( ( * ? ? ?
* corollary 2.4 ) , ( * ? ? ?
* corollary 2.7 ) ) _ let @xmath25 , @xmath35 and @xmath36 be monomial ideals in @xmath21 such that @xmath37 and @xmath32 be the disjoint union of @xmath38 and @xmath39 .
let @xmath4 be a variable of @xmath21 . then * if @xmath35 and @xmath36 have a linear resolution , then @xmath37 is a betti splitting of @xmath25 . *
if @xmath45 is an @xmath4-partition and @xmath35 has a linear resolution , then @xmath37 is a betti splitting of @xmath25 .
since we are using properties of ideals with linear quotients , we recall here some useful facts .
ideals with linear quotients have been introduced by herzog and takayama in @xcite . for further details see @xcite and @xcite .
_ let @xmath23 be a monomial ideal .
we say that the ideal @xmath25 has _ linear quotients _ if there exists an order @xmath46 of the monomials of @xmath32 such that , for every @xmath47 , the colon ideal @xmath48 are generated by a subset of @xmath49 .
_ an ideal with linear quotients is componentwise linear ( see ( * ? ? ?
* corollary 2.8 ) ) and , if generated in a single degree , it has a linear resolution ( see @xcite )
. the following characterization of square - free monomial ideals with linear quotients will be useful later ( see for instance ( * ? ? ?
* corollary 8.2.4 ) ) .
[ linquot ] let @xmath23 be a square - free monomial ideal .
then @xmath25 has linear quotients with respect to the order @xmath50 of the monomial generators if and only if for each @xmath51 and for all @xmath52 there exists a variable @xmath53 and an integer @xmath54 such that @xmath55 we now recall some definitions about clutters , for further details see @xcite . _ a _ clutter _
@xmath56 is a pair where @xmath57 is a finite set , called _ vertex set _ , and @xmath58 is a collection of non - empty subsets of @xmath57 , called _ edges _ , such that if @xmath59 and @xmath60 are distint edges of @xmath11 , then @xmath61 .
the clutter @xmath11 is _
@xmath0-uniform _ if all edges have exactly @xmath0 vertices .
vertex cover _ of a clutter @xmath11 is a subset @xmath62 such that @xmath63 , for every @xmath64 . a vertex cover is _ minimal _ if none of its proper subsets is a vertex cover . _ for @xmath65
this is the definition of simple graph . in literature clutters
are also known as _
simple hypergraphs_. let @xmath11 be a clutter on the vertex set @xmath66 .
the _ edge ideal _ of the @xmath11 is defined by @xmath67 the edges of a clutter can be seen as the facets of a simplicial complex
. then the edge ideal of a clutter coincides with the facet ideal introduced by faridi in @xcite .
the alexander dual ideal @xmath68 of @xmath10 is called the _ cover ideal _ of @xmath11 .
its minimal monomial generators correspond to the vertex covers of @xmath11 .
the clutter @xmath11 is called _ cohen - macaulay _ if @xmath69 is cohen - macaulay .
_ let @xmath70 .
the collection @xmath71 is _ not _ a clutter .
the clutter @xmath11 whose edge set is @xmath72 is not uniform .
its edge ideal in @xmath73 $ ] is @xmath74 its cover ideal is @xmath75 and in fact the minimal vertex covers of @xmath11 are @xmath76 .
_ _ let @xmath56 be a @xmath0-uniform clutter .
we say that @xmath11 is _
@xmath0-partite _ if there is a partition @xmath77 such that all the edges of @xmath11 have the form @xmath78 , where @xmath79 for @xmath80 . _ for @xmath65 , this is the definition of bipartite graph .
+ let @xmath81 be a finite _ poset _ , with partial order @xmath82 .
we label each element of @xmath2 in such a way that if @xmath83 , then @xmath84 a subset @xmath85 is called a _ chain _ if @xmath86 is a totally ordered set with respect to the induced order . a _ multichain _ of length @xmath0 in @xmath2 is a chain @xmath87 , in which we allow repetitions .
let @xmath88 by our labeling convention , we have @xmath89 a _ poset ideal _ @xmath90 in @xmath2 is a subset of @xmath2 with the following property : given @xmath91 and @xmath92 such that @xmath93 , then @xmath94 we may consider each poset ideal as a subposet of @xmath2 with respect to the induced order .
notice that , given @xmath95 , then @xmath96 is again a poset ideal of @xmath2 .
given @xmath97 elements of @xmath2 , denote by @xmath98 the smallest poset ideal containing @xmath99 let @xmath100 be _ the set of all poset ideals of @xmath2_. [ poset1 ] _ let @xmath101 be the poset with @xmath102 elements , whose hasse diagram is given in figure 1 . _
\(a ) at ( 0,0 ) @xmath103 ; ( b ) at ( -2,2 ) @xmath104 ; ( c ) at ( 0,2 ) @xmath105 ; ( d ) at ( 2,2 ) @xmath106 ; ( e ) at ( 0,4 ) @xmath107 ; ( f ) at ( 2,4 ) @xmath108 ; ( f ) ( d ) ( a ) ( c ) ( e ) ( a ) ( b ) ; examples of chains in @xmath2 are @xmath109 and @xmath110 .
examples of multichains of length @xmath111 and @xmath112 are @xmath113 , @xmath114 .
clearly @xmath115 .
the subposet @xmath116 is a poset ideal of @xmath2 and it is equal to @xmath117 .
the subposet @xmath118 is _ not _ a poset ideal of @xmath2 , because @xmath119 but @xmath120 .
let @xmath2 be a poset and @xmath1 be an integer .
we recall here the construction of the @xmath0-partite clutter @xmath3 .
let @xmath121 and @xmath122 $ ] a polynomial ring on @xmath123 variables .
the clutter @xmath3 is the clutter with vertex set @xmath124 whose edges are @xmath125 then the edge ideal of @xmath3 is @xmath126 we give a simple example of this construction .
[ poset2 ] _ let @xmath2 be the poset of example [ poset1 ] and @xmath127 . to avoid double indices , we set @xmath128 , @xmath129 and @xmath130 .
let @xmath131 $ ] .
to each multichain @xmath132 of length @xmath133 in @xmath2 , we associate the monomial @xmath134 . for instance , the multichain @xmath135 is associated with the monomial @xmath136 .
then the edge ideal of @xmath137 in @xmath138 is _
@xmath139 @xmath140 let @xmath141 for @xmath142 then @xmath143 . by construction
, @xmath3 is @xmath0-partite .
note that for @xmath65 this is the construction of a bipartite graph starting from a poset @xmath2 , given by herzog and hibi ( see @xcite and @xcite ) . for @xmath144 ,
_ not _ all cohen - macaulay @xmath0-uniform @xmath0-partite clutters arise from a poset ( see e.g. ( * ? ? ?
* example 3.4 ) ) .
+ following ( * ? ? ?
* theorem 1.1 ) , we describe explicitly the generators of the ideal @xmath16 .
a _ poset multideal _ of degree @xmath0 in @xmath2 is a @xmath0-tuple @xmath145 , such that @xmath146 notice that @xmath147 .
let @xmath148 be a poset multideal of degree @xmath0 in @xmath149 we define the monomial @xmath150 each monomial @xmath151 has degree @xmath152 here and in what follows , @xmath153 denotes the cardinality of a finite set . by (
* theorem 1.1 ) , we have @xmath154 to simplify the notation , from now on we denote by @xmath155 the ideal @xmath16 . _ let @xmath2 be the poset of example [ poset1 ]
. examples of poset multideal of degree @xmath133 and @xmath111 in @xmath2 , respectively , are @xmath156 consider the notation of example [ poset2 ] to avoid double indices , setting @xmath157 .
we have the monomials @xmath158 and @xmath159 . _ in ( * ? ? ?
* theorem 2.4 ) , the authors proved that @xmath155 is a weakly polymatroidal ideal .
weakly polymatroidal ideals have been introduced by hibi and kokubo in @xcite . from (
* theorem 1.3 ) , it follows that a weakly polymatroidal ideal has linear quotients .
then the ideal @xmath155 has linear quotients .
in particular it has a linear resolution .
hence @xmath160 is cohen - macaulay , by ( * ? ? ?
* theorem 3 ) .
our main result is a recursive formula for the betti numbers of @xmath155 , extending a recent result due to francisco , h and van tuyl , ( * ? ? ?
* theorem 3.8 ) .
we recall that the ideal @xmath155 is generated in a single degree @xmath161 and that it has a linear resolution .
[ cmdip ] let @xmath162 be a finite poset , @xmath1 be an integer .
then , for @xmath163 , @xmath164.\ ] ] before proving the theorem , we explain the situation in some degenerate cases .
[ degenerate ] * if @xmath165 , then @xmath166 . * if @xmath167 , then @xmath168 .
it is well - known that this ideal has a linear resolution and its betti numbers are given by @xmath169 * if @xmath162 is a finite poset and @xmath170 , then @xmath171 is generated by a single monomial @xmath172 . then @xmath173 and @xmath174 for each @xmath175 .
we now prove some technical results , which will be useful later .
[ splitinside ] let @xmath176 $ ] be a square - free monomial ideal with a @xmath0-linear resolution and @xmath177 .
let @xmath178 be a positive integer and @xmath179 $ ] .
then the ideal @xmath180 $ ] has a @xmath181-linear resolution and @xmath182 _ proof .
_ we first prove that the ideal @xmath183 has a @xmath181-linear resolution .
notice that @xmath184 is an @xmath185-regular sequence . by (
* theorem 2.2 ) , @xmath183 has a @xmath181-linear resolution , since @xmath25 has a @xmath0-linear resolution . to prove the formula we proceed by induction on @xmath186 if @xmath187 , we have nothing to show , since @xmath188 , for every @xmath189 let @xmath190 one has @xmath191 .
notice that @xmath192 is the disjoint union of @xmath193 and @xmath194 . by proposition [ urur ] ( ii ) , this decomposition is a betti splitting of @xmath183 , because @xmath195 has a @xmath181-linear resolution . then @xmath196 since @xmath197 and @xmath198 , for every @xmath163 , by induction @xmath199 [ final1 ] let @xmath200 be integers .
let @xmath201 be an ordered collection of monomial ideals with a @xmath0-linear resolution in @xmath202 $ ] .
set @xmath203 .
assume @xmath32 is the disjoint union of @xmath204 and that : * for every @xmath205 , the ideal @xmath206 have a @xmath0-linear resolution ; * for every @xmath205 , there exists an ideal @xmath207 , generated by a subset of the variables @xmath208 , such that @xmath209
. then @xmath210 , \text { for } i \geq 0.\ ] ]
we may assume @xmath211 , since for @xmath212 the formula is clear .
consider the following splitting of @xmath25 : @xmath213 by our assumption , @xmath32 is the disjoint union of @xmath214 and @xmath215 .
moreover the ideals @xmath216 and @xmath217 have a @xmath0-linear resolution by ( i ) , with @xmath218 . by proposition [ urur ] ( i ) ,
the splitting above is a betti splitting of @xmath25 . by assumption ( ii ) , we have @xmath219
. then @xmath220 by proposition [ splitinside ] , we get @xmath221 we conclude by setting @xmath222 and by induction on @xmath223 .
let @xmath81 be a poset and recall that , for our labeling convention , @xmath224 . let @xmath225 the smallest poset ideal of @xmath2 containing @xmath226 .
define @xmath227 let @xmath1 be an integer . for every @xmath228 , we define the ideal @xmath229 in order to define the ideal @xmath230 , we consider @xmath90 as a subposet of @xmath2 with respect to the induced order . notice that @xmath230 is generated in a single degree @xmath161 . since the ideal @xmath231 has a linear resolution , it follows that @xmath230 has a linear resolution .
we now give an order of the ideals @xmath232 .
let @xmath233 , with @xmath234
. we set @xmath235 if one of the following conditions is satisfied : * @xmath236 ; * @xmath237 and @xmath238 .
note that , with respect to the above order , @xmath239 , for every @xmath228 .
[ poset3 ] _ let @xmath2 be the poset of example [ poset1 ] . clearly @xmath240 and @xmath241 .
the poset ideals containing @xmath108 are @xmath242 consider @xmath243 , @xmath127 and set of @xmath2 with respect of the induced order : _
@xmath244 @xmath245 then @xmath246 notice that the given order of the elements of @xmath247 is the order @xmath248 defined above .
[ partialsumlinear ] let @xmath2 be a poset , @xmath224 and fix @xmath228 . then the ideal @xmath249 has linear quotients .
in particular @xmath35 has a linear resolution .
. then @xmath251 .
this ideal has linear quotients , since @xmath252 has linear quotients .
we may assume @xmath253 . to prove that @xmath35 has linear quotients , we use proposition [ linquot ] .
we want to define an ordering @xmath254 of the monomials in @xmath38 . for each @xmath255 , @xmath256 has linear quotients with respect to an ordering @xmath257 of the monomials in @xmath258 .
let @xmath259 . then @xmath260 and @xmath261 , for some @xmath262 .
we set @xmath263 if one of the following conditions is satisfied : * @xmath264 and @xmath265 ; * @xmath266 and @xmath267 . with respect to the above notation ,
let @xmath259 such that @xmath268 .
assume first @xmath266 .
since @xmath256 has linear quotients , by proposition [ linquot ] , there exists a variable @xmath269 and a monomial @xmath270 , with @xmath271 such that @xmath272 . by definition of @xmath254 , we have @xmath273 and we are done .
assume now @xmath264 .
we have @xmath265
. then there exist @xmath274 .
in fact assume , by contradiction , that @xmath275 .
it follows that @xmath276 , since @xmath277 is a poset ideal of @xmath2 .
this is a contradiction , since @xmath278 and @xmath264 .
then @xmath279 , because @xmath280 . since @xmath281 , then @xmath282 . by definition of @xmath283
, there exists a poset multideal @xmath284 of degree @xmath285 in @xmath286 such that @xmath287 since @xmath288 , there exists @xmath289 such that @xmath290 .
consider the monomial @xmath291 where @xmath292 .
clearly @xmath293 is a poset ideal of @xmath2 , because @xmath294 and we have @xmath295 , since @xmath296 .
moreover @xmath297 is a poset multideal of degree @xmath285 in @xmath293 . then @xmath298 . since @xmath299
, one has @xmath300 and @xmath273 . from the fact that @xmath301 , we conclude by proposition [ linquot ] . in the next lemma , we summarize some important properties of the ideals @xmath230 and @xmath302 , for @xmath228 .
we recall that the intersection @xmath303 of two monomial ideals is generated by @xmath304 , where @xmath305 denotes the least common multiple of the monomials @xmath306 and @xmath307 .
clearly @xmath308 , see @xcite .
[ inclusion ] let @xmath2 be a poset , @xmath224 , @xmath309 , @xmath1 an integer
. then 1 .
@xmath310 ; 2 .
@xmath311 and , in particular , @xmath312 ; 3 .
@xmath313 .
first we prove ( i ) .
let @xmath314 by definition there exists a poset multideal @xmath315 of degree @xmath285 in @xmath90 , such that @xmath316 , where @xmath317 .
since @xmath318 , then there exists @xmath319 such that @xmath320 note that @xmath321 is a poset multideal of degree @xmath0 in @xmath322 , because @xmath224 .
hence @xmath323 now we prove ( ii ) .
we first show the inclusion from left to right .
let @xmath324 there exist @xmath325 and @xmath326 poset multideals of degree @xmath0 in @xmath90 and @xmath327 , respectively , such that @xmath328 .
define @xmath329 we claim that @xmath330 is a poset multideal of @xmath331 .
fix @xmath319 .
notice that @xmath332 , because @xmath333 . if @xmath334 , we have nothing to prove , then we may assume @xmath335 .
let @xmath336 and @xmath337 , with @xmath93 .
we prove that @xmath338 .
assume first @xmath339 . since @xmath90 is a poset ideal of @xmath2 and @xmath91 , we have @xmath340 . on the other hand ,
@xmath341 and @xmath342 is a poset ideal of @xmath343 , then @xmath344 .
hence @xmath338 .
the proof is the same if @xmath345 . from the fact that @xmath346 , for every @xmath347 , we have that @xmath348 .
then @xmath349 , since @xmath350 and @xmath31 are square - free monomials .
we now prove the reverse inclusion .
let @xmath351 where @xmath330 is a poset multideal of degree @xmath0 in @xmath352 let @xmath353 and @xmath354 , for @xmath142 clearly @xmath355 we prove that @xmath356 is a poset multideal of degree @xmath0 in @xmath90 , proving that @xmath357 is a poset ideal of @xmath358 for every integer @xmath319 .
let @xmath359 and @xmath360 such that @xmath361 since in particular @xmath362 then @xmath363 since @xmath340 , then @xmath364 .
the proof for @xmath327 is the same .
then @xmath365 eventually we prove ( iii ) .
we have @xmath366 by part ( ii ) and since @xmath367 , the result follows .
[ final ] let @xmath81 be a poset . then @xmath368 , \text { for every } i \geq 0.\ ] ] to prove the formula , we use proposition [ final1 ] .
we consider the order @xmath248 defined above on the ideals @xmath232 .
we recall that the ideals @xmath232 have a linear resolution and that @xmath369 is the disjoint union of @xmath370 . by proposition [ partialsumlinear ] , the ideals @xmath371 have a linear resolution , for every @xmath228
. then condition ( i ) of proposition [ final1 ] is fulfilled .
we verify now condition ( ii ) of proposition [ final1 ] .
fix @xmath228 , with @xmath253 .
we define the following set : @xmath372 we claim that @xmath373 the inclusion from right to left is clear , since for every @xmath374 , we have @xmath375 , @xmath376 and then @xmath377 . to prove the other inclusion , let @xmath255 , such that @xmath378 . since @xmath379 and @xmath380 , then there exists @xmath381 ( by the same argument in the proof of proposition [ partialsumlinear ] ) . since @xmath382
, then @xmath383 then the poset ideal @xmath384 is an element of @xmath247 , such that @xmath385 and @xmath386 , i.e. @xmath374 .
notice that @xmath387 , then @xmath388 from this and from lemma [ inclusion ] ( ii)-(iii ) , it follows that @xmath389 so we proved ( [ jj ] ) . for @xmath228 and @xmath374 ,
let @xmath390 such that @xmath391 by ( [ jj ] ) and lemma [ inclusion ] ( iii ) it follows
@xmath392 so we proved that condition ( ii ) in proposition [ final1 ] is fulfilled .
one has @xmath393 by construction .
recall that @xmath394 for @xmath163 . then by proposition [ final1 ] , for every @xmath163 @xmath395.\ ] ] since the poset ideal @xmath396 is such that @xmath397 , the result follows .
now we are able to prove theorem [ cmdip ] . .
we proceed by induction on @xmath398 . for @xmath399 ,
the formula holds by remark [ degenerate ] .
we may assume @xmath400 .
we prove that @xmath401 let @xmath402 and @xmath330 be a poset multideal of degree @xmath0 in @xmath2 such that @xmath403 assume first @xmath404 , then @xmath405 divides @xmath406 notice that @xmath407 is a poset multideal of degree @xmath0 in @xmath322 and @xmath408 .
assume @xmath409 , for @xmath410 let @xmath411 then @xmath412 hence @xmath413 is a poset multideal of degree @xmath285 in @xmath90 and @xmath414 conversely , let @xmath415 .
let @xmath416 be a poset multideal of degree @xmath0 in @xmath322 such that @xmath417 since @xmath418 is a poset multideal of degree @xmath0 in @xmath419 one has @xmath420 let @xmath421 such that @xmath412 let @xmath284 a poset multideal of degree @xmath285 in @xmath90 such that @xmath422 .
one has @xmath402 since @xmath423 is a poset multideal of degree @xmath0 in @xmath149 recall that the ideal @xmath424 has a linear resolution .
clearly @xmath425 is the disjoint union of @xmath426 and @xmath369 . by proposition [ urur ] ( ii ) ,
the splitting above is an @xmath405-splitting .
notice that @xmath427 has a linear resolution and @xmath428 for every @xmath163 . by lemma [ inclusion ] ( i )
@xmath429 then @xmath430 by proposition [ final ] one has , for every @xmath163 , @xmath431.\ ] ] notice that @xmath432 . by induction
we conclude . _ for @xmath65 , it can be easily proved that this result is equivalent to ( * ? ? ?
* theorem 3.8 ) ( see equation ( [ ok ] ) ) . _ _ notice that the betti numbers @xmath433 do not depend on the characteristic of the base field @xmath5 .
_ the splitting formula can be simplified if @xmath2 has a unique maximal element .
[ simpler ] let @xmath2 be a poset with a unique maximal element @xmath226 and let @xmath1 be an integer
. then @xmath434 by assumption @xmath435 , where @xmath226 is the unique element in @xmath247 .
the splitting formula follows immediately by equation ( [ ok ] ) in the proof of theorem [ cmdip ] .
let @xmath6 be a cohen - macaulay bipartite graph . by ( * ? ?
* theorem 3.4 ) , there exists a poset @xmath2 such that @xmath436 . by using theorem [ cmdip ]
, we are able to give another formulation of ( * ? ? ?
* theorem 3.8 ) .
[ cmbip ] let @xmath6 be a cohen - macaulay bipartite graph and @xmath2 be the poset such that @xmath436 .
then @xmath437 the formula follows immediately by theorem [ cmdip ] with @xmath65 and remark [ degenerate ] .
in this section we present some interesting examples about betti splittings of alexander dual ideals of cohen macaulay simplicial complexes . by ( * ? ? ? * theorem 3 ) such ideals have a linear resolution . for more definitions about simplicial complexes , their properties and the stanley - reisner correspondence
we refer to ( * ? ? ?
* chapter 1 ) and ( * ? ? ?
* chapter 3 ) .
a _ facet _ is a maximal face of @xmath19 with respect to the inclusion of sets .
denote by @xmath442 the collection of facets of @xmath19 .
a simplicial complex @xmath19 is called _ pure _ if all its facets have the same cardinality .
the _ alexander dual ideal _
@xmath18 of @xmath19 is defined by 1.5 mm in the following example we show an ideal with a linear resolution that does not admit _ any _ betti splitting . in this example
we also point out that the existence of betti splittings of @xmath20 and @xmath18 are completely unrelated .
it is contractible but not collapsible ( see @xcite ) .
moreover @xmath19 is cohen - macaulay and the graded betti numbers of @xmath18 are given by the following linear resolution : @xmath448 in particular @xmath18 has @xmath449 minimal generators of degree @xmath112 . using cocoa @xcite or macaulay 2 @xcite
, it can be shown that for _ every _ monomial ideals @xmath450 such that @xmath451 being @xmath452 is the disjoint union of @xmath38 and @xmath39 , the decomposition @xmath453 is _ not _ a betti splitting of @xmath18 .
this example answers ( * ? ? ?
* question 4.3 ) , showing that , in general , there is no relation between the existence of betti splitting for a square - free monomial ideal and the characteristic dependence of its graded betti numbers , since in this case the ideal @xmath18 has characteristic - independent resolution . _
( ( * ? ? ?
* definition 3.1 ) ) _ _ a pure non - empty simplicial complex @xmath19 on @xmath7 vertices is called _ weakly vertex decomposable _
( wvd ) if there exists a vertex @xmath307 of @xmath19 fulfilling exactly one of the following conditions :
_ notice that provan and billera @xcite use the same name for a different class of complexes .
every weakly vertex decomposable simplicial complex is cohen - macaulay and every pure vertex decomposable simplicial complex is weakly vertex decomposable , see @xcite .
let @xmath19 be a vertex decomposable complex . by ( * ?
* theorem 2.8 , corollary 2.11 ) , the ideal @xmath18 admits @xmath4-splitting , for some @xmath4 .
weakly vertex decomposability of @xmath19 is a sufficient condition for the existence of an @xmath4-splitting of @xmath18 , for a suitable @xmath4 , generalizing the result cited above .
if @xmath19 is a cone over @xmath464 for some @xmath51 , then @xmath18 and @xmath465 have the same graded betti numbers .
then we may assume that @xmath19 is not a cone .
let @xmath51 be the vertex in the definition of weakly vertex decomposable simplicial complex . since @xmath464 is cohen - macaulay , then it is pure and @xmath51 is a shedding vertex ( see for instance @xcite ) . by (
* lemma 2.2 ) , we have @xmath466 .
recall again that @xmath464 is cohen - macaulay . then , by ( * ? ? ?
* theorem 3 ) , @xmath465 has a linear resolution and the result follows by proposition [ urur ] ( ii ) . [ hach]_(hachimori s example ) _ _ let @xmath19 be the simplicial complex in figure 3 , due to hachimori @xcite .
it is shellable , hence cohen - macaulay . ,
but not weakly vertex decomposable . in fact
@xmath19 is not a cone and @xmath464 is not cohen - macaulay for @xmath467 . _ by (
* theorem 6.2 ) , every vertex decomposable ball is non evasive . in the next example we present a non evasive ball @xmath19 such that @xmath18 does not admits @xmath4-splitting , showing that ( * ? ? ?
* theorem 2.8 , corollary 2.11 ) does not hold for the alexander dual ideal of a non evasive simplicial complex .
[ rud]_(rudin s ball ) _ _ let @xmath19 be the _
rudin s ball _ , a non - shellable triangulation of the tetrahedron with @xmath477-vertex introduced in @xcite .
since @xmath19 is a ball , it is cohen - macaulay and @xmath18 has a @xmath478-linear resolution : @xmath479 by ( * ? ? ? * theorem 6.3 ) , @xmath19 is non evasive .
it is not difficult to show that @xmath18 does not admit @xmath51-splitting , for @xmath480 .
nevertheless @xmath481 is a betti splitting of @xmath18 , where @xmath482 and @xmath483 are given by the following facets : @xmath484\{1,3,7,13},\{1,3,9,13},\{1,5,7,11},\{1,5,9,11},\{1,7,11,13},\{1,9,11,13},\{3,4,7,11 } , _ \{6,8,11,12},\{6,10,11,12},\{8,12,13,14},\{10,11,12,14}@xmath485 . +
both @xmath487 and @xmath488 are computed in @xmath489 $ ] .
the minimal graded free resolutions of @xmath487 , @xmath488 and @xmath490 are 10 b. benedetti , f. lutz , _ knots in collapsible and non - collapsible balls _ , electr
. j. comb .
* 20 * , 3 , research paper p31 , 29 ( 2013 ) .
m. caboara , s. faridi , _ odd - cycle - free complexes and the knig property _ , rock . mount .
j.l of math . ,
* 41 * , 4 ( 2011 ) pp .
1059 - 1079 .
cocoateam , _ cocoa 5 : a system for doing computations in commutative algebra _ , available at http://cocoa.dima.unige.it .
a. conca , j. herzog , _ castelnuovo - mumford regularity of products of ideals _ ,
math . * 54 * , 2 ( 2003 ) , pp .
137 - 152 .
j. a. eagon , v reiner , _ resolutions of stanley - reisner rings and alexander duality _ ,
j.pure and applied algebra * 130 * , 3 ( 1998 ) , pp .
265 - 275 .
s. eliahou , m. kervaire , _ minimal resolutions of some monomial ideals _
, j. algebra * 129 * ( 1990 ) , pp . 1 - 25
. v. ene , j. herzog , f. mohammadi , _ monomial ideals and toric rings of hibi type arising from a finite poset _
, european j. comb . * 2 * , 3 ( 2011 ) , pp .
404 - 421 .
a. engstrm , _ discrete morse functions from fourier transform _ , experimental mathematics * 18 * ( 2009 ) , pp .
45 - 53 . o. fernandez - ramos , p. gimenez , _ regularity 3 in edge ideals associated to bipartite graphs _ , journal of algebraic combinatorics , * 39 * , 4 ( 2014 ) , pp .
919 - 937 .
s. faridi , _ the facet ideal of a simplicial complex _ , manuscripta math .
* 109 * ( 2002 ) , pp .
159 - 174 . c. a. francisco , h. t. h , a. van tuyl , _ splittings of monomial ideals _
soc . * 137 * ( 2009 ) , pp .
3271 - 3282 . c. a. francisco , a. van tuyl , _ sequentially cohen - macaulay edge ideals _ , proc .
soc . * 135 * ( 2007 ) , pp .
2327 - 2337 .
d. r. grayson , m. e. stillman , _
macaulay2 , a software system for research in algebraic geometry _ , available at
http://www.math.uiuc.edu / macaulay2/. m. kokubo , t. hibi , _ weakly polymatroidal ideals _ , alg.colloq . * 13 * ( 2006 ) pp .
711 - 720 .
h. t. h , s. morey , r. h. villareal , _ cohen - macaulay admissible clutters _ , j. comm .
algebra * 1 * , 3 ( 2009 ) , pp .
463 - 480 .
h. t. h , a. van tuyl , _ resolutions of square - free monomial ideals via facet ideals : a survey _ , contemporary mathematics , * 448 * ( 2007 ) , pp .
91 - 117 . h. t. h , a. van tuyl , _ monomial ideals , edge ideals of hypergraphs , and their graded betti numbers _
comb * 27 * , 2 ( 2008 ) .
m. hachimori , _ simplicial complex library _ , http://infoshako.sk.tsukuba.ac.jp/ hachi / math / library / index eng.html . j. herzog , t. hibi , _ componentwise linear ideals _ , nagoya math .
j * 153 * ( 1999 ) , pp .
141 - 153 .
j. herzog , t. hibi , _ distributive lattices , bipartite graphs and alexander duality _
, j. algebr .
combinatorics , * 22 * ( 2005 ) , pp . 289 - 302 .
j. herzog , t. hibi , _ monomials ideals _ , graduate texts in mathematics , springer ( 2011 ) .
j. herzog , y. takayama , _ resolutions by mapping cones _ , the roos festschrift volume 2 , homology , homotopy and applications * 4 * , 2 , part 2 ( 2002 ) pp .
277 - 294 .
a. soleyman jahan , x. zheng , _ ideals with linear quotients _ , j. comb .
theory , series a , * 117 * , 1 ( 2010 ) , pp .
104 - 110 .
j. jonsson , _ simplicial complexes of graphs _ , lecture notes in mathematics * 1928 * , springer - verlag ( 2008 ) .
s. moradi , f. khosh - ahang , _ on vertex decomposable simplicial complexes and their alexander dual _ , arxiv:1302.5947v4 ( to appear in math.scand . )
f. mohammadi , s. moradi , _ weakly polymatroidal ideals ith applications to vertex cover ideals _ , osaka j. math . * 47 * , ( 2010 ) pp .
627 - 636 .
s. morey , e. reyes , r. h. villareal , _ cohen - macaulay , shellable und unmixed clutters with a perfect matching of knig type _ ,
j. pure appl .
algebra * 212 * ( 2008 ) , pp . 1770 - 1786 .
s. morey , r. h. villarreal , _ edge ideals : algebraic and combinatorial properties _ , progress in comm .
alg . comb . and homology * 1 * ( 2012 ) , pp .
u. nagel , t. rmer , _ glicci simplicial complexes _ , journal of pure and applied algebra * 212 * , 10 ( 2008 ) , pp .
2250 - 2258 .
j. s. provan , l. j. billera , _ decompositions of simplicial complexes related to diameters of convex polyhedra _
operations research * 5 * ( 1980 ) , pp .
576 - 594 .
m. e. rudin , _ an unshellable triangulation of a tetrahedron _ , bull .
( 1958 ) , pp .
r. p. stanley , _ balanced cohen - macaulay complexes _ trans .
* 249 * ( 1979 ) , pp .
139 - 157 .
a. van tuyl , r. h. villarreal , _ shellable graphs and sequentially cohen - macaulay bipartite graphs _ , j. comb .
theory , series a * 115 * , 5 ( 2008 ) pp .
799 - 814 .
r. h. villarreal , _ cohen - macaulay graphs
_ , manuscripta math .
* 66 * ( 1990 ) , pp .
277 - 293 .
r. woodroofe , _ chordal and sequentially cohen - macaulay clutters _ , electr
. j. comb . * 18 * , 1 ( 2011 ) .
e. c. zeeman , _ on the dunce hat _ ,
topology * 2 * ( 1963 ) , pp .
341 - 358 . | a natural extension of bipartite graphs are @xmath0-partite clutters , where @xmath1 is an integer . for a poset @xmath2 ,
ene , herzog and mohammadi introduced the @xmath0-partite clutter @xmath3 of multichains of length @xmath0 in @xmath2 , showing that it is cohen - macaulay .
we prove that the cover ideal of @xmath3 admits an @xmath4-splitting , determining a recursive formula for its betti numbers and generalizing a result of francisco , h and van tuyl on the cover ideal of cohen - macaulay bipartite graphs .
moreover we prove a betti splitting result for the alexander dual of a cohen - macaulay simplicial complex .
* keywords * : resolution of edge ideals , cohen - macaulay clutters , betti splittings , posets .
+ _ ams mathematics subject classification 2010 _ : 13d02 , 13a02 , 05e40 , 05e45 . |
Subsets and Splits