article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
it was argued long ago that melting of quarkonia above the deconfinement transition can serve as a signature of quark - gluon plasma formation in heavy ion collisions @xcite .
the basic idea behind this proposal was that due to color screening the potential between a quark and anti - quark will not provide sufficient binding at high temperature .
this problem can be formulated more rigorously in terms of quarkonium spectral functions , which can be , in principle , extracted from euclidean - time meson correlation functions calculated on the lattice .
attempts doing this based on the maximum entropy method ( mem ) have been discussed over the last few years .
the initial interpretation of data led to the conclusion that the @xmath0 charmonium states survive in the deconfined medium up to temperatures of about @xmath1 , with @xmath2 being the transition temperature @xcite .
recent analysis , however , has shown that , although mem can be used to extract reliably quarkonium spectral functions at zero temperature , at finite temperature it has severe limitations @xcite . at zero temperature quarkonium spectrum
is well described in non - relativistic potential models .
since the seminal paper by matsui and satz the problem of charmonium dissolution has been studied in potential models @xcite .
while the early studies used phenomenological potential , more recent studies rely on static quark anti - quark free energy calculated on the lattice .
in fact significant progress has been made in understanding the in - medium modification of inter - quark forces via lattice calculations of the free energy of static quark anti - quark pair .
calculations have been done in pure gluodynamics , 3-flavor and 2-flavor qcd @xcite , and preliminary results are also available in the physically relevant case of one heavy strange quark and two light quarks @xcite ( the light quark masses correspond to pion mass of about @xmath3 mev ) .
recently attempts to calculate quarkonium properties at finite temperature using resummed perturbation theory have been made @xcite .
resummed perturbation theory appears to be successful in calculations of bulk thermodynamics properties @xcite .
since the lattice calculations of spectral functions have severe limitations , in @xcite it has been pointed out , that comparison between the lattice data and potential models should be done in terms of the euclidean time correlators , for which the numerical results are much more reliable .
recent studies following this line have also been presented in refs . .
in this contribution we discuss the calculation of quarkonium spectral functions in a potential model , which uses the lattice data of the free energy of a static quark anti - quark pair .
since reliable calculations of the quarkonium correlators are available only in quenched approximation we consider qcd with only heavy quarks .
further details about this approach can be found in ref .
for heavy quarks the spectral function can be related to the non - relativistic green s function @xmath4 & \displaystyle \sigma(\omega)=k \frac{6}{\pi}\frac{1}{m^2 } { \rm i m } \vec{\nabla } \cdot \vec{\nabla ' } g^{nr}(\vec{r},\vec{r'},e)|_{\vec{r}=\vec{r'}=0}\ , , \label{green_sc } \end{aligned}\ ] ] for @xmath5-wave , and @xmath6-wave charmonia , respectively . here
@xmath7 . at leading order @xmath8 .
relativistic and higher order perturbative corrections will lead to a value different from unity @xcite .
the non - relativistic green s function satisfies the schrdinger equation @xmath9 g^{nr}(\vec{r},\vec{r'},e ) = \delta^3(r - r')\ , .
\label{schroedinger}\ ] ] the numerical method for solving this equation is presented in @xcite . at zero temperature
we use the cornell potential @xmath10 with parameters motivated by lattice results on static potential : @xmath11 and @xmath12 . in the actual calculations we use a potential which is screened beyond some distance @xmath13 with screening length @xmath14 to mimic many body effects at large energies ( see ref .
@xcite for further details ) . at finite temperature
we use a potential motivated by lattice results on the singlet free energy of a static quark anti - quark pair and which is defined in section iv of ref .
@xcite . at large energies , away from the threshold ,
the non - relativistic treatment is not applicable . the spectral function in this domain ,
however , can be calculated using perturbation theory .
we smoothly match the non - relativistic calculation of the spectral function to the relativistic perturbative result @xcite .
euclidean time correlators @xmath15 at some temperature @xmath16 can be calculated from the spectral functions using the integral representation @xmath17 here the integration kernel is @xmath18
in this section we discuss the comparison of the model calculations with zero temperature lattice data from isotropic lattices @xcite .
the lattice spacing has been fixed using the sommer - scale @xmath19fm .
its value is slightly larger than the one used in ref .
@xcite , since there the string tension of @xmath20mev has been used to set the scale .
calculations have been done at the charm quark mass which corresponds to an @xmath21 mass of about @xmath22gev .
the renormalization constants of the lattice operators have been calculated in 1-loop tadpole improved perturbation theory ( see ref .
@xcite for further details ) .
the @xmath23 factors in eq .
( [ green_sc ] ) have been chosen such that at large distances the correlators calculated in potential models agree with the lattice results . in fig .
[ fig : pscorr ] we show the pseudo - scalar correlator calculated on the lattice and in the potential model for several screening parameters , together with the corresponding spectral functions . as one can see from the figure the choice of the ad - hoc screening parameters have almost no effect on the euclidean correlator .
we see a reasonably good agreement between the lattice data and potential model calculations .
also shown in the figure is the correlator corresponding only to the non - relativistic spectral function . at small euclidean times
this falls below the lattice data by more than an order of magnitude .
thus correlators calculated on the lattice are sensitive to the relativistic continuum part of the spectral functions
. similar analysis of the correlators have been done in the vector , axial vector and scalar channels .
are shown.,width=343 ]
in this section we study the temperature - dependence of quarkonium spectral functions and correlators .
since the correlators depend on the temperature through the integration kernel and the spectral functions , it is customary to study the temperature dependence of the correlators in terms of the ratio @xmath24 , where @xmath25 this way the trivial temperature dependence due to the integration kernel is taken care of .
also many uncertainties of the lattice calculations cancel out in this ratio .
the finite temperature spectral functions are shown in fig .
[ fig : res_v_ps ] for the pseudo - scalar channel .
the 1s charmonium state is melted at @xmath26 .
we see , however , a large enhancement of the spectral functions near the threshold . note , that the height of the spectral functions near the threshold is comparable to the height of the bump in the spectral function calculated from mem @xcite .
it is therefore possible that the bump of the spectral functions calculated from lattice correlators using mem actually corresponds to a threshold enhancement , and was mistakenly interpreted as the @xmath0 state . in the case of bottomonium all states , except the @xmath0 state , are dissolved above the deconfinement transition .
the @xmath0 state can survive as a resonance until temperatures of about @xmath27 .
.the mass and the binding energy of the @xmath0 bottomonium state at different temperatures , and the continuum threshold . [ cols="^,^,^,^,^",options="header " , ] note , however , that the binding energy of the @xmath0 bottomonium is significantly reduced due to color screening , as shown in table [ tab ] .
the binding energy is defined as the distance between the continuum threshold @xmath28 and the bound state peak . due to the small binding energy
the 1s state will acquire a sizable thermal width , and may not show up as the resonance in the corresponding spectral function .
therefore , the actual dissolution temperature of the @xmath0 bottomonium will be smaller then the one estimated based on the simple potential model calculations which do not include the effect of the thermal width @xcite . .
the error - bars on the lattice spectral function correspond to the statistical error of the spectral function integrated in the @xmath29-interval corresponding to the horizontal error - bars .
the insets show the corresponding ratio @xmath30 together with the results from anisotropic lattice calculations @xcite . for charmonium ,
lattice calculations of @xmath30 are shown for @xmath31 ( squares ) , @xmath32 ( circles ) , and @xmath33 ( triangles ) . for bottomonium lattice data are shown for @xmath34 ( circles ) and @xmath35 ( triangles ) .
, title="fig:",width=291 ] .
the error - bars on the lattice spectral function correspond to the statistical error of the spectral function integrated in the @xmath29-interval corresponding to the horizontal error - bars .
the insets show the corresponding ratio @xmath30 together with the results from anisotropic lattice calculations @xcite . for charmonium ,
lattice calculations of @xmath30 are shown for @xmath31 ( squares ) , @xmath32 ( circles ) , and @xmath33 ( triangles ) . for bottomonium lattice data are shown for @xmath34 ( circles ) and @xmath35 ( triangles ) .
, title="fig:",width=291 ] in the insets of fig .
[ fig : res_v_ps ] we also show the corresponding ratio @xmath30 .
the large changes in the spectral functions are not visible in the correlation functions .
these agree quite well with the lattice results .
this is because even in the absence of bound states the spectral function is significantly larger than the spectral function corresponding to a freely propagating quark anti - quark pair . in the scalar and axial - vector channel at @xmath34 for charmonium ( left ) and bottomonium ( right ) .
lattice calculation on isotropic lattices @xcite are shown as filled symbols .
open symbols refer to results from anisotropic lattice calculations of ref .
@xcite.,title="fig:",width=287 ] in the scalar and axial - vector channel at @xmath34 for charmonium ( left ) and bottomonium ( right ) .
lattice calculation on isotropic lattices @xcite are shown as filled symbols .
open symbols refer to results from anisotropic lattice calculations of ref .
@xcite.,title="fig:",width=287 ] in the vector , scalar and axial - vector channels significant temperature dependence has been found @xcite .
it has been shown that this is due to the zero mode contribution , i.e. due to the @xmath36-like contribution to the quarkonium spectral functions @xcite . in the vector channel
the zero mode contribution corresponds to the heavy quark transport @xcite .
the zero mode contribution can be estimated in the free case .
it can also be shown that the zero mode contribution is absent in the pseudo - scalar channel .
if we add the free theory result for the zero mode contribution to the spectral function calculated in the potential model we can reproduce the temperature dependence of the scalar and axial - vector correlator both for charmonium and bottomonium . this is demonstrated by fig .
[ fig : resp ] where @xmath30 for the scalar and axial - vector channels is shown .
we discussed the calculations of quarkonium spectral functions and the corresponding euclidean time correlators in a potential model .
we have found that all quarkonium states , except the @xmath0 bottomonium state , dissolve at temperatures smaller than about @xmath26 .
this , however , does not lead to significant change of the correlators .
zero mode contribution on the other hand could give a large change in the correlators above the deconfinement transition .
we have found that the spectral functions calculated in our model can explain quite well the temperature dependence of the quarkonium correlators obtained in lattice qcd .
this work has been supported by u.s .
department of energy under contract no .
de - ac02 - 98ch10886 . t. matsui and h. satz , _ phys .
b _ * 178 * ( 1986 ) 416 t. umeda , k. nomura and h. matsufuru , _ eur .
j. c _ * 39s1 * ( 2005 ) 9 m. asakawa and t. hatsuda , _ phys .
lett . _ * 92 * ( 2004 ) 012001 s. datta , f. karsch , p. petreczky and i. wetzorke , _ phys .
d _ * 69 * ( 2004 ) 094507 g. aarts , c. r. allton , r. morrin , a. p. o. cais , m. b. oktay , m. j. peardon and j. i. skullerud , arxiv : hep - lat/0610065 ; _ nucl . phys .
a _ * 785 * ( 2007 ) 198 ; [ arxiv : hep - lat/0608009 ] ; arxiv:0705.2198 [ hep - lat ] a. jakovc , p. petreczky , k. petrov and a. velytsky , _ phys .
d _ * 75 * ( 2007 ) 014506 s. digal , et al .
, _ phys .
b _ * 514 * ( 2001 ) 57 ; _ phys . rev .
d _ * 64 * ( 2001 ) 094015 c. y. wong , _ phys .
c _ * 72 * ( 2006 ) 034906 .
mcsy and p. petreczky , _ eur .
j. c _ * 43 * ( 2005 ) 77 .
mcsy and p. petreczky , _ phys .
d _ * 73 * ( 2006 ) 074007 d. cabrera , r. rapp , arxiv : hep - ph/0610254 w. m. alberico , a. beraudo , a. de pace and a. molinari , _ phys .
d _ * 75 * ( 2007 ) 074009 c. y. wong and h. w. crater , _ phys .
d _ * 75 * ( 2007 ) 034505 o. kaczmarek , et al . , _ phys .
b _ * 543 * ( 2002 ) 41 ; _ phys . rev
. d _ * 70 * ( 2004 ) 074505 p. petreczky , k. petrov , _ phys . rev . d _ * 70 * ( 2004 ) 054503 o. kaczmarek , f. zantow , _ phys . rev .
d _ * 71 * ( 2005 ) 114510 k. petrov [ rbc - bielefeld coll .
] , arxiv : hep - lat/0610041 .
f. karsch , a. patks and p. petreczky , _ phys .
b _ * 401 * ( 1997 ) 69 j. o. andersen , e. braaten and m. strickland , _ phys . rev .
* 83 * ( 1999 ) 2139 j. p. blaizot , e. iancu and a. rebhan , _ phys .
* 83 * ( 1999 ) 2906 | we discuss the calculations of quarkonium spectral functions in potential models and their implications for the interpretation of the lattice data on quarkonium correlators .
in particular , we find that melting of different quarkonium states does not lead to significant change in the euclidean time correlators .
the large change of the quarkonium correlators above deconfinement observed in the scalar and axial - vector channels appears to be due to the zero mode contribution . |
`` polyelectrolytes are polymers bearing ionizable groups , which , in polar solvents , can dissociate into charged polymer chains ( macroions ) and small counterions '' @xcite . the combination of macromolecular properties and long - range electrostatic interactions results in an impressive variety of phenomena .
it makes these systems interesting from a fundamental as well as a technological point of view .
a thorough understanding of polyelectrolytes has become increasingly important in biochemistry and molecular biology .
this is due to the fact that virtually all proteins , as well as the dna , are polyelectrolytes .
their interactions with each other and with the charged cell - membrane are still far from being fully understood , which is partly due to the intricate coupling between ion distribution and chain conformation .
a first approach to the problem is to fix the conformation of the chain and to focus on a detailed description of the counterion distribution .
usually polyelectrolytes stretch due to the electrostatic repulsion of their charged groups .
moreover , many important polyelectrolytes have a large intrinsic stiffness ( e.g. , dna , actin filaments or microtubules ) .
therefore , a rod - like conformation is an obvious first choice .
the remaining problem of charged rods immersed into solution is much easier , but is still far from being exactly solvable .
an additional approximation , which is frequently used in theoretical descriptions , is to completely integrate out the counterionic degrees of freedom . on a linearized mean - field level
this yields a debye - hckel - like theory characterized by a screened coulomb potential between charged monomers . to obtain the correct physical properties
, one uses an _ effective _ yukawa potential , which in turn requires adjustable parameters such as an effective polyelectrolyte radius and charge . in this way
the dependence of the ionic structure , hereafter called electrical double layer ( edl ) , around two or more polyelectrolytes , as a function of the polyelectrolyte - polyelectrolyte distance , is not accounted for .
this information , however , is most relevant for the understanding of polyelectrolyte aggregation or self - assembling .
this effect is more important for polyelectrolytes in low concentration added salt solution , since for this case the ionic screening is weak .
this has been shown to be relevant for charged plates and charged spherical macroions @xcite .
a common further approximation assumes that the investigation of a small sub - volume containing only one rod and its counterions will suffice to unveil much of the interesting physics .
the main justification for this approach is that the sub - volume has zero net charge .
moreover , the counterions will also efficiently screen higher order multipoles .
hence , the interactions between two such sub - volumes , which are neglected when focusing on just one rod , will be fairly weak .
this approximation is called cylindrical cell model and it provides the framework for our simulation calculation .
one attractive feature of the cell model is that in the salt - free case the nonlinear pb equation can be solved _
exactly _ in this geometry @xcite .
it also displays in a clear fashion the effect of partial counterion _ condensation _ @xcite .
while a charged sphere loses all its counterions upon dilution , a charged plane keeps all of them . for a charged cylinder
the fraction of ions , which upon dilution remain in the vicinity of the macroion , can be anywhere between @xmath2 and @xmath3% .
addition of salt increases the screening of the charged rod .
if the salt content is large enough , the electric field will have decayed to zero before the cell radius is reached .
it is then permissible to extend the latter to infinity .
this is the approach that we use for our integral equation calculations .
the pb theory shows its limitations for systems where ion correlations become important @xcite .
it can not predict the attractive forces which are seen experimentally in dna solutions @xcite and which have been also found in various simulation @xcite and integral equation @xcite studies in the presence of multivalent counterions . in this paper
we present studies for a rod immersed in a salt solution which pb theory fails to describe correctly , namely , the possibility of overcharging a single rod , resulting in an effective charge reversal , and the appearance of a non - monotonic zeta - potential @xcite .
this effect is most relevant in electrophoresis experiments @xcite .
the overcharging of a macroion modifies in a non - linear fashion the electrophoretic mobility as a function of the polyelectrolyte charge and salt concentration . in the classical electrophoresis theory of wiersema , obrien and white @xcite , which is a linear theory based on the poisson - boltzmann ( pb ) description of the edl around the polyelectrolyte ,
this effect is not included .
the edl obtained through the well established integral equation approach does predict overcharging @xcite .
the overcharging effect has been observed in monte carlo ( mc ) simulations for planar @xcite , cylindrical @xcite and spherical @xcite geometries . moreover ,
several years ago , gonzales - tovar _
et al._@xcite predicted that for some macroion charge and electrolyte concentration the overcharging produces an electrophoresis mobility reversal .
this mobility reversal has been experimentally observed @xcite .
this effect has been taken into account in the electrophoresis theory by lozada - cassou _
et al._@xcite .
an excellent qualitative agreement with experimental results @xcite is found .
this overcharging effect and the predicted mobility reversal have been recently addressed by deserno _
et al._@xcite and shklovskii @xcite . in ref .
@xcite ion distributions are studied via the integrated charge distribution function , @xmath4 , and overcharging is addressed based on md simulations . in the present paper
we extend this research by approaching the overcharging effect through a well established integral equation theory @xcite as well as more detailed molecular dynamics ( md ) simulations .
we compute @xmath4 , the concentration profiles @xmath5 , the mean electrostatic potential profile @xmath6 , and the @xmath1-potential .
a comparison of the integral equation results with those from md is presented .
placed parallel to an edge of a cube of side length @xmath7 yields an infinite square array of infinitely long rods upon periodic replication of the cubic box .
the dark particles are the counterions or positive salt ions and the bright particles are the negative salt ions .
this particular snapshot contains 36 divalent counterions , 220 positive and 220 negative divalent salt ions.,width=264 ] compared to the spherical cell model , the cylindrical one presents one additional but crucial complication : the charged rod is infinitely long .
several methods have been proposed in the literature to handle this problem .
they essentially all use as a unit cell a hexagonal prism with a certain height .
this approximates the cylindrical cell by the `` most round '' space - filling object . in this publication
we take a computationally even simpler approach by using a _
cubic _ unit box of side - length @xmath7 and placing the dna parallel to one of the edges ( see fig . [
pic : cellpicture ] ) .
the justification for this approach is as follows : in a high - salt environment the charged rod will be screened within a distance much smaller than the size of the box
. the potential will therefore become constant long before the deviations from a cylindrical cell will be felt .
it is thus largely irrelevant , whether these deviations are of hexagonal or of cubic type .
when comparing with the cylindrical cell model , a radius @xmath8 is appropriate , since it yields the same area per rod .
the main advantage of this approach is that such a system can be treated with the plain cubic ewald sum or one of its mesh - upgrades @xcite .
this permits a very efficient way for computing the long - range electrostatic interactions .
a specification of two interaction potentials is necessary to describe the model system : _ (
i ) _ an excluded volume interaction preventing two particles from occupying the same position in space and _ ( ii ) _ the long - range coulomb potential . for the excluded volume interaction between the ions we use the following potential : @xmath9 & 0 < r \le r_{{\operatorname{cut}}}\equiv 2^{1/6}a \\[2ex ] 0 & r_{{\operatorname{cut } } } < r. \end{array } \right .
\label{eq : shifted_lj_potential}\ ] ] without the cutoff this would be the common lennard - jones potential describing particles with a diameter @xmath10 and an attractive potential minimum of @xmath11 . in order to achieve pure repulsion
, we cut the potential at the minimum and shift it up such that it smoothly goes to zero there . as a consequence , @xmath11 becomes largely irrelevant .
since two ions a distance @xmath10 apart have the ( repulsive ) interaction energy @xmath11 , we make the natural choice @xmath12 . in order to give the rod a larger radius than the ions
, we employ an ion - rod potential similar to eq .
( [ eq : shifted_lj_potential ] ) , in which @xmath13 is replaced by @xmath14 .
this shifts the hard core a distance @xmath15 towards larger radii and gives a distance of closest approach of @xmath16 .
since the ions have diameter @xmath10 this corresponds to a rod radius @xmath17 of @xmath18 .
of course , in this case @xmath13 is a cylindrical and not a spherical radial coordinate .
the electrostatic interaction energy between two ions with charge @xmath19 and @xmath20 , is given by @xmath21 where @xmath22 is the valence of the @xmath23 species and @xmath24 the elemental charge . by definition the _ bjerrum length _
@xmath25 is the distance at which two unit charges have interaction energy @xmath26 . with the vacuum dielectric constant @xmath27 and the dielectric coefficient @xmath28
applying to water at room temperature , one gets a bjerrum length of approximately @xmath29 .
the line charge density @xmath30 of the rod is modeled by placing unit charges along the rod axis at the distance @xmath31 .
the number of charges along the rod per bjerrum length is an important dimensionless measure of the line charge density and is often referred to as the manning parameter : @xmath32 its relevance lies in the fact that for @xmath33 the phenomenon of counterion `` condensation '' is observed @xcite . within the periodic boundary conditions employed during the simulations ,
the presence of such long - range interactions poses both mathematical and technical difficulties .
we use an efficient fft accelerated ewald sum , the p@xmath34 m algorithm , which scales almost linearly with the number of charges @xcite .
the final step consists of explicitly mapping the parameters to a dna system in aqueous solution .
this affects ion diameter @xmath10 , rod radius @xmath17 , bjerrum length @xmath25 and line charge density @xmath30 .
our choice is presented in table [ tab : dna_param ] .
we performed molecular - dynamics simulations of the rod - systems using a langevin thermostat in order to drive the system into the canonical state @xcite .
the real - space and fourier - space part of the electrostatic energy was used to check for equilibration .
the presented observables originate from averaging over roughly 1500 independent configurations .
the integral equation formalism is a well established statistical mechanical approach that has been shown to be reliable when it is applied to simple microscopic models of inhomogeneous fluids @xcite .
a particularly successful integral equation theory for inhomogeneous , charged fluids is the so - called hypernetted - chain / mean spherical approximation ( hnc / msa ) @xcite .
the hnc / msa theory for a model charged rod , immersed in an electrolyte , has been derived in the past @xcite . in this paper
, we solve the hnc / msa equation for an infinitely long , hard , charged cylinder of radius @xmath17 , with uniform line charge density @xmath30 .
the rod is immersed in a two - component restricted primitive model electrolyte ( rpm ) , i.e. , a fluid of charged hard spheres of diameter @xmath10 with a centered point charge @xmath35 .
the fluid electroneutrality condition is @xmath36 where @xmath37 is the bulk concentration of species @xmath38 . in order to satisfy the necessary condition of zero electrical field at infinity ,
the rod charge is compensated by the charge induced in the fluid , @xmath39 : @xmath40 the local concentration profile of the species @xmath38 is denoted as @xmath41 .
the solvent is taken as a dielectric continuum of dielectric constant @xmath42 .
for simplicity , rod and solvent are assumed to have the same dielectric constant , in order to avoid complications with dielectric boundaries .
it has long been recognized in physics that particles and fields are equivalent in the sense that both are defined through their interaction potentials .
this simple fact has been applied in the past to derive , in a straightforward manner , inhomogeneous integral equation theories from the ornstein - zernike ( oz ) equation for homogeneous fluids @xcite .
the homogeneous three - component oz equation is @xmath43 where @xmath44 and @xmath45 are the total and direct correlation functions , between particle @xmath46 located at @xmath47 and particle @xmath48 at @xmath49 of species @xmath23 and @xmath50 , respectively , @xmath51 . in consequence , one can consider a particle of a fluid as a source of an external field or an external field as a particle of the fluid . applying this simple idea to eqn .
( [ eq : ozh ] ) , one can think of one of the species , say @xmath52 , as made of infinite cylindrical rods and the remaining two species as ions . in @xmath52
s infinite dilution limit the cylinders are uncorrelated .
hence , letting @xmath53 , the ornstein - zernike equation for an ionic solution next to a charged cylinder reads @xmath54 in the past , several approximations ( or `` closures '' ) for the direct correlation function have been suggested .
two of them are : @xmath55 eqn .
( [ eq : chnc ] ) and eqn .
( [ eq : cmsa ] ) are known as the hypernetted chain ( hnc ) equation and the mean spherical approximation ( msa ) , respectively ; @xmath56 is the direct interaction potential between species @xmath23 and @xmath50 , and @xmath57 .
if the hnc closure is used for the direct correlation function between the rod particle and the @xmath50 species , equation ( [ eq : oz ] ) becomes @xmath58 in this scheme , the two - particles correlation functions @xmath59 and @xmath60 correspond to the one - particle total and direct inhomogeneous correlation functions @xmath61 and @xmath62 for species @xmath50 of a fluid under the influence of an external field produced by a rod particle .
the local concentration for the @xmath50 species is given by @xmath63 , where @xmath64 is called the reduced concentration profile .
therefore , the charge concentration profile in eqn .
( [ eq : lambda ] ) is given by @xmath65 for this model the direct interaction potential between the rod and the @xmath50 species of the fluid , @xmath66 , can be separated into two parts : the hard sphere - hard rod term @xmath67 and the electrostatic interaction potential @xmath68 .
the first takes into account the fact that ions can not penetrate or deform the cylinder @xmath69 0 & r > r_0+a/2 \end{array } \right .
\label{eq : hard_term}\ ] ] the second can be found from gauss law and is given by @xmath70 in the msa closure the homogeneous direct correlation function for a rpm electrolyte has an analytical expression .
this function can be written as @xmath71 in the first term appears the direct electric interaction potential between the species of the fluid , which is given by eqn .
( [ eq : electric ] ) , the second term is an electrical short range function and the third is the direct correlation function for a hard sphere fluid .
when the msa closure is used in the integral of eqn .
( [ eq : hnc ] ) , one obtains the hnc / msa equation for an ionic fluid next to a charged rod . taking advantage of the cylindrical geometry and the fact that the direct correlation function between the ions depends only on their relative distance @xmath72 where @xmath6 is the mean electrostatic potential @xmath73 the @xmath74 terms are integrals of the short range terms of the direct correlation function @xmath75 with @xmath76 and @xmath77 where @xmath78 . in the limit @xmath79 of point ions , @xmath80 and @xmath81 .
thus @xmath82 @xcite and eqn .
( [ eq : hnc2 ] ) becomes @xmath83 where @xmath6 is given by eqn .
( [ eq : meanelectric ] ) , which is the solution of the well known pb differential equation for point ions around a charged cylindrical electrode , i.e. , @xmath84 therefore , the integral equation version of the poisson - boltzmann differential equation is @xcite @xmath85 which is also the hnc / msa equation in the point ions limit .
the differences between these two theories are due to the ionic size correlations which are partially taken into account in hnc / msa theory but which are ignored in the pb equation . in the past
, it has been shown that the size effects become important for strong field interaction , poly - valent ions and high salt concentration @xcite .
the hnc / msa and pb integral equations are numerically solved with efficient finite element methods @xcite .
the solution of eqn .
( [ eq : hnc2 ] ) takes one minute in a r12000 processor of an sgi machine . in our theoretical calculation
we use the same input parameters as in our md calculations .
the salt concentration is obtained through @xmath86 where @xmath87 is the number of salt molecules used in the simulation .
nevertheless , we have to point out the differences in the short range interaction used in both models . while in hnc / msa we use hard particles , the md for technical reasons uses the potential from eq .
[ eq : shifted_lj_potential ] , which has some surface softness .
this results in an effective excluded volume , whose actual value depends on the interaction strength . at an interaction energy of @xmath26
the particle diameter is @xmath10 .
however , if the particles attract each other strongly , they can come closer to each other .
.system parameters for the dna simulations . [ cols="<,^,^,^",options="header " , ] the first maximum seen in the coion reduced concentration profile implies a coion concentration above its bulk value and indicates that the local electric field is attractive to coions .
it is also observed that hnc / msa overestimates the contact value of the distribution function with respect to molecular dynamics predictions .
the overestimation of hnc / msa can be associated with two facts : _ ( i ) _ the difference in excluded volume used in both models ; and _ ( ii ) _ hnc / msa theory does not take into account all the size and charge correlations . as a function of surface charge density @xmath88 , for a cylinder immersed in a @xmath89 m , 2:2 electrolyte .
the model parameters are given in tab .
[ tab : dna_param ] .
the dotted line is the prediction of pb theory , the dots are the md results and the solid line is the results from hnc / msa calculations.,width=264 ] . the distance @xmath13 is measured with respect to the cylinder surface in units of the ionic radius . in this plot a comparison between a low and a high excluded volume systems is presented .
the system where overcharging occurs corresponds to a monovalent electrolyte with @xmath90 m , @xmath91 ( @xmath92 c / m@xmath93 ) , and large ions ( @xmath94 ) .
the system where overcharging is not observed corresponds to a divalent electrolyte with @xmath95 m , @xmath96 ( @xmath97 c / m@xmath93 ) , and small ions ( @xmath98).,width=264 ] ( left ) and mean electrostatic potential @xmath6 ( right ) results , for the system from tab . [
tab : dna_param ] with manning parameter @xmath99 .
the distance @xmath13 is measured with respect to the cylinder surface in units of the ionic radius .
the line styles have the same meaning as in fig .
[ fig : dna_pr_mep_lambda ] .
the curves correspond to different salt concentration , @xmath100 m. in the charge distribution function the salt content increases from bottom to top , in the mean electrostatic potential it increases from top to bottom.,title="fig:",width=264 ] ( left ) and mean electrostatic potential @xmath6 ( right ) results , for the system from tab . [
tab : dna_param ] with manning parameter @xmath99 .
the distance @xmath13 is measured with respect to the cylinder surface in units of the ionic radius .
the line styles have the same meaning as in fig .
[ fig : dna_pr_mep_lambda ] .
the curves correspond to different salt concentration , @xmath100 m. in the charge distribution function the salt content increases from bottom to top , in the mean electrostatic potential it increases from top to bottom.,title="fig:",width=264 ] another important phenomenon is indicated by the behavior of the mean electrostatic potential in fig .
[ fig : dna_pr_mep_lambda]b .
it concerns the value of this potential at the distance of closest approach between ions and the rod , @xmath101 which can be identified with the zeta - potential of electrophoresis and which is of importance in the computation of electrophoretic mobilities @xcite . in the pb theory for a charged rod immersed into an ionic solution , and its linearized version , @xmath1 depends monotonically on @xmath102 @xcite
. however , fig .
[ fig : dna_pr_mep_lambda ] shows that , in the presence of overcharging , @xmath6 can become negative .
this calls the monotonic relation between @xmath1 and @xmath102 into question .
table [ tab : dna_zeta ] summarizes various predictions for the @xmath1-potential as a function of manning parameter together with results from the molecular dynamics simulations .
a graphical illustration is given in fig .
[ fig : zeta ] .
indeed , @xmath1 is found to first increase with @xmath102 , but from a certain value on it decreases and would finally even become negative . note that this would reverse the drift direction in electrophoresis measurements , as first predicted by gonzales - tovar _
et al._@xcite and recently demonstrated , for spherical macroions , by lozada - cassou _
et al._@xcite .
while nonlinear and linearized pb theory coincide with the data and with each other for small manning parameter , they completely fail to predict the back - bending , which already sets in at comparatively small values of @xmath102 .
the hnc / msa theory captures this effect , but it underestimates the value of the potential .
however , comparisons for the @xmath1-potential , as a function of @xmath88 , between hnc / msa and mc simulations for planar and spherical hard , charged macroions , immersed into a rpm electrolyte , have been made in the past @xcite .
for these two geometries , for 1:1 electrolytes and for 2:2 , at around 0.5 m , as in our fig .
[ fig : zeta ] , the hnc / msa - mc agreement is excellent .
the disagreement seen in fig .
[ fig : zeta ] can be related to the different short - range potentials used in our md and hnc / msa calculations , which gives for the md not a sharply defined ion - cylinder contact .
this is important , since the @xmath1-potential is evaluated exactly there .
for planar and spherical macroions , pb @xmath1 vs. charge curves are also above the hnc / msa and mc curves , which is consistent with our present observations . for a given rod charge and diameter , for any electrolyte solution parameters ,
a higher @xmath1-potential always implies a lower counterions adsorption @xcite . a soft - repulsive short - range potential , as in our md calculations , allows the counterions to get closer to the rod .
this produces a higher @xmath1-potential in the md simulation when compared to the hnc / msa results . in the case of the pb theory
, the neglect of the ionic - size correlations allows a higher density of counterions next to the rod , which in turn can never lead to overcharging . from eqn .
( [ eq : overcharging_radius ] ) at which overcharging sets in , for a dna - sized and -charged cylinder ( @xmath103 and @xmath104 c / m@xmath93 ) , as a function of salt concentration @xmath105 .
the distance @xmath106 is measured with respect to the cylinder surface in units of the ionic radius .
the dots are the results of molecular dynamics simulations while the solid line is the prediction of hnc / msa.,width=264 ] in all cases , the hnc / msa qualitative agreement with md is very good .
the success of hnc / msa theory indicates that local ion - size correlations are responsible for both phenomena . as it is concluded by lozada - cassou and jimnez - ngeles in ref .
@xcite , overcharging increases with an increase of the system excluded volume : i.e. , higher ionic size and/or concentration .
if size correlations are not considered ( as in pb theory ) or are negligible , overcharging does not occur , even for divalent ions . on the other hand
, overcharging can occur in a high excluded volume system even for monovalent ions . to illustrate this fact we have simulated two systems , for which the distance of closest approach to the charged rod is in both cases @xmath107 , hence dna - like .
the first rod system has a charge parameter of @xmath91 ( @xmath92 c / m@xmath93 ) and is immersed in a 1.0 molar 1:1 electrolyte solution with ions of diameter @xmath94 .
the second rod system has an even higher charge parameter of @xmath96 ( @xmath97 c / m@xmath93 ) and is immersed in a 0.5 molar 2:2 electrolyte , but , this time , the diameter is only @xmath98 .
the results are shown in fig .
[ fig : energy - entropy ] , where the charge distribution function is plotted .
contrary to the general believe , the system with the large monovalent ions shows overcharging , both in the hnc / msa and the md results , and the location of the peaks coincide , whereas we see small differences in the height .
further , strikingly , the system with small divalent ions shows no sign of overcharging , although the charge parameter for this system is even higher .
the shoulder in the distribution of the md simulation indicates that the second ionic layer is almost neutral .
a possible explanation is the following : the small ions have an energy of attraction of almost @xmath108 , hence the salt is actually present in the form of little overall neutral salt clusters .
this is supported by visual inspection of the md configurations .
these mostly neutral clusters are polarizable and hence tend to accumulate in the large field gradient surrounding the rod .
the free counterions left can come only from the rods charge and can in no way lead to an overcharging . a key prerequisite for this scenario
is that the interaction energy between ions exceeds the interaction between ions and the rod . also consistent is , that the hnc / msa calculation as well shows the vanishing of the overcharging .
finally , one might ask the question of how the effect of overcharging depends on salt concentration . obviously there can not be any overcharging in the absence of salt , since then even a complete `` condensation '' of all ions
would merely neutralize the rod .
however , it is not clear from the beginning whether an addition of just _ some _ salt will immediately lead to overcharging .
figure [ fig : dna_pr_mep_rho ] shows hnc / msa and md distribution functions @xmath4 and mean electrostatic potentials @xmath6 for a dna - sized and charged rod , immersed into a 2:2 electrolyte , for varying salt concentrations .
the phenomenon of overcharging is indeed observed , but only at sufficiently high salt concentration .
these results are consistent with the fact that higher electrolyte concentration increases the excluded volume , thus increases overcharging @xcite .
therefore , as proposed in ref .
@xcite , higher excluded volume implies higher overcharging .
this may be also illustrated by defining @xmath106 to be the radius at which overcharging sets in , i.e. , @xmath109 fig .
[ fig : dna_r1_rho ] illustrates the measured @xmath106 together with a hypernetted - chain prediction as a function of salt concentration @xmath105 .
clearly , @xmath106 must increase with decreasing @xmath105 , since overcharging is reduced and consequently the size of the charge - compensating layer must increase .
in fact , hypernetted - chain theory predicts that @xmath110 diverges at some _ finite _ density @xmath111 corresponding to a salt debye length of @xmath112 , or roughly 200 salt molecules within the simulation box . for the lower concentrations , the simulated @xmath106 values lie below the hypernetted - chain prediction , but this is not a feature generally to be expected . in fig . [
fig : dna_pr_mep_rho ] it can be seen that the stronger overcharging in hypernetted - chain theory should normally lead to a value of @xmath106 smaller than in the simulation .
but with decreasing density the finite radius of the simulation cell becomes relevant .
indeed , zero - salt distribution functions reach the value @xmath48 at the cell boundary , and not at infinity .
a reduction of the amount of added salt at fixed cell radius must necessarily lead to values of @xmath106 smaller than the hypernetted chain prediction for the bulk , since within a finite cell @xmath106 can not diverge .
on the other hand , for technical reasons , the md cell radius can not be increased indefinitely . in any case , our md results have not been able to unambiguously detect overcharging at densities equal to or lower than @xmath113 , which is in excellent agreement with the hnc / msa prediction for the minimum amount of salt needed to induce overcharging . from figs .
( [ fig : dna_pr_mep_rho ] ) and ( [ fig : dna_r1_rho ] ) it is clear that excluded volume effects determine the occurrence of overcharging , as proposed by some of us @xcite .
theoretical and numerical studies of stiff linear polyelectrolytes , immersed into a 2:2 rpm electrolyte have been presented .
the qualitative agreement between the md and hnc / msa results is excellent . for the particular case of dna parameters ,
there is also a very good quantitative agreement .
we argued that the better agreement in the planar and spherical case reported in refs .
@xcite is most probably due to an implementation of the short range interactions that is identical to the theoretical model and not merely very close . at low salt concentrations
the difference between a finite cell used in simulations and the @xmath114 system used in the hnc / msa calculations becomes important .
also , the macroion and ions electrical fields are much less screened and , hence , the charge correlations become more important @xcite .
both these points are responsible for the disagreement between theory and simulations at low salt .
while the former can in principle be resolved by increasing the size of the simulation cell , the latter can be addressed by the three point extension hnc / msa theory @xcite , which gives a better account of these correlations @xcite . in the past
the hnc / msa theory for an electrolyte next to a charged rod has been derived @xcite and applied @xcite to simple dna models . in this paper
we have shown this theory to be qualitatively correct .
in particular , for dna parameters the quantitative agreement is excellent . to the best of our knowledge , in gonzales - tovar _
et al._@xcite overcharging has first been predicted theoretically .
it was point out that this should entail an electrophoresis mobility reversal . in the pb theory the @xmath1-potential as
a function of the cylinder charge density is a monotonic function , while in the hnc / msa theory it is not . in this paper
we have shown the hnc / msa prediction to be in agreement with md calculations .
this non - monotonic behavior of the @xmath1-potential has important implications in electrophoresis calculations .
the standard electrophoresis theory @xcite is based on the pb prediction for @xmath1 and , hence , important differences should be found if the hnc / msa theory is applied to the electrophoresis problem .
recently , lozada - cassou _ et al._@xcite extended the electrophoresis theory to include ionic size effects , through the hnc / msa theory .
it was applied to spherical macroions . a mobility reversal and a very non - linear behavior of the mobility , as a function of the @xmath1-potential , were predicted .
these predictions were found to be in agreement with experimental results .
contrary to the standard electrophoresis theory , the mobility was shown to be non - universal .
this is due , precisely , to the non - monotonic behavior of the @xmath1-potential .
the hnc / msa and mc simulations for spherical and planar macroions do not show a maximum in @xmath115 as pronounced as that reported here for the cylindrical geometry . therefore , we believe that our hnc / msa and md results are particularly relevant . in view of the qualitative agreement between hnc / msa and md and the quantitative and qualitative disagreement of these with poisson - boltzmann theory , we conclude that for concentrated and divalent electrolytes , influenced by charged polyelectrolytes , the size correlations must be taken properly into account to describe the cylindrical double layer .
our md results clearly indicate that overcharging strongly depends on the system excluded volume , as proposed by some of us @xcite .
while maximum overcharging does not seem to have a limit , the minimum conditions to have overcharging depend on many different ways in which ionic charge , size and concentration , and surface charge density participate in a system , as we showed in figs .
( [ fig : dna_pr_mep_lambda ] ) , ( [ fig : energy - entropy ] ) and ( [ fig : dna_pr_mep_rho ] )
ch and mlc wish to thank w. gelbart and the institute for theoretical physics , university of california at santa barbara , where this project was started , for their hospitality .
mlc which to thank c. holm and k. kremer for their hospitality at mainz .
fja and mlc gratefully acknowledge the financial support of industrias negromex .
ch and md acknowledge a large computer time grant hkf06 from nic jlich and financial support by the german science foundation . | a study of a model rod - like polyelectrolyte molecule immersed into a monovalent or divalent electrolyte is presented .
results for the local concentration profile , mean electrostatic potential , charge distribution function and @xmath0potential are obtained from hypernetted - chain / mean spherical approximation ( hnc / msa ) theory and compared with molecular dynamics ( md ) simulations . as a particular case ,
the parameters of the polyelectrolyte molecule are mapped to those of a dna molecule . both , hnc / msa and md , predict the occurrence of overcharging , which is not present in the poisson - boltzmann theory .
futher an excellent qualitative , and in some cases quantitative , agreement between hnc / msa and md is found .
oscillations observed in the mean electrostatic potential , local concentration profiles , as well as the curvature of the @xmath1-potential are discussed in terms of the observed overcharging effect .
particularly interesting results are a very non - monotonic behavior of the @xmath1-potential , as a function of the rod charge density , and the overcharging by _ monovalent _ counterions . |
one way to describe the content of these lecture notes is to say that they give a proof of the following statement ( up to certain technical details that can be looked up in suitable articles ) .
[ theorem1.1 ] let @xmath1 be independent standard brownian motions .
define @xmath2 then , for any @xmath3 , @xmath4 with certain explicit @xmath5-dependent constants @xmath6 , where @xmath7 is the distribution function of the gue tracy - widom distribution .
the quantity @xmath8 was introduced by oconnell - yor @xcite , and it can be viewed as the partition function of a semi - discrete brownian polymer ( also sometimes referred to as the `` oconnell - yor polymer '' ) .
the limit relation above shows that this polymer model belongs to the celebrated kardar - parisi - zhang ( kpz ) universality class , see corwin @xcite for details on the kpz class and 1.6 of @xcite for more explanations , consequences , and references concerning the polymer interpretation .
the exact value of @xmath9 was conjectured by oconnell - yor @xcite and proven by moriarty - oconnell @xcite , and the above limit theorem was proven by borodin - corwin @xcite for a restricted range of @xmath5 and by borodin - corwin - ferrari @xcite for all @xmath3 .
a nice physics - oriented explanation of @xmath10 was given by spohn @xcite .
although the most direct proof of this theorem would likely be quite a bit shorter than these notes , brevity was not our goal . despite the probabilistic appearance of the statement ,
any of the known approaches to the proof would involve a substantial algebraic component , and the appearance of algebra at first seems at least slightly surprising .
the goal of these lectures notes is to suggest the most logically straightforward path ( in authors opinion ) that leads to the desired result , minimizing as much as possible the number of _ ad hoc _ steps one takes along the way .
( for an interested reader we remark that a shorter proof of theorem [ theorem1.1 ] can be obtained via combining corollary 4.2 of @xcite , theorem 2 of @xcite , and asymptotic analysis of @xcite . ) as we travel along our path ( that naturally starts on the algebraic side in representation theory of unitary groups ) , we encounter other probabilistic models that are amenable to similar tools of analysis .
the approach that we develop has a number of other applications as well .
it was so far used for ( we refer the reader to the indicated references for further explanations ) * asymptotics of the kpz equation with a certain class of initial conditions @xcite ; * asymptotics of log - gamma fully discrete random directed polymers @xcite ; * asymptotics of @xmath11-tasep and asep @xcite , @xcite ; * analysis of new integrable ( 1 + 1)d interacting particle systems discrete time @xmath11-taseps of @xcite , @xmath11-pushasep @xcite , @xcite , and @xmath11-hahn tasep @xcite ; * establishing a law of large numbers for infinite random matrices over a finite field @xcite ( conjectured by vershik and kerov , see @xcite ) ; * gaussian free field asymptotics of the general beta jacobi corners process @xcite ; * developing spectral theory for the @xmath11-boson particle system @xcite and other integrable particle systems @xcite ; * asymptotics of probabilistic models originating from representation theory of the infinite - dimensional unitary group @xmath12 @xcite , @xcite , @xcite , @xcite . the emerging domain of studying such models that enjoy the benefits of a rich algebraic structure behind , is sometimes called _ integrable probability _ , and we refer the reader to the introduction of borodin - gorin @xcite for a brief discussion of the domain and of its name ( the integrable nature of the semi - discrete polymer of theorem [ theorem1.1 ] was first established by oconnell @xcite ) .
to a certain extent , the present text may be considered as a continuation of @xcite , but it can be also read independently .
in contrast with @xcite , in our exposition below we do not shy away from the representation theoretic background and intuition that were essential in developing the subject .
we also focus on proving a single theorem , rather than describing the variety of other related problems listed above , in order to discuss in depth the analytic difficulties arising in converting an algebraic formalism into analytic statements .
these difficulties are related to the phenomenon of _ intermittency _ and popular yet highly non - rigorous and sometimes dangerous _
replica trick _ favoured by physicists , and one of our goals is to show how raising the amount of `` algebraization '' of the problem can be used to overcome them .
the notes are organized as follows . in section
[ sec : lozenge_tilings_and_representation_theory ] , we explain how lozenge tilings of a class of polygons on the triangular lattice can be interpreted via representation theory of the unitary groups , and how this leads to contour integral formulas for averages of various observables . in section [ sec : asymptotics ] , we show , in a specific example , how the steepest descent analysis of the obtained contour integrals yields meaningful probabilistic information about lozenge tilings .
section [ sec : markov_dynamics ] describes an approach to constructing local markov dynamics on lozenge tilings , and how ( 1 + 1)-dimensional interacting particle systems ( like usual and long range totally asymmetric simple exclusion processes ( taseps ) ) arise as marginals of such dynamics .
the approach we describe is relatively recent ; it was developed in borodin - petrov @xcite ( an extension of the method will appear in @xcite ) . section [ sec:_q_t_generalization ] deals with a two - parameter ( macdonald , @xmath0- ) generalization of the previous material . in section [ sec : asymptotics_of_q_deformed_growth_models ] we show how simple - minded asymptotics of the contour integrals in the @xmath11-deformation of lozenge tilings leads to semi - discrete brownian polymers .
the contour integrals describe the @xmath11-moments of the @xmath11-tasep , an integrable deformation of the usual tasep .
section [ sec : moments_for_q_whittaker_processes ] explains difficulties which arise if one straightforwardly tries to describe the distribution of the polymer partition function using its moments .
the latter come out naturally as limits of the @xmath11-tasep s @xmath11-moments . in final section [ sec : laplace_transforms ]
we demonstrate how those difficulties can be overcome through considering the laplace transform of the polymer partition function and its @xmath11-analog for the @xmath11-tasep particle locations . these are notes for lectures delivered at the 2013 cornell probability summer school , and we would like to thank the organizers for the invitation and warm hospitality .
we are also very grateful to ivan corwin and vadim gorin for numerous valuable comments , and we thank the anonymous referee for several helpful remarks .
ab was partially supported by the nsf grant dms-1056390 .
lp was partially supported by the rfbr - cnrs grants 10 - 01 - 93114 and 11 - 01 - 93105 .
we begin with a discussion of a well - known probabilistic model of randomly tiling a hexagon drawn on the triangular lattice , and explain its relation to representation theory of unitary groups .
this relation produces rather natural tools for analysis of uniformly random lozenge tilings of the hexagon .
[ scale=.7 , ultra thick ] at ( -.5-.3,-.3 ) @xmath13 ; at ( -.5-.3,3*+.3 ) @xmath14 ; at ( 6-.2,3*+.3 ) @xmath13 ; at ( 6-.2,-.3 ) @xmath14 ; at ( 2.5,-.5 ) @xmath15 ; at ( 2.5,4*+.5 ) @xmath15 ; at ( 2.5,4*+.8 ) ; ( 0,0 ) ++ ( 5,0 ) ++ ( 1,2 * ) ++ ( -1,2 * ) ++ ( -5,0 ) ++ ( -1,-2 * ) ++ ( 1,-2 * ) cycle ; in ( + 3 , ) , ( + 3.5,+ ) , ( + .5,+ ) , ( + 2,+2 * ) ( 0,0 ) ( .5 , ) ( 1,0 ) ( .5,- ) cycle ; in ( , ) , ( + 1,),(+2 , ) , ( -1/2,+ ) , ( + 1,+2 * ) , ( + 4,+2*),(+5,+2 * ) , ( + 2.5,+3*),(+3.5,+3*),(+4.5,+3 * ) ( 0,0 ) ( -.5 , ) ( .5 , ) ( 1,0 ) cycle ; in ( + 3,),(+4 , ) , ( + .5,+),(+1.5,+),(+3.5,+),(+4.5,+ ) , ( -1,+2*),(+2,+2 * ) , ( -.5,+3*),(+.5,+3 * ) ( 0,0 )
( 1.5 , ) ( 1,0 ) cycle ; consider the problem of tiling a hexagon with sides of length @xmath16 drawn on the triangular lattice by _ lozenges _ that are defined as pairs of triangles glued together ( see fig .
[ fig : tiling_abc]a ) . here
@xmath13 , @xmath15 , and @xmath14 are any positive integers , and we assume that the side of an elementary triangle has length @xmath17 .
there are three different types of lozenges : ll ( 1)&(2 ) + [ scale=.7 , ultra thick ] at ( 2.5,4*+.8 ) ; ( 0,0 ) ++ ( 5,0 ) ++ ( 1,2 * ) ++ ( -1,2 * ) ++ ( -5,0 ) ++ ( -1,-2 * ) ++ ( 1,-2 * ) cycle ; in ( + 3 , ) , ( + 3.5,+ ) , ( + .5,+ ) , ( + 2,+2 * ) ( 0,0 )
( .5 , ) ( 1,0 ) ( .5,- ) cycle ; ( .5,-/3 ) ( .5,/3 ) ; ( .5,-/3 ) circle ( .06 ) ; ( .5,/3 ) circle ( .06 ) ; in ( , ) , ( + 1,),(+2 , ) , ( -1/2,+ ) , ( + 1,+2 * ) , ( + 4,+2*),(+5,+2 * ) ( 0,0 ) ( -.5 , )
( 1,0 ) cycle ; ( .5,/3 ) ++ ( -.5,/3 ) ; ( 0,2*/3 ) circle ( .06 ) ; ( .5,/3 ) circle ( .06 ) ; in ( , ) , ( + 1,),(+2 , ) , ( -1/2,+ ) , ( + 1,+2 * ) , ( + 4,+2*),(+5,+2 * ) , ( + 2.5,+3*),(+3.5,+3*),(+4.5,+3 * ) ( 0,0 ) ( -.5 , )
( .5 , ) ( 1,0 ) cycle ; ( .5,/3 ) ++ ( -.5,/3 ) ; ( 0,2*/3 ) circle ( .06 ) ; ( .5,/3 ) circle ( .06 ) ; in ( + .5,+),(+1.5,+),(+3.5,+),(+4.5,+ ) , ( -1,+2*),(+2,+2 * ) ( 0,0 ) ( .5 , ) ( 1.5 , ) ( 1,0 ) cycle ; ( .5,/3 ) ++ ( .5,/3 ) ; ( 1,2*/3 ) circle ( .06 ) ; ( .5,/3 ) circle ( .06 ) ; in ( -.5,+3*),(+.5,+3 * ) ( 0,0 ) ( .5 , ) ( 1.5 , ) ( 1,0 ) cycle ; ( .5,/3 ) ++ ( .5,/3 ) ; ( 1,2*/3 ) circle ( .06 ) ; ( .5,/3 ) circle ( .06 ) ; in ( + 3,),(+4 , ) ( 0,0 ) ( .5 , ) ( 1.5 , ) ( 1,0 ) cycle ; ( .5,/3 ) ++ ( .5,/3 ) ; ( 1,2*/3 ) circle ( .06 ) ; ( .5,/3 ) circle ( .06 ) ; in ( 0,2*/3),(1,2*/3),(2,2*/3),(4,2*/3),(5,2*/3 ) , ( -.5,2*/3+),(2-.5,2*/3+),(3-.5,2*/3+),(5-.5,2*/3+),(6-.5,2*/3 + ) , ( 0,2*/3 + 2*),(1,2*/3 + 2*),(3,2*/3 + 2*),(4,2*/3 + 2*),(5,2*/3 + 2 * ) ( 0,0 ) ( 0,2*/3 ) ; in ( .5,/3),(1.5,/3),(2.5,/3 ) , ( 0,/3+),(3,/3 + ) , ( 0.5,/3 + 2*),(1.5,/3 + 2*),(3.5,/3 + 2*),(4.5,/3 + 2 * ) , ( 2,/3 + 3*),(3,/3 + 3*),(4,/3 + 3 * ) ( 0,0 ) ( .5,/3 ) ; in ( 3.5,/3),(4.5,/3 ) , ( 1,/3+),(2,/3+),(3,/3+),(4,/3+),(5,/3 + ) , ( .5,/3 + 2*),(2.5,/3 + 2*),(3.5,/3 + 2 * ) , ( 1,/3 + 3*),(2,/3 + 3 * ) ( 0,0 ) ( -.5,/3 ) ; & [ scale=.7 , ultra thick ] at ( 2.5,4*+.8 ) ; ( 0,0 ) ++ ( 5,0 ) ++ ( 1,2 * ) ++ ( -1,2 * ) ++ ( -5,0 ) ++ ( -1,-2 * ) ++ ( 1,-2 * ) cycle ; in ( + 3 , ) , ( + 3.5,+ ) , ( + .5,+ ) , ( + 2,+2 * ) ( 0,0 ) ( .5 , ) ( 1,0 ) ( .5,- ) cycle ; in ( , ) , ( + 1,),(+2 , ) , ( -1/2,+ ) , ( + 1,+2 * ) , ( + 4,+2*),(+5,+2 * ) , ( + 2.5,+3*),(+3.5,+3*),(+4.5,+3 * ) ( 0.5,0 ) ++ ( -.5 , ) ; ( 0,0 ) ( -.5 , )
( .5 , ) ( 1,0 ) cycle ; in ( + 3,),(+4 , ) , ( + .5,+),(+1.5,+),(+3.5,+),(+4.5,+ ) , ( -1,+2*),(+2,+2 * ) , ( -.5,+3*),(+.5,+3 * ) ( 0.5,0 ) ++ ( .5 , ) ; ( 0,0 ) ( .5 , )
( 1.5 , )
( 1,0 ) cycle ; + ll ( 3)&(4 ) ' '' '' + [ scale=.7 , ultra thick ] at ( 2.5,4*+.8 ) ; ( 0,0 ) ++ ( 5,0 ) ++ ( 1,2 * ) ++ ( -1,2 * ) ++ ( -5,0 ) ++ ( -1,-2 * ) ++ ( 1,-2 * ) cycle ; in ( + 3 , ) , ( + 3.5,+ ) , ( + .5,+ ) , ( + 2,+2 * ) ( 0,0 ) ( .5 , ) ( 1,0 ) ( .5,- ) cycle ; ( 0,0 ) ( .5 , ) ( 1,0 ) ( .5,- ) cycle ; in ( , ) , ( + 1,),(+2 , ) , ( -1/2,+ ) , ( + 1,+2 * ) , ( + 4,+2*),(+5,+2 * ) , ( + 2.5,+3*),(+3.5,+3*),(+4.5,+3 * ) ( 0,0 ) ( -.5 , ) ( .5 , ) ( 1,0 ) cycle ; in ( + 3,),(+4 , ) , ( + .5,+),(+1.5,+),(+3.5,+),(+4.5,+ ) , ( -1,+2*),(+2,+2 * ) , ( -.5,+3*),(+.5,+3 * ) ( 0,0 ) ( .5 , ) ( 1.5 , )
( 1,0 ) cycle ; ( 0,0 )
( 1.5 , )
( 1,0 ) cycle ; & [ scale=.7 , ultra thick ] at ( 2.5,4*+.8 ) ; ( 0,0 ) ++ ( 5,0 ) ++ ( 1,2 * ) ++ ( -1,2 * ) ++ ( -5,0 ) ++ ( -1,-2 * ) ++ ( 1,-2 * ) cycle ; in ( + 3 , ) , ( + 3.5,+ ) , ( + .5,+ ) , ( + 2,+2 * ) ( 0,0 ) ( .5 , ) ( 1,0 ) ( .5,- ) cycle ; ( .5,0 ) circle ( .17 ) ; in ( , ) , ( + 1,),(+2 , ) , ( -1/2,+ ) , ( + 1,+2 * ) , ( + 4,+2*),(+5,+2 * ) , ( + 2.5,+3*),(+3.5,+3*),(+4.5,+3 * ) ( 0,0 ) ( -.5 , )
( .5 , ) ( 1,0 ) cycle ; in ( + 3,),(+4 , ) , ( + .5,+),(+1.5,+),(+3.5,+),(+4.5,+ ) , ( -1,+2*),(+2,+2 * ) , ( -.5,+3*),(+.5,+3 * ) ( 0,0 )
( 1.5 , ) ( 1,0 ) cycle ; @xmath18 \draw [ thick ] ( 0,0 ) -- ( .5,0.866 ) -- ( 1,0 ) -- ( .5,-0.866 ) -- ( 0,0 ) -- ( .5,0.866 ) ; \draw [ thick , densely dotted ] ( 0,0 ) -- ( 1,0 ) ; \end{tikzpicture}}$ ] , @xmath18 \draw [ thick ] ( 0,0 ) -- ( -.5,0.866 ) -- ( .5,0.866 ) -- ( 1,0 ) -- ( 0,0 ) -- ( -.5,0.866 ) ; \draw [ thick , densely dotted ] ( 0,0 ) -- ( .5,0.866 ) ; \end{tikzpicture}}$ ] , and @xmath18 \draw [ thick ] ( 0,0 ) -- ( .5,0.866 ) -- ( 1.5,0.866 ) -- ( 1,0 ) -- ( 0,0 ) -- ( .5,0.866 ) ; \draw [ thick , densely dotted ] ( 1,0 ) -- ( .5,0.866 ) ; \end{tikzpicture}}$ ] .
such tilings ( that are in a bijective correspondence with _ boxed plane partitions _ ) can be interpreted in a variety of ways ( see fig . [
fig : tiling_abc1 ] ) : 1 . as _ dimers _ ( or _ perfect matchings _ ) on the dual hexagonal lattice .
2 . as sets of nonintersecting bernoulli paths following lozenges of two types ( + ( 0.5,0 ) ++ ( -.5 , ) ; ( 0,0 ) ( -.5 , )
( 1,0 ) cycle ; + and + ( 0.5,0 ) ++ ( .5 , ) ; ( 0,0 )
( 1.5 , ) ( 1,0 ) cycle ; + with prescribed beginnings and ends .
3 . as stepped surfaces made of @xmath19 cubes .
4 . as interlacing configurations of lattice points
centers of lozenges of one of the types , say ,
@xmath18 \draw [ thick ] ( 0,0 ) -- ( .5,0.866 ) -- ( 1,0 ) -- ( .5,-0.866 ) -- ( 0,0 ) -- ( .5,0.866 ) ; \end{tikzpicture}}$ ] , as on fig .
[ fig : tiling_abc1 ] . such configurations must have a prescribed number of points in each _ horizontal _ section that may depend on the section .
our first goal is to match this combinatorial object with a basic representation theoretic one .
denote by @xmath20 the ( compact lie ) group of all the unitary matrices , where @xmath21 is the conjugate transpose .
] of size @xmath22 . a ( finite - dimensional ) _ representation _ of @xmath20 is a continuous map @xmath23 ( for some @xmath24 ) which respects the group structure : @xmath25 , @xmath26 .
a representation is called _ irreducible _ if it has no _ invariant _ subspaces @xmath27 ( @xmath28 or @xmath29 ) , i.e. , such that @xmath30 . the classification of irreducible representations of @xmath20 ( equivalently , of @xmath31 by analytic continuation
`` unitary trick '' of h. weyl ) is one of high points of the classical representation theory .
it is due to hermann weyl in mid-1920 s . in order to understand how it works ,
let us restrict @xmath32 to the abelian subgroup of diagonal unitary matrices @xmath33 any commuting family of ( diagonalizable , is unitary in a suitable basis ( e.g. , see @xcite ) , hence all our matrices are diagonalizable . ] ) matrices can be simultaneously diagonalized . in particular ,
this is true for @xmath34 .
hence , for @xmath35 , @xmath36 where each @xmath37 is a continuous homomorphism @xmath38 .
any such homomorphism has the form @xmath39 each @xmath22-tuple @xmath40 for @xmath41 , @xmath35 ,
is called a _ weight _ of the representation @xmath32 .
there is a total of @xmath42 weights ( which is the dimension of the representation ) .
irreducible representations of @xmath20 are in one - to - one correspondence with ordered @xmath22-tuples @xmath43 .
the correspondence is established by requiring that @xmath44 is the unique highest ( in lexicographic order ) weight of the corresponding representation .
then the generating function of all weights of this representation @xmath45 can be written as @xmath46_{i , j=1}^{n } } { \det\big[z_i^{n - j}\big]_{i , j=1}^{n}}. \label{schur } \end{aligned}\ ] ] note that the denominator in is the vandermonde determinant which evaluates to @xmath47_{i , j=1}^{n}=\prod_{1\le i < j\le n}(z_i - z_j).\end{aligned}\ ] ] the numerator in is necessarily divisible by the denominator because of its skew - symmetry with respect to @xmath48 , and thus the ratio is a finite linear combination of the monomials of the form @xmath49 , @xmath50 ( i.e. , an element of @xmath51^{{s}(n)}$ ] ) .
the polynomials @xmath52 are called _ schur polynomials _ , after issai schur , who used them in the representation theory of the symmetric group in his thesis around 1900 . however , one of the earliest appearances of them dates back to cauchy @xcite and jacobi @xcite , over 100 years before weyl s work .
the schur polynomials are denoted by @xmath53 .
schur polynomials are , generally speaking , symmetric homogeneous laurent polynomials in @xmath22 variables .
while the ratio of determinants formula is beautiful and concise ( it is a special case of _ weyl s character formula _ which works for any compact semi - simple lie group ) , is does not describe the set of weights explicitly . to do that , we need an elementary [ lemma : schur_branching ] for any @xmath43 , @xmath54 where the sum is taken over @xmath55 , the notation @xmath56 means the _ interlacing _ @xmath57 and @xmath58 , @xmath59 .
clear the denominators in and compare coefficients by each of the monomials @xmath49 .
applying this lemma @xmath22 times , we see that the weights are in one - to - one correspondence with interlacing triangular arrays of integers @xmath60 \def\h{0.8 } \def\x{2 } \node at ( -5,5 ) { $ { \lambda}_n$ } ; \node at ( -3,5 ) { $ { \lambda}_{n-1}$ } ; \node at ( 0-\x/2,5 ) { $ \ldots\ldots\ldots\ldots$ } ; \node at ( 3-\x,5 ) { $ { \lambda}_{2}$ } ; \node at ( 5-\x,5 ) { $ { \lambda}_{1}$ } ; \node at ( -4,5-\h ) { $ \mu_{n-1}$ } ; \node at ( -2,5-\h ) { $ \mu_{n-2}$ } ; \node at ( 0-\x/2+.1,5-\h ) { $ \ldots$ } ; \node at ( 2-\x,5-\h ) { $ \mu_{2}$ } ; \node at ( 4-\x,5-\h ) { $ \mu_{1}$ } ; \node at ( -3,5 - 2*\h ) { $ \nu_{n-2}$ } ; \node at ( 3-\x,5 - 2*\h ) { $ \nu_{1}$ } ; \foreach \lepoint in { ( 4.5-\x,5-\h/2),(2.5-\x,5-\h/2),(-3.5,5-\h/2 ) , ( -2.5,5 - 3*\h/2),(3.5-\x,5 - 3*\h/2),(2.5-\x,5 - 5*\h/2),(1.7-\x,5 - 7*\h/2 ) } { \node [ rotate=45 ] at \lepoint { $ \le$ } ; } ; \foreach \gepoint in { ( 3.5-\x,5-\h/2),(-4.5,5-\h/2),(-2.5,5-\h/2 ) , ( -3.5,5 - 3*\h/2),(-2.5,5 - 5*\h/2),(.5,5 - 3*\h/2),(-1.7,5 - 7*\h/2 ) } { \node [ rotate=135 ] at \gepoint { $ \ge$ } ; } ; \node at ( 0-\x/2,5 - 3*\h ) { $ \ldots\ldots\ldots\ldots$ } ; \node at ( 0-\x/2+.08,5 - 4*\h ) { $ \omega_1 $ } ; \end{tikzpicture } \end{array } \label{gt_scheme}\end{aligned}\ ] ] such arrays are called _ gelfand tsetlin schemes / patterns _ , and they will play a prominent role in what follows . observe that if we shift all leftmost entries of a gelfand tsetlin scheme of depth ( or height ) @xmath22 by @xmath61 , the second to left ones by @xmath17 , etc .
, then in the end we obtain a similar array where some of the inequalities become strict : @xmath60 \def\h{0.85 } \def\x{.7 } \def\y{.15 } \node at ( -5,5 ) { $ { \lambda}_n$ } ; \node at ( -3+\y,5 ) { $ { \lambda}_{n-1}+1 $ } ; \node at ( 0-\x/2,5 ) { $ \ldots\ldots\ldots$ } ; \node at ( 3-\x+0*\y,5 ) { $ { \lambda}_{2}+n-2 $ } ; \node at ( 5-\x+9*\y,5 ) { $ { \lambda}_{1}+n-1 $ } ; \node at ( -4,5-\h ) { $ \mu_{n-1}$ } ; \node at ( -2+\y,5-\h ) { $ \mu_{n-2}+1 $ } ; \node at ( 0-\x/2,5-\h ) { $ \ldots$ } ; \node at ( 2-\x+0*\y,5-\h ) { $ \mu_{2}+n-3 $ } ; \node at ( 4-\x+5*\y,5-\h ) { $ \mu_{1}+n-2 $ } ; \node at ( -3,5 - 2*\h ) { $ \nu_{n-2}$ } ; \node at ( 3-\x+2*\y,5 - 2*\h ) { $ \nu_{1}+n-3 $ } ; \foreach \lepoint in { ( 4.5-\x+5*\y,5-\h/2),(2.5-\x,5-\h/2),(-3.5,5-\h/2 ) , ( -2.5,5 - 3*\h/2),(3.5 + 3*\y-\x,5 - 3*\h/2),(2.5-\x,5 - 5*\h/2),(1.7-\x,5 - 7*\h/2 ) } { \node [ rotate=45 ] at \lepoint { $ < $ } ; } ; \foreach \gepoint in { ( 3.5 + 4*\y-\x,5-\h/2),(-4.5,5-\h/2),(-2.5,5-\h/2 ) , ( -3.5,5 - 3*\h/2),(1.5 + 2*\y,5 - 3*\h/2),(-2.5,5 - 5*\h/2),(-1.7,5 - 7*\h/2 ) } { \node [ rotate=135 ] at \gepoint { $ \ge$ } ; } ; \node at ( 0-\x/2,5 - 3*\h ) { $ \ldots\ldots\ldots\ldots\ldots\ldots$ } ; \node at ( 0-\x/2+.08,5 - 4*\h ) { $ \omega_1 $ } ; \end{tikzpicture } \end{array } \label{shifted_gt_scheme}\end{aligned}\ ] ] [ prop : u_a+c ] lozenge tilings of the hexagon with sides @xmath16 ( fig .
[ fig : tiling_abc ] ) are in one - to - one correspondence with weights of the irreducible representation of the unitary group @xmath62 with highest weight @xmath63 by picture , see fig . [
fig : tilings_weights ] .
[ scale=.8 , ultra thick , block/.style = rectangle , thick , draw = black , text width=13em , align = center , rounded corners , minimum height=2em ] ( -1,2*)(0,0 ) ( 7,0 ) node[below , xshift=3 ] @xmath64 ; ( 0,0 ) ( -3.3,6.6 * ) node[left ] @xmath65 ; in ( -1.5,+3 * ) , ( -2,+4*),(-1,+4 * ) , ( -2.5,+5*),(-1.5,+5*),(-.5,+5 * ) , ( + 4.5,+5*),(+5.5,+5 * ) , ( + 5,+4 * ) ( 0,0 ) ( .5 , ) ( 1,0 ) ( .5,- ) cycle ; ( .5,0 ) circle ( .17 ) ; ( -1.5,2 * ) ++(-1.5,3 * ) ; ( 5,3 * ) ++(1,2 * ) ; ( -.5,0 ) ++ ( -1,2 * ) ++(1.5,3 * ) ++(4,0 ) ++(1,-2 * ) ++(-1.5,-3*)++(-4,0)cycle ; ( 0,0 ) circle ( .1 ) node [ below , yshift=-2 ] 0 ; at ( 2,5*+.1 ) @xmath66 ; at ( -.7,3.5 * ) @xmath67 ; at ( 4.3,4 * ) @xmath68 ; ( 6,0.2 )
( 6,5 * -.2 ) node [ black , midway , xshift=50 ] @xmath69 ; if we coordinatize by taking the centers of the vertical lozenges in the coordinate system on the picture , then we read off the shifted gelfand tsetlin schemes .
the total number of weights of @xmath45 ( or , equivalently , the dimension of the representation ) was denoted by @xmath70 above , and it is given by [ prop : s_la_1111 ] for any @xmath43 ,
@xmath71 this is a special case of _ weyl s dimension formula _ ( which again works for any compact semi - simple lie group )
. follows from either directly ( by the lhpital s rule ) , or through the substitution @xmath72 and the limit @xmath73 .
consider the uniform probability measure @xmath74 on the space of all lozenge tilings of the hexagon with sides @xmath16 ( see fig .
[ fig : tiling_abc ] ) . the normalizing factor in the measure @xmath74 ( the so - called _ partition function _ ) is given in with @xmath44 as in .
for tilings of the hexagon with sides of length @xmath16 the partition function was first computed in a nicer product form @xmath75 by macmahon @xcite .
let us focus on what happens when we consider one horizontal slice of our uniformly random lozenge tiling .
as above , we coordinatize it by locations of centers of the vertical lozenges .
it is easiest to assume that the slice is close enough to one of the two horizontal boundaries , say , the lowest one ( see fig . [
fig : slice_h ] ) .
[ scale=.7 , ultra thick ] ( 0,0 ) ++ ( 5,0 ) ++ ( 1.5,3 * ) ++(-2,4*)++(-5,0)++(-1.5,-3 * ) ++(2,-4*)cycle ; at ( -2.1,3 * ) @xmath13 ; at ( 2.5,-.45 ) @xmath15 ; at ( 5.85,.6 * ) @xmath14 ; ( -2.5,2 * ) ++(10.5,0 ) node [ right ] @xmath76 ; in ( + 4,+ ) , ( + .5,+ ) , ( + 5,+ ) ( 0,0 )
( .5 , ) ( 1,0 )
( .5,- ) cycle ; ( .5,0 ) circle ( .15 ) ; at ( .3,2*+.3 ) @xmath77 ; at ( 4.8,2*+.3 ) @xmath78 ; at ( 3.8,2*+.3 ) @xmath79 ; at ( 2,2*+.3 ) @xmath80 ; [ prop : horizontal_slice_distribution ] for any @xmath81 , @xmath82 , the distribution of lozenges on the horizontal slice at height @xmath81 has the form @xmath83 where @xmath84 probability measures of the form with arbitrary positive weight function @xmath85 are known as _ orthogonal polynomial ensembles _ as they are closely related to the orthogonal polynomials with weight @xmath86 .
the measure itself is often referred to as the hahn orthogonal polynomial ensemble , as this particular weight @xmath86 corresponds to the classical hahn orthogonal polynomials .
see , e.g. , @xcite and references therein for details .
we can cut the enumeration problem into two that look like those on fig .
[ fig : fill_fill ] , and then multiply the results .
each of the two problems ( compute the number of tilings of the corresponding region with fixed top row ) is solved by the dimension formula .
[ scale=.65 , ultra thick ] ( -1,2 * ) (0,0)(5,0)++(1,2 * ) ++(-1,0 ) ++(-.25,-/2 ) ++(-.25,/2 ) ++(-.5,0 ) ++(-.25,-/2 ) ++(-.25,/2 ) ++(-3,0 ) ++(-.25,-/2 ) ++(-.25,/2 ) ++(-1,0)cycle ; at ( 2.3 , ) ; in ( + 4,+ ) , ( + .5,+ ) , ( + 5,+ ) ( 0,0 ) ( .5 , ) ( 1,0 ) ( .5,- ) cycle ; ( .5,0 ) circle ( .12 ) ; at ( .3,2*+.3 ) @xmath77 ; at ( 4.8,2*+.3 ) @xmath78 ; at ( 3.8,2*+.3 ) @xmath79 ; at ( 2,2*+.3 ) @xmath80 ; in ( + 1.5,+ ) , ( + .5,+ ) , ( + 5,+ ) , ( + 6.5,+),(+7,+),(+7.5,+),(+8,+ ) , ( + 6.75,+/2),(+7.25,+/2),(+7.75,+/2 ) , ( + 7,),(+7.5 , ) , ( + 7.25,-/2 ) , ( -1,+),(-1.5,+),(-1.25,+/2 ) ( 0,0 ) ( .5 , ) ( 1,0 ) ( .5,- ) cycle ; ( .5,0 ) circle ( .12 ) ; at ( .3,2*+.3 ) @xmath78 ; at ( 4.8,2*+.3 ) @xmath77 ; at ( 1.3,2*+.3 ) @xmath79 ; at ( 3,2*+.3 ) @xmath80 ; ( 0,2 * ) ++(.25,-/2 ) ++(.25,/2 ) ++(.5,0 ) ++(.25,-/2 ) ++(.25,/2 ) ++(3,0 ) ++(.25,-/2 ) ++(.25,/2 ) ++(1,0 ) ++(1,-2 * ) ++(-1.5,-3 * ) ++(-5,0 ) ++(-2,4 * ) ++(.5,)cycle ; at ( 3,-.5 * ) ; the computation of proposition [ prop : horizontal_slice_distribution ] already allows to see asymptotic transitions for a fixed @xmath81 ( for example , @xmath87 ) .
we can rewrite as @xmath88 .
\nonumber\end{aligned}\ ] ] one can consider the following limit regimes : 1 . if @xmath89 so that @xmath90 , the first term just contributes to a constant , while the second one converges to @xmath91 .
2 . in a similar way ,
if we keep @xmath13 finite and send @xmath92 in such a way that @xmath93 , @xmath94 , then we see that the relevant part of converges to @xmath95 3 . a slightly more complicated limit transition would be to take @xmath89 so that the triple ratio @xmath96 has a finite limit . then stirling s formula shows that after proper shifting and scaling of @xmath97 , which would not affect the factor @xmath98 , the nontrivial part of @xmath99 converges to a gaussian weight @xmath100 , @xmath101 . there is also a representation - theoretic way to view these results . restricting to a fixed horizontal slice
means that we only care about the restriction of our representation of @xmath102 ( recall proposition [ prop : u_a+c ] ) to the subgroup @xmath103 of matrices which are nontrivial ( i.e. , different from @xmath104 ) only in the top - left @xmath105 corner . in terms of weights
, we only care about powers of @xmath106 and substitute @xmath107 .
this is equivalent to saying that the probability of @xmath108 is the normalized coefficient of @xmath109 in the identity @xmath110 where @xmath44 is as in , and we are dividing by the normalizing constants to have the `` @xmath111 '' coefficients add up to 1 .
this corresponds to looking at relative dimensions of _ isotypical _ subspaces ( i.e. , those that transform according to fixed irreducible representation ) rather than the actual ones .
the first two of the above three limit transitions turn into @xmath112 in fact , these two limits @xmath90 and @xmath113 correspond to certain infinite - dimensional representations of the infinite - dimensional unitary group @xmath114 . the third ( gaussian ) limit is the eigenvalue projection of the matrix fourier transform identity @xmath115 where @xmath116 is the space of @xmath117 hermitian ( @xmath118 ) matrices , @xmath119 , and @xmath120 is the probability measure on @xmath116 with the density @xmath121 also known as the _ gaussian unitary ensemble _ ( or _ gue _ ) .
this limit is a special case of the so - called quasi - classical limit in representation theory that degenerates `` large '' representations to probability measures on ( co - adjoint orbits of ) the associated lie algebra , e.g. , see @xcite , @xcite , @xcite . for a broad survey of quantization ideas in representation theory see e.g. @xcite and references therein .
we are interested in more complex limit transitions than those in [ sub : distribution_of_lozenges_on_a_horizontal_slice ] , and for accessing them the following representation theoretic thinking is useful .
our probability weights arise as relative dimensions of the isotypical subspaces in the representation space for @xmath20 . moreover , these subspaces are blocks of identical irreducibles with respect to the action of the smaller group @xmath103 . the problem of decomposing a representation on irreducible components is often referred to as the _ problem of _ ( _ noncommutative _ ) _ harmonic analysis_. it can be viewed as a noncommutative fourier transform an analogue of the classical fourier transform when @xmath122 acts by shifts on @xmath123 .
the `` best '' way to solve such a problem would be to find operators in the representation space which project to a given isotypical component .
for the classical fourier transform , these operators have the form @xmath124 for the action of the symmetric group , such operators are known under the name _ young symmetrizers _ , they date back to the earliest days of representation theory .
however , even if one can construct such operators , they are quite complicated .
the `` next best '' thing is to find operators which are scalar in each irreducible representation ( the projection operators take value @xmath17 in one irreducible representation , and @xmath61 in all other irreducible representations ) . by a simple schur s lemma ,
such operators are exactly those that commute with the action of the group .
observe that @xmath103 has a nontrivial center scalar matrices of the form @xmath125 , @xmath126 .
their action on elements @xmath127 amounts to multiplying each @xmath128 by @xmath129 , and their action on a vector of weight @xmath130 is the multiplication by @xmath131 ( see [ sub : representations_of_unitary_groups ] ) . hence , using the homogeneity of the schur polynomials we see that on an irreducible representation of @xmath103 with highest weight @xmath132 such an operator acts as the scalar operator @xmath133 .
let us now apply such an operator , viewed simply as the dilation operator @xmath134 , to the decomposition identity defining @xmath135 : @xmath136 ( clearly , @xmath137 ) . setting @xmath138 above , we get @xmath139 this immediately tells us ( `` for free '' ) , that @xmath140 has the poisson distribution with parameter @xmath141 because the left - hand side is the characteristic function of that distribution .
going further , the first nontrivial example of an operator which commutes with the action of @xmath103 is the so - called _
quadratic casimir laplace operator _ @xmath142 .
its action on functions on @xmath143 is given by @xmath144 such operators exist for all semi - simple lie groups and are one of the basic representation - theoretic objects . also , @xmath145 is the ( projection to eigenvalues of the ) generator of the brownian motion on @xmath103 . in other words , is the generator of the circular dyson brownian motion @xcite , @xcite .
see also [ sub : dyson_brownian_motion ] below for a related markov dynamics .
it is immediate to see ( using the ratio of determinants formula and the fact that @xmath146 ) that the action of the quadratic casimir
laplace operators on the schur polynomials is diagonal , and @xmath147 we could now proceed with the application of @xmath142 to .
however , let us first note that the dilation operators @xmath148 can be written in a form rather similar to @xmath142 : @xmath149 indeed , the desired eigenrelation @xmath137 again follows from and the fact that @xmath150 .
let us now note that we have a general recipe on our hands of constructing operators which have schur functions as their eigenfunctions .
namely , for any operator of the form @xmath151 with @xmath152 we have @xmath153 for example , we can take @xmath154 where @xmath155 is a parameter . then using we
obtain a @xmath11-difference operator that can be rewritten in the form @xmath156 and gives @xmath157 we can now apply @xmath158 to the identity defining the measure @xmath159 .
this gives @xmath160 as before , we would like to substitute @xmath138 in the above identity
. however , observe that the left - hand side is not well - suited for that .
a standard trick helps the left - hand side can be rewritten as a simple contour integral : [ lemma : contour_integral ] let @xmath161 be a holomorphic function .
then @xmath162 where the integration contour goes around @xmath106 is the positive direction . using the above lemma and setting @xmath138
, we read from : @xmath163 the quantity in the right - hand side of does not seem very probabilistic , but we can now use the arbitrariness of the parameter @xmath11 . for any @xmath164 , we can compare the coefficients of @xmath165 in both sides of .
this amounts to integrating the left - hand side again ( with @xmath166 ) , and thus yields : [ thm : tiling_density ] for any @xmath167 and @xmath168 , @xmath169 the left - hand side of is a very meaningful probabilistic quantity it is the probability of seeing a vertical lozenge at any given location on the horizontal slice ( cf . fig . [
fig : slice_h ] ) .
this is the so - called _ density function _ of the measure @xmath159 .
furthermore , we see that the right - hand side of is well - suited for asymptotics .
we perform the asymptotic analysis in the next section .
theorem [ thm : tiling_density ] is a special case of a more general formula that represents correlation functions of the so - called _ schur measures _ as multiple contour integrals .
see @xcite and references therein for details .
here we perform an asymptotic analysis of the density function of the measure @xmath159 on the @xmath81th horizontal slice in the regime @xmath170 where @xmath171 is the point of observation in , which also must be scaled to yield nontrivial asymptotics .
the limit regime is quite nontrivial and is not achievable via elementary tools ( in contrast with the limit transitions in [ sub : distribution_of_lozenges_on_a_horizontal_slice ] ) .
the reader may want to peek at figures [ fig : asymp_picture ] and [ fig : sim_asymp_picture ] below to see what type of description we are aiming at .
changing variables @xmath172 , @xmath173 in the left - hand side of gives @xmath174 here by @xmath175 and @xmath176 we have denoted small positively oriented contours around @xmath61 and @xmath17 , respectively .
further analysis uses the original idea of okounkov @xcite and largely follows @xcite .
we observe that the integrand above has the form @xmath177 if we manage to deform the contours in such a way that @xmath178 on them except for possibly finite number of points ( where @xmath179 denotes the real part ) , then our integral would asymptotically vanish as @xmath180 .
the deformation depends on the location of the critical points of @xmath181 , i.e. , of the roots of the equation @xmath182 the discriminant of the numerator has the form @xmath183 we will now consider all possible cases one by one .
* case 1 . @xmath184 . * in this case both roots of are real and greater than @xmath17 .
the plot of @xmath185 looks as on fig .
[ fig : nu_bigger ] ( top ) . cc [ scale=1.2 , ultra thick ] ( 0,0 ) ( 6.8,0 ) ; ( 0,0 ) ( 0,2.5 ) ; ( 0,0 ) ( -2,-2 ) ; ( 0,0 )
( 0,-2.5 ) ; ( 2,2.5 )
( 2,-2.5 ) ; ( 4.95,-1.7 )
( 4.95,0 ) ; ( 3.55,.9 )
( 3.55,0 ) ; ( 0,0 ) circle ( .08 ) node[below right ] 0 ; ( 2,0 ) circle ( .08 ) node[below right ] 1 ; plot [ smooth , tension=.7 ] coordinates ( 1.8,-2.5 ) ( 1.5,-1 ) ( .6,1 ) ( .3,2.5 ) ; plot [ smooth , tension=1 ] coordinates ( -.3,2.5 ) ( -.9,.4 ) ( -2,-1 ) ; ( 4.95,-1.7 ) circle ( .08 ) node[below ] @xmath186 ; ( 3.55,.9 ) circle ( .08 ) node[above , yshift=4 ] @xmath187 ; plot [ smooth , tension=.8 ] coordinates ( 2.2,-2.5 ) ( 3.5,.9 ) ( 5,-1.7 ) ( 6.8,2 ) ; plot ( -.02+.475*cos ( ) , 1.5+.18*sin ( ) ) ; plot ( -.02+.475*cos ( ) , 1.5+.18*sin ( ) ) ; plot ( 2.04+.534*cos ( ) , -1+.2*sin ( ) ) ; plot ( 2.04+.534*cos ( ) , -1+.2*sin ( ) ) ; plot ( 3.529+.36*cos ( ) , .55+.11*sin ( ) ) ; plot ( 3.529+.36*cos ( ) , .55+.11*sin ( ) ) ; plot ( 5.02+.73*cos ( ) , -.7+.21*sin ( ) ) ; plot ( 5.02+.73*cos ( ) , -.7+.21*sin ( ) ) ; + ' '' '' [ scale=9 ] ( -.36,0)(.7,0 ) ; ( 0,-.25)(0,.25 ) ; ( 0,0 ) circle();at ( 0,0 ) @xmath61 ; ( .3,0 ) circle();at ( .3,0 ) @xmath17 ; ( .14,0 ) ellipse ( .44 and .18 ) node[yshift=38,xshift=90 ] @xmath188 ; ( .3,0 ) ellipse ( .1 and .08 ) node[yshift=10,xshift=12 ] @xmath86 ; ( .4,0 ) circle ( ) node[below right , xshift=-1 ] @xmath187 ; ( .58,0 ) circle ( ) node[below right , xshift=-1 ] @xmath186 ; moving the @xmath188 contour to the level line @xmath189 and the @xmath86 contour to the level line @xmath190 , we achieve that @xmath191 , which implies the desired vanishing .
however , in the process of deformation , the @xmath188 contour , which was originally a small circle around the origin , has swallowed the @xmath86 contour , see fig .
[ fig : nu_bigger ] ( bottom ) .
because of @xmath192 in the integrand , we have to compensate the result of moving the contours by subtracting the residue @xmath193 thus , we see that for @xmath184 , the density of vertical lozenges asymptotically vanishes . * case 2 .
@xmath194 . * in this case , two critical points solutions of are complex conjugate . consider the contour plot of @xmath195 , where we have shifted @xmath181 by the value of @xmath196 at the upper critical point , @xmath197 , @xmath198 ( @xmath199 denotes the imaginary part ) .
this contour plot looks like fig .
[ fig : contour_plot ] ( left ) .
[ cols="^,^ " , ] in a similar way , using products of operators @xmath158 with different values of @xmath11 , one can extract integral representations for higher _ correlation functions _ of vertical lozenges ( i.e. , probabilities that a given set of locations is occupied by vertical lozenges ) .
those integral representations can be analyzed exactly in the same fashion as above , this was done in @xcite . indeed ,
if one knows ( here @xmath81 is fixed , but one can also handle different @xmath81 s ) @xmath200 for any @xmath201 , one can extract the order @xmath202 correlation function by looking at coefficients of monomials @xmath203 .
the result reproduces known formulas for the correlation functions of the so - called _ schur processes _ , e.g. , see @xcite and references therein .
it should also be possible to carry out a similar program for the case of the growing hexagon with sides @xmath16 when the triple ratio @xmath96 remains constant .
this would require analyzing the asymptotics of ratios of the form @xmath204 with growing @xmath44 as in , which can probably be done via recently developed techniques of @xcite .
in a different way , integral representations for the correlation functions in the hexagon were recently obtained and asymptotically analyzed in @xcite , @xcite , @xcite .
our next goal is to add an extra dimension to our probabilistic models by introducing suitable markov evolutions on them .
this is not obvious and requires preliminary work .
a hint at the existence of a nontrivial markov dynamics comes from the relation to random matrices mentioned before ( in particular , see the third limit regime in [ sub : distribution_of_lozenges_on_a_horizontal_slice ] ) . indeed ,
a gue matrix of size @xmath117 has density with respect to the lebesgue measure on the linear space @xmath116 of hermitian @xmath117 matrices given by @xmath205 where @xmath206_{i , j=1}^{n}$ ] .
thus , the @xmath207 quantities @xmath208 are independent identically distributed standard normal random variables . following dyson @xcite , one can replace these variables by standard brownian motions .
a nontrivial computation shows that the corresponding markov process on hermitian matrices projects to a _ markov _ process on the spectra of matrices .
the generator of the process on the spectra is given by ( here @xmath209 ) : @xmath210 here on the right , @xmath211 means composition of operators : first , we multiply by the vandermonde determinant @xmath212 , then apply the laplacian , and after that divide by the vandermonde determinant , similarly to above .
the projection of the ( random ) matrix @xmath213 ( evolving according to standard brownian motions of its elements ) to the spectrum then has the distribution density ( see for example @xcite ) @xmath214 the dynamics with generator ( called the _ dyson brownian motion _ ) can be easily mimicked for all the ensembles of the form @xmath215 considered in
[ sub : distribution_of_lozenges_on_a_horizontal_slice ] .
let us focus on the poisson ( @xmath90 ) case , when @xmath216 , @xmath217 .
consider a markov jump process with generator @xmath218 where @xmath219 is the generator of the standard poisson process , and @xmath220 acts as @xmath221 on the @xmath222th coordinate .
one easily checks that the measures with @xmath216 are generated by the above markov process started from the initial condition @xmath223 .
the process with generator can be obtained by conditioning independent poisson processes not to intersect until time @xmath224 , and also to grow at the same rate : @xmath225 ( different growth rates of different @xmath226 s will result in conjugating @xmath227 by a different function , cf .
@xcite , @xcite . )
this is similar to the stationary version of the dyson brownian motion being obtained from independent one - dimensional standard brownian motions by conditioning on the event that they never intersect , and , moreover , stay within the distance @xmath228 from the origin as time goes to plus or minus infinity .
there is also another `` perpendicular '' markovian structure on the measures from [ sub : distribution_of_lozenges_on_a_horizontal_slice ] .
observe that the uniform measure on lozenge tilings has the following property : if we pick a domain inside the hexagon , then fixing the boundary lozenge configuration induces the uniform measure on tilings of the interior .
this seemingly trivial observation becomes useful when the hexagon becomes infinitely large in some way ( as in [ sub : distribution_of_lozenges_on_a_horizontal_slice ] ) .
then the global uniform measure makes no sense , but this property survives
. we will refer to it as to the _ gibbs property_. in particular , fixing @xmath81 vertical lozenges on the horizontal slice of height @xmath81 ( as on fig .
[ fig : slice_h ] ) induces the uniform measure on the set of all configurations of lozenges between this slice and the lower border ( height zero ) .
thus , given locations @xmath229 of the vertical lozenges on the @xmath81-th slice , the distribution of @xmath230 vertical lozenges at height @xmath230 is given by the ratio ( assuming that @xmath231 interlaces @xmath232 ) @xmath233 ( we have used proposition [ prop : s_la_1111 ] ) .
we will denote the above probabilities by @xmath234 .
note that the horizontal slices of measures that we obtain in [ sub : distribution_of_lozenges_on_a_horizontal_slice ] by taking limits @xmath90 and @xmath93 of the hexagon are also related by these _
_ @xmath235 .
in the gue limit , the formula remains the same , except that the @xmath236 are now reals , not integers . in this case
the above formula gives the density of a markov kernel with respect to the lebesgue measure .
the two markov processes discussed above ( the dyson brownian motion and its discrete analogue @xmath237 ) are quite canonical , but they have one deficiency they are one - dimensional ( in the sense that the state space consists of particle configurations in @xmath238 or @xmath239 ) .
we would like to construct a two - dimensional process which has interlacing two - dimensional arrays as its state space , and that `` stitches together '' the above one - dimensional processes in a natural way .
we begin by considering one such process which is constructed as follows .
consider random words built from the alphabet @xmath240 as follows : each letter @xmath241 is appended at the end of the word according to a standard (= rate 1 ) poisson process jumps , and different letters appear independently .
we can encode this as on fig .
[ fig : rsk_lpp ] : we draw a star ( @xmath242 ) in row @xmath241 at the time moment when a new letter @xmath241 is added . the stars in each row
form a poisson process , and different rows are independent .
cc [ scale = 1.1 , thick ] ( 0,3.5)(0,0 ) ( 6.5,0 ) node[below right , xshift=-15 , yshift=-2 ] time ; in 1,2,3,4,5,6,7 ( 6.2,/2)(0,/2 ) ; in 1,2,3 at ( -.3 , /2-.25 ) ; at ( -.3 , 7/2-.25 ) @xmath22 ; at ( -.3 , 2.13 ) @xmath243 ; at ( -.3 , 2.63 ) @xmath243 ; at ( .3,.23 ) @xmath242 ; at ( .9,2.23 ) @xmath242 ; at ( 1.4,2.23 ) @xmath242 ; at ( 2.2,.73 ) @xmath242 ; at ( 2.5,1.73 ) @xmath242 ; at ( 3,1.23 ) @xmath242 ; at ( 3.7,3.23 ) @xmath242 ; at ( 4.1,2.73 ) @xmath242 ; at ( 5.2,1.23 ) @xmath242 ; at ( 5.7,2.23 ) @xmath242 ; from these data we construct a gelfand
tsetlin scheme of depth @xmath22 written as @xmath244 as follows ( see fig . [
fig : rsk_paths ] ) : @xmath245{0.58\textwidth } { the maximal number of ( $ * $ ) one can collect on fig.~\ref{fig : rsk_lpp } along $ j$ nonintersecting up - right paths that connect points $ ( 1,2,\ldots , j)$ on the left border ( $ \mathrm{time}=0 $ ) , and $ ( h - j+1,h - j+2,\ldots , h)$ on the right border ( $ \mathrm{time}=t>0 $ ) } \right ) .
\label{rsk_gt_definition}\end{aligned}\ ] ] cc [ scale = 1.1 , thick ] ( 0,3.5)(0,0 ) ( 6.5,0 ) node[below right , xshift=-15 , yshift=-2 ] time ; in 1,2,3,4,5,6,7 ( 6.2,/2)(0,/2 ) ; in 1,2,3 at ( -.3 , /2-.25 ) ; at ( -.3 , 5/2-.25 ) @xmath81 ; at ( -.3 , 7/2-.25 ) @xmath22 ; ( 0,.73 ) ++(.6,0)++(0,1.5 ) (6.2,2.23 ) ; ( 0,.23 ) ++(1.2,0)++(0,.5)++(1.4,0)++(0,.5 ) ++(3,0)++(0,.5)(6.2,1.73 ) ; ( 0,.73 ) circle(.1 ) ; ( 0,.23 ) circle(.1 ) ; ( 6.2,2.23 ) circle(.1 ) ; ( 6.2,1.73 ) circle(.1 ) ; at ( .3,.23 ) @xmath242 ; at ( .9,2.23 ) @xmath242 ; at ( 1.4,2.23 ) @xmath242 ; at ( 2.2,.73 ) @xmath242 ; at ( 2.5,1.73 ) @xmath242 ; at ( 3,1.23 ) @xmath242 ; at ( 3.7,3.23 ) @xmath242 ; at ( 4.1,2.73 ) @xmath242 ; at ( 5.2,1.23 ) @xmath242 ; at ( 5.7,2.23 ) @xmath242 ; in particular , we see that @xmath246 , @xmath247 , is the length of the longest increasing subsequence of letters in the subword made of letters @xmath248 .
moreover , @xmath249 is the total number of letters @xmath250 in our word .
after time @xmath251 , the distribution of the gelfand
tsetlin scheme @xmath44 defined by is the same as the @xmath90 limit of the uniform measure on tilings of hexagon .
that is , to obtain the measure on gelfand
tsetlin schemes , one takes the following distribution of the top row @xmath252 : @xmath253 and projects it down by the stochastic links , i.e. , multiplies it by @xmath254 this is essentially greene s theorem for the robinson schensted correspondence coupled with explicit formulas for the number of standard and semistandard young tableaux .
see , e.g. , @xcite , @xcite , @xcite .
as we are interested in time evolution , the following statement is relevant : [ prop : rsk_dyn ] the markov process on random words ( i.e. , the process of adding new letters according to standard poisson processes ) projects to a markov process on gelfand
tsetlin schemes defined above .
it can be described by the following rules : 1 .
each `` particle '' @xmath246 has an independent poissonian clock of rate 1 . when the clock rings , the particle jumps by 1 , i.e. , @xmath255 .
2 . when any particle @xmath256 moves by 1 , it triggers either the move @xmath257 , or @xmath258 ( exactly one of them ) , see fig .
[ fig : moves_lr ] .
the second one is chosen generically , while the first one is chosen only if @xmath259 , i.e. , if the move @xmath260 violated the interlacing constraint .
see @xcite ( in particular , 7 ) and references therein . ;
( , 1 ) circle ( .12 ) node [ above right , xshift=-3 ] @xmath261 ; ( -,1 ) circle ( .12 ) node [ above left , xshift=3 ] @xmath262 ; ; ( - , ) + + ( -+2*,1 - 2 * ) ; ( , )
+ + ( -2*,1 - 2 * ) ; ( 0,0 ) circle ( .12 ) node [ below right ] @xmath256 ; the markov dynamics on gelfand
tsetlin schemes from proposition [ prop : rsk_dyn ] turns out to have the following properties : a. for each @xmath263 and any initial condition , the evolution of @xmath264 is markovian ( i.e. , lower rows do not care about the upper ones ) .
b. for each @xmath263 , the evolution preserves the gibbs property ( [ sub : gibbs_property ] ) on @xmath264 .
that is , if one starts with an initial condition of the form @xmath265 then after running the dynamics for any time @xmath266 , the distribution of @xmath264 will be of the same form with a different probability measure @xmath267 .
c. for each @xmath263 , the map @xmath268 is the time @xmath251 evolution of the markov process with the generator @xmath269 . while ( i ) is obvious , ( ii ) and ( iii ) are not ; they follow e.g. from theorem [ thm : schur_system_of_eqns ] below .
there is one more property which can be easily observed from the random words description of the dynamics .
namely , the projection of the process of proposition [ prop : rsk_dyn ] to the rightmost particles @xmath270 is _ markov_. it is more convenient to describe it in shifted strictly ordered coordinates @xmath271 ( cf . ) .
then each @xmath272 jumps to the right by 1 independently with rate 1 , and pushes @xmath273 over by 1 if @xmath273 occupies the target location of @xmath272 ( i.e. , if we had @xmath274 before the jump ) .
we call this process the _ pushtasep _ , i.e. , the _ pushing totally asymmetric simple exclusion process _
( it was introduced in @xcite under the name _ long - range tasep _ , see also @xcite ) .
definition is powered by what is known as robinson schensted algorithm in combinatorics .
questions related to application of various insertion algorithms ( including the general robinson schensted algorithm ) to random input were considered in , e.g. , @xcite , @xcite , @xcite , and @xcite , and can be traced back to the work of vershik and kerov @xcite in mid-1980 s .
the dynamical perspective has been substantially developed by oconnell @xcite , @xcite , @xcite and biane bougerol
oconnell @xcite ( see also chhaibi @xcite ) .
the existence of the markov dynamics ( of proposition [ prop : rsk_dyn ] ) satisfying ( i)(iii ) is remarkable , yet its above construction is fairly complicated .
we would like to access it in a different way .
let us search for all continuous - time markov jump processes on gelfand
tsetlin schemes which satisfy conditions ( i)(iii ) of [ sub : an_example_of_a_two_dimensional_dynamics ] .
they must have the following structure : each particle @xmath256 jumps to the right by 1 with a certain rate ( potentially dependent on @xmath275 ) , and its jump triggers further moves on the higher levels @xmath276 .
indeed , because of ( iii ) and the fact that @xmath269 moves one particle at a time , no two particles on the same level can jump simultaneously . moreover , because of ( i ) , moves can propagate only upwards . in order to reach a reasonable classification , we need to restrict the class further by requiring _
nearest neighbor interactions _ : a move of @xmath256 can only trigger ( potentially with certain probabilities ) moves of the immediate top right neighbor @xmath261 and the the immediate top left neighbor @xmath262 ( see fig . [
fig : moves_lr ] ) , which can trigger moves on level @xmath277 , and so on .
actually , it is better to extend the notion of the top right nearest neighbor from @xmath261 to the first particle in the sequence @xmath278 whose jump does not violate interlacing .
we will additionally assume ( extending the nearest neighbor hypothesis ) that the individual jump rates of particles at level @xmath81 may only depend on @xmath279 and @xmath280 , and that the same is true for left and right probabilities of move propagation from level @xmath230 to level @xmath81 .
let us now parametrize our possibilities .
fix @xmath281 , and denote by @xmath282 the jump rate of @xmath256 , @xmath283 . also , denote by @xmath284 the _ conditional _ probabilities that , given that the @xmath241th part of @xmath279 has just increased by 1 , this move propagates to the top left neighbor @xmath285 of @xmath286 : @xmath287 where @xmath288 is the vector having zeros at each position except the @xmath241th where it has 1 .
similarly , let @xmath289 where @xmath290 is the lower index of the nearest top right neighbor of @xmath286 that is free to jump ( typically , @xmath291 ) .
[ prop : apart ] assume that @xmath279 and @xmath280 interlace with strict inequalities instead of weak ones .
then properties ( ii ) , ( iii ) imply @xmath292 where we set @xmath293 .
consider the gibbs measure on the first @xmath81 levels that projects to the delta measure at @xmath280 on level @xmath81 .
the rate of any jump @xmath294 , @xmath295 , can be computed in two different ways , using @xmath269 or using the two - dimensional dynamics on the array .
the projection of the array dynamics to levels @xmath296 looks as follows : on level @xmath230 we have the process driven by @xmath297 whose jumps may propagate to level @xmath81 with probabilities @xmath298 .
moreover , particles on level @xmath81 can jump independently according to the jump rates @xmath299 . comparing the two ways to describe the rate of @xmath294 yields the desired relations .
the system of equations of proposition [ prop : apart ] needs to be modified if the inequalities between parts of @xmath279 and @xmath280 are not strict . indeed , if @xmath256 is `` blocked '' by @xmath300 , i.e. , @xmath301
, then @xmath299 must be zero , and also @xmath302 and @xmath303 make no sense as @xmath279 could not have just come from the jump @xmath304 .
the modification looks as follows .
[ thm : schur_system_of_eqns ] for any @xmath305 , let @xmath306 be the set of indices such that each particle @xmath307 is free to move , i.e. , @xmath308
. then @xmath309 with the agreement that @xmath310 .
solving these equations for all pairs @xmath305 , @xmath311 , under the conditions @xmath312 , @xmath313 , @xmath314 , and @xmath315 , is equivalent to constructing a nearest neighbor markov dynamics as defined above satisfying conditions ( i)(iii ) , with an additional `` forced move '' rule : if @xmath316 and @xmath286 moves ( by 1 ) , then @xmath256 also moves . and
@xmath280 not interlacing after a move on level @xmath230 . ]
similar to proposition [ prop : apart ] , cf .
@xcite .
it is easy to describe linear spaces of solutions to the above linear systems .
any combination of them , for every pair @xmath305 , gives us a markov process with desired properties .
one can choose such combinations to design different processes .
we give three examples below , see @xcite for more .
* example 1 . * all @xmath317 , all @xmath318 , and @xmath319 this is the dynamics described above in [ sub : an_example_of_a_two_dimensional_dynamics ] via nonintersecting paths .
* example 2 . *
all @xmath320 , all @xmath321 , and @xmath322 this dynamics can be viewed as coming from the column insertion algorithm ( as opposed to the row insertion algorithm corresponding to the dynamics of [ sub : an_example_of_a_two_dimensional_dynamics ] ) .
observe that the restriction of this dynamics to the left - most particles @xmath323 is markovian . via the shift @xmath324 this restriction matches the well - known _
totally asymmetric simple exclusion process _
( _ tasep _ ) .
this dynamics was first introduced in @xcite .
* example 3 . * all @xmath318 , all @xmath321 , and all @xmath325 .
this dynamics has minimal pushing and maximal `` noise '' ( coming from individual jumps ) . it can be viewed as a two - dimensional growth model ; in terms of the stepped surfaces interpretation , independently with rate one this dynamics adds all possible `` sticks '' ( directed columns ) of the form [ scale=.5 ] ; ; in ( 0,0 ) ( 0,0 ) ( .5,0.866 ) ( 1,0 ) ( .5,-0.866 ) cycle ; in ( , - ) ( 0,0 ) ( .5,0.866 ) ( 1.5,0.866 ) ( 1,0 ) cycle ; in ( 2*,0 ) ( 0,0 ) ( -.5,0.866 ) ( .5,0.866 ) ( 1,0 ) cycle ; [ scale=.5 ] ; ; in ( 0,0),(-,- ) ( 0,0 ) ( .5,0.866 ) ( 1,0 ) ( .5,-0.866 ) cycle ; in ( , -),(0,-2 * ) ( 0,0 ) ( .5,0.866 ) ( 1.5,0.866 ) ( 1,0 ) cycle ; in ( 2*,0 ) ( 0,0 ) ( -.5,0.866 ) ( .5,0.866 ) ( 1,0 ) cycle ; [ scale=.5 ] ; ; in ( 0,0),(-,-),(-2*,-2 * ) ( 0,0 ) ( .5,0.866 ) ( 1,0 ) ( .5,-0.866 ) cycle ; in ( , -),(0,-2*),(-,-3 * ) ( 0,0 ) ( .5,0.866 )
( 1.5,0.866 )
( 1,0 ) cycle ; in ( 2*,0 ) ( 0,0 ) ( -.5,0.866 ) ( .5,0.866 )
( 1,0 ) cycle ; [ scale=.5 ] ; ; in ( 0,0),(-,-),(-2*,-2*),(-3*,-3 * ) ( 0,0 ) ( .5,0.866 )
( 1,0 ) ( .5,-0.866 ) cycle ; in ( , -),(0,-2*),(-,-3*),(-1,-4 * ) ( 0,0 )
( .5,0.866 )
( 1.5,0.866 )
( 1,0 ) cycle ; in ( 2*,0 ) ( 0,0 ) ( -.5,0.866 ) ( .5,0.866 )
( 1,0 ) cycle ; directed as shown ( no overhangs allowed ) .
projection of this dynamics to the leftmost particles @xmath323 gives tasep , and projection to the rightmost particles @xmath326 gives pushtasep .
see @xcite for more details on this dynamics .
pictorially , the three examples can be represented as on fig .
[ fig:3ex ] .
ccc ; ; ; in ( 0,0 ) , ( - , ) , ( , ) , ( 2*,2*),(-2*,2*),(0,2 * ) , ( 3*,3*),(1*,3*),(-,3*),(-3*,3 * ) circle ( ) ; in ( 0,0 ) , ( , ) , ( 2*,2 * ) , ( 3*,3 * ) circle ( 1.8 * ) ; in ( 0,0 ) , ( , ) , ( - , ) , ( 2*,2*),(0,2*),(-2*,2 * ) + + ( -.9*,.9 * ) ; & ; ; ; in ( 0,0 ) , ( - , ) , ( , ) , ( 2*,2*),(-2*,2*),(0,2 * ) , ( 3*,3*),(1*,3*),(-,3*),(-3*,3 * ) circle ( ) ; in ( 0,0 ) , ( - , ) , ( -2*,2 * ) , ( -3*,3 * ) circle ( 1.8 * ) ; in ( 0,0 ) , ( , ) , ( - , ) , ( 2*,2*),(0,2*),(-2*,2 * ) + + ( .9*,.9 * ) ; & ; ; ; in ( 0,0 ) , ( - , ) , ( , ) , ( 2*,2*),(-2*,2*),(0,2 * ) , ( 3*,3*),(1*,3*),(-,3*),(-3*,3 * ) circle ( ) ; in ( 0,0 ) , ( - , ) , ( , ) , ( 2*,2*),(-2*,2*),(0,2 * ) , ( 3*,3*),(1*,3*),(-,3*),(-3*,3 * ) circle ( 1.8 * ) ; pictures like fig .
[ fig : any_ex ] provide other interesting examples of two - dimensional dynamics , cf .
@xcite .
cc ; ; ; in ( 0,0 ) , ( - , ) , ( , ) , ( 2*,2*),(-2*,2*),(0,2 * ) , ( 3*,3*),(1*,3*),(-,3*),(-3*,3 * ) circle ( ) ; in ( 0,0 ) , ( , ) , ( -2*,2 * ) , ( 1*,3 * ) circle ( 1.8 * ) ; in ( 0,0 ) , ( 0,2*),(-2*,2 * ) + + ( -.9*,.9 * ) ; in ( , ) , ( - , ) , ( 2*,2 * ) + + ( .9*,.9 * ) ; & ; ; ; in ( 0,0 ) , ( - , ) , ( , ) , ( 2*,2*),(-2*,2*),(0,2 * ) , ( 3*,3*),(1*,3*),(-,3*),(-3*,3 * ) circle ( ) ; in ( 0,0 ) , ( , ) , ( -2*,2*),(0,2 * ) , ( 3*,3*),(-1*,3 * ) circle ( 1.8 * ) ; in ( 0,0 ) , ( -2*,2 * ) + + ( -.9*,.9 * ) ; in ( , ) , ( 0*,2 * ) + + ( .9*,.9 * ) ; note that due to , pictures corresponding to various `` fundamental '' dynamics for which two of the three quantities @xmath327 , @xmath328 , and @xmath329 in each of the equations are zero ( and the remaining quantity is @xmath17 ) can be transformed one into the other by a sequence of local `` flips '' as on fig .
[ fig : flips ] .
we have seen how to construct random growth models in @xmath330 dimension ( tasep , pushtasep ) and @xmath331 dimension , and in [ sec : asymptotics ] we have seen how these models can be analyzed at large times .
we will now move on to a @xmath11-deformation of this picture , which will eventually lead us to directed polymers in random media .
the developments of the previous sections were heavily based on properties of the schur polynomials @xmath332_{i , j=1}^{n } } { \det\big[x_i^{n - j}\big]_{i , j=1}^{n } } \\&=\sum_{{\lambda}^{(1)}\prec{\lambda}^{(2)}\prec \ldots\prec{\lambda}^{(n ) } } x_1^{|{\lambda}^{(1)}| } x_2^{|{\lambda}^{(2)}|-|{\lambda}^{(1)}|}\ldots x_n^{|{\lambda}^{(n)}|-|{\lambda}^{(n-1)}|}.\end{aligned}\ ] ] we would like to add parameters to the theory .
it is easy to deform (= add parameters to ) our model viewed as a probabilistic object .
however , most such deformations would lack solvability properties of the original model based on schur polynomials .
the reason is that the schur polynomials are algebraic objects , and algebraic structures ( in contrast with probabilistic ones ) are usually very rigid .
thus , to find meaningful ( solvable ) deformations of the model requires nontrivial algebraic work .
historically , first two different one - parameter deformations of the schur polynomials were suggested : around 1960 by algebraists philip hall and d.e .
littlewood , and around 1970 by a statistician henry jack .
the _ hall littlewood polynomials _ naturally arose in finite group theory and were later shown to be indispensable in representation theory of @xmath333 over finite and @xmath334-adic fields .
the _ jack polynomials _ extrapolated the so - called zonal spherical functions arising in harmonic analysis on riemannian symmetric spaces from three distinguished parameter values that correspond to spaces over @xmath335 , and @xmath336 .
they are also known as eigenfunctions of the trigonometric calogero
sutherland integrable system . in mid-1980
s , in a remarkable development ian macdonald united the two deformations into a two - parameter deformation known as _ macdonald polynomials_. the two parameters are traditionally denoted as @xmath11 and @xmath251 .
we will soon set @xmath251 to @xmath61 , so it will not interfere with the time variable in our markov processes . the hall
littlewood polynomials arise when @xmath337 , and the jack polynomials correspond to the limit regime @xmath338 , where @xmath339 .
schur polynomials correspond to @xmath340 .
other significant values are : schur s q - functions ( for @xmath337 , @xmath341 ) ; monomial symmetric functions ( @xmath337 , @xmath342 ) ; and ( the most important for us ) @xmath11-whittaker functions arising for @xmath343 .
the shortest way to define macdonald polynomials is to say that these are elements of @xmath344^{s(n)}$ ] ( this is the algebra of symmetric polynomials in variables @xmath345 whose coefficients are rational functions in @xmath11 and @xmath251 ) , that diagonalize the following first order @xmath11-difference operator : @xmath346\operatorname{\mathscr{t}}_{q , x_i}\\&=&\displaystyle \sum_{i=1}^{n}\prod_{j\ne i}\frac{tx_i - x_j}{x_i - x_j } \operatorname{\mathscr{t}}_{q , x_i } , \end{array}\end{aligned}\ ] ] where , as before , @xmath347 .
it is immediately recognized as a deformation of the @xmath340 operator from [ ssub : general_recipe ] . the operator @xmath158 from is called the _ first macdonald difference operator_. there are also higher order ones , @xmath348 the operators @xmath349 are diagonalized by the same polynomial basis ( * ? ? ?
vi ) . as schur polynomials ,
the macdonald polynomials in @xmath22 variables are parametrized by @xmath350 .
we denote the ( _ monic _ ,
i.e. , with coefficient 1 of the lexicographically largest monomial , which is @xmath351 ) macdonald polynomials by @xmath352 .
they satisfy @xmath353 where @xmath354 are the elementary symmetric polynomials .
the eigenrelation for @xmath355 and @xmath340 was used in
[ sec : lozenge_tilings_and_representation_theory ] to compute the density function of the vertical lozenges . developing the ( beautiful ) theory of macdonald polynomials requires significant efforts , and we will not pursue this here . an excellent resource is the macdonald s book @xcite .
instead , we will focus on the @xmath11-whittaker ( @xmath343 ) case , where , for a story parallel to the schur case ( [ sec : lozenge_tilings_and_representation_theory][sec : markov_dynamics ] ) , we need the following facts .
[ prop : fact1_qwhit ] for any @xmath43 , we have @xmath356 where @xmath357 is the @xmath11-analogue of the factorial .
we thus see that the interlacing structure ( gelfand tsetlin schemes ) remains intact , but the gibbs property is @xmath11-deformed .
we will now say that a probability measure on gelfand
tsetlin schemes @xmath358 is _ gibbs _
if for any @xmath359 , @xmath360 with the @xmath11-deformed stochastic links @xmath361 recall that in the schur case we had simply @xmath362 which was then explicitly evaluated in .
[ prop : fact2_qwhit ] if we define the coefficients @xmath135 by expanding form a basis in the linear space of symmetric polynomials in @xmath363 .
thus , is a variant of the taylor expansion . ]
@xmath364 then @xmath365 for any @xmath366
. moreover , these probability measures on @xmath367 are time @xmath251 distributions of a jump markov process with jump rates @xmath368 where for @xmath369 the last factor is omitted . a proof ( in the more general @xmath0-setting ) can be found in @xcite . in the schur case ,
reduces to the ratio of vandermonde determinants , cf . above .
propositions [ prop : fact1_qwhit ] and [ prop : fact2_qwhit ] give us sufficient data to run the same search ( as in
[ sec : markov_dynamics ] ) for nearest neighbor markov processes which preserve the gibbs measures , and which on each level are described by @xmath370 .
this immediately leads to the following @xmath11-analogue of theorem [ thm : schur_system_of_eqns ] : [ thm : q_dyn_main_thm ] for any @xmath305 , define @xmath371 with @xmath372 , @xmath373 .
if there is a @xmath374 whose indices make no sense , then the corresponding factor is omitted : @xmath375 let @xmath376 be all the indices such that particle @xmath307 is free to move , i.e. , @xmath377 . then @xmath378 with agreement @xmath379 , and also @xmath380 .
solving these equations for all pairs @xmath305 is equivalent to constructing nearest neighbor markov dynamics satisfying the @xmath11-versions of conditions ( i)(iii ) of [ sub : an_example_of_a_two_dimensional_dynamics ] . using theorem [ thm : q_dyn_main_thm ]
, we can now explore the same three examples as in [ sub : examples_of_two ] : * example 1 .
* we enforce the almost sure move propagation ( i.e. , @xmath381 ) , and also @xmath319 this gives a unique solution @xmath382 for all @xmath241 such that @xmath285 is free .
in fact , this expression telescopes to give @xmath383 we observe that for @xmath384 , all the probabilities @xmath385 and @xmath284 are nonnegative , and the projection to the rightmost particles @xmath326 is _
markovian_. in the shifted variables @xmath386 , it can be described as follows : each particle jumps to the right by 1 independently with poisson clock of rate 1 .
if the @xmath241th particle moved , it triggers the move of @xmath387st one with probability @xmath388 , where @xmath389 is the number of empty spots in front of the @xmath241th particle before the move ( which in its turn may trigger the move of the @xmath390nd particle , etc . ) .
note that the probability @xmath388 is 1 if @xmath391 .
we call this particle system the _ @xmath11-pushtasep _ , it was first introduced in @xcite . its generalization ( called _ @xmath11-pushasep _ ) with particles moving in both directions can be found in @xcite
. * example 2 .
* now we again enforce @xmath392 , and @xmath322 this gives @xmath393 for all @xmath241 such that @xmath285 is free .
obviously , this gives negative probabilities , and we do not pursue this example further .
* example 3 . * here we enforce @xmath394 .
this clearly gives @xmath395 , and for @xmath384 this is a well - defined markov process without long - range interactions .
it was first constructed in @xcite , and it is closely related to the @xmath11-boson stochastic particle system of @xcite , see also @xcite , @xcite . while the projection of this process to the rightmost particles @xmath326
does not appear to be markovian ( because @xmath396 depends on @xmath397 ) , the projection to the leftmost particles @xmath323 _ is _ markovian . in
the shifted coordinates @xmath324 it can be described as follows : each particle jumps to the right by 1 independently of the others with rate @xmath398 , where @xmath389 is ( as before ) the number of empty spaces in front of @xmath399 before the jump .
note that this rate vanishes when @xmath391 , which correspond to a tasep - like blocking of the move .
we call this interacting particle system the _ @xmath11-tasep_. obviously , as @xmath400 , the @xmath11-pushtasep turns into the usual pushtasep , and @xmath11-tasep becomes the usual tasep .
we have thus obtained @xmath11-deformations of the random growth models from the schur case .
our next task will be to investigate their asymptotic behavior at large times .
our main tool in studying asymptotics will be the macdonald difference operators ( [ sub : definition_of_macdonald_polynomials ] ) .
by @xmath159 we mean the measure defined in proposition [ prop : fact2_qwhit ] . for any @xmath401 , @xmath402 where all the integrals are taken over small positively oriented closed contours around 1 . in the proof
we need to use the second macdonald parameter @xmath403 . for this ,
let us in this proof denote the time variable by @xmath404 to avoid the confusion .
we apply the @xmath405th order macdonald operator @xmath349 to the series expansion defining our measures , which now looks as @xmath406 we then replace the sum in the left - hand side by the residue expansion of the integral ( see @xcite for more detail ) @xmath407 where the contours encircle @xmath408 and no other poles ( i.e. , the residues are taken at @xmath409 , @xmath410 , @xmath411 ) .
note that in fact @xmath412_{a , b=1}^{k } \end{aligned}\ ] ] via the cauchy determinant formula . in the right - hand side of we use the eigenrelation ( see [ sub : definition_of_macdonald_polynomials ] ) : @xmath413 we then divide both sides of by @xmath414 , take the limit as @xmath415 , and also set @xmath416 .
there are two simple limit transitions that one can observe in the right - hand side of the contour integral formula .
we consider them in [ sub : gaussian_limit ] and [ sub : polymer_limit ] below .
the first limit regime is @xmath417 , @xmath418 , @xmath128 s do not change .
looking at the left - hand side of , which is @xmath419 , it is natural to expect that each @xmath420 grows as @xmath421 so that the quantities @xmath422 have finite limits ( which may still be random variables ) .
looking at higher powers of macdonald operators indeed reveals that this is a gaussian limit : @xmath423 has a law of large numbers with gaussian fluctuations of size @xmath424 , and the markov dynamics we constructed converge to gaussian processes .
we do not pursue this limit regime here , its detailed exposition will appear in @xcite .
another , structurally similar appearance of gaussian processes can be found in @xcite .
the second limit is a bit more complicated .
we again take @xmath417 , but now @xmath425 ( i.e. , we wait for a longer time than in the gaussian limit of [ sub : gaussian_limit ] ) . then to see a nontrivial limit
, we have to take the @xmath128 s of distance @xmath426 from 1 : @xmath427
. then we have @xmath428 and we see that the right - hand side of becomes @xmath429 times an asymptotically finite expression . to figure out the limiting behavior of @xmath430 , we now have to take @xmath431 of this expression , or take the natural logarithm and multiply by @xmath432 .
this gives @xmath433 where the remainder @xmath434 is supposed to be a finite random variable .
equivalently , @xmath435 with some limiting random variables @xmath436 .
( note that at this moment this is simply a _ guess _ ! ) we can now test what is happening with our dynamics under this conjectural scaling . for example , consider the @xmath11-pushtasep .
the asymptotics of the pushing probability is @xmath437 the increment of @xmath438 over time @xmath439 must then be ( 1 ) the increment coming from its own jumps , which is @xmath440 , where @xmath441 is a brownian motion , and ( 2 ) the increment coming from pushing , which is @xmath442 times the increment of @xmath443 . collecting terms of order @xmath421 , we conclude that @xmath444 where @xmath445 are independent standard brownian motions ( for @xmath87 the last term is omitted ) .
this system of stochastic differential equations ( sdes , for short ) is solved by @xmath446 the integral in the right - hand side can be viewed as the logarithm of the partition function ( i.e. , the _ free energy _ ) of a semi - discrete brownian polymer , see fig .
[ fig : polymer ] , and also ( * ? ? ?
* chapter 5 ) for a general discussion of directed polymers in random media .
cc [ scale=1 ] in 1,2,3,4,5,6 ( 7,)(0 , ) ; ( 4,1)(7,1 ) ; in 1,2,3 at ( -.35 , ) @xmath447 ; at ( -.35,6 ) @xmath22 ; at ( 7.1,.65 ) @xmath404 ; at ( 7.2,4.4 ) @xmath448 ; ( 0,1)(1.2,1)(1.2,2)(1.6,2)(1.6,3 ) (3.8,3)(3.8,4)(4.4,4)(7,4 ) ; ( 0,1 ) circle ( .12 ) ; ( 7,4 ) circle ( .12 ) ; in 1, ... ,80 + + ( 0.02,rand*0.06 ) ; in 1, ... ,80 + + ( 0.02,rand*0.06 ) ; in 1, ... ,80 + + ( 0.02,rand*0.06 ) ; more precisely , to any poisson - type up - right path @xmath449 that travels from @xmath17 to @xmath81 during time @xmath404 with jumps at moments @xmath450 , assign the energy @xmath451 then @xmath452 is the logarithm of the integral of the boltzmann factor @xmath453 over the lebesgue measure on all such paths ( the inverse temperature can be absorbed into the rescaling of @xmath404 with the help of the brownian scaling ) .
similar empirical scaling arguments show that the markov process on gelfand
tsetlin schemes of depth @xmath22 that lead to the @xmath11-pushtasep ( example 1 in [ sub : examples_of_q_deformed_two_dimensional_dynamics ] ) converges to a solution of the following system of sdes : @xmath454 [ thm : polymer_limit ] under the above scaling , the measure @xmath455 weakly converges to a probability measure on arrays @xmath456 that can be written in the form @xmath457 the integral taken over the lebesgue measure on the polytope of @xmath405-tuples of nonintersecting poisson - type paths @xmath458 joining @xmath459 with @xmath460 , see fig .
[ fig : nonint_polymer ] .
the limit measure is invariant under the flip @xmath461 .
the markov dynamics related to the semi - discrete directed brownian polymer was introduced in @xcite .
the weak convergence was proven in @xcite .
[ scale=1 ] in 1,2,3,4,5,6,7 ( 7,)(0 , ) ; ( 4,1)(7,1 ) ; in 1,2,3 at ( -.35 , ) @xmath447 ; at ( -.35,6 ) @xmath81 ; at ( 7.1,.65 ) @xmath404 ; at ( 6.5,4.4 ) @xmath462 ; at ( 6.5,6.4 ) @xmath448 ; ( 0,1)(2.2,1)(2.2,2)(2.6,2)(2.6,3 ) (3.8,3)(3.8,4)(4.4,4)(7,4 ) ; ( 0,2)(1.7,2)(1.7,3)(2.3,3)(2.3,4 ) (3.3,4)(3.3,5)(7,5 ) ; ( 0,3)(.7,3)(.7,4)(1.4,4)(1.4,5 ) (2,5)(2,6)(7,6 ) ; ( 0,1 ) circle ( .12 ) ; ( 0,2 ) circle ( .12 ) ; ( 0,3 ) circle ( .12 ) ; ( 7,4 ) circle ( .12 ) ; ( 7,5 ) circle ( .12 ) ; ( 7,6 ) circle ( .12 ) ; in 1, ... ,80 + + ( 0.02,rand*0.06 ) ; in 1, ... ,80 + + ( 0.02,rand*0.06 ) ; in 1, ... ,80 + + ( 0.02,rand*0.06 ) ; in 1, ... ,80 + + ( 0.02,rand*0.06 ) ; it is also known that the right - hand side of satisfies the system of sdes .
thus , the convergence in the above theorem should extend to a trajectory - wise statement , but , to our best knowledge , this has not been worked out in full detail yet .
observe the similarity of and the row robinson schensted construction of [ sub : an_example_of_a_two_dimensional_dynamics ] ( in particular , see figures [ fig : rsk_paths ] and [ fig : nonint_polymer ] ) .
in fact , arises via a _
geometric _ lifting of the robinson
schensted(knuth ) correspondence , see @xcite , @xcite , @xcite .
it is worth noting that other ( brownian - type ) scaling limits of growth models discussed here are considered in @xcite , @xcite .
from now on we focus on the asymptotic behavior of the free energy @xmath463 ( equality in distribution ) of the semi - discrete brownian polymer , see [ sub : polymer_limit ] ( recall that this is the logarithm of the polymer s partition function @xmath464 ) .
the free energy can be viewed as a limit of either @xmath465 ( thus , @xmath11-pushtasep ) , or of @xmath466 ( corresponding to the @xmath11-tasep ) , and either one can be used for the analysis .
( note that to obtain the polymer s partition function , @xmath22 must remain fixed . )
we will employ the @xmath11-tasep , as this is a bit more straightforward , and there are more details on the @xmath11-tasep in the literature .
we start by employing products of first order macdonald operators rather than a single one to obtain moments ( of all orders ) of the @xmath11-tasep particle locations .
[ prop : moments_of_the_qtasep ] consider the random gelfand tsetlin schemes of depth @xmath467 distributed according to @xmath468 ( this is the gibbs measure with the top row distributed according to @xmath469 , see [ sub:_q_whittaker_facts ] ) .
then for any @xmath470 , @xmath471 where the integral is taken over positively oriented , _ nested _ contours around 1 : the @xmath472 contour encircles 1 and no other poles , the @xmath473 contour encircles 1 and the contour @xmath474 , and so on ; the @xmath475 contour encircles @xmath476 , see fig .
[ fig : qtasep_contours ]
. [ scale=8 ] ( -.1,0)(1.4,0 ) ; ( 0,-.25)(0,.25 ) ; ( 0,0 ) circle();at ( 0,0 ) @xmath61 ; ( 1,0 ) circle();at ( 1,0 ) @xmath17 ; ( , 0 ) circle();at ( , 0 ) @xmath11 ; ( 1,0 ) circle(.08 ) node[above right ] @xmath472 ; ( , 0 ) circle(.08 * ) node[above left , xshift=-12 , yshift=3 ] @xmath477 ; ( + .01,0 ) ellipse ( .24 and .14 ) node[yshift=26,xshift=10 ] @xmath473 ; ( -.05,0 ) ellipse ( .34 and .18 ) node[yshift=11,xshift=-60 ] @xmath478 ; at ( .42,.1 ) @xmath479 ; at ( 1.17,.1 ) @xmath80 ; ( -.13,0 ) ellipse ( .54 and .22 ) node[yshift=31,xshift=-110 ] @xmath480 ; we consider @xmath481 , for larger @xmath405 the argument is similar .
we start with the defining identity : @xmath482 where @xmath483 means the normalization of @xmath352 by itself evaluated at all @xmath484 .
apply the first macdonald operator in @xmath485 variables to ( note that the second macdonald parameter @xmath251 is zero , and it has no relation to the time @xmath251 in ) .
this operator has the form @xmath486 because of the eigenrelation @xmath487 , in the right - hand side of we observe @xmath488 on the other hand , the left - hand side of is , by residue expansion , equal to the integral @xmath489 over a positively oriented contour around the simple poles @xmath490 .
the above argument works for applying @xmath491 to any multiplicative function @xmath492 with the exponentials @xmath493 in replaced by @xmath494 . in the next step ,
we apply @xmath495 to the result of application of @xmath491 to after setting @xmath496 ( the order @xmath497 is important ) . for the right - hand side of the resulting expression ,
we use proposition [ prop : fact1_qwhit ] that gives @xmath498 here @xmath499 means the matrix product @xmath500 .
this implies , together with the eigenrelation @xmath501 , that the right - hand side ( after setting @xmath502 ) gives @xmath503 . on the other hand , in the left - hand side
, the @xmath504-dependence in , after setting @xmath496 , is in the form of @xmath505 with @xmath506 .
hence , we can apply the same residue expansion ( for computing the application of @xmath495 to ) using the fact that @xmath507 here @xmath86 is the new integration variable whose contour has to encircle @xmath508 , but not any other poles ( in particular , not @xmath509 ) . renaming @xmath510 , we obtain the desired formula for @xmath481 . for larger @xmath405
the proof is similar .
the above iterative argument is due to v. gorin , see @xcite for a more general version
. the original proof of in @xcite involved a discretization of the quantum delta bose gas , see also @xcite and [ sub : moments_of_semi_discrete_brownian_polymer ] below .
since we already know the scaling which takes us from @xmath511 to the polymer partition function ( [ sub : polymer_limit ] ) , we can immediately do the limit in the integral of proposition [ prop : moments_of_the_qtasep ] .
this is very similar to the limit that we took in
[ sub : polymer_limit ] .
that is , let us use @xmath512 this leads to the following formula for the moments of the polymer partition function : for @xmath513 ( as in proposition [ prop : moments_of_the_qtasep ] ) , we have @xmath514 where the integrals are now over nested contours around @xmath61 : the @xmath515 contour contains only @xmath61 , the @xmath516 contour contains @xmath517 and @xmath61 , and so on ; the @xmath518 contour contains @xmath519 .
this limit transition from , however , is _ not _ a proof of the formula . indeed ,
theorem [ thm : polymer_limit ] only claims weak convergence , and we have exponential moments under the expectation ( that is , expectations of _ unbounded _ functions )
. _ sketch of the proof of . _
observe that if we define ( for _ any _ , not necessarily ordered @xmath520 ) @xmath521 ( with contours as in ) , then @xmath522 further , for @xmath523 , @xmath196 vanishes ( because there are no poles inside the smallest contour ) , and for @xmath524 , @xmath525 because when we write out the integral for this linear combination , the integrand will be skew - symmetric in @xmath526 and @xmath527 , and the two corresponding contours can be taken to be the same ( the obstacle to both these properties in is the factor @xmath528 , and the linear combination above exactly cancels this factor out ) . these properties together with initial condition @xmath529
uniquely determine @xmath530 for @xmath531 .
thus , it suffices to check that the moments of the polymer partition function ( see ) satisfy the same properties .
this check is fairly straightforward .
a discussion of how the evolution equation and the boundary conditions at @xmath524 used above relate to a discrete quantum bose gas can be found in @xcite , @xcite , @xcite .
there is a further limit that takes the semi - discrete polymer to a fully continuous one . in that case
, one defines @xmath532 where @xmath533 is the space - time white noise , and @xmath534 means normally ordered exponential , e.g. , see @xcite for an explanation .
equivalently ( via the feynman - kac formula ) , @xmath535 solves the _ stochastic heat equation with multiplicative noise _
@xmath536 then , defining @xmath537 we have @xmath538 away from diagonal subset , and @xmath539 the observations are not hard and were recorded at least as far back as the end of 1980 s by kardar @xcite and molchanov @xcite .
we believe that a rigorous proof can be extracted from the results of @xcite .
in particular , case @xmath540 was treated in ( * ? ? ? * i.3.2 ) . to the best of our knowledge
, the general case has not been worked out in full detail yet . a concise way to write is @xmath541 @xmath542 is called the _ delta - bose gas _ ( or _ lieb liniger _ ) _ hamiltonian_. the nested contour integral then takes the form @xmath543 where @xmath544 , and @xmath545 ( the above formula is not true without the order assumption on the @xmath546 s ) . for more detail ,
e.g. , see @xcite and references therein . by setting @xmath547 in , we see that the nested contour integrals provide us with all moments of the polymer partition function @xmath464 .
one might expect that this is sufficient to find its distribution or , equivalently , the distribution of the free energy @xmath548 .
it turns out that in this particular situation this is not true .
the distribution of the polymer partition function displays _ intermittency _ , which we now discuss .
this term appeared in studying the velocity and temperature fields in a turbulent medium @xcite , and describes structures that appear in random media having the form of peaks that arise at random places and at random time moments .
the phenomenon is widely discussed in physics literature , with magnetic hydrodynamics ( like on the surface of the sun ) and cosmology ( theory of creation of galaxies ) being two well - known examples , e.g. , see @xcite , @xcite .
the main property that allows one to detect an intermittent distribution is anomalous behavior ( as compared to the gaussian case , for example ) of ratios of successive moments .
* toy example . *
consider a sequence of independent identically distributed random variables @xmath549 , each taking value @xmath61 or @xmath550 with probability @xmath551 .
set @xmath552 .
then , clearly , @xmath553 the growth speed can be measured by the quantities @xmath554 on the other hand , computing a similar quantity for the sum of the @xmath555 s ( which is asymptotically gaussian ) gives because the random variable in question grows roughly linearly in @xmath22 , and @xmath533 above has exponential growth . ]
@xmath556 the key difference of these two cases is that @xmath557 in the first case , while @xmath558 in the second one . in general
, imagine that one has a time - dependent nonnegative random variable @xmath559 which grows in @xmath251 roughly exponentially ( or @xmath560 grows roughly linearly ) .
there are many ways to measure such growth ; we mostly follow @xcite in the exposition below .
define : 1 .
_ almost sure lyapunov exponent _
@xmath561 if the a.s . limit exists ( the `` a.s . ''
requirement can be weakened to convergence in probability ) .
moment _ ( or _ annealed _ ) _ lyapunov exponents _
@xmath562 ( assuming that limits exist ) .
hlder s inequality implies that @xmath563 ( because @xmath564 ) .
the strict inequalities will be referred to as _ intermittency_. note that we also obviously have @xmath565 . in a typical situation , when the distribution of @xmath559 does not deviate much from its mean , @xmath566 if there exists @xmath567 such that @xmath568 then for all @xmath569 , @xmath570 hlder s inequality with @xmath571 gives @xmath572 which implies that @xmath573 .
replacing @xmath405 by @xmath574 and taking @xmath87 , we have @xmath575 ( we used the hypothesis of the lemma ) .
rearranging terms gives the needed inequality for @xmath576 .
repeating inductively , we obtain the desired claim .
let us now show how the definition of intermittency relates to peaks .
if we pick @xmath577 such that @xmath578 , then for large enough @xmath251 ( below we omit @xmath251 in the notation for @xmath559 ) : 1 .
@xmath579 , because otherwise we would have @xmath580 .
an overwhelming contribution to @xmath581 comes from the region where @xmath582 .
indeed , @xmath583 the first term is @xmath584 , and we know that the left - hand side behaves exactly as @xmath585 .
3 . @xmath586 because @xmath587 .
hence , we observe a hierarchy of higher and higher peaks concentrated on smaller and smaller sets ( that are actually exponentially small in probability ) , and higher peaks contribute overwhelmingly to high enough moments .
in the situation of random fields when ergodicity allows to replace computing expectations by space averaging , at each fixed large time one can then observe a hierarchy of islands with exponentially ( in time ) high values that dominate moment computations .
intermittency is a characteristic feature of products of a large number of independent random variables ( cf .
the toy example above ) .
indeed , by the central limit theorem , let us check that random variables of the form @xmath588 ( for example , with independent identically distributed @xmath555 s ) are intermittent .
we have @xmath589^{p}\right)= e^{t\left(p\mu+\frac{p^{2}}2\sigma^{2}\right)},\end{aligned}\ ] ] which implies that @xmath590 since under intermittency the moments are dominated by increasingly atypical behavior ( i.e. , observed with small probability ) , it is hard to expect that the moments would determine the distribution .
for example , for the exponential of the standard gaussian @xmath591 they do not : any distribution with density @xmath592 with @xmath593 , @xmath594 , gives the same moments .
when @xmath595 , this is the density of the log - normal random variable @xmath596 .
see @xcite for more detail .
we will now check if the polymer partition function is intermittent .
[ thm : polymer_intermittency ] the moment lyapunov exponents for the brownian polymer are given by 1 .
in the semi - discrete case ( [ sub : polymer_limit ] and [ sub : moments_of_semi_discrete_brownian_polymer ] ) , @xmath597 , where ( for @xmath598 ) @xmath599 and @xmath600 is the unique solution to @xmath601 on @xmath602 .
2 . in the fully continuous case ( [ sub : continuous_brownian_polymer ] ) , for paths between @xmath603 and @xmath604 , @xmath605 steepest descent for contour integral representations of moments given before ( [ sub : moments_of_semi_discrete_brownian_polymer ] and [ sub : continuous_brownian_polymer ] ) . see @xcite for details . in the fully continuous setting , the above result is due to kardar @xcite ( non - rigorously ) , and bertini
cancrini @xcite .
given our previous discussion on peak domination in moments , it seems hopeless that this limit behavior of moments would carry any information about the behavior of the main bulk of the distribution .
this is , however , a suitable moment to demonstrate the ( in)famous _ replica trick _ widely used in physics . note that , at least formally , @xmath606 averaging both sides , dividing by @xmath251 and taking @xmath607 suggests @xmath608 where we tacitly use a ( non - unique ! ) analytic continuation of @xmath609 off nonnegative integers . for the semi - discrete polymer , @xmath610 , and @xmath611 taking the value of @xmath612 at the only critical point of this function on @xmath602 ,
gives @xmath613 this actually _ is _ the correct answer !
it was conjectured in @xcite and proven in @xcite .
similarly , in the fully continuous case , @xmath614 which is also correct @xcite , @xcite .
we thus see that this very nonrigorous procedure , quite remarkably , lead us to the correct almost sure behavior ! in the next section we show how to access these results rigorously , and our approach will also explain in a way why the replica trick worked in this particular situation .
as we have seen in [ sec : moments_for_q_whittaker_processes ] , the intermittency phenomenon prevents us from recovering the distribution of the polymer partition function from its moments .
however , this is not so in the @xmath11-setting .
namely , the @xmath11-moments @xmath615 , @xmath616 , uniquely determine the distribution of @xmath466 ( because @xmath617 and @xmath618 , so these are moments of a bounded random variable ) .
our plan is thus to convert the @xmath11-moment formulas that we have ( proposition [ prop : moments_of_the_qtasep ] ) into a formula for the expectation of a one - parameter family of observables that remain bounded ( unlike the moments @xmath615 ) in the @xmath619 which leads to polymers .
since this will involve @xmath11-moments with @xmath620 , it is inconvenient to use nested contours in integral representations as their positions depend on @xmath405 .
there are two ways to `` un - nest '' the contours : ( 1 ) to deform all of them to identical large concentric circles @xmath621 ; or ( 2 ) to deform all of them to identical small concentric circles @xmath622 .
the first way is easier to realize , but it is harder to turn the result into a meaningful asymptotic information .
thus , we proceed with the second one .
the following lemma is nontrivial and very useful : [ lemma : unnesting ] let @xmath623 be a meromorphic function and @xmath624 be its singular set which must not include @xmath61 .
assume that @xmath625 is disjoint from @xmath624 for all integers @xmath626 .
then @xmath627_{i , j=1}^{\ell}\times\\ & \hspace{120pt } \times \prod_{j=1}^{\ell}f(w_j)f(qw_j ) \ldots f(q^{{\lambda}_j-1}w_j)dw_j , \label{unnesting } \end{aligned}\ ] ] where each @xmath628 contour contains @xmath629 and @xmath624 , but not @xmath61 ( thus , the @xmath630 contours are nested ) , and the @xmath299 contours contain @xmath624 and no other poles ( so , the @xmath86 contours can be taken to be all the same ) .
this lemma is the result of mere bookkeeping of the residues when we shrink the contours and take into account the poles at @xmath633 .
the fact that the result is rather nice is nontrivial , and takes origin in harmonic analysis on riemannian symmetric spaces and hecke algebras , cf .
a proof of the lemma can be found in ( * ? ? ?
3.2.1 ) , see also ( * ? ? ?
* lemma 3.3 and prop .
7.4 ) .
the form of the right - hand side of suggests that one could take a generating function of such expressions over different @xmath405 .
more exactly , it easily implies that @xmath636_{i , j=1}^{\ell } \times\\&\displaystyle\hspace{50pt}\times\prod_{j=1}^{\ell } ( 1-q)^{n_j}\zeta^{n_j}f(w_j)\ldots f(q^{n_j-1}w_j)dw_j .
\end{array } \label{getting_qlaplace}\end{aligned}\ ] ] we will now take @xmath631 , with all the integration contours above being small enough positively oriented contours around 1 . here
@xmath637 are simply permuted values of @xmath638 in , and the change of the combinatorial factor from @xmath639 to @xmath640 is due to that un - ordering . is the number of different ways to obtain a given @xmath641 with @xmath642 from @xmath643 . ] now , using the @xmath11-exponential identity ( e.g. , see @xcite , @xcite ) @xmath644 we can rewrite the left - hand side of as @xmath645 one should expect that in a suitable scaling limit as @xmath619 ( which we can predict by looking at the moment asymptotics ) , the @xmath11-moment generating function would converge to the laplace transform of the polymer partition function . the latter _ does _ define the distribution uniquely , with or without intermittency .
the real question now is how to take a similar limit in the right - hand side of .
observe that termwise limit would produce a moment generating series , and we already know that it diverges !
let us consider the case @xmath646 in which the problem of convergence is already there .
then @xmath647 s are the @xmath11-moments of the simple continuous - time one - sided random walk started from @xmath61 at @xmath343 .
we expect their @xmath11-generating function to converge ( as @xmath73 ) to the laplace transform of the lognormal distribution ( i.e. , @xmath648 ) .
indeed , we should expect that because @xmath649 where @xmath650 .
observe that we only have first order poles at @xmath651 in the right - hand side of for @xmath646 .
because of vanishing of the @xmath652_{i , j=1}^{\ell}$ ] for equal values of the @xmath299 s , we conclude that only @xmath653 give a nontrivial contribution .
this contribution is @xmath654 we now need to take the @xmath619 limit in the above sum , and we can not do that termwise as this would result in a divergent series .
a standard tool of the theory of special functions used for dealing with such a limit is the _ mellin barnes integral representation _ which dates back to the end of the 19th century . in its simplest incarnation
, it says that @xmath655 where the integral in the right - hand side goes in the negative direction around the poles @xmath656 .. ] indeed , @xmath657 where we assume @xmath658 to be defined with the branch cut @xmath659 . omitting convergence and contour deformation justifications ( which can be performed ) , we rewrite the series in as @xmath660 where @xmath661 and the integration is taken over a contour as on fig .
[ fig : mellin_barnes ] .
we can now take the needed limit .
we note that @xmath663 is the @xmath11-analogue of the euler @xmath664-function , and that ( e.g. , see @xcite ) @xmath665 we take the scaling @xmath666 so that @xmath667 and @xmath668 we see that the limit of the integral in is @xmath669 which is a correct expression for the laplace transform of the lognormal random variable @xmath648 , as we expected . the same mellin barnes integral representation works for any @xmath467 .
the summations over @xmath670 are replaced by integrals over @xmath671 with @xmath672 inside , and using scaling of [ sub : polymer_limit ] together with theorem [ thm : polymer_limit ] ( which guarantees convergence of expectations of bounded functions ) , as well as the asymptotic relations above ( setting @xmath673 ) , we obtain the following generating function for the semi - discrete brownian polymer s partition function ( [ sub : polymer_limit ] ) : [ thm : laplace_polymer ] fix @xmath467 , @xmath674 , and @xmath675
. then in was denoted by @xmath404 in [ sub : polymer_limit ] and [ sub : case_n_1_and_the_mellin_barnes_integral_representation ] . ]
@xmath676_{i , j=1}^{\ell}. \end{aligned}\ ] ] note that theorem [ thm : tw_convergence ] proves the value of the almost sure lyapunov exponent @xmath684 that we guessed ( for @xmath685 , but this could have been for any @xmath5 ) using replica trick in [ sub : replica_trick ] .
the tracy
widom distribution in the right - hand side of arises as the series @xmath686_{i , j=1}^{\ell } , \label{tw}\end{aligned}\ ] ] where the @xmath687 and the @xmath688 contours are as on fig .
[ fig : airy_cont ] .
[ scale=1 , ultra thick ] ( -1.5,0)(.7,0 ) ; ( 0,-)(0 , ) node [ below , yshift=-4,xshift=0 ] @xmath61 ; ( -1,-.866)(-.5,0 ) node [ above , xshift=5 ] @xmath687 ; ( -.5,0)(-1,.866 ) ; ( -.75,0 ) arc ( -180:-120:.25 ) ; at ( -1,-.3 ) @xmath689 ; [ scale=1 , ultra thick ] ( -.6,0)(1.5,0 ) ; ( 0,-)(0 , ) node [ below , yshift=-4,xshift=0 ] @xmath61 ; ( 1,-.866)(.5,0 ) node [ above , xshift=-5 ] @xmath688 ; ( .5,0)(1,.866 ) ; ( .75,0 ) arc ( 0:-60:.25 ) ; at ( 1,-.3 ) @xmath689 ; the way one reaches from the right - hand side of is fairly straightforward . by changing the variables @xmath690 , one rewrites the part of the integrand that depends on the large parameter @xmath22 as @xmath691 since @xmath692 we take @xmath693 , and then we see that @xmath694 the analysis then follows the scheme explained in
[ sec : asymptotics ] , with @xmath188 contours being deformed to the domain with @xmath695 , and @xmath696 contours to the domain with @xmath697 .
the limiting expression arises in the situation when @xmath698 has a double critical point @xmath699 , and through a local change of integration variables near the critical point ; the constant @xmath700 is actually @xmath701 .
details can be found in @xcite and @xcite .
let us conclude by observing that if we expand the right - hand side of into residues at @xmath702 , we get back the divergent generating series for the moments of the polymer partition function that we found before .
this shows that a more sophisticated replica trick than the one from [ sub : replica_trick ] can actually be used to obtain the limiting distribution , and not only the law of large numbers ( i.e. , @xmath684 ) .
namely , one can obtain the moments by solving the equations ( the delta bose gas of [ sub : continuous_brownian_polymer ] ) that they satisfy , write down the series for the laplace transform through moments ( despite the fact that this series diverges ) , make sense of this series via the mellin
barnes integral representation , and then proceed with the asymptotic analysis . this approach was successfully carried out in physics papers @xcite , @xcite .
however , the only plausible explanation we have at the moment as to why such an approach leads to the correct answer , is that it is a limiting case of the @xmath11-deformed situation , where all the steps are legal and indeed lead to a proof of the gue edge fluctuations .
a. okounkov .
symmetric functions and random partitions . in s.
fomin , editor , _ symmetric functions 2001 : surveys of developments and perspectives_. kluwer academic publishers , 2002 .
arxiv : math/0309074 [ math.co ] . | these are lecture notes for a mini - course given at the cornell probability summer school in july 2013 .
topics include lozenge tilings of polygons and their representation theoretic interpretation , the @xmath0-deformation of those leading to the macdonald processes , nearest neighbor dynamics on macdonald processes , their limit to semi - discrete brownian polymers , and large time asymptotic analysis of polymer s partition function . |
it has recently been demonstrated how an open - ended antiferromagnetic heisenberg spin chain can be used as an adiabatic quantum data bus @xcite ( see sec [ sub : db_review ] for a brief review ) .
this data bus takes advantage of antiferromagnetic couplings to transfer qubits of information adiabatically .
first a single qubit , encoded in a single spin is joined to an even length heisenberg spin chain slowly enough such that the adiabatic theorem applies .
then a single spin on the other end of the chain is separated , again slowly enough for the adiabatic theorem to apply . as long as
the interactions between the spins on the chain are predominately antiferromagnetic , the qubit will be successfully transferred from one end of the chain to the other .
this protocol is illustrated in fig .
[ fig : transport cartoon ] .
antiferromagnetic spin clusters have been studied for there potential usefulness in quantum computing in other contexts , for example in refs .
@xcite .
cartoon of an adiabatic quantum bus protocol for the heisenberg spin chain @xcite .
a spin with the encoded qubit is connected to one end of an even length antiferromagnetic chain . afterwards
, the spin on the opposite end is removed adiabatically .
as long as the chain interactions are predominately antiferromagnetic and the adiabatic theorem is satisfied the qubit will be transferred . ]
this paper demonstrates how by applying particular unitary operations to spin chains for single qubit gates , and by using a specific spin network for a cnot gate , one can achieve universal holonomic quantum computation .
this method uses open - loop holonomies , meaning that the hamiltonian is not necessarily returned to the same state after the adiabatic process .
the methods used here can also be extended to non - adiabatic implementations of geometrical quantum computing .
holonomic quantum computation ( hqc ) was conceived and shown to be universal by zanardi and rasetti @xcite and was formulated in terms of a non - abelian berry phase .
hqc is considered to be an appealing method for achieving fault tolerant quantum computing because of its geometrical nature and because it can be implemented adiabatically , and therefore has all of the advantages of adiabatic quantum computation @xcite .
although many implementations of holonomic and geometric quantum computation are adiabatic , there are examples which are not @xcite . other proposed architectures for holonomic quantum computation use a variety of architectures , including superconducting systems with josephson junctions @xcite .
further examples propose using quantum dots @xcite .
single molecule magnets have been another system of interest @xcite .
a recent proposal has also been made for using holonomies which involve attaching and removing a spin from a spin 1 chain@xcite .
this architecture , although it looks superficially very similar to ours , performs computations locally , at the site of the spin rather than by transport as ours does .
@xcite also proposes an implementation on ultracold polar molecules .
another interesting proposal involves using a quantum wire with a twisted cluster state hamiltonian@xcite .
this proposal is similar to ours , but implements the twist in a fundamentally different way .
most other approaches to hqc involve building a system , and explicitly calculating the holonomies caused by various manipulations of the system . in our proposal
we start with a process which has a trivial berry phase .
real space twists are then performed on the spin chain used in this process .
unlike most examples of hqc , all results can be derived without explicitly considering the curvature of the underlying manifold of states , all single qubit gates result from the same underlying hamiltonian with basis rotations applied to it . in principle
this idea could be used for more traditional , closed - loop holonomic quantum computation by twisting a spin chain used to transport the qubit adiabatically slowly , performing an adiabatic transport protocol , untwisting the chain adiabatically slowly , and performing the transport protocol in reverse to return to the initial hamiltonian .
the net result of this process would be to perform a rotation on the initial spin which the qubit is encoded and return the spin chain to its initial state . in this case spin rotation would be achieved by a non - abelian berry phase acquired via transport on a closed adiabatic loop in parameter space , and would therefore be a closed - loop holonomic gate .
such a process involves unnecessarily many steps however , so it is more convenient to think of this process in terms of the framework of open - loop holonomic quantum computation .
the mathematical differences of this approach from others affords us the advantage that the spectrum of the underlying hamiltonian is the same for all twists , meaning that , by construction , all single qubit gates can be implemented in a way which requires the same annealing time to reach a given accuracy .
this architecture has the advantage that the only operation it ever requires to be adiabatically performed is the joining or removal of a spin from a chain or cluster .
the nature of the twists used here also means that a non - adiabatic transport protocol could be used instead , and universal computation would still be achieved .
this paper also outlines an implementation of the necessary components of this design using superconducting flux qubits .
superconducting flux qubits are a popular architecture for implementing scalable adiabatic quantum computing @xcite , and therefore are a natural choice for designing a scalable holonomic quantum computer .
an additional advantage of the use of superconducting flux qubits is that the designs tend to have spatially extended qubits and a high degree of connectivity@xcite .
the large spatial extent of the qubits means that a design could be implemented in which a qubit would only need to be transferred across a small number of spins to be moved from one location in a computer to any other arbitrary location . for this reason it is only necessary that the transport protocol be efficient for short chains , as has already been demonstrated in @xcite , rather than in the thermodynamic limit .
there has been recent experimental work involving quantum annealing to degenerate ground state manifolds using currently available superconducting flux qubit hardware@xcite . in this paper
it was demonstrated experimentally that signatures of quantum behaviors can be observed in the final state within a degenerate ground state manifold .
this provides an indication that a ground state manifold can be produced accurately enough on current hardware that quantum effects dominate over classical effects and design inaccuracies .
although the architecture proposed here can not be implemented on the hardware used in @xcite , this experiment does provide proof of principle for the use of degenerate manifolds in superconducting flux qubit systems . while it is not the main focus of this paper , we would also like to point out that there are other potential methods of implementing this architecture .
one example of such an implementation would be to use a coupled cavities scheme similar to the one explored in @xcite .
for such an implementation long spin chains may be required , and as a result properties in the thermodynamic limit may be important .
for such an implementation , the architecture given in this paper could easily be generalized to a j1-j2 spin chain with @xmath0 which is know to be gaped in the thermodynamic limit@xcite . in this case
, another option would be to implement the architecture non - adiabatically using the methods described in @xcite .
consider initially an antiferromagnetic heisenberg spin chain .
it has been shown that such a chain can act as a quantum data bus , both adiabatically@xcite and by using the dynamics of its excitations @xcite .
the initial hamiltonian is given by @xmath1
now imagine that one inserts a twist into the spin chain by applying a local unitary transformation of the form @xmath2 on @xmath3 spins , where @xmath4 are all mutually orthogonal to each other .
this yields a new hamiltonian of the form @xmath5 such a twist does not effect the spectrum of the hamiltonian , and therefore the dynamics of the adiabatic quantum bus protocol , or other quantum bus protocols which may make use of the unitary dynamics of the hamiltonian .
it is important to note , however , that after transfer across the chain , the spin will be rotated into the @xmath4 basis . as we will demonstrate later , transfer through this twisted spin chain
can perform any desired unitary rotation on the qubit being transferred , and thus can be used to implement any single qubit gate , see fig .
[ fig : twist_protocol ] .
( color online ) illustration of a single unitary gate implemented by adiabatic transport on a twisted chain . ]
one should note that while in this example we consider a simple heisenberg spin chain , gates can be implemented in this way on an xyz spin chain or a j1-j2 spin chain , or other sufficiently complex quantum spin hamiltonians . figuring out which twist to use to perform a given gate can be done easily and will be illustrated in the next section .
the local twist to implement the hadamard gate is @xmath6 , @xmath7and @xmath8 .
this twist can be calculated without difficulty , for details see sec .
[ sub : hadamard_calc ] .
one can therefore conclude that the hadamard gate can be implemented by performing the quantum bus protocol on the hamiltonian @xmath9 @xmath10 one can perform similar twists to implement any given single qubit gate .
the calculation to find @xmath4 in eq .
[ eq : htwist ] for other gates can be performed in the same way as the one in the previous section for the hadamard .
table [ tab : twists ] shows how to implement single qubit gates .
these gates are sufficient to perform an arbitrary unitary operation on a single spin .
it is shown in @xcite that any unitary rotation can be approximated to arbitrary precision with the gates given in table [ tab : twists ] .
we have already shown how to build an adiabatic quantum bus to move qubit states to arbitrary locations in the system .
next we discuss that a cnot gate can be implemented under this architecture .
then we have demonstrated a universal quantum computer . .
twists for implementing various single qubit gates[tab : twists ] [ cols="^,^,^,^,^",options="header " , ] we need to test how this design for a cnot gate performs because we can not rely on previous work to show that the qubit is actually transferred accurately . the two free parameters in eq .
[ eq : cnot hamiltonian ] are the strength of the ising bonds and the fields which we denote by h , and the time for the protocol to be performed , @xmath11 we now examine whether this hamiltonian actually implements a cnot gate effectively for reasonable values of h and @xmath12 .
to test this we need to answer 2 questions .
firstly , is the system close enough to the adiabatic limit for reasonable values of @xmath12 ?
secondly , is the desired effect of shutting off one possible path for the information achieved for reasonable values of h ? to answer these questions ,
we examine the overlap of the final output state ( final state of the `` out '' qubit in fig .
[ fig : cnot_design ] ) with the expected output state from a controlled not gate ( fig.[fig : cnot_performance ] ) .
note that because the case where a not gate is performed , and the case where the gate acts trivially are related by a simple unitary transformation on the hamiltonian , acceptable performance in one of these cases implies acceptable performance in the other .
averaging over different initial states is therefore unnecessary as it would yield the exact same result as any particular choice of control and input states .
( color online ) measures of performance of the cnot gate a ) one minus fidelity of the output spin versus @xmath12 for h=10 b ) one minus fidelity of the output spin versus h for @xmath12=10 c ) output fidelity for initial up spin with a not performed versus h and @xmath12 light is larger ( more positive ) , dark is smaller ( more negative ) d ) gap versus @xmath13 for various values of h. the field h and the gap are given in units of the heisenberg bond energy . ]
[ fig : cnot_performance ] demonstrates the effectiveness of this gate .
[ fig : cnot_performance ] a ) shows that for a moderate field and ising bond strength the gate can be made to perform well as long as the annealing time is sufficient .
this figure also shows that the gate can continually be made more effective by running it longer without having to raise h. the oscillations in the fidelity are related to the time scale of small excitations produced during the annealing process . in many applications one may have enough control over @xmath12that the annealing time
can be chosen in a way that the process lies near one of the local minima of error shown in fig .
[ fig : cnot_performance ] a ) .
[ fig : cnot_performance ] b ) shows the effect of h on fidelity for an annealing time of 10 ( in units of inverse heisenberg couplings ) . in this figure
one can see that increasing h is ineffective at improving performance above a certain value .
this indicates that at this point the field is already effectively completely blocking one path that the information transport can take .
[ fig : cnot_performance ] c ) shows the combined effects of h and @xmath12on the output polarization .
it is consistent with the conclusions we have reached from a ) and b ) .
finally , fig .
[ fig : cnot_performance ] d ) shows that the system gap remains quite large throughout the process .
it also demonstrates that beyond a certain value of h the gap does not increase significantly with changing h , which is consistent with the picture of one path being completely closed to information transfer .
it is interesting to note that increasing h increases the gap throughout the process .
to build an implementation of this holonomic architecture based on superconducting flux qubits one needs to design circuits that implement heisenberg spins and the appropriate gate couplings between spins . fortunately significant
work has already been done , for example in refs .
@xcite , towards the design of couplers for superconducting flux qubits .
however since the previously discussed schemes were based on ising spin systems , we still need to establish a method for designing circuits which emulate heisenberg spins .
it is interesting to point out that this computational architecture works with the limited connectivity of the designs proposed in refs .
specifically the cnot gate we have proposed fits in a single 2x4 chimera lattice cell like the one used in ref .
@xcite .
it has been shown in @xcite that a qubit which behaves like an ising spin can be constructed from a hamiltonian of the form @xmath14 where the applied fluxes @xmath15 act as effective magnetic fields , altering the shape of a potential well for the system in a way that mimics a spin constrained to move in a plane .
the constants @xmath16 simply act to scale the effect of the flux . to construct a heisenberg qubit one needs to add a third direction , leading to a hamiltonian of the form @xmath17 @xmath18 such a hamiltonian would allow the shape of the potential well to be changed along 3 directions and would therefore mimic a heisenberg spin rather than an ising spin .
previously proposed designs also only couple qubits along one direction .
if a new type of coupler were added to the currently implemented circuits which couple the spins in the y direction in addition the z direction , then an xy model could be implemented . to implement an xyz heisenberg model
, one needs to both design qubits which are not constrained to lie in a plane and build 3 types of couplers , one for each direction in space .
first consider a ccjj ( compound - compound josephson junction ) circuit as defined in @xcite .
the effective hamiltonian of this circuit is given by @xcite @xmath19 where @xmath20 .
note that there are more indeces than in @xcite because we do not have the condition that @xmath21 or @xmath22 .
the first two terms of the hamiltonian in eq.[eq : ham ] are not important for what we are trying to demonstrate here @xcite . the definition of all of these terms can be found in eq .
b4b - f in @xcite , and in sec .
[ sub : qubit_asssumptions ] of this paper .
let us make the simplifying assumption that all of the critical currents are equal for all junctions .
in practice there is variability in junction fabrication , but this error can be compensated by building a cccjj device ( see fig .
[ fig : cccjj ] ) .
let us also assume that our circuit is designed in such a way that we can inductively couple the left and right loop to each other very strongly such that @xmath23 and @xmath24 .
these assumptions cause the equations to simplify greatly ( see sec . [ sub : qubit_asssumptions ] ) , yielding @xmath25 @xmath26 @xmath27 ) one can effectively change the critical current of the junction pair and compensate for manufacturing errors .
a similar example with a cjj and a ccjj can be found in @xcite[fig : cccjj ] . ]
this leads to an effective hamiltonian of the form @xmath28 when we substitute in @xmath29from eq . [ eq : blrsimp ] , this hamiltonian becomes @xmath30 @xmath31 which is of the form given in eq .
[ eq : heisenberg_ham ] . the corresponding circuit is shown in fig .
[ fig : circ_ind ] . ]
we have demonstrated an architecture for a universal quantum computer using heisenberg spin chains and clusters . this architecture has the advantage that it can be implemented adiabatically and therefore has all of the advantages of adiabatic quantum computing .
it has already been demonstrated in @xcite that the single qubit gates and data bus used in this computer can be implemented with high fidelity for reasonable annealing time .
we further demonstrate that a controlled not gate can be implemented at high fidelity with a reasonably short annealing time and reasonable hamiltonian parameters .
in addition to suggesting an architecture , we also propose a design to physically realize this architecture .
we suggest a method for building superconducting flux qubit systems which model the low energy degrees of freedom of a heisenberg model .
because the architecture we propose can be implemented adiabatically , only the low energy degrees of freedom need to be reproduced .
we choose a superconducting flux qubit implementation because of the experimental success of these systems in performing non - universal adiabatic quantum computing with an ising spin glass model .
furthermore it has been shown that these ising spin glass models can be realized accurately enough that the degeneracy of the ground state manifolds is not broken .
the universal heisenberg spin based computer would represent a significant improvement over the current non - universal ising systems because it would allow these computers to implement important algorithms which the ising spin glass system has not been able to , such as shors algorithm for factoring large numbers @xcite .
the authors would like to thank p. zanardi , i. marvian , s. boixo , d. a. lidar , l. c. venuti , and s. takehashi for helpful conversations .
some numerical calculations were performed on the usc high performance computing cluster .
this research is partially supported by the aro muri grant w911nf-11 - 1 - 0268 .
the adiabatic data bus protocol discussed here is a process in which an even length anti - ferromagnetic heisenberg spin chain prepared in its ground state is joined adiabatically slowly to a single spin in a state @xmath32 .
a spin is also removed from the opposite side of the chain , again adiabatically slowly .
the final removed spin will again be in the state @xmath32 .
the hamiltonian for this protocol is where @xmath35 and @xmath36 are the annealing schedules , functions chosen such that @xmath37 , @xmath38 , @xmath39 , and @xmath40 .
this protocol can also be generalized to an xyz and j1-j2 spin chain ( see @xcite for more details ) . for simplicity
let us consider this protocol with a linear annealing schedule such that [ [ detailed - discussion - of - simplifying - assumptions - for - superconducting - flux - qubitssubqubit_asssumptions ] ] detailed discussion of simplifying assumptions for superconducting flux qubits[sub : qubit_asssumptions ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ let us make the simplifying assumption that all of the critical currents are equal , @xmath66
. in practice there is variability in junction fabrication , but this error can be compensated by building a cccjj device ( see fig .
[ fig : cccjj ] ) .
this assumption causes the equations to simplify greatly because @xmath67 , which has the consequence that @xmath68 and @xmath69 .
we can further assume that in our design that @xmath70 , this additional assumption causes @xmath71 and @xmath72 , which in turn causes @xmath73 .
after this simplification we now have | in this paper we examine the use of an adiabatic quantum data transfer protocol to build a universal quantum computer .
single qubit gates are realized by using a bus protocol to transfer qubits of information down a spin chain with a unitary twist .
this twist arises from altered couplings on the chain corresponding to unitary rotations performed on one region of the chain .
we show how a controlled not gate can be realized by using a control qubit with ising type coupling .
the method discussed here can be extended to non - adiabatic quantum bus protocols .
we also examine the potential of realizing such a quantum computer by using superconducting flux qubits . |
tries and suffix trees are the most popular data structures on words @xcite .
trie _ is a digital tree built over , say @xmath0 , strings ( the reader is referred to @xcite for an in depth discussion of digital trees . )
a string is stored in an external node of a trie and the path length to such a node is the shortest prefix of the string that is not a prefix of any other strings ( cf.=1000 figure [ fig - trie ] ) . throughout ,
we assume a binary alphabet .
then each branching node in a trie is a binary node .
a special case of a trie structure is a _
suffix trie _ ( tree ) which is a trie built over suffixes of a _ single _ string .
since 1960 tries were used in many computer science applications such as searching and sorting , dynamic hashing , conflict resolution algorithms , leader election algorithms , ip addresses lookup , coding , polynomial factorization , lempel - ziv compression schemes , and molecular biology .
for example , in the internet ip addresses lookup problem @xcite one needs a fast algorithm that directs an incoming packet with a given ip address to its destination . as a matter of fact , this is the _ longest matching prefix _ problem , and standard tries are well suited for it .
however , the search time is too large .
if there are @xmath0 ip addresses in the database , the search time is @xmath17 , and this is not acceptable . in order to improve the search time , andersson and nilsson @xcite introduced a novel data structure called the _ level compressed trie _ or in short lc trie ( cf.=1000 figure [ fig - trie ] ) . in the lc trie
we replace the root with a node of degree equal to the size of the largest _ full subtree _ emanating from the root ( the depth of such a subtree is called the _ fillup level _ ) .
this is further carried on recursively throughout the whole trie ( cf .
figure [ fig - trie ] ) .
some recent experimental results reported in @xcite indicated a `` dramatic improvement '' in the search time when full subtrees are replaced by `` partially fillup subtrees '' . in this paper , we provide a theoretical justification of these experimental results by considering @xmath4-lc tries in which one replaces a subtree with the last level only @xmath4-filled by a node of degree equal to the size of such a subtree ( and we continue recursively ) . in order to understand theoretically the @xmath4-lc trie behavior , we study here the so called _
level @xmath5 and the _ typical depth _ or the search time @xmath10 .
the @xmath4-fillup level is the last level in a trie that is @xmath4-filled , i.e.=1000 filled up to a fraction at least @xmath4 ( e.g. , in a binary trie level @xmath18 is @xmath4-filled if it contains @xmath19 nodes ) .
the typical depth is the length of a path from the root to a randomly selected external node ; thus it represents the typical search time . in this paper
we analyze the @xmath4-fillup level and the typical depth in an @xmath4-lc trie in a probabilistic framework when all strings are generated by a memoryless source with @xmath20 and @xmath21 . among other results , we prove that the @xmath4-lc trie shows a rather moderate improvement over the original lc tries
. we shall quantify this statement below .
tries were analyzed over the last thirty years for memoryless and markov sources ( cf.=1000 @xcite ) .
pittel @xcite found the typical value of the fillup level @xmath22 ( i.e. , @xmath23 ) in a trie built over @xmath0 strings generated by mixing sources ; for memoryless sources with high probability ( whp ) @xmath24 where @xmath25 is the smallest probability of generating a symbol and @xmath26 is the rnyi entropy of infinite order ( cf.=1000 @xcite ) .
we let @xmath27 . in the above ,
we write @xmath28 to denote @xmath29 in probability , that is , for any @xmath30 we have @xmath31 as @xmath32 .
this was further extended by devroye @xcite , and knessl and szpankowski @xcite who , among other results , proved that the fillup level @xmath22 is concentrated on two points @xmath33 and @xmath34 , where @xmath33 is an integer @xmath35 for @xmath11 .
the depth in regular tries was analyzed by many authors who proved that whp the depth is about @xmath36 ( where @xmath37 is the shannon entropy rate of the source ) and that it is normally distributed when @xmath11 @xcite .
the original lc tries were analyzed by andersson and nilsson @xcite for unbiased memoryless source and by devroye @xcite for memoryless sources ( cf.=1000 also @xcite ) .
the typical depth ( search time ) for regular lc tries was only studied recently by devroye and szpankowski @xcite who proved that for memoryless sources with @xmath11 @xmath38 in this paper we shall prove some rather surprising results .
first of all , for @xmath39 we show that the @xmath4-fillup level @xmath5 is whp equal either to @xmath33 or @xmath34 where @xmath40 as a consequence , we find that if @xmath11 , the depth @xmath10 of the @xmath4-lc is for large @xmath0 typically about @xmath41 the ( full ) @xmath2-fillup level @xmath22 shown in should be compared to the @xmath4-fillup level @xmath5 presented in .
observe that the leading term of @xmath5 is _ not _ the same as the leading term of @xmath22 when @xmath11 .
furthermore , @xmath4 contributes only to the second term asymptotics .
when comparing the typical depths @xmath42 and @xmath10 we conclude that both grow like @xmath43 with two constants that do not differ by much .
this comparison led us to a statement in the abstract that the improvement of @xmath4-lc tries over the regular lc tries is rather moderate .
we may add that for relatively slowly growing functions such as @xmath44 the constants in front of them do matter ( even for large values of @xmath0 ) and perhaps this led the authors of @xcite to their statements .
the paper is organized as follows . in the next section we present our main results which are proved in the next two sections .
we first consider a poissonized version of the problem for which we establish our findings .
then we show how to depoissonize our results completing our proof .
consider tries created by inserting @xmath0 random strings of 0 and 1 .
we will always assume that the strings are ( potentially ) infinite and that the bits in the strings are independent random bits , with @xmath20 and thus @xmath21 ; moreover we assume that different strings are independent .
we let @xmath45 : = # @xmath46 and @xmath47 , i.e.=1000 the proportion of nodes filled at level @xmath18 .
note that @xmath45 may both increase and decrease as @xmath18 grows , while @xmath48 recall that the fillup level of the trie is defined as the last full level , i.e. @xmath49 , while the height is the last level with any nodes at all , i.e.@xmath50 .
similarly , if @xmath51 , the @xmath52-fillup level @xmath5 is the last level where at least a proportion @xmath52 of the nodes are filled , i.e.@xmath53 we will in this paper study the @xmath52-fillup level for a given @xmath52 with @xmath54 and a given @xmath1 with @xmath55 . we have the following result , where whp means with probability tending to 1 as @xmath56 , and @xmath57 denotes the normal distribution function . theorem [ t1 ] is proved in section [ sdepo ] , after first considering a poissonized version in section [ spo ] . [ t1 ] let @xmath52 and @xmath1 be fixed with @xmath54 and @xmath55 , and let @xmath58 be the @xmath52-fillup level for the trie formed by @xmath0 random strings as above .
then , for each @xmath0 there is an integer @xmath59 such that whp @xmath60 or @xmath34 .
moreover , @xmath61 for @xmath62 .
thus the @xmath4-fillup level @xmath58 is concentrated on at most two values ; as in many similar situations ( cf.=1000 @xcite ) , it is easily seen from the proof that in fact for most @xmath0 it is concentrated on a single value @xmath33 , but there are transitional regimes , close to the values of @xmath0 where @xmath33 changes , where @xmath58 takes two values with comparable probabilities .
note that when @xmath63 , the second term on the right hand side disappears , and thus simply @xmath64 ; in particular , two different values of @xmath65 have their corresponding @xmath33 differing by @xmath66 only .
when @xmath11 , changing @xmath52 means shifting @xmath33 by @xmath67 . by theorem [ t1 ] ,
whp @xmath58 is shifted by the same amounts . to the first order
, we thus have the following simple result . [ ct1 ] for any fixed @xmath52 and @xmath1 with @xmath54 and @xmath55 , @xmath68 in particular , @xmath69 as @xmath56 .
surprisingly enough , the leading terms of the fillup level for @xmath23 and @xmath70 are quantitatively different for @xmath11 .
it is well known , as explained in the introduction , that the regular fillup level @xmath22 is concentrated on two points around @xmath71 , while the partial fillup level @xmath5 concentrates around @xmath72 .
secondly , the leading term of @xmath5 does not depend on @xmath4 and the second term is proportional to @xmath73 , while for the regular fillup level @xmath22 the second term is of order @xmath74 .
theorem [ t1 ] yields several consequences for the behavior of @xmath4-lc tries .
in particular , it implies the typical behavior of the depth , that is , the search time .
below we formulate our main second result concerning the depth for @xmath4-lc tries delaying the proof to section [ slc ] ; cf.=1000 and @xcite for lc tries . [ t3 ] for any fixed @xmath75 and @xmath76 we have @xmath77 as @xmath32 where @xmath78 is the entropy rate of the source . as a direct consequence of theorem [ t3 ] we can numerically quantify experimental results recently reported in @xcite where a `` dramatic improvement '' in the search time of @xmath4-lc tries over the regular lc tries was observed . in a regular lc trie
the search time is @xmath79 with the constant in front of @xmath44 being @xmath80 @xcite . for @xmath4-lc
tries this constant decreases to @xmath81 .
while it is hardly a `` dramatic improvement '' , the fact that we deal with a slowly growing leading term @xmath44 , may indeed lead to experimentally observed significant changes in the search time .
in this section we consider a poissonized version of the problem , where there are @xmath82 strings inserted in the trie .
we let @xmath83 denote the @xmath52-fillup level of this trie .
[ t2 ] let @xmath52 and @xmath1 be fixed with @xmath54 and @xmath55 , and let @xmath83 be the @xmath52-fillup level for the trie formed by @xmath84 random strings as above . then , for each @xmath85 there is an integer @xmath86 such that whp ( as @xmath87 ) @xmath88 or @xmath89
. we shall prove theorem [ t2 ] through a series of lemmas .
observe first that a node at level @xmath18 can be labeled by a binary string of length @xmath18 , and that the node is filled if and only if at least two of the inserted strings begin with this label .
for @xmath90 , let @xmath91 be the number of ones in @xmath92 , and let @xmath93 be the probability that a random string begins with @xmath92 .
then , in the poissonized version , the number of inserted strings beginning with @xmath90 has a poisson distribution @xmath94 , and these numbers are independent for different strings @xmath92 of the same length .
consequently , @xmath95 where @xmath96 are independent indicators with @xmath97 hence , @xmath98 so @xmath99 and , by chebyshev s inequality , @xmath100 consequently , @xmath101 is sharply concentrated , and it is enough to study its expectation .
( it is straightforward to calculate @xmath102 more precisely , and to obtain a normal limit theorem for @xmath45 , but we do not need that . ) assume first @xmath103 .
[ l1 ] if @xmath103 and @xmath104 then @xmath105 .
let @xmath106 and define @xmath107 by @xmath108 , i.e. , @xmath109 which leads to @xmath110 let @xmath111 . by and , @xmath112 if @xmath113 , then @xmath114 and @xmath115 if @xmath116 , then @xmath117 and @xmath118 hence yields , using @xmath119 , @xmath120 by the berry esseen theorem ( * ? ?
* theorem xvi.5.1 ) , @xmath121 by and the assumption , @xmath122 this finally implies @xmath123 and the lemma follows by and .
[ l2 ] fix @xmath103 . for every @xmath124
, there exists @xmath125 such that if @xmath126 , then @xmath127 .
a string @xmath90 has two extensions @xmath128 and @xmath129 in @xmath130 .
clearly , @xmath131 , and if there are exactly 2 ( or 3 ) of the inserted strings beginning with @xmath92 , then @xmath132 .
hence @xmath133 let @xmath134 and @xmath135 be as in the proof of lemma [ l1 ] , and let @xmath136 .
then @xmath137 $ ] and thus @xmath138 .
moreover , by and the assumption , @xmath139 thus , if @xmath18 is large enough , we have by the standard normal approximation of the binomial probabilities ( which follows easily from stirling s formula , as found already by de moivre @xcite ) @xmath140 for some @xmath141 .
hence , by , @xmath142 as needed .
- now assume @xmath103 . starting with any @xmath18 as in ,
we can by lemmas [ l1 ] and [ l2 ] shift @xmath18 up or down @xmath66 steps and find @xmath143 as in such that , for a suitable @xmath125 , @xmath144 and @xmath145 .
it follows by that whp @xmath146 and @xmath147 , and hence @xmath148 or @xmath149 .
this proves theorem [ t2 ] in the case @xmath103 .
the case @xmath150 follows by symmetry , interchanging @xmath1 and @xmath151 . in the remaining case @xmath63 , all @xmath152 are equal .
thus , by and , @xmath153 given @xmath65 , there is a @xmath154 such that @xmath155 .
we take @xmath156 .
then , @xmath157 and thus @xmath158 for some @xmath159 .
similarly , @xmath160 for some @xmath161 , and the result follows in this case too .
to complete the proof of theorem [ t1 ] we must depoissonize the results obtained in theorem [ t2 ] , which we do in this section .
given an integer @xmath0 , let @xmath33 be as in the proof of theorem [ t2 ] with @xmath162 , and let @xmath163
. then @xmath164 and @xmath165 as @xmath56 . by monotonicity
, we thus have whp @xmath166 , and by theorem [ t2 ] it remains only to show that we can take @xmath167 .
let us now write @xmath168 and @xmath169 , since we are working with several @xmath170 .
[ l3 ] assume @xmath11 .
then , for every @xmath18 , @xmath171 we have @xmath172 and thus , by and the argument in , @xmath173 which completes the proof . - by lemma [ l3 ] , @xmath174 . hence , by the proof of theorem [ t2 ] , for large @xmath0 , @xmath175 and @xmath176 , and thus whp @xmath177 or @xmath34 .
moreover , the estimate @xmath61 follows easily from the similar estimate for the poisson version in lemma [ l1 ] ; we omit the details .
this completes the proof of theorem [ t1 ] for @xmath103 .
the case @xmath150 is again the same by symmetry .
the proof when @xmath63 is similar , now using .
first , let us explain heuristically our estimate for @xmath10
. by the asymptotic equipartition property ( cf.=1000 @xcite ) at level @xmath33 there are about @xmath178 strings with the same prefix of length @xmath33 as a randomly chosen one , where @xmath179 is the entropy . that is , in the corresponding branch of the @xmath52-lc trie
, we have about @xmath180 strings ( or external nodes ) , where for simplicity @xmath181 . in the next level
, we shall have about @xmath182 external nodes , and so on .
in particular , at level @xmath10 we have approximately @xmath183 external nodes . setting this @xmath184 leads to our estimate ( [ et3 ] ) of theorem [ t3 ] .
we now make this argument rigorous .
we construct an @xmath52-lc trie from @xmath0 random strings @xmath185 and look at the depth @xmath10 of a designated one of them . in principle , the designated string should be chosen at random , but by symmetry , we can assume that it is the first string @xmath186 . to construct the @xmath52-lc trie , we scan the strings @xmath185 in parallel one bit at a time , and build a trie level by level . as soon as the last level is filled less than @xmath52 , we stop ; we are now at level @xmath187 , just past the @xmath52-fillup level . the trie above this level , i.e.=1000 up to level @xmath5 , is compressed into one node , and we continue recursively with the strings attached to each node at level @xmath188 in the uncompressed trie , i.e.=1000 the sets of strings that begin with the same prefixes of length @xmath188 .
to find the depth @xmath189 of the designated string @xmath186 in the compressed trie , we may ignore all branches not containing @xmath186 ; thus we let @xmath190 be the number of the @xmath0 strings that agree with @xmath186 for the first @xmath188 bits .
note that we have not yet inspected any later bits .
hence , conditioned on @xmath58 and @xmath190 , the remaining parts of these @xmath190 strings are again i.i.d.=1000 random strings from the same memoryless source , so we may argue by recursion .
the depth @xmath189 equals the number of recursions needed to reduce the number of strings to 1 .
we iterate , and let @xmath196 be the number of strings remaining after @xmath197 iterations ; this is the number of strings that share the first @xmath197 levels with @xmath186 in the compressed trie .
we have @xmath198 and @xmath199 .
we stop the iteration when there are less than @xmath200 strings remaining ; we thus let @xmath201 be the smallest integer such that @xmath202 . in each iteration before @xmath201 , holds with error probability @xmath203 .
hence , for any constant @xmath204 , we have whp for every @xmath205 , with @xmath206 , @xmath207 or equivalently @xmath208 iterating this result once , we see that whp at most @xmath219 further iterations are needed to reduce the number to less than @xmath220 . finally , the remaining depth then whp is @xmath221 even without compression .
hence we see that whp@xmath222 since @xmath223 is arbitrary , theorem [ t3 ] follows .
it remains to prove lemma [ la ] .
let @xmath224 be the number of the strings @xmath185 that are equal to @xmath186 for at least their first @xmath18 bits .
the @xmath225 , and thus , for any @xmath124 , @xmath226 lemma [ la ] thus follows from the following two lemmas , using the observation that @xmath227 .
the first lemma is a large deviation estimate corresponding to corollary [ ct1 ] .
we begin with the poissonized version , with @xmath84 strings as in section [ spo ] .
let @xmath230 , and let @xmath231 be fixed with @xmath232 .
then , by lemma [ l1 ] , if @xmath228 is large enough , @xmath233 and @xmath234 for all large @xmath170 . by a chernoff bound , can be sharpened to @xmath235 and thus @xmath236 similarly , @xmath237 . to depoissonize ,
let @xmath163 as in section [ sdepo ] and note that , again by a chernoff estimate , @xmath238 and @xmath239 .
thus , with probability @xmath240 , @xmath241 and the result follows ( if we increase @xmath228 ) . since @xmath248 , there exists @xmath249 such that if @xmath250 , then @xmath251 , and thus @xmath252 noting that @xmath253 and @xmath254 , we see that , provided @xmath223 is small enough , @xmath255 for some @xmath256 , and then and a chernoff estimate yields , when @xmath250 , @xmath257 and thus @xmath258 y. reznik , on the average density and selectivity of nodes in multi - digit tries , _ proceedings of the seventh workshop on algorithm engineering and experiments and the second workshop on analytic algorithmics and combinatorics ( alenex / analco ; vancouver , 2005 ) _ , siam , 230239 , 2005 . | andersson and nilsson introduced in 1993 a _ level - compressed trie _
( in short : lc trie ) in which a full subtree of a node is compressed to a single node of degree being the size of the subtree .
recent experimental results indicated a `` dramatic improvement '' when full subtrees are replaced by `` partially filled subtrees '' . in this paper , we provide a theoretical justification of these experimental results showing , among others , a rather moderate improvement of the search time over the original lc tries .
for such an analysis , we assume that @xmath0 strings are generated independently by a binary memoryless source ( a generalization to markov sources is possible ) with @xmath1 denoting the probability of emitting a `` @xmath2 '' ( and @xmath3 ) .
we first prove that the so called @xmath4-fillup level @xmath5 ( i.e. , the largest level in a trie with @xmath4 fraction of nodes present at this level ) is concentrated on two values whp ( with high probability ) ; either @xmath6 or @xmath7 where @xmath8 is an integer and @xmath9 denotes the normal distribution function .
this result directly yields the typical depth ( search time ) @xmath10 in the @xmath4-lc tries with @xmath11 , namely we show that whp @xmath12 where @xmath13 and @xmath14 is the shannon entropy rate .
this should be compared with recently found typical depth in the original lc tries which is @xmath15 where @xmath16 . in conclusion
, we observe that @xmath4 affects only the lower term of the @xmath4-fillup level @xmath5 , and the search time in @xmath4-lc tries is of the same order as in the original lc tries .
[ cols= " < , < " , ] * key words * : digital trees , level - compressed tries , partial fillup , probabilistic analysis , poissonization . |
recently the question was addressed whether one can construct initial states that affect the _ entire _ temporal evolution of critical non - equilibrium systems .
this is the case of systems that display a phase transition between an active state and a phase with infinitely many absorbing states .
non - universality of dynamic properties , associated with the initial configuration dependence of the survival probability of clusters started from a single active site , has been reported @xcite .
similar changes in the critical spreading behaviour have been observed by grassberger et al @xcite in a model where long - time memory effects are explicitly introduced . in the system studied by these authors , the susceptibility to the spreading of an active agent changes after the first encounter , remaining constant afterwards . despite the observed non - universal dynamical critical behaviour , susceptibility in the first encounter does not affect the critical point location .
grassberger et al argue that these results apply to models with multiple absorbing states where an effective memory - dependent susceptibility is present .
however , different behaviour is predicted in the case of slowly decaying memory effects , in which case both the value of the critical point and the exponents are expected to depend on the initial state .
a dependence on the initial configuration has also been found @xcite in the case of long - range spatial correlations , in which case the dynamical critical exponents change continuously as a function of initial correlation length .
the critical relaxation from an initial homogeneous state in systems with multiple absorbing states was recently examined from a field theoretical ( langevin equations ) approach @xcite .
the evolution equation for the order parameter density was found to include a memory term which is not present in the simpler case of directed percolation ( dp ) . in the present study
we have investigated memory effects in two 1-d models with multiple absorbing states , the pcp @xcite and ttp @xcite models . in both models there
is a non - order parameter field , dynamically coupled to the order parameter field , that gives rise to an effective susceptibility for spreading .
we show that the non - ordering field relaxes to its steady state value by the same power - law time dependence as the order parameter field and this is characterised by the natural , long - time behaviour exponent of the density decay of a dp process .
this is clear evidence of slowly decaying memory .
this power - law boundary condition ( in time ) is similar to that of the long - range power - law boundary condition ( in space ) of @xcite and we can see the emergence of continuously changing dynamical critical exponents .
the small shift of the critical point as a function of initial conditions , shown by high precision simulations for the pcp model , is in agreement with the above arguments .
damage spreading ( ds ) simulations invented in biology @xcite and later in physics @xcite are useful to show the stability of the systems with respect to small perturbations .
the spreading behaviour has been shown to be sensitive to the dynamics leading to the same steady states .
an `` objective '' definition of ds has been proposed @xcite according to which the phase diagrams of the steady states of non - equilibrium models can be divided to sectors in which all , none or parts of the physically possible dynamical rules generate stable damages .
the phase transitions between the phases can be continuous and usually belong to the dp universality class @xcite .
however , if the damage variables possess conservation and the ds absorbing states exhibit symmetries - which usually happens when the ds transition point coincides with the ordinary critical point the ds transition can belong to a different universality class @xcite .
the ds transition can not be in the passive phase of the replicas , but if it occurs in the active phase , the fluctuating replicas at the ds absorption point exclude the non - dp ds behaviour @xcite . we have investigated the ds properties of the pcp and ttp models ; the ds transition is shown to coincide with the ordinary critical point , and the non - universal spreading exponents
have been inherited as well . in section @xmath0
, we give a brief introduction to the pcp and ttp models .
time - dependent and critical relaxation studies are described in @xmath1 , whereas section @xmath2 is devoted to damage spreading simulations . some comments and conclusions
are presented in @xmath3 .
both of these models have a single control parameter @xmath4 and qualitatively similar phase diagrams , displaying an active state ( @xmath5 in case of pcp , @xmath6 for ttp ) and ( infinitely many ) absorbing phases ( @xmath6 in case of pcp , @xmath5 for ttp ) . in the ttp model ,
each site may be vacant , single or doubly ( active ) occupied , and this can be described by a @xmath7-state variable @xmath8 . in each time step , a site is chosen at random . in the absence of active sites , the dynamics is indeed trivial : if @xmath9 ( or @xmath10 ) , then @xmath11 with probability @xmath4 ( or @xmath12 ) . the system relaxes exponentially to a steady state where a fraction @xmath4 of sites have @xmath13 and the others are vacant . if @xmath14 , then @xmath15 if @xmath16 and @xmath17 are both @xmath18 and @xmath19 if only one of the nearest neighbours of site @xmath20 ( @xmath21 or @xmath22 ) has @xmath23 , in which case @xmath24 .
as can be easily seen , the number of active sites either decreases or remains the same in all processes other then @xmath25 ; the frequency of these processes depends on the concentration of 1-s , which is controlled by the parameter @xmath4 . any configuration consisting of only 0-s or 1-s is absorbing in what concerns the active sites .
the absorbing states in this model are fluctuating - in the respective sector of phase space , ergodicity is not broken . as we show below ,
the dynamics of 1-s is however strongly affected by the presence of active sites . at the critical point
, the concentration of 1-s relaxes to its steady state value ( equal to @xmath26 ) by a power - law .
the pcp is a @xmath27-state variable model with multiple absorbing states , each one of them completely frozen in time , contrary to what happens in the ttp . in the pcp ,
nearest - neighbour pairs of particles ( dimers or active sites ) annihilate each other with probability @xmath4 or create , with probability @xmath12 , a particle at one of the adjacent ( vacant ) sites to the dimer .
dimers can not be generated spontaneously and therefore play the role of the 2-s in the ttp .
there is a natural configuration to which the system at criticality evolves after all activity has died out ; the relationship between the natural particle density and @xmath28 is not a simple one unlike the ttp case . as shown below , the relaxation to this natural state is a slow process - a power - law in time is also found in this case and slowly - decaying memory effects arise as a result of the coupling between the local density of dimers and the local density of isolated occupied sites .
time dependent simulations @xcite have become an effective tool to explore dynamical critical exponents of systems at non - equilibrium phase transition points .
the simulations are started from a single active seed embedded in a sea of inactive sites and followed up to some @xmath29 time such that the cluster size can not exceed the system size @xmath30 .
the quantities usually investigated are the mean number of active sites @xmath31 ( pairs in case of pcp and 2-s in case of ttp model ) averaged over all trial samples , the survival probability @xmath32 of the clusters and the mean spreading size @xmath33 of the surviving clusters . at the critical point and for asymptotically
long times these quantities exhibit power - law behaviour like @xmath34 @xmath35 @xmath36 which define the exponents @xmath37 , @xmath38 and @xmath39 respectively .
the cluster size exponent @xmath39 characterising the linear scale is related to the anisotropy exponent of the system @xmath40 by @xmath41 .
the order - parameter density inside the surviving clusters can be expressed in terms of these exponents as @xmath42 and is expected to show the same long - time decay as from an arbitrary bulk configuration with @xmath43 , i.e. @xmath44 where @xmath45 is the steady state order - parameter exponent .
exponents @xmath37 and @xmath38 are found to depend on the initial concentration of particles @xmath46 , and are related to static exponents @xmath45 and
@xmath47 by the hyperscaling relation @xcite @xmath48 seed growing simulations for the pcp and ttp models have been carried out up to @xmath49 time steps for @xmath50 trial runs .
we measured the order parameter density @xmath51 as well as the relaxation of the non - order field density @xmath52 towards the natural values @xmath53 of the models . in case of the pcp we used @xmath54 @xcite while for the ttp model @xmath55 @xcite .
the densities were measured inside the `` infected '' regions of surviving clusters only . to estimate the critical exponents and the transition points together
, we determined the local slopes of the scaling variables .
for example , in the case of the order parameter density we computed @xmath56 \over \log(m)}\ ] ] with @xmath57 .
when @xmath58 , one should see a horizontal straight line as @xmath59 .
the off - critical curves should possess curvature : curves corresponding to @xmath60 should veer upward , curves with @xmath61 should veer downward .
figures [ pcp - tds-1 ] and [ pcp - tds-2 ] show the local slopes of @xmath62 for the pcp in case of @xmath63 and @xmath64 respectively . for the order parameter density we obtained the same results within numerical accuracy . as one can read off , the particle density exhibits long - time power - law behaviour with dp exponent , but the critical point is slightly lower in case of @xmath63 than in case of @xmath64 .
[ pcp - tds-1 ] this however agrees with the slow - relaxing susceptibility picture , because @xmath63 is smaller than the natural value @xmath65 and increases very slowly in the bulk , therefore a smaller annihilation probability is enough to drive the system to the absorbing state .
this dependence can also be the result of the lattice version of the pcp model , where , in contrast to the field theory where there is always some finite density of fields , the isolated 1-s are frozen .
both densities exhibit the same exponents in good agreement with @xmath66 of the @xmath67 d dp class .
the survival probability exponent measured in our high precision simulation is in agreement with the value of @xcite within numerical accuracy . in the critical point estimates
we could see a tiny difference depending on whether we measured the survival probability or the density in the same simulation . namely the @xmath28 appears to be nearer to the results of @xcite if we estimated it from the @xmath32 data ( @xmath68 instead of @xmath69 in case of @xmath63 and @xmath70 instead of @xmath71 in case of @xmath64 ) .
this small offset can be understood on the basis of our density measuring method plus the slow , frozen relaxation in pcp that seems to cause a crossover effect .
the density has been averaged in the infected regions , where frozen `` islands '' can appear which do nt evolve at all but keep the non - natural densities for long times .
therefore we overestimate the size of the region in which densities really relax . to verify this picture
we have performed simulations where we averaged the 1 s and 2 over fixed size . in this case
we started the pcp process from randomly distributed pairs in a system of size @xmath72 and followed the evolution up to @xmath73 .
in cases of @xmath63 and @xmath74 we found that the critical point is about @xmath75 nearer to the @xmath32 and the @xmath31 ds results .
the scaling exponent was again dp like : @xmath76 .
[ pcp - tds-2 ] in the case of ttp model we have performed simulations for @xmath77 , @xmath78 and @xmath79 initial particle densities . in this model
, we did not observe any shift of @xmath28 , as figure [ ttp - tds-1 ] shows , and all the densities scale with the @xmath66 exponent .
[ ttp - tds-1 ] the similar power - law behaviours of @xmath51 and @xmath80 can be understood if one considers the coupled langevin equations describing the time evolution of the processes .
for example , in case of the pcp model , they look like @xcite : @xmath81 @xmath82 where @xmath83 and @xmath84 are gaussian uncorrelated noise terms proportional to @xmath85 .
one can see that the equations are coupled strongly by @xmath86-terms ; since the right hand sides of the equations contain the same powers of the scaling fields , the time derivatives are expected to have the same scaling too . in reference
@xcite it was argued that @xmath87 ( @xmath88 is the natural concentration in this formalism ) may be taken as an approximate solution of equation ( [ eq8 ] ) , in which case the @xmath89 cross - term in ( [ eq9 ] ) has the form @xmath90 the power - law time dependence of @xmath91 is obvious , because an exponential relaxation to @xmath92 would just give in ( [ eq9 ] ) a term similar to @xmath93 that can shift the critical point but not the critical indices ( from the dp values ) .
the long - range scaling behaviour of the non - order density suggests that the @xmath91 field possesses critical fluctuations .
to test this , we have performed steady state simulations as well .
we have measured the 0 , 1 and 2 densities in case of the ttp model just above the critical point .
we considered @xmath94 systems and let them evolve from random initial conditions with @xmath4 slightly above @xmath95 ; about 40000 mc lattice updates were necessary to reach the steady state .
as one can see in figure [ steady ] , least - square fits of @xmath96 v.s .
@xmath97 resulted in regular @xmath98 scaling exponent @xmath99 @xcite .
the other two densities ( 0-s and 2-s ) exhibited the same steady state exponents too . for the fluctuation of @xmath100 we observed the scaling @xmath101 with @xmath102 exponent , which agrees with dp universality class value again @xcite .
the critical behaviour of the @xmath91 field demands that extra care is taken when dealing with truncated versions of equations ( [ eq8 ] ) and ( [ eq9 ] ) . a numerical integration of ( [ eq9 ] ) , including the non - markovian term ( [ eq11 ] ) , was carried out by lopez and muoz @xcite and revealed that the presence of the memory term is responsible for scaling up to some time with non - universal values of @xmath37 and @xmath38 .
however , these authors did not find the linear relation between the shift from the dp values ( @xmath103 and @xmath104 ) and @xmath105 that our results for pcp show ( figure [ pcp_res ] ) and was also found in previous ttp studies @xcite .
we think one has to take into account the omitted terms in eq .
( [ eq10 ] ) , which we have shown to exhibit power law in time and therefore give a relevant contribution to the renormalization of @xmath106 .
we have also investigated the early stages of the relaxation of the order parameter .
+ in reference @xcite it was suggested that in the short time regime ( @xmath107 ) one might observe the dynamic percolation scaling of grassberger @xcite . early time scaling - currently referred to as _ critical initial slip _ - was introduced by jansen et al @xcite and recently investigated by van wijland et al @xcite for a reaction - diffusion model with two kinds of particles , using rg analysis .
they found non - universal dependence ( on the initial particle distribution ) of the slip exponent in the case of unequal diffusion coefficients .
+ the system was prepared by adding a few 2-s ( uniform density @xmath108 ) to a random uniform background of 0-s and 1-s ( density @xmath109 ) .
the system evolution was recorded up to @xmath110 and averages were performed over independent runs ( @xmath111 typically ) .
log - log plots of @xmath112 - see fig .
[ slip ] - show a linear region ( lasting for @xmath113 for @xmath114 equal to a few percent ) with a slope that is independent of @xmath114 but depends on @xmath115 .
this may be evidence for the critical initial slip with a non - universal slip exponent @xmath116 equal to that slope . unlike the dp case , where @xmath117 @xcite , the values we found for @xmath116 differ significantly from the corresponding @xmath37 values .
the damage spreading simulations have been initialised by two replicas of states with identical , but random uniform distribution of single 1-s of given concentrations .
then a seed ( a pair in case of pcp and a 2 in case of ttp ) is added to each replica such that they become nearest neighbours and the initial difference is 2 ( see fig.[dsini ] ) .
the order parameter characterising the damage is the hamming distance between replicas @xmath118 where @xmath119 denote the pairs in case of pcp and the variable 2-s in case of the ttp model . at the ds critical point ( @xmath120 )
we expect that the order parameter scales as @xmath121 similarly the survival probability of damage variables behaves as @xmath122 and the average mean square distance of damage spreading from the center scales as @xmath123 averages were performed over @xmath124 independent runs for each value of @xmath4 in the vicinity of @xmath120 ( but for @xmath125 only over the surviving runs ) .
the @xmath29 was @xmath126 in these simulations .
figures [ dam_0_eta ] , [ dam_2_eta ] , [ dam_242_eta ] and [ dam_432_eta ] show the local slopes results of the hamming distance for the pcp model and different initial concentration of 1-s . the @xmath120 transition points are found to coincide with the ordinary critical points of the replicas within numerical accuracy .
a small , but monotonic tendency in the variation of @xmath120 ( as in case of @xmath28 ) with initial conditions can be observed in table i : ._damage spreading simulation results in the pcp model _ [ cols="^,^,^,^,^",options="header " , ] the ds critical @xmath37 and @xmath38 exponents show the same non - universal behaviour as the corresponding ordinary critical exponents ( see also @xcite ) and coincide with them within numerical precision .
the exponent @xmath39 is constant within numerical accuracy . in the case of @xmath127 , we performed runs with uniform initial distributions and with system generated configurations of isolated 1-s with the same average concentration .
the latter , non - uniform distribution was generated by letting a single replica to run at @xmath128 until it reached the absorbing state ; then the infected area of the system was used as initial state for ds simulations , similarly to what was done in ref .
we do nt see significant differences ( fig .
[ dam_242_eta ] ) between the two cases for @xmath129 large , both of them result in exponents in agreement with the best dp class @xmath130 value @xcite . in case of the ttp model we performed ds simulations for @xmath77 only .
we could see analogous ds behaviour as in the case of pcp .
again the critical point and exponents coincide with the corresponding critical values .
similarly to what has been observed in models that belong to the pc universality class @xcite , we can also conclude that if the ds transition point coincides with @xmath28 the scaling behaviour is inherited .
+ an interesting implication of this result can be stated exploiting the possible mapping of these models to soc models @xcite .
the corresponding critical sandpile models are not chaotic in the sense that the avalanches ( or clusters ) arising from the dropping of pairs ( or seeds ) to the lattice result in trajectories with power - law increasing differences only .
this does not exclude the possibility that other soc models generated by the way of ref .
@xcite are chaotic , since if @xmath120 happens to be in the active phase , the perturbations in the soc model generate differences that increase faster than a power law .
two representatives of systems with a continuous phase transition to infinitely degenerate absorbing states ( pcp and ttp models ) have been investigated numerically in @xmath131 . in order to clarify the influence of the initial condition on the dynamic properties , we have performed time - dependent simulations and analysed the evolution of both the order parameter and the non - order field densities .
we gave numerical evidence that the non - order field is in a critical state simultaneously with the order parameter field .
the isolated particles density exhibits a continuous phase transition with dp exponents to a non - absorbing state , therefore its fluctuations can not be neglected when one tries to understand the non - universal behaviour of critical exponents .
due to the dynamic coupling between the two fields , the slow ( power - law ) decay of the background particle density induces a long - time memory of the susceptibility to spreading of the order parameter @xcite .
we have found that the estimates of @xmath28 obtained from the time dependence of the survival probability are consistent with those given by ds studies ; a small shift as a function of @xmath46 is exhibited , which we interpret as due to the slowly decaying memory @xcite .
the study of the density in seed - growing simulations produced values of @xmath28 slightly off the former ones ; we think this is probably a crossover effect .
we do nt see such a @xmath28 shift for the ttp model so this maybe specific of the pcp model , connected to the non - ergodicity of its absorbing states . in both models
the critical exponents @xmath37 and @xmath38 have been found to behave linearly as a function of the initial particle concentration .
preliminary studies of the early time critical regime suggest the existence of a ( short ) initial slip regime characterised by an exponent which depends on the initial particle concentration @xmath46 , in agreement with rg predictions for a similar model @xcite .
this topic needs however additional investigation .
the damage spreading investigations have shown that the ds point coincides with the critical point and so the critical indices `` inherit '' the same non - universal scaling behaviour similarly what was found in an earlier study @xcite .
we thank r. dickman for very stimulating correspondence .
+ support from the nato grant crg-970332 is acknowledged .
this work was partially financed by praxis xxi ( portugal ) within project praxis/2/2.1/fis/299/94 .
g. dor gratefully acknowledges support from the hungarian research fund otka ( nos . t025286 and t023552 ) .
+ the simulations were performed partially on the fujitsu ap-1000 + and ap-3000 parallel supercomputer . | we have investigated the effect of the initial condition on the spreading exponents of the one - dimensional pair contact process ( pcp ) and threshold transfer process ( ttp).the non - order field was found to exhibit critical fluctuations , relaxing to its natural value with the same power - law as the order parameter field .
we argue that this slow relaxation , which was not taken into account in earlier studies of these models , is responsible for the continuously changing survival probability exponent .
high precision numerical simulations show evidence of a ( slight ) dependence of the location of the transition point on the initial concentration , in the case of pcp .
the damage spreading ( ds ) point and the spreading exponents coincide with those of the ordinary critical point in both cases . |
rare decays due to the flavor changing neutral current @xmath2 transitions constitute one of the most important classes of decays in carefully checking the predictions of the standard model ( sm ) at tree level , since they are forbidden in sm at loop level . in sm
the flavor changing neutral current ( fcnc ) processes @xmath3 proceed through the electroweak penguin and box diagrams .
these decays are also very suitable in looking for new physics ( np ) beyond the sm , via contributions of the new particles to the penguin and box diagrams , that are absent in the sm . the @xmath4 decay has been observed in @xcite .
moreover , the forward
backward asymmetry has been measured in @xcite . the longitudinal polarization and forward
backward asymmetry of @xmath4 and the isospin asymmetry of @xmath5 and @xmath6 are also measured by babar collaboration in @xcite and @xcite , respectively .
the experimental results are more or less in agreement with the predictions of sm . however , the precision of experiments is currently too low to make the final conclusion
. the situation should considerably be improved at lhcb .
the radiative decays of @xmath0 meson , involving @xmath7 , where @xmath8 is the orbitally excited state , is observed by belle .
the other radiative and semileptonic decay modes involving @xmath7 and @xmath9 are hopefully expected to be measured soon .
similar to the @xmath4 decay the @xmath10 decay is also a very good object for probing the new physics effects beyond the sm . here
the problem becomes more sophisticated due to the mixing of @xmath11 and @xmath12 state .
the physical states @xmath7 and @xmath9 are determined by [ eaasy01 ] ( l k_1(1400 ) .
+ k_1(1270 ) . )
= ( lr & - + & ) ( l k_1a . + k_1b . ) . in the present work we calculate the tensor form factors for the @xmath13 transition in the framework of the light cone qcd sum rules method ( lcsr ) ( for more about lcsr see @xcite ) .
the paper is organized in the following way . in section 2
we derive the lcsr for the tensor form factors describing the @xmath14 transition .
section 3 is devoted to the numerical analysis of the sum rules for the form factors .
we also summarize our results in this section .
the @xmath15 decay is described by @xmath16 transition at quark level .
the effective hamiltonian responsible for the @xmath3 transition is given by , [ eaasy02 ] h = - 4 g_f v_tb v_ts^ * _ i=1 ^ 10 c_i ( ) o ( ) , where the form of the local wilson operators @xmath17 is given in @xcite .
this effective hamiltonian leads to the following result for the @xmath3 decay amplitude [ eaasy03 ] m v_tb v_ts^ * \{c_9^eff |s _ ( 1-_5 ) b | _ + c_10 |s _ ( 1-_5 ) b | _ _ 5 + 2 m_b q^2 c_7 |s i _ q^(1 + _ 5 ) | _ } , where the wilson coefficient @xmath18 , with @xmath19 , contains both the perturbative and the long distance contribution parts .
the explicit expression of @xmath20 , @xmath21 , @xmath22 and @xmath23 are given in @xcite .
the long distance effects generated by the four quark operators with the @xmath24quark have recently been calculated for the @xmath25 and @xmath26 decays in @xcite and it is obtained that below the charmonium region of @xmath27 this effect can change the value of @xmath21 around 5% and 20% for @xmath28 and @xmath29 transitions , respectively .
similar calculations for @xmath30 transition has not yet been calculated . for simplicity , in the following discussions
we denote @xmath31 and @xmath32 as @xmath8 .
it follows from eq.([eaasy03 ] ) that for the calculation of the @xmath33 transition , the matrix elements @xmath34 and @xmath35 are needed . for the @xmath36 transition
, these matrix elements are defined in terms of the form factors as follows : [ eaasy05 ] k_1(p , ) |s _ ( 1-_5 ) b b(p_b ) - i 2 m_b+m_k_1 _ ^
( ) * p_b^p^a^k_1 ( q^2 ) + + 2 m_k_1 ( ^()*p_b ) q^2 q _ , + + [ eaasy06 ] k_1(p , ) |s _ q^(1+_5 ) b b(p_b ) 2 t_1^k_1 ( q^2 ) _
^ ( ) * p_b^p^ + i t_2^k_1 ( q^2 ) + i t_3^k_1 ( q^2 ) ( ^ ( ) * q ) , where @xmath37
. there are the following relations between the form factors : [ eaasy07 ] v_3^k_1(q^2 ) v^k_1_1(q^2 ) - m_b - m_k_1 2 m_k_1 v_2^k_1(q^2 ) , + v_3^k_1(0 ) v^k_1_0(0 ) , , + t_1^k_1(0 ) t^k_1_2(0 ) . to be able to calculate the form factors responsible for the @xmath33 transition we consider the following two correlation functions : [ eaasy08 ] _
i d^4x e^iqx k_1 ( p , ) t\{|s(x ) _ ( 1-_5 ) b(x )
|b(0 ) i _ 5 d(0 ) } 0 , + [ eaasy09 ] _
i d^4x e^iqx k_1 ( p , ) t\{|s(x ) _
|b(x ) b(0 ) i _ 5 d(0 ) } 0 . in order to construct the sum rules for the form factors responsible for the @xmath36 transition these correlation functions
should be calculated in two different languages , in terms of hadrons and quark and gluon degrees of freedom .
the calculation of the correlation function in terms of quark and gluon degrees of freedom is carried out at virtualities @xmath38 and @xmath39 . using the operator product expansion ,
the sum rules are obtained by equating these two representations through the dispersion relations .
phenomenological parts of the correlation functions ( [ eaasy08 ] ) and ( [ eaasy09 ] ) can be obtained by inserting complete set of hadrons with the same quantum numbers as the interpolating current , and separating the ground state one can easily obtain [ eaasy10 ] _
-k_1 ( p , ) |s _ ( 1-_5 ) b b(p_b ) b(p_b ) |b i _ 5 d 0 p_b^2 - m_b^2 + , + [ eaasy11 ] _ - k_1 ( p , ) |s _ b b(p_b ) b(p_b ) |b i _ 5 d 0 p_b^2 - m_b^2 + , where @xmath40 " describes the contributions coming from higher states and continuum , and the matrix element @xmath41 is given in eq .
( [ eaasy05 ] ) . the second matrix element in eq .
( [ eaasy10 ] ) is expressed in the standard way [ eaasy12 ] b(p_b ) |b i _ 5 d 0 = f_b m_b^2 m_b , where @xmath42 is the @xmath0decay constant and @xmath43 is the @xmath44quark mass .
the matrix element @xmath45 is defined as [ eaasy13 ] k_1 ( p , ) |s _ b b(p_b ) -i a(q^2 ) [ _ ^ ( ) * ( p+p_b)_- _ ^ ( ) * ( p+p_b ) _ ] + i b(q^2 ) ( _ ^ ( ) * q_- _ ^ ( ) * q _ ) + i ( p_q_- p_q _ ) .
contracting eq .
( [ eaasy13 ] ) with the momentum @xmath46 and using the relation [ nolabel ] _ _ 5 = - i2 _
^ , the following relations among @xmath47 and @xmath48 can easily be obtained : [ eaasy14 ] t_1^k_1(q^2 ) a(q^2 ) , + t_2^k_1(q^2 ) a(q^2)- q^2m_b^2-m_k_1 ^ 2 b(q^2 ) , + t_3^k_1(q^2 ) b(q^2 ) + c(q^2 ) .
using eqs .
( [ eaasy12 ] ) and ( [ eaasy13 ] ) , for the phenomenological parts of the correlation functions we get [ eaasy15 ] _
-f_b m_b^2 m_b 1p_b^2 - m_b^2 \ { - 2i m_b^2 - m_k_1 ^ 2 _ ^ ( ) * p_b^p^a^k_1(q^2 ) + + 2 m_k_1 ^ ( ) * q q^2 q _ } , + + [ eaasy16 ] _ - f_b m_b^2 m_b 1p_b^2 - m_b^2 \ { - i a(q^2 ) ( _ ^ ( ) * p_- _ ^ ( ) * p _ ) + i b(q^2 ) ( _ ^ ( ) * q_- _ ^ ( ) * q _ ) + 2i c(q^2 ) ^ ( ) * q m_b^2-m_k_1 ^ 2 ( p_q_- p_q _ ) } , where @xmath49 .
we now proceed to calculate the theoretical part of the correlation functions .
the calculation is performed by using the background field approach @xcite . in the large virtuality region , where @xmath50 and @xmath51 , the operator product expansion is applicable to the correlation functions . in light cone sum rules the method is based on the expansion of the non local quark
antiquark operators in powers of the deviation from the light cone . in obtaining the expression of the correlation functions the propagator of heavy quark and the matrix elements of the non local operators @xmath52 and
@xmath53 between the vacuum and axial vector meson are needed , where @xmath54 are the dirac matrices ( in our case @xmath55 or @xmath56 ) , and @xmath57 is the gluon field strength tensor .
the expression of the heavy quark operator is given in @xcite [ eaasy17 ] s_q = s_q^free ( x ) + ig_s e^-ikx du , where @xmath58 is the free quark operator and we adopt the convention for covariant derivative @xmath59 .
two particle distribution amplitudes for the axial vector mesons are presented in @xcite [ eaasy18 ] k_1(p , ) |s_(x ) q_(0 ) 0 -i4 du e^iupx + & & \ { f_k_1 m_k_1 + f_^a + & + & o(x^2 ) } _ , where [ eaasy1818 ] |g_3 ( u ) g_3(u ) + _ - 2 g_^(a ) ( u ) , + @xmath60 are the twist2 , @xmath61 , @xmath62 , @xmath63 and @xmath64 are twist3 , and @xmath65 and @xmath66 are twist4 functions .
the three particle distribution amplitudes are define as [ eaasy19 ] k_1(p , ) |s(x ) _ _ 5 g_s g _
( ux ) q ( 0 ) 0 p_(p_^()*_- p_^ ( ) * _ ) f_3 k_1^a a + , + k_1(p , ) |s(x ) _ g_s _ ( ux ) q ( 0 ) 0 i p_(p_^()*_- p_^ ( ) * _ ) f_3 k_1^v v + , where [ nolabel ] a = e^ipx ( _ 1 + u _ 3 ) a(_1,_2,_3 ) , and [ nolabel ] = _ 0 ^ 1 d _ 1 _ 0 ^ 1 d _ 2 _ 0 ^ 1 d _ 3 ( 1-_1-_2-_3 ) . here
@xmath67 , @xmath68 , and @xmath69 are the respective momentum fractions carried by @xmath70 , @xmath71 quarks and gluon in the meson . using these definitions , and after lengthy calculations for the theoretical parts of the correlation functions , we obtain [ eaasy20 ] & & correlation function = + & & 14 du \ { f_k_i m_k_i + & & + f^_k_i [ 12 _ ( u ) [ s_q ( /p /^ ( ) * - /^ ( ) * /p ) ] - 12 m_k_i^2 & & + h_3^(i)4 m_k_i^2 q _ [ s_q _ 5 ( /^ ( ) * _ - _ /^ ( ) * ) ] + 12 h_^(p ) m_k_i^2 ^ ( ) * _ [ s_q ] } + & & + 14 dv _
i 1\{m_b^2 - [ q + ( _ 1 + v _ 3 ) p]^2 } ^2 \ { 2 v pq ( /^ ( ) * /p ) } , + where [ nolabel ] s_q , with , q = q+pu , + _
a^(i ) _ 0^u dv , + f^(i ) _ 0^u f(v ) dv , + f^(ii ) _ 0^u dv _ 0^v
dv^f(v^ ) , and @xmath72 correspond to @xmath73 , respectively , @xmath74 is equal to @xmath55 or @xmath56 . after taking derivatives and traces , equating expressions of correlation functions ( [ eaasy15 ] ) , ( [ eaasy16 ] ) and ( [ eaasy20 ] ) , and performing borel transformation with respect to the variable @xmath75 in order to suppress the higher states and continuum contributions , one can obtain the sum rules for the transition form factors .
here we present the sum rules only for the tensor form factors , since @xmath76 , @xmath77 , @xmath78 and @xmath79 are calculated within the same framework in @xcite : [ eaasy21 ] a_i(q^2 ) - m_b^2 f_i 2 m_b^2 f_b e^m_b^2/m^2 _ 0 ^ 1 du \{1u e^-s(u)/m^2 - 14 e^-s(u)/m^2 m_i f_i u m_b f_i ( m_b^2 + q^2 ) g_^v ( u ) + ( u m^2 + s_0-q^2 ) } + e^m_b^2/m^2 _ 0 ^ 1 v dv e^-s(k)/m^2 d_1 d_3 f_3i^a a(_i ) + f_3i^v v(_i ) ( _ 1 + v _ 3)^2 + \ { - ( m_b^2 - q^2 ) ( ( _ 1 + v _ 3 ) m^2 + s_0-q^2 ) } , + + b_i(q^2 ) - m_b^2 f_i 2 m_b^2 f_b e^m_b^2/m^2 _ 0 ^ 1 du \ { 1u e^-s(u)/m^2 + e^-s(u)/m^2 14 m_i f_i m_b f_i g_^(v ) ( u ) + ( u m^2 + s_0-q^2 ) } + e^m_b^2/m^2 _ 0 ^ 1 v dv e^-s(k)/m^2 d_1 d_3 f_3i^a a(_i ) + f_3i^v v(_i ) ( _ 1 + v _ 3)^2 + \ { - ( m_b^2 - q^2 ) ( ( _ 1 + v _ 3 ) m^2 + s_0-q^2 ) } , + + c_i(q^2 ) e^m_b^2/m^2 _ 0 ^ 1 du \{1u e^-s(u)/m^2 [ ( 2 _ a^(i ) ( u ) - g_^(v ) ( u ) 2 ) + ( u m^2 + s_0-q^2 ) } , where @xmath80 or @xmath0 , and [ nolabel ] s(n ) = m_b^2 - ( 1-n ) q^2 n , k_1 + v _ 3 , , u_0 = m_b^2-q^2 s_0-q^2 . in these expressions , we neglect terms @xmath81 . using eq .
( [ eaasy14 ] ) one can easily obtain the corresponding sum rules for @xmath82 , @xmath83 and @xmath84 tensor form factors .
a few words about the form factors responsible for the @xmath36 transition , in the large recoil region in the heavy quark limit , are in order .
it can be shown that , similar to the @xmath85 ( vector meson ) case , all seven form factors responsible for the @xmath36 transition can be expressed in terms of the independent functions @xmath86 and @xmath87 , in the large recoil region and in the heavy quark limit .
indeed we find that , for to the @xmath36 transition these form factors can be written in terms of @xmath86 and @xmath87 as : [ eaasy22 ] v_0^k_1(q^2 ) ( 1 - m_k_1 ^ 2 m_b e ) _
^k_1(q^2 ) + m_k_1 m_b _
^k_1(q^2 ) , + v_1^k_1(q^2 ) ( 2 e m_b + m_k_1 ) _
^k_1(q^2 ) , + v_2^k_1(q^2 ) ( 1 + m_k_1 m_b ) _
^k_1(q^2)- m_k_1 e _
^k_1(q^2 ) , + a^k_1(q^2 ) ( 1 + m_k_1 m_b ) _
^k_1(q^2 ) , + t_1^k_1(q^2 ) _
^k_1(q^2 ) , + t_2^k_1(q^2 ) ( 1 - q^2 m_b^2 - m_k_1 ^ 2 ) _
^k_1(q^2 ) , + t_3^k_1(q^2 ) _
^k_1(q^2 ) - ( 1 - m_k_1 ^ 2 m_b^2 ) m_k_1 e _
^k_1 , where [ nolabel ] e = m_b^2 + m_k_1 ^ 2 - q^22 m_b , is the energy of @xmath8 meson .
explicit expressions of the functions @xmath86 and @xmath87 can be obtained following the same steps of calculation as is given in @xcite .
these expressions are quite lengthy , and therefore we do not present them here in this work .
our final remark in this section is as follows .
the physical @xmath7 and @xmath9 are the mixing states of @xmath31 and @xmath32 , and the form factors for the @xmath88 and @xmath89 transitions can be obtained from @xmath90 and @xmath91 transition form factors with the help of following transformations , [ nolabel ] ( c |k_1(1270 ) |s _ ( 1-_5 ) b b + ) = m ( c |k_1a |s _ ( 1-_5 ) b b + ) , [ nolabel ] ( c |k_1(1270 ) |s _ q^(1+_5 ) b b + ) = m ( c |k_1a |s _ q^(1+_5 ) b b + ) , where = ( lr & + & - ) . from the analysis of @xmath92 and @xmath93
decays , the mixing angle @xmath94 is obtained to have the value @xmath95 , where the minus sign is related to the relative phases of @xmath96 and @xmath97 .
the phases are fixed by adopting the conventions @xmath98 and @xmath99 .
in this section we present our numerical analysis of the sum rules for the form factors .
the main parameters entering to the sum rules for the @xmath100 transition form factors are the leptonic decay constant of the @xmath101 meson , das of the @xmath8 meson , the mass of the b
quark , borel parameter @xmath102 and and the continuum threshold @xmath103 .
the explicit expressions of the das for @xmath31 and @xmath32 mesons , as well as the parameters in the das are given in @xcite and their properties are presented them in the appendix .
few words about the value of the leptonic decay constant @xmath42 are in order .
it is shown in @xcite that the pole mass of the @xmath44quark produces rather large higher order radiative nlo corrections in the results for @xmath42 .
moreover , it is shown in this work that in the @xmath104 scheme the higher order corrections are under control , and therefore , the predictions for @xmath42 is more reliable . for this reason in further numerical analysis
we use the value @xmath105 as is obtained in @xcite within the framework of @xmath104 scheme , at @xmath106 scale . for the @xmath104 mass @xmath107
we use @xmath108 which is obtained from @xmath109 @xcite . for the strange quark mass we use the value @xmath110 , which is obtained from the analysis of pseudoscalar qcd sum rules @xcite . as has already been noted ,
the sum rules for the form factors also contain two auxiliary parameters : the borel mass parameter @xmath102 and the continuum threshold @xmath103 , and obviously , any physical quantity should be independent of them .
for this reason , we try to find such regions these parameters where physically measurable quantities are independent of them . in determining the value of the continuum threshold @xmath103
, we require that the prediction of the mass sum rules for the @xmath0 meson coincides with the experimental data .
we also require that @xmath103 must not be far from the reliable " region , i.e. , it should be below the next resonance in this channel .
these conditions lead to the result that , the expected values of @xmath103 lie in the interval @xmath111 .
the upper bound of @xmath102 is determined by demanding that the total contributions of higher states and continuum threshold should be less than half of the dispersion integral .
the lower limit of @xmath102 is found if we require that the highest power @xmath112 term contributes less than , say , 25% of the sum rules .
both these conditions are satisfied when @xmath102 varies in the region @xmath113 . in numerical calculations
we have used @xmath114 and @xmath115 .
unfortunately , the sum rules can not predict the dependence of the form factors on @xmath27 in the relevant physical region @xmath116 .
the sum rules results are not reliable when @xmath117 . in order to extend the results for the form factors coming from qcd sum rules predictions to cover the whole physical region , we look for a parametrization of the form factors in such a way that , the results obtained for the region @xmath118 can be extrapolated to whole physical region .
to extend our calculations to whole physical region , we use the @xmath119series parametrization ( for more about this parametrization , see @xcite and references therein ) , which is based on the analyticity of the form factors on @xmath27 . before presenting the series expansion of the tensor form factors of the @xmath36 transition , following the work @xcite , we present the helicity amplitudes as the linear combination of the form factors as are defined in eq .
( [ eaasy05 ] ) , which are more convenient for the analysis . after a simple calculation we obtain the following helicity amplitudes for the @xmath36 transition
( see also @xcite ) [ eassy23 ] h_0 ( q^2 ) t_2(q^2 ) - 2 m_k_1 ( m_b^2 - m_k_1 ^ 2 ) t_3 ( q^2 ) , + h_1 ( q^2 ) t_1(q^2 ) , + h_2(q^2 ) ( m_b^2 - m_k_1 ^ 2 ) t_2 ( q^2 ) , where @xmath120 , and subscripts @xmath121 and @xmath122 correspond to the longitudinal and linear combinations of the transversal polarizations , [ nolabel ] _
1,2^= 1 , of the virtual axial meson , respectively .
we define the following parametrization for the tensor form factors based on the z series expansion , [ eaasy24 ] h_0(q^2 ) _ k=0^n-1 _ k^(0 ) z^k , + h_1(q^2 ) _ k=0^n-1 _ k^(1 ) z^k , + h_2(q^2 ) _ k=0^n-1 _ k^(2 ) z^k , where @xmath123 is defined as , [ eaasy25 ] z(q^2,q_0 ^ 2 ) = - + , and @xmath124 and @xmath125 ) .
the parameter @xmath126 is chosen from the solution of the equation @xmath127 , which gives @xmath128 .
the mass of the resonances entering to the factor @xmath129 are @xmath130 for @xmath131 , and @xmath132 for @xmath133 , respectively ( see @xcite ) . the function @xmath134 is given in eq .
( 39 ) of the work @xcite . in order to perform the numerical analysis for the helicity amplitudes , and derive
the unitary bound one needs to calculate the two point correlation function of the tensor current .
this calculation is done in @xcite and we will use the results of this work . for the @xmath119series expansion parametrization , the unitarity constraint leads to the result [ eaasy26 ] _
k=0^n-1 \ { _ k^(0)2 + _ k^(1)2 + _ k^(2)2 } 1 . in figs .
( 1)(3 ) we present the fits of the series expansion parametrization to lcsr results for the helicity amplitudes @xmath135 , @xmath133 and @xmath136 , respectively . in these figures , we take into account the uncertainties coming from gegenbauer moments of the axial vector meson , mass of @xmath44 and @xmath70 quark , borel mass @xmath102 and the threshold parameter @xmath103 . in table 1
we present the values of the coefficients @xmath137 and @xmath138 , @xmath139 , entering to the series expansion of the helicity amplitudes @xmath135 , @xmath133 and @xmath136 , and the unitarity constraint @xmath140 given in eq .
( [ eaasy26 ] ) . due to the large cancellation among different terms in @xmath119series expansion
, we could not fix all coefficients @xmath141 from the fit .
therefore , we keep only the first two terms in the expansion , i.e. , we take @xmath142 ( see also @xcite ) .
@xmath143 our final remark is as follows : as has already been noted , in the numerical calculations we use @xmath144 . if @xmath145 @xcite had been used the values of all form factors presented in this work increase by a factor 1.4 . in summary , we calculate the tensor form factors of @xmath0 decays into @xmath146wave axial vector meson .
these form factors are relevant to the studies for the exclusive fcnc transitions .
the sum rules obtained could further be improved by including the @xmath147 corrections , as well as , improving the values of the input parameters involving the das .
this research of k. c. y. is supported in part by the national center for theoretical sciences and the national science council of r.o.c . under grant no .
nsc99 - 2112-m-003 - 005-my3 .
99 a. ishikawa _
et al . _ ,
belle collaboration , phys .
rev . lett . * 91 * , 261601 ( 2003 ) . b. aubert _
et al . _ ,
babar collaboration , phys .
lett . * 91 * , 221802 ( 2003 ) .
a. ishikawa _
_ , belle collaboration , phys .
lett . * 96 * , 251801 ( 2006 ) . b. aubert _
et al . _ ,
babar collaboration , phys .
d * 73 * , 092001 ( 2006 ) .
b. aubert _
et al . _ ,
babar collaboration , phys .
d * 79 * , 031102 ( 2009 ) .
b. aubert _
_ , babar collaboration , phys .
lett . * 102 * , 091803 ( 2009 ) .
i. i. balitsky , v. m. braun , v. kolesnichenko , nucl .
b * 312 * , 509 ( 1989 ) .
v. l. chernyak and i. r. zhitnitsky , nucl . phys .
b * 345 * , 137 ( 1990 ) .
g. buchalla , a. j. buras and m. e. lautenbacher , rev .
* 68 * , 1125 ( 1996 ) .
a. khodjamirian , t. mannel , a. a. pivovarov , and y m. yang , jhep * 089 * , 1009 ( 2010 ) .
i. i. balitsky , v. m. braun , nucl . phys .
b * 311 * , 541 ( 1989 ) .
v. m. belyaev , v. m. braun , a. khodjamirian and r. rckl , phys .
d * 51 * , 6177 ( 1995 ) .
k. c. yang,_et al .
_ , jhep * 051 * , 108 ( 2005 ) .
k. c. yang , nucl .
b * 776 * , 187 ( 2007 ) .
k. c. yang , phys .
d * 78 * , 034018 ( 2008 ) .
a. khodjamirian , t , mannel and n. offen , phys .
d * 75 * , 0754013 ( 2007 ) .
m. jamin , and b. o. lange , phys .
d * 65 * , 056005 ( 2002 ) . c. dominguez , int .
a * 25 * , 5223 ( 2010 ) .
a. bharucha , th .
feldmann , m. wick , jhep * 1009 * , 090 ( 2010 ) .
t. m. aliev , v. l. eletsky , sov .
phys . * 38 * , 936 ( 1983 ) .
the two parton chiral even lcdas are given by @xmath148 where @xmath149 , @xmath150 , and @xmath151 is the momentum fraction carried by the @xmath70 in the @xmath152 meson .
the two parton chiral odd lcdas are defined by @xmath153 where the functions @xmath154 , @xmath155 and @xmath156 are given in eq .
( [ eaasy19 ] ) . in su(3 ) limit , due to @xmath157-parity , @xmath158 , @xmath62 , and @xmath65 are symmetric ( antisymmetric ) under the replacement @xmath159 for the @xmath160 ( @xmath161 ) states , whereas @xmath162 , @xmath64 , and @xmath66 are antisymmetric ( symmetric ) .
up to twist3 , we adopt the following normalization conventions @xcite , @xmath163 for @xmath31 , but becomes @xmath164 for @xmath32 , where @xmath165 and @xmath166 are defined through @xmath167 with @xmath168 and @xmath169 . @xmath170 and @xmath171 are the g parity violating zeroth gegenbauer moments and vanish in the su(3 ) limit .
we use the twist2 distributions @xcite @xmath172~ , \label{eq : lcda-3p1-t2 - 1}\\ \phi_\perp(u ) & = & 6 u \bar u \left [ a_0^\perp + 3 a_1^\perp\ ,
\xi + a_2^\perp\ , \frac{3}{2 } ( 5\xi^2 - 1 ) \right]~ , \label{eq : lcda-3p1-t2 - 2}\end{aligned}\ ] ] for the @xmath31 and @xmath173~ , \label{eq : lcda-1p1-t2 - 1}\\ \phi_\perp(u ) & = & 6 u \bar u \left [ 1 + 3 a_1^\perp\ , \xi + a_2^\perp\ , \frac{3}{2 } ( 5\xi^2 - 1 ) \right]~ , \label{eq : lcda-1p1-t2 - 2}\end{aligned}\ ] ] for the @xmath32 , where @xmath174 . for the relevant three - parton twist3 chiral even lcdas , we use the contributions up to terms of conformal spin @xmath175 and take into account the corrections arising from the strange quark mass : @xmath176~ , \label{eq : lcda-3p1-t3 - 1}\\ { \cal v}&=&360\alpha_1\alpha_2\alpha_3 ^ 2 \big [ 1 + \omega^v_{k_{1a}}\frac{1}{2}(7\alpha_3 - 3)\big ] + 5040 ( \alpha_1-\alpha_2)\alpha_1\alpha_2\alpha_3 ^ 2 \sigma^v_{k_{1a}}~,\label{eq : lcda-3p1-t3 - 2 } \end{aligned}\ ] ] for the @xmath31 , and @xmath177 + 5040 ( \alpha_1-\alpha_2)\alpha_1\alpha_2\alpha_3 ^ 2 \sigma^a_{k_{1b}}~ , \label{eq : lcda-1p1-t3 - 1}\\ { \cal v}&=&5040 ( \alpha_1-\alpha_2)\alpha_1\alpha_2\alpha_3 ^ 2 + 360\alpha_1\alpha_2\alpha_3 ^ 2 \big
[ \lambda^v_{k_{1b}}+ \sigma^v_{k_{1b}}\frac{1}{2}(7\alpha_3 - 3)\big]~ , \label{eq : lcda-1p1-t3 - 2 } \end{aligned}\ ] ] for the @xmath32 , where @xmath178 s correspond to conformal spin 7/2 , while @xmath179 s and @xmath180 s are parameters with conformal spin 9/2 .
as the su(3)symmetry is restored , we have @xmath178s=@xmath180s=0 . for the relevant two
parton twist3 chiral even lcdas , we take the approximate expressions up to conformal spin @xmath175 and of order @xmath181 @xcite : @xmath182\xi(5\xi^2 - 3 ) \nonumber\\ & & { } -\frac{9}{2 } { a}_1^\perp \,\widetilde{\delta}_+\,\left(\frac{3}{2}+\frac{3}{2}\xi^2+\ln u + \ln\bar{u}\right ) - \frac{9}{2 } { a}_1^\perp\,\widetilde{\delta}_-\ , ( 3\xi + \ln\bar{u } - \ln u)~ , \label{eq : ga-3p1}\\ g_\perp^{(v)}(u ) & = & 6 u \bar u \bigg\ { 1 + \bigg(a_1^\parallel + \frac{20}{3 } \zeta_{3,k_{1a}}^a \lambda_{k_{1a}}^a\bigg ) \xi\nonumber\\ & & + \bigg[\frac{1}{4}a_2^\parallel + \frac{5}{3}\ , \zeta^v_{3,k_{1a } } \left(1-\frac{3}{16}\ , \omega^v_{k_{1a}}\right ) + \frac{35}{4 } \zeta^a_{3,k_{1a}}\bigg ] ( 5\xi^2 - 1 ) \nonumber\\ & & + \frac{35}{4}\bigg(\zeta_{3,k_{1a}}^v \sigma_{k_{1a}}^v -\frac{1}{28}\zeta_{3,k_{1a}}^a \sigma_{k_{1a}}^a \bigg ) \xi(7\xi^2 - 3 ) \bigg\}\nonumber\\ & & { } -18 \ , a_1^\perp\widetilde{\delta}_+ \ , ( 3u \bar{u } + \bar{u } \ln \bar{u } + u \ln u ) - 18\ , a_1^\perp\widetilde{\delta}_- \ , ( u \bar u\xi + \bar{u } \ln \bar{u } - u \ln u)~ , \label{eq : gv-3p1 } \end{aligned}\ ] ] for the @xmath31 , and @xmath183 \xi\left(5\xi^2 - 3\right ) \nonumber\\ & & { } + \frac{3}{16}\ , a_2^\parallel \left(15\xi^4 -6 \xi^2 -1\right ) + 5\ , \zeta^v_{3,k_{1b}}\lambda^v_{k_{1b}}\left(3\xi^2 -1\right ) \nonumber\\ & & { } + \frac{105}{16}\left(\zeta^a_{3,k_{1b}}\sigma^a_{k_{1b } } -\frac{1}{28 } \zeta^v_{k_{1b}}\sigma^v_{k_{1b}}\right ) \left(35\xi^4 -30 \xi^2 + 3\right)\nonumber\\ & & { } -15 { a}_2^\perp \bigg [ \widetilde{\delta}_+ \xi^3 + \frac{1}{2}\widetilde{\delta}_-(3\xi^2 - 1 ) \bigg ] \nonumber\\ & & { } -\frac{3}{2}\,\bigg[\widetilde{\delta}_+\ , ( 2 \xi + \ln\bar{u } -\ln u ) + \ , \widetilde{\delta}_-\,(2+\ln u + \ln\bar{u})\bigg](1 + 6a_2^\perp ) ~,\label{eq : ga-1p1}\\ g_\perp^{(v)}(u ) & = & 6 u \bar u \bigg\ { a_0^\parallel + a_1^\parallel \xi + \bigg[\frac{1}{4}a_2^\parallel + \frac{5}{3 } \zeta^v_{3,k_{1b } } \bigg(\lambda^v_{k_{1b } } -\frac{3}{16 } \sigma^v_{k_{1b}}\bigg ) + \frac{35}{4 } \zeta^a_{3,k_{1b}}\sigma^a_{k_{1b}}\bigg](5\xi^2 - 1 ) \nonumber\\ & & { } + \frac{20}{3}\ , \xi \left[\zeta^a_{3 , k_{1b } } + \frac{21}{16 } \bigg(\zeta^v_{3,k_{1b}}- \frac{1}{28}\ , \zeta^a_{3,k_{1b}}\omega^a_{k_{1b } } \bigg ) ( 7\xi^2 - 3)\right]\nonumber\\ & & { } -5\ , a_2^\perp [ 2\widetilde\delta_+ \xi + \widetilde\delta_- ( 1+\xi^2 ) ] \bigg\}\nonumber\\ & & { } - 6 \bigg[\ , \widetilde{\delta}_+ \ , ( \bar{u } \ln\bar{u } -u\ln u ) + \ , \widetilde{\delta}_- \ , ( 2u \bar{u } + \bar{u } \ln \bar{u } +
u \ln u)\bigg ] ( 1 + 6 a_2^\perp)~ , \label{eq : gv-1p1}\end{aligned}\ ] ] for the @xmath32 , where @xmath184 & @xmath185 & @xmath186 & @xmath187 & @xmath188 & @xmath189 & @xmath190 + @xmath31 & @xmath191 & @xmath192 & @xmath193 & @xmath194 & @xmath195 & @xmath196 + @xmath32 & @xmath197 & @xmath198 & @xmath199 & @xmath191 & @xmath200 & @xmath201 + | the tensor form factors of @xmath0 into p wave axial vector meson transition are calculated within light cone qcd sum rules method .
the parametrizations of the tensor form factors based on the series expansion are presented .
# 1#2#3 @xmath1 # 1#2#3 @xmath1 0^*0 5_5 o _ ^0 _ pacs number(s ) : 11.55.hx , 13.75.gx , 13.75.jz |
the computing requirements for high energy physics ( hep ) projects like the large hadron collider ( lhc ) @xcite at the european laboratory for particle physics ( cern ) in geneva , switzerland are larger than can be met with resources deployed in a single computing center .
this has led to the construction of a global distributing computing system known as the worldwide lhc computing grid ( wlcg ) @xcite , which brings together resources from nearly 160 computer centers in 35 countries .
computing at this scale has been used , for example , by the cms @xcite and atlas @xcite experiments for the discovery of the higgs boson @xcite . to achieve this and other results the cms experiment , for example , typically used during 2012 a processing capacity between 80,000 and 100,000 x86 - 64 cores from the wlcg .
further discoveries are possible in the next decade as the lhc moves to its design energy and increases the machine luminosity .
however , increases in dataset sizes by 2 - 3 orders of magnitude ( and commensurate processing capacity ) will eventually be required to realize the full potential of this scientific instrument .
building the software to run on and operate such a computing system is a major challenge .
the distributed nature of the system implies that ownership and control of the resources is also distributed , and thus the resources are by necessity heterogeneous in nature .
this heterogeneity appears both in terms of specific x86 hardware generations and in patch levels of the deployed linux operating systems .
as these resources are at times shared with other projects custom modifications of systems for hep - specific or experiment - specific reasons are in general not possible .
the very large number of cpu hours used also introduces significant reliability requirements on the software .
the software itself is non - trivial : each experiment typically is dependent on many millions of lines of code , written for the most part in c++ with contributions from up to a thousand physicists . given the upgrade and evolution plans for the lhc , these software projects , begun in the late 1990 s , will likely need to evolve with computing technology through the 2020 s .
such an environment however provides significant opportunities for innovation . in this paper
we examine one interesting technology , _ process checkpoint - restart _ , which has great potential for use in hep workflows in such a system .
we first describe the specific use cases of interest and the requirements to make the technology useful in the hep environment .
we then provide some benchmarks for the use of checkpoint - restart with the cms software on today s x86 - 64 processors . and finally , we examine aspects of this technology which are of interest given the possible evolution of processor technologies and resource availability in the coming years .
we look in particular at the use of checkpointing with architectures like intel s xeon phi , a member of intel s mic ( many integrated core ) architecture .
in many circumstances it is desirable to `` checkpoint '' the state of a unix process , or a set of processes , to disk with the possibility of restarting it at a later time .
there are a number of interesting use cases for this functionality : * debugging : * the very large number of jobs and cpu hours required for hep computing makes high reliability of the software quite important .
the distributed and heterogeneous nature of the computing system however makes debugging problems somewhat difficult , as the first step to resolving most code behavior problems is being able to reproduce the problem . while a traditional `` core '' file may provide information about the process state after a crash has happened , it does nt allow one to step through the program to see the behavior leading to the crash
. in the case where the crash happens after a job has run for many hours , reproducing a problem by rerunning it from the beginning can also be quite expensive . if however a job were to checkpoint its state from time to time , it would be possible to use the last checkpointed state before a crash to reproduce and replay the problem quickly .
* avoiding cpu - intensive initialization steps : * many applications are constructed such that they have cpu - intensive initialization steps .
examples include in - memory geometry construction from a simplified geometry description , physics cross section table calculations , etc .
typically this is done to allow generality of software implementation for multiple possible job configurations : it is easier to calculate quantities derived from job configurations on the fly at the beginning of each job than to store those quantities for a possibly infinite number of potential job configurations .
hep workflows are however constructed such that a particular job configuration may be used for a very large number of jobs , where the geometry or physics process configuration are the same and only random number seeds or input files change from job to job .
the result is that the same calculation is done in every single job instance in a particular workflow . in most cases where jobs run for a long time
this job initialization time is negligible relative to the total running time .
however two cases exist where job initialization time can be problematic .
first , very short duration jobs can sometimes be required for other software reasons or for operational reasons related to resource availability or input dataset structure . in this case
the overhead from long startup initialization times can be a significant fraction of the cpu utilization .
second , as will be described later , there are strong reasons to consider the use of multi - threaded applications in the future . in the case where the startup initialization itself can not be easily parallelized and will be executed sequentially on a single core
, the initialization itself may effectively idle a large number of other cores eventually needed for the event processing . in this case , a single instance of the job can be run and checkpointed just after the initialization phase .
that checkpoint can then be used to restart a much larger number of instances of the application in batch , with only minor reconfiguration to set input files and/or random number seeds .
the instances running in batch can thus avoid the startup cpu cost .
* allowing preemption during opportunistic resource use : * from time to time , it is possible to `` opportunistically '' use computing resources which belong to some other organization when the other organization does not have enough work to keep the processors fully utilized for some period of time . in some of these cases the period of time for which the resources are made available may not be well defined in advance and the resources may need to be handed back to their owner in an unscheduled fashion . in this case
it is useful to be able to `` preempt '' running opportunistic jobs , checkpoint their state to disk and restart them when opportunistic use is again possible .
* interactive `` workspaces '' : * in interactive programs , such as event displays and analysis tools , the user provides inputs which lead to particular state of the program at a given time . being able to save that state out , for example before going home for the day , and restart later is often desirable . *
very long running jobs : * in situations where jobs must run for an extremely long time , sometimes days or weeks , they can be sensitive to hardware or infrastructure failures or interfere with required site maintenance .
in these cases it can be quite useful to checkpoint periodically the program to avoid losing and needing to repeat the calculations from scratch after such failures .
* managing `` tails '' for multi - threaded applications * : several hep experiments are moving in the direction of multi - threaded frameworks , which ( initially ) process events on different threads . as the cpu time per event can vary significantly ( with long `` tails '' to the distribution ) at the end of the job , one thread may still be processing an event which takes a long time while the other threads / cores are idled .
one possibility for managing such situations would be to checkpoint the job with a single active thread and restart a number of such jobs at a later time together , to keep the full set of cpu cores active .
a rudimentary `` checkpoint - restart '' can sometimes be achieved in an application - specific fashion .
for example a typical hep event processing framework can be constructed to perform a simple `` checkpoint '' by flushing completed output events to its output file after every n input events have been processed .
in addition it is necessary to write some sort of `` metadata '' to track any other relevant internal state needed to restart the job , e.g. how far the job had progressed through its input events , the state of random number generators , etc . in this example
a `` restart '' would then be performed by restarting the framework and passing it information to allow it to reconfigure itself to match the state it was in at the time of the output checkpoint .
this requires however the addition and maintenance of dedicated code , both in the framework itself and externally in the workflow management system . in some cases , where third party libraries are used which also maintain state , it can be quite complex to truly restore the same state .
if the state is encapsulated within the code of the library , for software engineering reasons , it can also be impossible .
a much better , and more general , solution is true process - level checkpointing .
this is a technology which has existed for a long time , especially in high performance computing ( hpc ) and batch systems , however often the particular implementations are tied to specific environments .
thus the technology has not seen general use in hep high throughput distributed computing . in this paper
we examine the use of a transparent , user - level checkpointing package for distributed applications called distributed multithreaded checkpointing ( dmtcp ) @xcite .
the features of dmtcp make it more appropriate for deployment in the hep distributed computing environment .
dmtcp ( distributed multithreaded checkpointing ) is free , open source software ( http://dmtcp.sourceforge.net , lgpl license ) .
the dmtcp project traces its roots to late 2004 .
a key feature of dmtcp for use in the heterogeneous hep computing environment is that it works in user space , with no kernel - level modifications required . as such
it is works with a wide range of linux kernel versions .
it also works with multi - threaded applications and compression of output checkpoint files is possible
. its usage can be as simple as : .... dmtcp_launch ./myapp arg1 ... dmtcp_command --checkpoint [ from a second terminal window ] dmtcp_restart ckpt_myapp_*.dmtcp .... dmtcp is also `` contagious '' .
if a process begins under dmtcp control , then any child processes will also be under dmtcp control , and any remote processes ( spawned through `` ssh '' ) will also be under dmtcp control . at the time of checkpoint , a script , dmtcp_restart_script.sh , is written , and the script can restart all processes across all nodes for the given computation . the newly released dmtcp version 2.0 ( as of oct
. 3 , 2013 ) , supports dmtcp plugins to flexibly adapt to external conditions .
for example , the dmtcp plugin interface permits application - initiated checkpoints , as well as application - delayed checkpoints during critical operations .
alternatively , the interval flag of dmtp_launch permits automatic periodic checkpointing .
dmtcp plugins make it easier to use event hooks to detach such external resources as a database prior to checkpoint , and to reconnect during restart .
while dmtcp will save and restore the file offset of open files , event hooks make available an alternative of cleanly closing valuable files during checkpoint , and re - opening them during restart . in another example , dmtcp virtualizes network addresses to enable transparent migration to a new cluster .
finally , if a large region of memory is not actively used at the time of checkpoint , then the size of the checkpoint image can be considerably reduced .
an event hook allows the application to write zeroes into the inactive memory at checkpoint time , and dmtcp will then replace the zeroes by zero - fill - on - demand pages ( empty pages to be recreated on demand ) . among the contributed plugins for dmtcp
is the rm ( resource manager ) plugin , which supports use of dmtcp within the torque and slurm batch queues .
plugin support for additional batch queues is planned .
similarly , future support for the intel scif network is planned , allowing one to checkpoint a computation over a network of intel mic cpus .
the scif plugin will be based on the existing contributed plugin to support checkpointing over infiniband .
other contributed plugins support checkpointing a network of virtual machines .
virtual machines ease the job of deploying complex software .
to investigate the characteristics of dmtcp with hep software , we chose to make tests using the cms simulation application .
the machine used to do the test was a dual quad - core intel xeon l5520 operating at 2.27ghz , with 24 gb of memory .
tests were done with checkpoint files written to a local disk as well as the normal job output files .
the test job had no input event file .
however it reads conditions via a web squid from a remote database .
dmtcp version 1.2.7 and cmssw version @xmath0 , compiled with gcc version 4.7.2 , were used for the tests . for simplicity
the test job used with checkpointing was the only cpu - intensive process running on the machine at the time the tests were performed .
the tests were done using the external dmtcp coordinator trigger to produce a checkpoint , rather than the api .
the cms application generated simple minimum bias events and simulated them with geant4 @xcite .
the particular test job itself has a 2 minute initialization phase and then takes an average time per event of @xmath113 seconds .
the memory footprint is @xmath11 gb vsize ( 750 mb rss ) .
a typical uncompressed checkpoint takes @xmath11.5s and the resulting size on disk of the checkpoint file was @xmath1750 mb . when triggering checkpointing with the compression on , @xmath110s was required .
the checkpoint image was however significantly smaller , at only @xmath1220 mb . in both cases
no problems were seen in restarting the application from the checkpoint files .
the construction of the wlcg was greatly facilitated by the convergence around the year 2000 on commodity x86 hardware and the standardized use of linux as the operating system for scientific computing clusters .
even if multiple generations of x86 hardware ( and hardware from both intel and amd ) are provided in the various computer centers , this was a far simpler situation than the previous typical mix of proprietary unix operating systems and processors . until around 2005
, a combination of increased instruction level parallelism and ( in particular ) processor clock frequency increases insured that performance gains expected from moore s law would be seen by single sequential applications running on a single processor . the combination of linux , commodity x86 processors and moore s law gains for sequential applications made for a simple software environment .
however since around 2005 processors have hit scaling limits , largely driven by overall power consumption @xcite .
the first large change in commercial processor products as a result of these limits was the introduction of `` multicore '' cpus , with more than one functional processor on a chip .
at the same time clock frequencies ceased to increase with each processor generation and indeed were often reduced relative to the peak .
the result of this was one could no longer expect that single , sequential applications would run faster on newer processors .
however in the first approximation , the individual cores in the multicore cpus appeared more or less like the single standalone processors used previously .
most large scientific applications ( hpc / parallel or high throughput ) run in any case on clusters and the additional cores are often simply scheduled as if they were additional nodes in the cluster .
this allows overall throughput to continue to scale even if that of a single application does not .
it has several disadvantages , though , in that a number of things that would have been roughly constant over subsequent purchasing generations in a given cluster ( with a more or less fixed number of rack slots , say ) now grow with each generation of machines in the computer center .
this includes the total memory required in each box , the number of open files and/or database connections , increasing number of independent ( and incoherent ) i / o streams , the number of jobs handled by batch schedulers , etc .
the specifics vary from application to application , but potential difficulties in continually scaling these system parameters puts some pressure on applications to make code changes in response , for example by introducing thread - level parallelism where it did not previously exist .
there is moreover a more general expectation that the limit of power consumption on future moore s law scaling will lead to more profound changes going forward . in particular ,
the power hungry x86 - 64 `` large '' cores of today will likely be replaced by simpler and less power hungry `` small '' cores .
one example of such a technology is the intel mic architecture , as implemented in the intel xeon phi coprocessor card .
to test the use of checkpointing on the xeon phi , we used a beta version of geant4 version 10 which provides support for event - based multi - threaded applications .
we did not use the full cms simulation for this , but instead a simpler benchmark application ( fullcms ) which uses the actual cms geometry imported from gdml file .
the experimental setup was a standard intel xeon box with 32 logical cores , equipped with an intel xeon phi 5120p coprocessor card . in our test
the application is started and checkpoints are triggered in different moments . then we have restarted the application from the checkpoint file and we have verified that the application resumes correctly from the saved state .
this condition is verified checking that the final output of the simulation equals the original one .
a comparison of random number engine status at the end of the job . since in a geant4 application
there is a very large use of random numbers ( billions of calls to the engine in typical application ) , if the status of the application from the checkpoint file does not match exactly the original one the sequence of the random number calls will be different , producing a different final state of the random number engine .
a typical geant4 application with multi - thread support consists of a sequential part in which the geometry of the experimental setup is built in memory and physics processes are initialized ( e.g. material - dependent cross sections are calculated ) .
threads are then spawned , initialized and they start to simulate events independently ( see figure [ fig : geant4-ckpt ] ) . to reduce the total memory footprint the most memory consuming objects are shared between threads .
the need for synchronization between threads ( locks , barriers ) is minimized since only read - only objects are shared .
we have performed tests to verify the correct behavior of dmtcp for two of the use cases described in section [ sec : usecase ] . in the first case
we have instrumented the geant4 application code with a call to dmtcp to trigger a checkpoint file at the end of the initialization phase ( figure [ fig : ckpt1 ] ) . on the intel xeon phi accelerator
the initialization takes about 5 minutes .
the checkpointing itself takes about 1 minute ( the application working directory , physically located on the host , was mounted by the coprocessor through nfs ) and the resulting checkpoint image file is 1.4 gb ( uncompressed ) . restarting from the checkpoint image file takes less than 10s .
functionally this appears to work as expected .
the resulting checkpoint image file can thus be distributed to other nodes and the simulation process `` cloned '' without the startup cost , simply by resetting the random number seeds . for the second test
we have emulated the use case of some threads being slower than other in producing results .
this can happen if one or more threads is simulating more complex - than - average events .
to control such behavior we have modified our application in a way that half of the threads were responsible of simulating simple and fast events ( low energy single particles ) while the second half of threads was responsible for longer ones ( ten times higher energy ) .
in current geant4 multi - threaded mode the application will wait for all threads before terminating thus leaving half of the mic cores unused .
we have instrumented the application code to verify when the number of active threads drops below a given value , in such a case a checkpoint is triggered ( figure [ fig : ckpt2 ] ) .
also in this case we have verified that the application was behaving as expected .
it is important to note that we did not have to modify geant4 `` kernel '' code to enable checkpointing , all code modifications were done at the application level . for the first test ( checkpointing at first event ) we have provided feedback to geant4 developers that have introduced a new user - hook , not present in the initial design of the code , that allows for the execution of ( optional ) user code just before the first event is simulated but after all threads have been fully initialized ( this guarantees that the checkpointing is performed in a reproducible state of the application ) .
both tests show that checkpointing can be used to increase the efficiency of resource usage also on accelerator technologies where the minimization of the time spent in sequential fractions of the code or with only few threads active is fundamental to efficiently use the hardware resources .
we have made investigations into the use of the checkpoint - restart technology dmtcp with hep applications from cms and geant4 .
we have reported on the performance seen , both on a traditional x86 - 64 architecture and on intel s xeon phi , for situations relevant for a number of interesting use cases for hep computing .
we believe that the results obtained are very encouraging and demonstrate the viability of the use of this technology in the hep environment .
this work was partially supported by the national science foundation under grant oci-0960978 ( arya , cooperman ) and cooperative agreement phy-1120138 ( elmer ) as well as by the u.s .
department of energy ( dotti ) .
9 evans l and bryant p 2008 lhc machine _ jinst _ * 3 * s08001 bird i 2011 computing for the large hadron collider _ annual review of nuclear and particle science _ * 61 * 99 - 118 chatrchyan s et al ( cms collaboration ) 2008 the cms experiment at the cern lhc _ jinst _ * 3 * s08004 aad g et al ( atlas collaboration ) 2008 the atlas experiment at the cern large hadron collider _ jinst _ * 3 * s08003 chatrchyan s et al ( cms collaboration ) 2012 observation of a new boson at a mass of 125 gev with the cms experiment at the lhc _ phys.lett . _
* b716 * 30 - 61 aad g et al ( atlas collaboration ) 2012 observation of a new particle in the search for the standard model higgs boson with the atlas detector at the lhc
_ phys.lett .
_ * b716 * 1 - 29 ansel j , arya k and cooperman g 2009 dmtcp : transparent checkpointing for cluster computations and the desktop _ proc .
ieee intl .
parallel and distributed processing symposium _
( rome ) agostinelli s et al 2003 geant4 - a simulation toolkit _ nuclear instruments and methods in physics research _ * a 506 * 250 - 303 fuller s h and millet l i ( editors ) 2011 _ the future of computing performance : game over or next level ? _ the national academies press . | process checkpoint - restart is a technology with great potential for use in hep workflows .
use cases include debugging , reducing the startup time of applications both in offline batch jobs and the high level trigger , permitting job preemption in environments where spare cpu cycles are being used opportunistically and efficient scheduling of a mix of multicore and single - threaded jobs .
we report on tests of checkpoint - restart technology using cms software , geant4-mt ( multi - threaded geant4 ) , and the dmtcp ( distributed multithreaded checkpointing ) package .
we analyze both single- and multi - threaded applications and test on both standard intel x86 architectures and on intel mic .
the tests with multi - threaded applications on intel mic are used to consider scalability and performance .
these are considered an indicator of what the future may hold for many - core computing . |
research into 5d transition metal compounds has blossomed in recent years due to the potential for new electronic and magnetic phenomena to emerge from the interplay between crystal field effects , electronic correlations ( hubbard u ) and strong spin - orbit coupling ( soc ) .
iridium is an appealing element in which to study these physics due to its tendency to adopt a 4 + oxidation state , where the ground electronic configuration 5d@xmath6 is predicted to yield a _
j _ = @xmath7 electronic state in the presence of strong soc @xcite .
further , iridates are known to crystallize in numerous structure types , each with unique ir - ir connectivity , and this fact has made them an active area of study@xcite . in aggregate ,
recent literature on the magnetic and electronic properties of iridates are unified by a desire to probe the influence of strong soc and u on electronic behavior@xcite .
based honeycombs , in particular the _ a_@xmath1iro@xmath0 ( _ a _
= li / na ) family , have fallen under intense scrutiny following predictions that they might host a spin - liquid ground state@xcite .
efforts to produce new iridium oxide honeycomb lattices have intensified@xcite , but 4d and 5d materials containing continuous honeycomb connectivity remain scarce . here
we report the synthesis and characterization of two new honeycomb iridates based on ir@xmath8 ( 5d@xmath5 ) .
the first , nairo@xmath0 is produced via _ chimie douce _ oxidative deintercalation of sodium from na@xmath1iro@xmath0 , which preserves the planar honeycomb network of edge - sharing iro@xmath3 octahedra .
the second , sr@xmath0cair@xmath1o@xmath2 , is a 2:1 ordered perovskite , which shares its structure with sr@xmath0caru@xmath1o@xmath2@xcite , and consists of layers of iro@xmath3 octahedra forming a buckled honeycomb lattice .
magnetic susceptibility data collected on both compounds demonstrate weak temperature - independent magnetism , and resistivity measurements show semiconducting behavior consistent with hopping conductivity .
these compounds are thus ideal to investigate the interplay between non - cubic crystal fields and soc . by comparison with related 4d and 5d honeycomb compounds with
partially filled t@xmath9 subshells , we find that d electron count correlates strongly with the magnetic behavior , and present a thorough structural analysis of sr@xmath0cair@xmath1o@xmath2 and sr@xmath0caru@xmath1o@xmath2 that ultimately suggests the correlation is driven by soc .
nairo@xmath0 was prepared _ via _
the oxidative deintercalation reaction : @xmath10 the 1:1 relationship between oxidant consumption and na@xmath11 deintercalation of equation 1 was confirmed by laboratory x - ray powder diffraction ( xrpd ) data collected on powders after vacuum evaporation of the solvent , which showed crystalline nabr as the only side product . in contrast to the precursor material na@xmath1iro@xmath0 which degrades rapidly in laboratory air@xcite , nairo@xmath0 is stable under ambient conditions , as xrpd data showed no evidence of decomposition after several months of air exposure .
however , nairo@xmath0 degrades rapidly upon heating above 200@xmath12c in air , as evidenced by the broadening of diffraction peaks and appearance of new reflections .
attempts at producing intermediate stoichiometries ( i.e. na@xmath13iro@xmath0 ) were unsuccessful , and instead produced two - phase samples containing a mixture of na@xmath1iro@xmath0 and nairo@xmath0 .
further , no phase width was observed in nairo@xmath0 , as reactions performed with excess bromine produced only nairo@xmath0 with no observable difference in lattice parameters or physical properties .
the structure of nairo@xmath0 was solved in space group @xmath14 _ via _ rietveld refinement to laboratory xrpd data ( fig .
[ nairo3_rietveld](a ) ) .
final structural parameters are available in the si .
solution of the structure in higher symmetry cells was prevented by the high degree of inter - plane stacking disorder , which arises due to the fact that there are several possible translations that can take place in the honeycomb plane from one layer to another , each of which generates a different stacking pattern at minimal energetic cost .
stacking disorder is a well known structural perturbation in many layered compounds , and evidence of this disorder can be plainly seen in the poorly - fit peak broadening and incorrect intensities present at low diffraction angles .
further evidence of this structural disorder is provided by synchrotron xrpd data ( fig .
[ nairo3_rietveld ] ( a , inset ) ) , which show highly asymmetric @xmath150 0 l@xmath16 peaks this may occur as a consequence of decomposition in the intense synchrotron beam . using diffax@xcite , we modeled the diffraction patterns for three distinct stacking variants . when viewed from perpendicular to the honeycomb plane ,
these stacking variants can be described based on which atoms in the honeycomb plane eclipse ( sit directly above ) the na ion in the adjacent plane .
the three primary stacking possibilities are : 1 ) na - ir eclipsed ; 2 ) na - na fully eclipsed , which produces infinite na channels ; 3 ) staggered , in which none of the atoms in the honeycomb plane eclipse those in the adjacent plane , which is the pattern adopted by the parent material na@xmath1iro@xmath0 .
the simulated diffraction patterns for these stacking arrangements , along with xrpd data collected on a sample of na@xmath1iro@xmath0 are shown in figure fig .
[ nairo3_rietveld](b ) .
the observed xrpd pattern closely resembles the simulation obtained for the na - ir eclipsed variant , but with noticeable broadening and attenuation of several peaks .
a fourth pattern was simulated based on a model consisting of 95% na - ir eclipsed stacking with a 5% probability of a fully - eclipsed stacking fault .
this model produces a good qualitative agreement with the raw xrpd data , suggesting that the fully eclipsed stacking variant may only be slightly less energetically favorable than the na - ir eclipsed variant .
the presence of the fully eclipsed stacking fault suggests that further synthetic work may yield control over stacking order in this compound .
for an in - depth discussion on stacking disorder in honeycomb iridates , see ref .
this is the second known structure with the formula nairo@xmath0-a post - perovksite structure with the same formula is obtained when synthesized under high - pressure and temperature @xcite .
neither structural polymorph of nairo@xmath0 can be synthesized using conventional solid - state techniques , a fact that underscores the novelty and relative instability of the ir@xmath4 oxidation state .
the structure of sr@xmath0cair@xmath1o@xmath2 was solved in space group _ p_2@xmath17/_c _ ( 14 ) _ via _ rietveld refinement to neutron powder diffraction ( npd ) and xrpd data ( fig .
[ srcairo_structure ] ( a ) ) , and was found to be isostructural to sr@xmath0caru@xmath1o@xmath2@xcite .
final structural parameters are available in the si . the monoclinic structure of sr@xmath0cair@xmath1o@xmath2 , visible in fig .
[ srcairo_structure ] ( b ) , can be described as a 2:1 ordered perovskite with the formula a@xmath0bb@xmath1o@xmath2 , where iro@xmath3 and cao@xmath3 octahedra share corners to form the perovskite lattice and sr@xmath18 cations occupy the 12-fold coordinate a sites .
in contrast to the ba@xmath0@xmath19ir@xmath1o@xmath2 ( _ m _ = mg , ca , sc , ti , zn , sr , zr , cd , in , etc . ) family of 6h - perovskite - like structures , which host face - sharing ir@xmath1o@xmath2 dimers@xcite , the iro@xmath3 octahedra of sr@xmath0cair@xmath1o@xmath2 share corners to form a buckled honeycomb lattice in the b c plane .
while the structure of sr@xmath0cair@xmath1o@xmath2 may seem odd given the apparent wealth of other barium - based compounds with analogous stoichiometry , it is actually unsurprising that a conventional perovskite lattice is formed in this case due to the improved match in ionic radii between sr and ir . the same is true of sr@xmath0caru@xmath1o@xmath2 , which also has a similar collection of ba - based 6h - perovskite - type cousins .
while many different perovskites containing iridium have been reported in addition to the widely studied sr@xmath1iro@xmath20 , this seems to be a rare example of an iridium - based perovskite containing honeycomb connectivity .
furthermore , the corner - sharing bonding motif of the honeycomb lattice is the first of its kind in iridates , and thus opens a new avenue along which to search for new honeycomb materials that are free from stacking disorder and are air stable .
the corefinement of the sr@xmath0cair@xmath1o@xmath2 structure to both xrpd and npd data yielded stable oxygen positions that can be used to investigate the honeycomb connectivity in detail ( fig .
[ srcairo_structure ] ( b ) ) . as a result of octahedral tilting , the ir o
ir bond angles between nearest - neighbor sites on the honeycomb lattice are significantly less than the ideal 180@xmath12 expected for an undistorted perovskite . the ir o bond lengths
are also distorted : for the first iridium site , three adjacent ir o bonds are significantly shorter than the overall average of 1.99 , while the opposite three are significantly longer . a similar `` three short , three long '' pattern is also observed to a lesser degree on the second iridium site , where the three short bonds form a plane that is roughly perpendicular to the plane formed by the longer three bonds .
this pattern results in significant variations in the bridging ir
o ir bond lengths : the shortest ir
ir distance ( 3.96(3 ) ) is formed via ir o bonds of very similar lengths ( 2.04(4 ) & 2.05(3 ) , respectively ) , while the longest ( d@xmath21 = 4.01(1 ) ) is formed via ir o bonds of very disproportionate lengths ( 2.14(3 ) &
1.92(4 ) , respectively ) .
these non - uniform exchange interactions result in an effective dimerization between ir sites even though the distances between adjacent iridium sites on the honeycomb lattice are remarkably similar .
these observations are consistent with a non - spherical perturbation to the d - electron states of ir 5 + , which will be discussed in detail in the next section .
[ honeycomb_magnetization ] shows magnetic susceptibility data collected on nairo@xmath0 and sr@xmath0cair@xmath1o@xmath2 , along with linear fits to the @xmath22-corrected inverse susceptibilities ( inset ) .
in contrast to na@xmath1iro@xmath0 , which has a magnetic moment of 2.0(1 ) @xmath23 per ir site and a weiss temperature @xmath24 = -159(3 ) k , honeycomb nairo@xmath0 shows only a small temperature - independent susceptibility @xmath22 = 3.29@xmath2510@xmath26 emu / mol ir in the high temperature regime and a curie tail at low temperatures .
a curie - weiss fit to the magnetic susceptibility data ( fig .
[ honeycomb_magnetization ] ( inset ) ) yields a curie constant of c = 4.3(1)@xmath2510@xmath27 emu / mol ir k , and a weiss temperature @xmath24 = 0(2 ) k. similar magnetic behavior is observed in sr@xmath0cair@xmath1o@xmath2 ( fig .
[ honeycomb_magnetization ] ) .
sr@xmath0cair@xmath1o@xmath2 was initially obtained as a phase - pure powder after 6 short heatings , and a curie - weiss fit to the magnetic susceptibility data collected on this sample ( fig .
[ honeycomb_magnetization ] ( inset ) ) yielded a temperature - independent susceptibility of @xmath22 = 2.6(1)@xmath2510@xmath27 emu / mol ir , a curie constant c = 9.3(1)@xmath2510@xmath27 emu / mol ir k , and a weiss temperature @xmath24 = -8(5 ) k. optimization of the stoichiometry and heating schedule for sr@xmath0cair@xmath1o@xmath2 ( see e.s.i . ) yielded a sample whose diffraction peaks were significantly sharper and more intense , consistent with an improvement in crystallinity and homogeneity .
this improvement resulted in a substantial reduction of c , @xmath28 and @xmath22 ( 4.6(1)@xmath2510@xmath27 emu / mol ir k , -8(5 ) k , and 3.29@xmath2510@xmath26 emu / mol ir , respectively ) . in all samples ,
the small observed magnetic moments are likely due to dilute magnetic impurities or orphan spins roughly 1 % of free _ s _
= @xmath7 spins in the bulk could account for the observed susceptibility .
furthermore , the small weiss temperatures observed from all datasets indicate that the magnetic electrons are non - interacting , which lends further weight to the argument for magnetic defects and impurities rather than intrinsic magnetism .
these results are thus consistent with a _ j _ = 0 state in the ir@xmath4 metal centers of both compounds .
furthermore , the magnetic moments observed for ir@xmath8 in these compounds are consistent with what is observed in many other ir@xmath8 compounds ( table 1 ) .
the observed temperature - independent susceptibility is not due to delocalization of charge carriers to form a metal , as is the case in other iridates such as [ rb / k]ir@xmath20o@xmath29@xcite .
[ resistivity ] shows electrical resistivity data collected on polycrystalline bar of nairo@xmath0 and sr@xmath0cair@xmath1o@xmath2 .
in both datasets , sample resistance increases on cooling , diverging rapidly below _ t_@xmath30100 k , and exceeding the detection limit of the instrument below _
t_@xmath3075 k in the case of nairo@xmath0 .
a simple arrhenius - like activation barrier did not yield a good fit to the resistivity data .
instead , the behavior of both nairo@xmath0
andsr@xmath0cair@xmath1o@xmath2 is well described by a variable - range hopping model .
the conductivity is consistent with hopping in two or three dimensions : a plot of ln@xmath31 vs. @xmath32 ( fig .
4(inset ) ) or ln@xmath31 vs. @xmath33 ( not shown ) yields a linear relationship .
this is similar to what is found for na@xmath1iro@xmath0 and indeed many other 4d and 5d honeycombs , and consistent with the materials structure@xcite .
both nairo@xmath0 and sr@xmath0cair@xmath1o@xmath2 are important new additions to a select few 4d and 5d compounds that exhibit honeycomb connectivity between mo@xmath3 octahedra , tabulated in table 2 .
for a given d electron count , there is little difference in observed magnetic moments between the 4d and 5d groups , despite the fact that both crystal field and soc energy scales change significantly when moving down the periodic table . while the magnetic moments observed for the d@xmath6 configurations are unsurprising given the expected magnetic moment of @[email protected]@xmath35 for a _ s _
= @xmath7 system , the weak temperature - independent magnetism observed for both 4d@xmath36 ru@xmath37 and 5d@xmath36 ir@xmath8 is an unexpected result in the low - spin octahedral crystal field case , which should have 2 unpaired electrons and total spin _ s _
= 1 . fig .
[ nrgscales ] shows three possible origins of a nonmagnetic state in an ir@xmath8o@xmath3 ( ru@xmath37o@xmath3 ) octahedron .
one possibility is that direct overlap between adjacent ir sites generates new molecular orbitals , which removes the threefold degeneracy of the t@xmath9 manifold and results in a _
= 0 state , a scenario that has been proposed for li@xmath1ruo@xmath0 based on the significant variation in ru - ru bondlengths@xcite .
the second possibility is that distortions from perfect octahedral symmetry remove the degeneracy of the t@xmath9 manifold locally ( i.e. a jahn - teller distortion ) , thus resulting in a completely filled , twofold degenerate ground state .
such distortions are thought to be highly important in the ir@xmath4 perovskites [ sr / ba]@xmath1iro@xmath20 , and may also be driven by extended crystal field effects ( i.e. interactions beyond nearest - neighbor oxygen and iridium atoms)@xcite .
the final possibility outlined here is that soc is the strongest perturbation to the cubic crystal field in ir@xmath8 , which is consistent with electronic structure calculations performed by phelan _
et al . _ on sr@xmath38la@xmath39ir@xmath20o@xmath40@xcite .
while the computational route invariably leads to the conclusion that soc produces a nonmagnetic state in ir@xmath8 , one can also reach this conclusion on paper using group theory . to understand how soc produces a _ j _
= 0 state in this case , one must generate appropriate term symbols for electronic states in the presence of soc by referencing the double group for o@xmath41 symmetry .
in addition to the five irreducible representations of standard character table for group o@xmath41 , the double group has six new irreducible representations ( @xmath42 , @xmath43 , and @xmath44)@xcite . by referencing to the o@xmath41 double group
, one finds that soc splits the @xmath45 ( t@xmath9 ) manifold into two sets of spin orbitals : a twofold degenerate set with @xmath46 symmetry and a fourfold degenerate set with @xmath47 symmetry .
the @xmath48 ( e@xmath49 ) manifold is not split in the presence of soc , but does acquire a new irreducible representation @xmath50 . because orbitals from the original @xmath45 and @xmath48 manifolds share the same irreducible representations , they interact to form low- and high - energy pairs , akin to bonding and antibonding orbitals in a conventional mo diagram .
the net result for the d@xmath36 case is a completely filled @xmath51 manifold , which gives rise to the _ j _ = 0 state .
while the first scenario ( direct metal - metal bonding ) is well supported for the case of li@xmath1ruo@xmath0 , it is not a likely explanation for our ir@xmath8 compounds , as nairo@xmath0 and sr@xmath0cair@xmath1o@xmath2 show similar magnetic properties despite having wildly different ir - ir connectivity and internuclear distances .
further information can be obtained by examining the ir
o bond lengths and octahedral distortions , as they provide direct signatures of orbital degeneracy . fig .
[ kde ] compares the ir
o bond lengths present in sr@xmath0cair@xmath1o@xmath2 and sr@xmath0caru@xmath1o@xmath2 using gaussian kernel density estimates and illustrations to provide a visual understanding of the distortions present in these octahedra .
both compounds exhibit similar `` three short , three long '' distortions of the two octahedral metal sites , as discussed in section 2.1 .
these distortions are distinctly asymmetric , and thus inconsistent with a jahn teller effect .
the asymmetry is also likely not due to extended crystal field effects , as both ir sites have similar extended coordination spheres ( proximity to ca@xmath52 and sr@xmath52 ions ) .
rather , the observed asymmetry is likely a direct consequence of soc , which , due to the mixing of bare - ion orbital identities , gives rise to asymmetric ( i.e. direction - dependent ) exchange .
the fact that the octahedral distortions are amplified in sr@xmath0cair@xmath1o@xmath2 compared to sr@xmath0caru@xmath1o@xmath2 provides further support for the influence of soc .
future spectroscopic experiments can directly probe for the excited states predicted by this model , and the optical signatures should also respond to application of a magnetic field , thus yielding a powerful method of investigating soc driven physics in these compounds .
two new insulating honeycomb iridates , nairo@xmath0 and sr@xmath0cair@xmath1o@xmath2 , have been synthesized and characterized _ via _
diffraction experiments , magnetometry , and resistivity . in both compounds
, iridium exists in a 5 + oxidation state , yielding a 5d@xmath36 electronic configuration .
magnetization measurements demonstrate that both compounds exhibit negligible magnetic susceptibility , implying that both of these compounds are in close proximity to either a _ s _ = 0 or a _ j _ = 0 magnetic state .
structural studies performed on sr@xmath0cair@xmath1o@xmath2 suggest soc is the dominant energy scale in determining the ground magnetic state of ir@xmath8 compounds .
further spectroscopic experiments on these compounds will shed light on the nature of the magnetism observed in iridates .
dcw and tmm gratefully acknowledge support from the david and lucile packard foundation and the american chemical society petroleum research fund .
the institute for quantum matter is supported by the u.s .
department of energy , office of basic energy sciences , division of material sciences and engineering under grant no .
de - fg02 - 08er46544 .
r. dcw and tmm are grateful to the nist center for neutron research for the neutron diffraction beam time obtained on bt-1 .
99 g. jackeli & g. khaliullin , mott insulators in the strong spin - orbit coupling limit : from heisenberg to a quantum compass and kitaev models .
lett . _ , 2009 , * 102 , * 017205 b .- j .
yang & y. b. kim , topological insulators and metal - insulator transition in the pyrochlore iridates .
b. _ , 2010 , * 82 * , 085111 y.singh & p. gegenwart , antiferromagnetic mott insulating state in single crystals of the honeycomb lattice material na@xmath1iro@xmath0 .
b. _ , 2010 ,
* 82 * , 064412 x. liu , vamshi m. katukuri , l. hozoi , wei - guo yin , m. p. m. dean , m. h. upton , jungho kim , d. casa , a. said , t. gog , t. f. qi , g. cao , a. m. tsvelik , jeroen van den brink , and j. p. hill , testing the validity of the strong spin - orbit - coupling limit for octahedrally coordinated iridate compounds in a model system sr@xmath0cuiro@xmath3 , _ phys .
lett . _ , 2012 , * 109 * 157401 b. f. phelan , j. krizan , w. xie , q. gibson , r. j. cava , a new material for probing spin - orbit coupling in iridates , _ phys .
rev . b. _ 2015 , * 91 * , 155117 w. witczak
krempa , g. chen , y. b. kim , & l. balents , correlated quantumphenomena in the strong spin - orbit regime , _ ann .
cond . mat .
phys . _ , 2013 , * 5 * , 57 - 82 a. shitade , h. katsura , j. kunes , x .-
qi , s.c .
zhang , n. nagaosa , quantum spin hall effect in a transition metal oxide na@xmath1iro@xmath0 .
lett . _ , 2009 , * 102 , * 256401 h. gretarsson et al . ,
crystal field splitting and correlation effect on the electronic structure of _ a_@xmath1iro@xmath0 .
lett . _ , 2013 , * 110 , * 076402 j. chaloupka , g. jackeli & g. khaliullin , kitaev heisenberg model on a honeycomb lattice : possible exotic phases in iridium oxides _ a_@xmath1iro@xmath0 . _
_ , 2010 , * 105 , * 027204 i. kimchi , j. j. analytis , a. vishwanath , three dimensional quantum spin liquid in a hyperhoneycomb iridate model and phase diagram in an infinite d approximation .
2014 , * 90 * , 205126 e. c. andrade , m. vojta , magnetism in spin models for depleted honeycomb
lattice iridates : spin glass order towards percolation .
b. _ 2014 , * 90 * , 205112 j. g. rau , e. k .- h .
lee , h .- y .
kee , generic spin model for the honeycomb iridates beyond the kitaev limit .
lett . _ , 2014 , * 112 , * 077204 h. lei , z. zhong , h. hosono , structural , magnetic , and electrical properties of li@xmath1ir@xmath53ru@xmath38o@xmath0 .
rev . b. _ 2014 , * 89 * , 020409 v. m. katukuri , _ et al .
_ , kitaev interactions between _ j _ = 1/2 moments in honeycomb na@xmath1iro@xmath0 are large and ferromagnetic : insights from _ ab initio _ quantum chemistry calculations .
_ new j. phys . _ , 2014 , * 16 * , 013056 d. hirai , j. matsuno , h. takagi , fabrication of ( 111)-oriented ca0.5sr0.5iro3/srtio3 superlattices a designed playground for honeycomb physics , _ apl materials _ , 2015 , * 3 * , 041508 j. t. rijssenbeek , s. malo , v. caignaert , k. r. poeppelmeier , site and oxidation - state specificity yielding dimensional control in perovskite ruthenates , _ journal of the american chemical society _ , 2002 ,
* 124 * , 2090 m. m. j. treacy , j. m. newsam & m. w. deem , a general recursion method for calculating diffracted intensities from crystals containing planar faults .
_ proceedings of the royal society , london _ , 1991 , * a433 * , 499 - 520 t. sakamoto , y. doi , y. hinatsu , crystal structures and magnetic properties of 6h - perovskite - type oxides ba@xmath0_m_ir@xmath1o@xmath2 ( _ m _
= mg , ca , sc , ti , zn , sr , zr , cd and in ) , _ journal of solid state chemistry _ , 2006 , * 179 * , 2595 - 2601 a. talanov , w. a. phelan , z. a. kelly , m. a. siegler , t. m. mcqueen , control of the iridium oxidation state in the hollandite iridate solid solution k@xmath53ir@xmath20o@xmath29 , _ j. inorg .
chem _ , 2014 ,
* 53 * , 4500 - 4507 s. a. j. kimber , i. i. mazin , j. shen , h. o. jeschke , s. v. streltsov , d. n. argyriou , r.valenti , and d. i. khomskii , valence bond liquid phase in the honeycomb lattice material li@xmath1ruo@xmath0 , _ phys .
b _ , 2014 , * 89 * , 081408 n. a. bogdanov , v. m. katukuri , j. romhanyi , v. yushankhai , v. kataev , b. buchner , j. van den brink , and l. hozoi , orbital reconstruction in nonpolar tetravalent transition - metal oxide layers , _ nat .
comm . _ , 2015 , * 6 * , 7306 v. m. katukuri , _ et al . _ , electronic structure of low - dimensional 4d@xmath57 oxides :
interplay of ligand distortions , overall lattice anisotropy , and spin orbit interactions , _ j. inorg .
chem _ , 2014 ,
* 53 * , 4833 - 4839 c. i. hiley , m. r. lees , j. m. fisher , d. thomspsett , s. agrestini , r. i. smith , r. i. walton , ruthenium(v ) oxides from low - temperature hydrothermal synthesis , it j. angew .
, 2014 , * 53 * , 4423 - 4427 c. i. hiley , d. o. scanlon , a. a. sokol , s. m. woodley , a. m. ganose , s. sangiao , j. m. de teresa , p. manuel , d. d. khalyavin , m. walker , m. r. lees , and r. i. walton , antiferromagnetism at t@xmath58500k in the layered hexagonal ruthenate srru@xmath1o@xmath3 , _ physical review b _ * 92 * 104413 ( 2015 ) j. c. want _ et al .
_ , lattice tuned magnetism of ru@xmath37(4d@xmath5 ) ions in single crystals of the layered honeycomb ruthenates li@xmath1ruo@xmath0 and na@xmath1ruo@xmath0 , _ phys .
b _ , 2014 , * 90 * , 161110(r ) k. w. plumb , j. p. clancy , l. j. sandilands , v. v. shankar , y. f. hu , k. s. burch , h .- y .
kee , y .- j .
kim , @xmath54rucl@xmath0 : a spin
orbit assisted mott insulator on a honeycomb lattice , _ phys .
b _ , 2014 , * 90 * 041112(r ) along with rietveld refinement of the structure in p@xmath59 to the data ( red ) and calculated difference between the model and the data ( blue ) are shown in the main frame . stacking
faults between honeycomb layers complicate the refinement , as can be seen from the missed intensities and peak shape mismatches between 2@xmath60 = 15@xmath12 to 35@xmath12 . in the inset , the observed profile of the ( 001 ) peak , as observed by laboratory xrpd data collected in reflection geometry ,
is compared to the profile observed from synchrotron xrpd data collected on aps 11-bm in transmission geometry . *
b ) * laboratory xrpd data ( black , top ) along with simulations for three distinct stacking arrangements of adjacent nair@xmath1 honeycomb layers ( red , blue and green ) are shown .
a fourth simulation ( purple ) shows the pattern that results from having a 95% probability of an na ir eclipsed stacking arrangement with a 5% probability of a fully eclipsed stacking fault , which provides a good description of the observed diffraction pattern . in the upper right - hand corner
the structure of nairo@xmath0 is shown , in which iro@xmath3 octahedra ( green ) share edges to form honeycomb layers with na cations ( orange ) occupying the honeycomb `` holes '' and partially occupying the spaces sandwiched between na and ir sites in the honeycomb layers.,width=336 ] cair@xmath1o@xmath2 to laboratory xrpd data ( top ) and ncnr bt-1 npd data ( bottom ) are shown .
data points are displayed in black , the fit in orange , and the difference between the data and fit , blue .
expected peak positions for the monoclinic cell are displayed as gray tick marks below the data in each pane . *
b ) * the unit cell of sr@xmath0cair@xmath1o@xmath2 is displayed along with a cutout of the honeycomb network formed by corner - sharing iro@xmath3 octahedra in the a - b plane .
ir distances and ir o ir bond angles are shown in the bottom left of the honeycomb structure , and ir o bond lengths for each of the two iridium sites are shown in the upper right hand corner . despite the uniformity of the ir
ir distances in the honeycomb lattice , the iro@xmath3 octahedra exhibit significant asymmetric distortions , and these distortions are illustrated by gray triangles in the plane formed by the three most closely bonded oxygen atoms of each octahedron .
these distortions are inconsistent with a jahn - teller effect , and instead are likely driven by strong soc of the ir@xmath8 ion as described in the text.,width=672 ] ( red squares ) , the first sample of sr@xmath0cair@xmath1o@xmath2 ( black crosses ) , and a sample of sr@xmath0cair@xmath1o@xmath2 produced _ via _ an optimized synthetic procedure ( black circles ) are shown .
all magnetic data were collected on polycrystalline samples , and a correction for the temperature - independent susceptibility of the sample holder was applied to the data in order to extract real values of @xmath22 for each sample .
two datasets are shown for sr@xmath0cair@xmath1o@xmath2 to illustrate that the observed sample @xmath22 s are due to magnetic defects , the concentration of which can be substantially reduced through correct sample preparation .
linear fits to the @xmath22-corrected inverse susceptibility data for nairo@xmath0 and sr@xmath0cair@xmath1o@xmath2 , visible in the inset , were used to extract curie constants c and weiss temperatures @xmath28 for nairo@xmath0 and the optimized sample of sr@xmath0cair@xmath1o@xmath2 .
both samples exhibit weak , temperature - independent susceptibility consistent with a small concentration of magnetic defects .
these data demonstrate that the ir@xmath8 species in the bulk of both compounds have no unpaired electrons.,width=336 ] ( red diamonds ) and sr@xmath0cair@xmath1o@xmath2 ( black circles ) are shown .
the increase in resistance of the pellets on cooling is a hallmark of electrically insulating materials . in the inset ,
the natural logarithm of the normalized resistance is plotted as a function of _ t_@xmath61 .
both datasets are linear in this plot in the high - temperature regime consistent with 2-dimensional variable - range hopping of impurity - localized charge carriers.,width=336 ] = 0 states in ir@xmath8 iridates . * a ) * direct @xmath62 and @xmath63 overlap between adjacent iridium sites yields a nonmagnetic ( _ s _
= 0 ) state by removing the threefold degeneracy of the isolated t@xmath9 manifolds .
the energy of the splitting , @xmath64 , is proportional to the distance between adjacent iridium sites , d@xmath65 . *
b ) * a local distortion of the cubic crystal field _ via _ jahn - teller distortion or interaction with cations outside the first coordination sphere ( extended crystal field ) splits the t@xmath9 manifold by an energy proportional to the axial elongation ( i@xmath66/i@xmath67 ) , which also yields a _ s _
= 0 state .
* c ) * the ground state of the d@xmath5 configuration in a cubic crystal field can be split by soc to yield a _
= 0 ground state , which originates from _ s _ = 1 and
_ l _ = 1 spin and orbital angular momenta .
the upper bound to the magnitude of this splitting is proportional to the bare ion soc constant @xmath68 .
relative to the crystal field @xmath69.,width=336 ] cair@xmath1o@xmath2 andsr@xmath0caru@xmath1o@xmath2 .
gaussian kernel density estimates ( kde ) computed on the ir
o bond lengths present in sr@xmath0cair@xmath1o@xmath2 ( red ) and sr@xmath0caru@xmath1o@xmath2 ( blue ) provide a visual comparison of the octahedral distortions present in both compounds .
the top and bottom panes contain data for the 1st and 2nd metal sites in each structure , respectively ; in each pane , the solid lines represent kde s and the small tick marks represent the actual m o bond lengths . a distinct `` 3short
, 3long '' bond length pattern is observed in both compounds .
model octahedra in each pane show the shape of the distortion in both cases , with the plane formed by the closely bonded oxygen atoms highlighted in gray.,width=336 ] | we report the structures and physical properties of two new iridates , nairo@xmath0 and sr@xmath0cair@xmath1o@xmath2 , both of which contain continuous two - dimensional honeycomb connectivity .
nairo@xmath0 is produced by room temperature oxidative deintercalation of sodium from na@xmath1iro@xmath0 , and contains edge - sharing iro@xmath3 octahedra that form a planar honeycomb lattice .
sr@xmath0cair@xmath1o@xmath2 , produced _ via _ conventional solid - state synthesis , hosts a buckled honeycomb lattice with novel corner - sharing connectivity between iro@xmath3 octahedra . both of these new compounds
are comprised of ir@xmath4 ( 5d@xmath5 ) and exhibit negligible magnetic susceptibility .
they are thus platforms to investigate the origin of the nonmagnetic behavior exhibited by ir@xmath4 oxides , and provide the first examples of a _ j
_ = 0 state on a honeycomb lattice . |
ultracold spinor bose gases provide a beautiful example to study fragmented bose - einstein condensates ( bec ) @xcite , where bose - einstein condensation occurs in two or more single particle states simultaneously .
this is an unusual scenario , in contrast with conventional bose - einstein condensation where bosons cluster together into a single state . for single - component bosons , condensation in a single state
is enforced by repulsive interactions : the energetic cost of fragmentation is too high because of the associated exchange energy @xcite . for bosons with an internal degree of freedom
, one can escape this mechanism by building correlations between the particles to cancel the exchange energy @xcite .
a spin-1 bec with antiferromagnetic interactions in a tight trap has been predicted to host such fragmented condensates for vanishing magnetic fields @xcite .
the atoms condense into a single spatial mode but there remains a large internal degeneracy at the single - particle level .
antiferromagnetic interactions lift this degeneracy , and lead to a total spin - singlet ground state which is completely fragmented between the three sublevels .
the many - body singlet state displays strong quantum correlations , and has attracted much theoretical interest .
this spin - singlet fragmented condensate is fragile against any perturbation lifting the single - particle degeneracy , such as external magnetic fields @xcite . in experiments with alkali atoms ,
the most relevant perturbation is the quadratic zeeman splitting between the zeeman sublevels @xmath0 and @xmath1 @xcite . for finite atom number @xmath2
, there is a small but non - vanishing window where the singlet state survives as the quadratic zeeman splitting increases from zero , before a crossover to a single @xmath0 condensate takes place ( `` single bec domain '' ) @xcite .
an appropriate witness of the transition is the variance @xmath3 which goes from @xmath4 in the spin - singlet state to @xmath5 in an uncorrelated many - body state @xcite . because of the sensitivity to external perturbations ,
the singlet state has so far eluded experimental observation .
the gap to the first excited states is low and scales as the inverse of the number of atoms @xcite .
evaporative cooling used to produce quantum gases is unable to reach such ultralow temperatures .
another procedure is to adiabatically produce the singlet state by first applying a magnetic field and condensing in the @xmath0 state , and then slowly remove the field to produce the desired singlet state see the sketch in fig . [
fig : rampq ] . in order to stay adiabatic
, the dynamics must be very slow in view of the small energy scales involved , making the procedure vulnerable to heating or inelastic losses . in this article
we introduce a way to shortcut the adiabatic following and thus produce the desired final state in times much shorter than those needed in adiabatic processes .
such methods have been recently derived for a number of quantum mechanical systems see for instance ref .
@xcite , and promise to provide important advances in actual implementations of quantum technologies , for instance trapped ions @xcite .
exact protocols have been derived for particular problems , e.g. the quantum harmonic oscillator @xcite . in other cases , approximate procedures , obtained by adapting exact ones ,
have been proven to be quite promising when applied to quantum many - body systems @xcite .
as will be shown , the approximate shortcut protocol will be obtained from a large @xmath2 limit of the quantum many - body system .
this limit will allow us to map our original many - spin problem into an effective harmonic oscillator , for which an exact solution is available @xcite .
interestingly , very recently a similar harmonic description of a spinor bec has allowed the authors in ref .
@xcite to prove parametric amplification of a spinor system .
this work proves experimentally the appropriateness of the harmonic description .
the article is organized as follows . in section [ sec2 ]
, we present the theoretical model to describe the spinor bec and discuss the adiabatic preparation of the ground state . in section [ sec : sa ] we obtain our protocol to shortcut the adiabatic evolution in the spinor system from a continuum approximation to the spin dynamics . in section [ sec3 ]
we apply our shortcut protocol to the bec regime ( dominated by the quadratic zeeman energy ) . in section [ sec4 ]
we consider a broader range of parameters , discussing the quality of our protocol to produce fragmented bec starting from the bec side .
in section [ sec5 ] we present results making use of current experimental setups @xcite . in section [ sec6 ] ,
we briefly summarize our work and present the main conclusions .
with all atoms in the @xmath0 state . in this regime ,
the initial state is very close to the ground state ( arbitrarily close as @xmath6 ) . in the adiabatic method ,
the qze is slowly reduced in such a way that the state of the system remains always close to the instantaneous ground state . stopping the qze ramp at @xmath7 ,
the system ends up in a total spin - singlet ground state with strong spin correlations . stopping the ramp at a large , negative value of @xmath8
, we prepare instead a twin fock state with half the atoms in the zeeman @xmath9 . in this paper
we target the production of the singlet state , and examine this procedure and alternative ramps which are not adiabatic but result in a state close to the ground state in a much faster time .
[ fig : rampq ] ]
we consider an ultracold gas of spin-1 bosons in a harmonic trap under the action of an external magnetic field .
we assume a single spatial mode in the trap , that is , all bosons condense in the same spatial orbit irrespective of their internal state . with this assumption
we are left with three single - particle states , @xmath10 , @xmath11 and @xmath12 , corresponding to the zeeman states with magnetic quantum numbers @xmath13 , respectively .
the linear zeeman effect acts only as a shift in the energy and does not contribute to determine the equilibrium state .
the main contribution of the magnetic field is the quadratic , or second order , zeeman ( qz ) effect @xcite . under these assumptions ,
the system is well described by the hamiltonian @xcite @xmath14 where @xmath15 is the spin interaction energy per atom , @xmath2 is the number of atoms , @xmath16 is the ( dimensionless ) total spin operator , @xmath8 is the quadratic zeeman energy and @xmath17 is the number operator of the zeeman state @xmath18 . the first term in the right - hand side of eq .
( [ eq : ham1 ] ) describes antiferromagnetic interactions between pairs of atoms , and favours configurations with low total spin @xmath19 . in absence of the quadratic zeeman term , @xmath7 ,
the eigenstates are known analytically and are given by the total spin eigenstates @xmath20 , where @xmath19 is the total spin and @xmath21 the eigenvalue of @xmath22 , the projection of the total spin on the @xmath23 axis .
this is the basis we will be using in the following sections .
low-@xmath19 configurations are obtained by putting many spin-1 atoms to form singlet state pairs , while the remaining atoms can occupy any zeeman sublevel . for practical convenience , from now on @xmath2
will be set to an even number .
the ground state for even @xmath2 is the total spin singlet @xmath24 .
this highly fragmented state , termed `` spin - singlet condensate '' ( ssc ) , takes the form of a condensate of delocalized spin - singlet pairs , @xmath25 where @xmath26 creates a pair of atoms in the two - particle singlet state , @xmath27 is an annihilation operator for a particle in the zeeman state with third component of the angular momentum equal to @xmath28 , and @xmath29 is the boson vacuum .
this fragmented state is characterized by three macroscopically populated states , @xmath30 , with large fluctuations of the individual components @xcite . the second term in eq .
( [ eq : ham1 ] ) describes the interaction of the system with the external magnetic field . in the non - interacting limit @xmath31 and for @xmath32 ,
the qze forces all the spins to occupy the state @xmath0 , thus forming a single bose - einstein condensate with @xmath33 and @xmath34 , the so - called @xmath23-polar state , @xmath35 it is worth noting that @xmath21 remains fixed when changing @xmath8 , because the hamiltonian commutes with @xmath22 .
this is a good approximation to the behavior due to the experimental conditions of the atomic quantum gases , which are highly isolated from the environment , and to the microscopic rotational invariance of the spin exchange interaction @xcite . for simplicity , we take @xmath36 .
also , as the number of particles @xmath2 is fixed during the evolution , we will omit it on the kets , thus , for now on we will use the notation @xmath37 . for generic values of @xmath39
, we write a general state @xmath40 with fixed @xmath2 and @xmath36 as @xmath41 .
the schrdinger equation @xmath42 in the @xmath19 basis reduces to the following discrete eigenvalue equation ( see appendix [ app : a ] ) , @xmath43 in the upper panel of fig . [
fig : cx ] we show the transition from the @xmath44-dominated fragmented regime to the single bec regime when varying the ratio @xmath45 .
the transition between the two regimes takes place at values @xmath46 and is seen in the behavior of the variance @xmath47 of the populations in the @xmath0 zeeman state . as explained in the introduction in the uncorrelated bec state ,
@xmath48 , while in the spin - singlet state the fluctuations are much larger , @xmath49 . experimentally , the value of the qze can be controlled easily in real time .
for instance , for sodium atoms with hyperfine spin @xmath50 in a magnetic field @xmath51 , the quadratic zeeman shift contributes a _ positive _ amount to @xmath8 .
it is also possible to achieve @xmath52 by using the differential level shift induced on the individual zeeman sublevels by a far off - resonant microwave field ( see @xcite for details ) . with a suitable choice of the microwave polarization , detuning and power ,
the sign and magnitude of @xmath8 can be changed at will .
this experimental control of the qze opens a way to the generation of strongly correlated states in spin-1 quantum gases .
the principle is the following . for zero magnetization and a large and positive qze ,
the ground state is very close to a single bec with all atoms in the @xmath0 zeeman state .
a good approximation of this state can be prepared `` by hand '' , _
e.g. _ by applying radio - frequency rf pulses with suitable frequency and polarization to a spin - polarized ensemble in @xmath53 , for instance .
starting from this initial state and decreasing slowly the value of @xmath8 , the system will adiabatically follow its ground state , and end up prepared in the ssc state given in eq .
( [ ssc ] ) when @xmath54 . we can estimate the speed at which the magnetic field should be decreased by the usual adiabatic criterion , @xmath55 , where @xmath56 and @xmath57 are two eigenstates of the hamiltonian .
the dangerous region is around @xmath58 , where the energy gap to the first excited state takes its minimum value @xmath59 . in this region ,
the qze ramp has to be very slow .
we make a crude estimate by assuming that @xmath8 decreases between @xmath44 and @xmath60 in a time @xmath61 .
also in this region , @xmath62 are on the order of @xmath63 .
this leads to @xmath64 and to the adiabaticity criterion , @xmath65 the catastrophic scaling @xmath66 shows that this method will be limited to small , mesoscopic samples . using very long ramp times to fulfill the adiabaticity criterion will make the protocol vulnerable to experimental limitations not captured by the single - mode hamiltonian , such as technical heating ( specific for each experimental setup ) and inelastic losses ( specific for each atom ) . inelastic atom losses destroy the rotational symmetry since atoms are lost at random from any zeeman state . a common source of inelastic losses is three - body recombination into a weakly - bound molecule and a fast atom , resulting in three atoms lost from the trap . the total rate of these events can be written as @xmath67 , where @xmath68 is determined by a species - dependent rate constant @xmath69 and by the spatial density @xmath70 .
demanding less than one single inelastic event ( on average ) during the entire adiabatic protocol gives a bound @xmath71 .
for illustrative purposes , we consider a gas of atoms condensing in the gaussian ground state of a tight harmonic trap of frequency @xmath72 .
the gaussian ground state of the trap is a good approximation of the actual condensate wavefunction for sufficiently low atom number @xmath73 , with @xmath74 the harmonic oscillator length , with @xmath75 the atomic mass and with @xmath76 the spin - independent @xmath77wave scattering length .
the spin - dependent scattering length @xmath78 is determined by the relation @xmath79
the bound @xmath71 can be written as a bound on the maximum affordable atom number in trap , written in compact form as @xmath80 where @xmath81 has the dimension of a length .
we specialize to the case of @xmath50 sodium atoms , where @xmath82 nm and @xmath83 nm @xcite and three - body loss rate constant @xmath84at.@xmath85/s @xcite . using @xmath86khz
, one finds @xmath87 for the parameters given above , showing that the adiabatic approach is reserved for mesoscopic samples containing only a few atoms .
this motivates us to find alternative solutions enabling a substantial speed - up of the dynamics , which is our main objective in the rest of this paper .
in view of the limitations of the adiabatic approach described above , we now examine a different method where the same final result can be reached in a much shorter time . in the literature , there are well - established shortcut protocols for one - body harmonic potentials @xcite . our strategy is to use these results to manipulate the many - spin system of interest by mapping it to an effective harmonic oscillator problem .
we show in this section how a reasonable harmonic approximation to the many - body problem can be derived . by means of such approximate equation
we map the shortcut protocol to the exact time dependent schrdinger equation built from eq .
( [ eq : ham2 ] ) .
the first step consists in deriving a continuum approximation to the hamiltonian , eq .
( [ eq : ham2 ] ) . for large @xmath2 and considering @xmath88 ,
the coefficients @xmath89 can be assumed to vary smoothly from @xmath19 to @xmath90 .
hence , @xmath89 can be approximated by a continuous function @xmath91 , where @xmath92 varies from @xmath60 to @xmath93 .
following the derivations in appendix [ app : b ] , we arrive at an effective schrdinger - like equation for a harmonic oscillator @xmath94 with the oscillator frequency given by @xmath95 and the oscillator `` mass '' by @xmath96 .
the ground state obeying the boundary condition @xmath97 is the wave function @xmath98 with energy @xmath99 manifold , @xmath47 computed for three different number of atoms .
middle panel : energy of the ground state of the exact system compared with the energy of the ground state obtained in eq .
( [ eq : en ] ) for different values of the parameter @xmath8 .
when @xmath100 , the system becomes @xmath44-dominated and , thus , the energy is constant [ see eq .
( [ eq : en ] ) ] .
it is worth noting that both in the fragmented ( @xmath44-dominated ) and bec ( @xmath8-dominated ) regimes , the approximate value of the energy agrees well with the exact one .
lower panel : wave function of the ground state compared to the one obtained with the continuous approximation described in the text .
the middle and lower panels are obtained for a system of @xmath101 spins . in all cases , @xmath102 hz . for the lower panel ,
we have used @xmath103 .
[ fig : cx ] ] in fig .
[ fig : cx ] we compare the approximated and the exact solutions of our hamiltonian . in the middle panel of fig .
[ fig : cx ] we see that the energy of the ground state is well reproduced by the harmonic approximation .
in particular it is interesting to note that the harmonic approximation works well both in the @xmath44-dominated regime and in the @xmath8-dominated one . comparing the actual wave functions in the lower panel of fig .
[ fig : cx ] , we can see that the solution of the approximate hamiltonian has a similar shape as the exact wave function although its maximum is slightly displaced towards higher values of @xmath104 .
the idea behind the shortcut to adiabaticity in the time - dependent evolution of an harmonic oscillator is the following .
first we consider that the system is initially in the ground state for a certain initial value @xmath105 of the control parameter .
then , we impose that at a given time @xmath106 the system must be exactly in the ground state for a different value of the control parameter , @xmath107 .
the goal is , thus , to find a function @xmath108 that does the job .
if the final time is sufficiently large , then any smooth ramp of the control parameter should work , since the evolution would be adiabatic . for short ramp times , an arbitrary ramp function would in general result in the excitation of many modes besides the ground state at the final time .
the goal is therefore to engineer the ramp function in such a way as to minimize the excitations at @xmath109 and beyond , _ i.e. _ one seeks to produce an almost stationary state once the ramp is completed .
the schrdinger - like equation in eq .
( [ eq : ham3 ] ) is already close to the one corresponding to a harmonic oscillator .
the control parameter is the qze , @xmath110 .
the term on the right - hand side ( a shift in the total energy ) does not have any effect on the dynamics .
we also limit ourselves to the regime @xmath111 .
the final schrdinger - like equation eq .
( [ eq : ham3 ] ) is that of a harmonic oscillator with time - dependent `` mass '' and frequency @xmath112 .
a similar equation was considered in ref .
@xcite to describe the dynamics of a two - mode bose - hubbard model .
following the same method , we look for a self - similar solution @xmath113 , with a scaling parameter @xmath114 .
such a solution exists if the scaling parameter obeys the so - called ermakov equation @xcite , @xmath115 the constant @xmath116 is just an integration constant that we set to @xmath117 .
this , together with the substitution @xmath118 gives @xmath119 @xmath120 is an arbitrary function that only has to satisfy the frictionless conditions @xmath121 where @xmath122 will always be zero . there is an infinite set of functions @xmath120 that can be used for this purpose , as the boundary conditions provide ample freedom to choose @xmath120 .
here we will use a simple polynomial ansatz , taken from ref .
@xcite , @xmath123 where @xmath124 .
thus , the function @xmath108 is @xmath125 the six frictionless conditions previously mentioned uniquely determine the fifth - order polynomial chosen .
we have tested that using a sixth - order polynomial or power - law functions did not improve the results , hence eq .
( [ eq : b ] ) is used in the rest of this work .
let us remark that the freedom in choosing @xmath120 can be used to design more constrained protocols depending on the specific needs , e.g. avoiding too large values of @xmath108 .
in this section , we consider the performance of our shortcut protocol in the bec regime .
that is , our main goal is to evolve from the ground state of eq .
( [ eq : ham1 ] ) for an initial value of @xmath126 to the corresponding ground state for @xmath107 such that @xmath127 . strictly speaking , the choice @xmath128 does not comply with the assumption used to derive the protocol , @xmath129 .
the reason why the protocol can be extended to this situation is that for short times @xmath130 , the protocol produces a very small variation of @xmath8 , and the dynamics thus remains mostly adiabatic . to judge the quality of the shortcut protocol we will compare it to two other ramps shapes , linear and exponential , @xmath131 the linear ramp is uniquely determined , and the exponential ramp was found to provide the best results for a decay constant @xmath132 , a value we have used in all the reported results .
we benchmark our shortcut protocol by numerically solving the full time dependent schrdinger equation with @xmath133 from eq .
( [ eq : ham1 ] ) for a particular ramp .
we used the mean - squared spin @xmath134 as a fidelity witness . note that in the regime we consider in this section , the target final value of @xmath135 is far above the value @xmath136 below which the ground state reduces to the total spin singlet state . as a result
, the value of @xmath134 in the ground state corresponding to @xmath135 fulfills @xmath137 . for each @xmath19 at each time @xmath138 .
blue circles correspond to a system evolved using the shortcut protocol and red circles correspond to a system evolved with an exponential ramp .
panels * ( a ) * , * ( c ) * show the evolution of a system from @xmath139 hz to @xmath140 in @xmath141 s and panels * ( b ) * , * ( d ) * show the evolution of a system from @xmath139 hz to @xmath142 in @xmath141 s , so this figure can be directly compared to fig .
[ fig : fid - s2 ] .
since only the even solution is considered , the wave function @xmath89 at odd @xmath19 is identically 0 .
[ fig : cs-3 ] ] in fig .
[ fig : fid - s2 ] we present the first results , corresponding to two different dynamical situations .
the first one shown in fig .
[ fig : fid - s2](a , c ) goes from @xmath143 to @xmath144 .
the second one shown in fig .
[ fig : fid - s2](b , d ) goes one order of magnitude smaller , to @xmath145 .
also we compare in the figure two different values of @xmath101 and @xmath146 .
several features can be observed . in all cases ,
the shortcut protocol performs clearly better than the other two ramps , while the exponential ramp performs better than the linear ramp .
the spin witness @xmath134 at the final time at @xmath147s is substantially lower for the shortcut protocol ( @xmath148 decreases by an order of magnitude from its initial value ) , and closer to the value expected in the final ground state for larger @xmath2 . in comparison , the other two protocols are only able to decrease it by at most a factor of 4 in the same time .
moreover , the final value of @xmath148 decreases with increasing atom number at a fixed @xmath106 .
equivalently , the final fidelities obtained with the shortcut protocol improve as we increase @xmath2 .
this could be expected as we have obtained our protocol in the large @xmath2 limit , thus making the protocol closer to an exact description as @xmath2 is increased .
the results obtained with the exponential and linear ramps are mostly independent of the number of particles . after the time evolution , i.e. @xmath109 , for the three different protocols considered : shortcut ( a ) , linear ( b ) and exponential ( c ) . in all cases
only the results that give @xmath149 are shown . finally , in ( d )
, we show the contour lines corresponding to @xmath150 . in all cases
we have @xmath151 particles and an initial value of @xmath152 hz .
[ fig : figmap1 ] ] it is also interesting to see how the wave function evolves in time , going from a state with large @xmath148 , where @xmath89 are centered around large @xmath19 , to a state with small @xmath148 , where the wave function takes substantial values around @xmath153 or @xmath154 . in fig .
[ fig : cs-3 ] we compare the wave functions at different times obtained with the shortcut ( a , b ) and exponential ( c , d ) protocols .
the calculations correspond to the @xmath101 ones reported in fig .
[ fig : fid - s2 ] . as can be clearly seen in all cases
the wave function for the shortcut is much more peaked around @xmath153 than the exponential one .
also , as expected , the final wave function is more concentrated at smaller values of @xmath19 as we target final states with smaller @xmath135 .
this can be seen comparing panels ( a , c ) , computed with @xmath155 with panels ( b , d ) , computed with @xmath156 .
finally , note that even though the shortcut performs quite well , the final wave function is peaked at @xmath154 rather than at @xmath153 , which reflects the fact that we are still on the single bec side of the crossover reported in fig .
[ fig : cx ] ( c ) .
for the same conditions as in fig .
[ fig : figmap1 ] .
the results obtained from the three different protocols , the shortcut , a linear ramp and an exponential one are given in panels ( a ) , ( b ) and ( c ) , respectively .
finally , in ( d ) we show the contour lines of the three cases above for @xmath157 .
all these plots have been realized using a system of @xmath151 particles and @xmath152 hz . [ fig : figfra1 ] ] in the previous section we have shown the superior performance of the shortcut protocol in comparison with exponential and linear ramps in the bec regime , @xmath158 . in this section
we explore the fragmented condensate domain , that is , qze ramps going from @xmath159 to @xmath160 . in figs .
[ fig : figmap1 ] and [ fig : figfra1 ] , we provide an extensive comparison between our shortcut protocol and linear and exponential ramps . the figures depict the final values of @xmath148 , fig .
[ fig : figmap1 ] , and @xmath161 , fig .
[ fig : figfra1 ] . in those figures we consider @xmath151 atoms , starting from the ground state corresponding to a value of @xmath162 .
the figures cover a broad range of final target values of @xmath135 ranging from deep in the bec sector into well below the transition to the fragmented condensate region , @xmath163 , see fig .
[ fig : cx ] ( upper panel ) .
results are also reported as a function of the desired final time , @xmath106 .
we take again @xmath164 hz and final times ranging from @xmath165 to @xmath166 seconds .
as found previously , the shortcut protocol performs better than the exponential and linear ramps in the bec region , as can be seen looking at the @xmath167 region in the three figures .
for instance the region in the @xmath168 map , where small final values of @xmath148 are larger for the shortcut protocol .
the exponential produces also relatively low values , with a result mostly independent of the value of @xmath135 , while the linear ramp fails to produce small final values , unless @xmath169 s. in situations in which the target final state is clearly in the fragmented domain , @xmath170 , the only method that produces sizeable fragmentation , as measured by @xmath171 , is the shortcut protocol ( see fig . [
fig : figfra1 ] ) .
the exponential ramp requires times almost two orders of magnitude larger to obtain the same level of fragmentation in the system . in line with the latter , lower final values for @xmath148
are obtained for the shortcut protocol for those cases in which the fragmentation is closer to the singlet value @xmath172 @xcite . for the parameters considered here ,
a shortcut ramp performed in @xmath173 is able to produce a state very close to the ground state , for @xmath151 .
note that , using the notations and assumptions of section [ xxx ] , the chosen value of @xmath174hz is achieved in a trap of frequency @xmath175hz for @xmath151 .
the corresponding three - body lifetime is @xmath176s@xmath177 , or @xmath178 : less than a single three - body loss event ( on average ) during the entire shortcut protocol .
losses should not be a concern for @xmath179 .
as a function of time computed for systems evolved following the shortcut protocol , a linear or an exponential ramp .
we compare together a system of @xmath180 spins ( exact solution ) and a system with @xmath181 spins ( approximate solution ) .
the system is evolved from a state with @xmath182 hz to @xmath142 in @xmath183 s. the shortcut protocol also works for this set of parameters , although it is , in principle , only valid for @xmath184 , in the sense that it provides a clear gain over simpler exponential or linear ramps . [ fig : comp ] ] we have been using , throughout the full manuscript , parameters taken from realistic proposals , most of them from @xcite . in this section , we explore different parameters taken from other experimental setups .
some experiments @xcite have been recently done with @xmath185rb bose condensates composed of @xmath181 atoms with @xmath186 hz and a @xmath108 around @xmath187 .
taking these parameters , we have calculated the evolution of a system with @xmath188 particles and @xmath186 hz to check whether the shortcut protocol still gives good results under these experimental conditions .
results have also been obtained for a system with @xmath181 spins using the shortcut protocol for the approximate hamiltonian in eq .
( [ eq : ham3 ] ) . in fig .
[ fig : comp ] both results are shown for comparison .
@xmath148 , shown in fig .
[ fig : comp ] , has been computed for these two systems and , although the initial values of @xmath108 are larger than @xmath44 , the shortcut protocol still drives the system to the ground state ( or close ) and improves the performance of the other two ramps . although the shortcut protocol is , in principle , only valid for @xmath184 , this and other calculations ( where we have driven a system from different values of @xmath189 , all between 10 and 1000 times larger that @xmath44 , to @xmath135 above and below @xmath44 ) show that the protocol can be successfully applied for larger @xmath189 values .
we have presented a method to prepare a spin-1 bec into a many - body spin singlet state by making use of an approximate protocol to shortcut the adiabatic following in the many - body system .
the protocol consists in specific functions @xmath108 which are constructed such that the time evolution of the system brings the many - body state from the ground state for @xmath190 to the ground state for @xmath191 .
the main aim is to produce the very fragmented ground state of the spinor system in absence of quadratic magnetic field , starting from a condensate in the @xmath0 manifold in a regime dominated by the quadratic zeeman term .
the performance of the shortcut protocol has been compared to both a linear and an exponential ramp of the parameter @xmath8 .
even though the protocol is only approximate , it is shown to provide a much better performance than the exponential and linear ones almost in all situations . in the bec side , that is , for @xmath192 , the method works almost perfectly for time intervals of the order of @xmath193 and larger .
the method works also better for cases in which the bec - fragmented transition is targeted .
in particular it works with similar accuracy as the exponential ramp up to times one order of magnitude smaller . to quantify the performance we have computed the achieved final value of @xmath134 and the value of @xmath194 .
we have obtained results for systems with different sizes and final and initial setups and we have seen that the protocol achieves better results for larger systems .
results have also been obtained from approximate solutions of the schrdinger equation [ using eq .
( [ eq : ham3 ] ) and ( [ eq : cx ] ) ] . based on these results
we have been able to extrapolate the method to larger systems and find that , with this protocol , a many - body spin singlet state can be obtained for many different systems sizes .
we have also shown the success of our method when applied to systems prepared with parameters taken from current experimental setups .
we believe that our method for preparing a bec into a singlet state with short times is experimentally realizable and efficient .
further improvements to the shortcut protocol profiting from the available freedom inherent to the presented procedure will be the object of forthcoming investigations .
we acknowledge stimulating discussions with members of the bose - einstein condensates group at lkb , in particular with bertrand evrard and jean dalibard , and with tommaso roscilde . this work has been partially supported by darpa ( optical lattice emulator grant ) .
we acknowledge financial support from the spanish mineco ( fis2014 - 54672-p ) , from generalitat de catalunya grant no . 2014sgr401 and the maria de maeztu grant ( mdm-2014 - 0369 ) .
lds ackowledges support from the eu ( ief grant no .
236240 ) and tz from the hamburg center for ultrafast imaging .
is supported by the ramn y cajal mineco program .
11 e. j. mueller , t. .l ho , m. ueda , and g. baym , phys . rev .
a * 74 * , 033612 ( 2006 ) .
p. nozires , in : _ bose - einstein condensation _
, ed . by a. griffin , d. w. snoke and s. stringari , cambridge university press ( 1995 ) . c. k. law , h. pu , and n. p. bigelow , phys .
lett . * 81 * , 5257 ( 1998 ) .
ho and s. k. yip , phys . rev .
84 * , 4031 ( 2000 ) .
m. koashi and m.ueda , phys .
lett . * 84 * , 1066 ( 2000 ) .
y.castin and c. herzog , cras paris , tome * 2 * , srie 4 , 419 ( 2001 ) .
f. zhou , int .
b * 17 * , 2643 ( 2003 ) .
r. barnett , j. d. sau , and s. das sarma , phys .
a * 82 * , 031602 ( 2010 ) .
l. de sarlo , l. shao , v. corre , t. zibold , d. jacob , j. dalibard , and f. gerbier , new j. phys . *
15 * , 113039 ( 2013 ) .
v. corre , t. zibold , c. frapolli , l. shao , j. dalibard , f. gerbier , epl * 110 * , 26001 ( 2015 ) .
d. m. stamper - kurn and m. ueda , rev .
* 85 * , 1191 ( 2013 ) .
e. torrontegui , s. ibaez , s. martnez - garaot , m. modugno , a. del campo , d. gury - odelin , a. ruschhaupt , xi chen , and j. g. muga , adv . at .
* 62 * , 117 ( 2013 ) .
s. an , d. lv , a. del campo , and k. kim , arxiv:1601.05551 .
x. chen , a. ruschhaupt , s. schmidt , a. del campo , d. gury - odelin , and j. g. muga , phys .
lett . * 104 * , 063002 ( 2010 ) .
b. juli - daz , e. torrontegui , j. martorell , j. g. muga , and a. polls , phys .
a * 86 * , 063623 ( 2012 ) .
a. yuste , b. juli - daz , e. torrontegui , j. martorell , j. g. muga , and a. polls , phys .
a * 88 * , 043647 ( 2013 ) .
s. campbell , g. de chiara , m. paternostro , g. massimo palma , and r. fazio , phys .
* 114 * , 177206 ( 2015 ) .
hoang , m. anquez , b.a .
robbins , x.y .
yang , b.j .
land , c.d .
hamley , and m. s. chapman , nat . commun .
* 7 * , 11233 ( 2016 ) .
f. gerbier , a. widera , s. flling , o. mandel , and i. bloch , phys .
a * 73 * , 041602(r ) ( 2006 ) .
s. knoop , t. schuster , r. scelle , a. trautmann , j. appmeier , m. k. oberthaler , e. tiesinga , and e. tiemann , phys .
a * 83 * , 042704 ( 2011 ) .
a. grlitz , t. l. gustavson , a. e. leanhardt , r. lw , a. p. chikkatur , s. gupta , s. inouye , d. e. pritchard , and w. ketterle , phys .
lett . * 90 * , 090401 ( 2003 ) .
the hamiltonian in eq .
( [ eq : ham1 ] ) is diagonalized by the total spin eigenstates @xmath20 , where @xmath19 is the total spin and @xmath21 the total projection of @xmath19 in the z - axis direction . a general state @xmath195 is written as , @xmath196 .
the construction of the angular momentum eigenstates is not trivial and these are built as follows @xcite , @xmath197 where @xmath198 , @xmath199 , @xmath200 and @xmath201 are the creation and annihilation operators of the state @xmath202 respectively , @xmath203 is the lowering total spin operator and @xmath204 is the singlet creation operator .
the expression of these states involves three operators .
the first one , @xmath205 is the creation operator of a spin @xmath93 particle with @xmath53 .
this operator acting @xmath19 times over the vacuum leads to the many - particle state @xmath206 .
the following acting operators @xmath207 and @xmath208 commute with the total spin momentum operator and therefore do not modify @xmath19 .
the singlet creator operator creates pairs with total spin @xmath60 , so it only changes the number of particles . then , @xmath209 repeated actions of this operator add singlet pairs until the state @xmath210 is obtained .
finally , the lowering angular momentum operator @xmath211 acts @xmath198 times without affecting @xmath2 or @xmath19 , leading to the final state @xmath212 .
the normalization factor is obtained after tedious calculations , @xmath213 the complete hamiltonian in eq .
( [ eq : ham1 ] ) can be computed in the @xmath214 basis .
the interaction term is diagonal , but the operator @xmath215 has matrix elements between states with @xmath19 and @xmath216 .
the total hamiltonian is thus tridiagonal and therefore easy to solve numerically .
the action of @xmath217 is explicitly given by @xcite , @xmath218 |n\ s\ m\rangle \ , , \end{aligned}\ ] ] where @xmath219 the resulting hamiltonian , given also in eq .
( [ eq : ham2 ] ) , reads , @xmath220 with @xmath221 \ , . \nonumber\end{aligned}\ ] ]
a continuum approximation of eq .
( [ eq : ham2 ] ) can be obtained by considering @xmath88 .
the wave function @xmath89 can , thus , be approximated by a continuous function @xmath91 , where @xmath92 and varies from @xmath60 to @xmath93 .
then , @xmath222 can be taken as a small parameter and a taylor expansion can be made , @xmath223 by substituting this expression into eq .
( [ eq : ham2 ] ) the following continuum schrdinger equation is obtained @xmath224 where @xmath225 taking into account that @xmath88 , we can set the order until which we want to approximate . performing a taylor expansion in @xmath104 , @xmath226 and @xmath227 , and substituting the resulting expressions in eqs .
( [ eq : abc ] ) , one finds @xmath228 keeping terms to leading order in @xmath229 we arrive at the schrdinger - like equation ( [ eq : ham3 ] ) . | we devise a method to shortcut the adiabatic evolution of a spin-1 bose gas with an external magnetic field as the control parameter . an initial many - body state with almost all bosons populating the zeeman sublevel @xmath0 ,
is evolved to a final state very close to a macroscopic spin - singlet condensate , a fragmented state with three macroscopically occupied zeeman states .
the shortcut protocol , obtained by an approximate mapping to a harmonic oscillator hamiltonian , is compared to linear and exponential variations of the control parameter .
we find a dramatic speedup of the dynamics when using the shortcut protocol . |
it has been long known that galaxies have evolved over the history of the universe , and a major branch of recent astronomy has been to quantify how the galaxies we see in the present day came to be as they are now .
it is impossible to observe how a given galaxy evolves over cosmologically - significant time scales , but its evolution can be inferred either on a galaxy by galaxy basis from the present state of individual galaxies for example the star formation history ( e.g. maraston et al .
2006 ; noeske et al .
2007b ) , or the merger history ( e.g. conselice 2003 ; conselice et al . 2009 ) ; or from the changes to the properties of a population of galaxies at different times .
massive galaxies are a major component of a number of branches of astronomy , being a pivotal test for , amongst others , dark matter theories and galaxy formation models .
multiple constraints have now been put on when the most massive ( m@xmath10 ) galaxies formed ( e.g. , fontana et al .
2004 ; glazebrook et al . 2004
; bundy et al .
2006 ; conselice et al .
2007 , 2011 ) and it is seen that @xmath11 is an era of transition when massive galaxies shut off their star formation ; their mass for the most part , already in place ( glazebrook et al .
2004 ; bundy et al . 2006 ; conselice et al . 2007 ;
prez - gonzlez et al .
2008a ; ilbert et al .
the star formation in these most massive galaxies was significantly higher before this epoch , and there appears to be evidence of star formation rates well in excess of those found in today s universe ( e.g. hopkins 2004 ; prez - gonzlez et al .
2008b ) . within this transition epoch , there is increasing evidence that the shut off in star formation is not triggered simultaneously across the universe , but is highly dependent on the masses of the galaxies . in a process that has been named ` downsizing '
, it has been suggested that the most massive galaxies finish their star formation earlier than less massive systems , a progression observed in broad - band colours ( e.g. bundy et al .
2006 ) , radio observations ( hopkins 2004 ) , infrared observations ( e.g. papovich et al . 2006 ; bell et al . 2007 ) and emission line surveys ( e.g. heavens et al . 2004 ; panter et al .
2007 ) .
measurements of high - redshift star formation rates ( hereafter sfrs ) have until recently , been dogged by many problems .
the more favoured emission line diagnostic , the h@xmath1 line is only available via optical spectroscopy at @xmath12 0.4 and so the more troublesome [ oii ] line has been regularly used at higher redshifts ( gallagher , hunter & bushouse 1989 ; kennicutt 1998a ; rosa - gonzlez , terlevich & terlevich 2002 ; charlot et al .
2002 ; kewley , geller & jansen 2004 ) . the [ oii ] line is affected more by metalicity and due to its bluer wavelength , is more influenced by dust attenuation than the h@xmath1 line ( jansen , franx & fabricant 2001 ; charlot et al .
2002 ) . as a result
, it is never entirely certain whether high-@xmath13 sfrs derived from [ oii ] are completely compatible with low-@xmath13 sfrs derived from other more accurate methods .
infrared ( ir ) sfr tracers , for example 24@xmath14 m emission ( lonsdale persson & helou 1987 ; rowan - robinson & crawford 1989 ; kennicutt 1998a ; calzetti et al .
2007 ) is also commonly used as a ` proxy ' sfr diagnostic .
emission from sfr regions is absorbed by dust and re - radiated as ir radiation , allowing a conversion to be calibrated to retrieve the original sfr .
this method has however , problems in measuring star formation rates .
for example , using current technology , spectral line tracers are also able to measure sfrs to an order of magnitude lower than the mips ir instrument , and are more sensitive at higher redshifts ( see for example noeske et al .
other problems include the fact that the extrapolation from a few sub - mm and radio luminosities to a bolometric ir luminosity is subject to many uncertainties , especially at high redshift ( smail , ivison & blain 1997 ; chapman et al .
2005 ) and systematic changes ( for example papovich et al . 2007 ) ; contamination by an active galactic nuclei ( agn ) ( see daddi et al .
2007b ) ; and possibly cirrus emission comming from older stars ( lonsdale persson & helou 1987 ; helou 1986 ; kennicutt 1998a ) .
radio sfrs , another sfr measure , often relies on stacking of data sources which may lead to complications in interpretting results ( e.g. ivison et al .
2010 ) .
each measure of star formation tracks a different element of the stellar evolution cycle , each with its own precise definitions ( see kennicutt 1998a for a review of what different wavelengths actually measure ) and therefore unique biases .
it is important therefore to get a measure of whether the apparent changes in global sfr are actually due to a change in the sfr of galaxies , or because the different measurements at different redshifts are measuring different processes .
the h@xmath1 line is the best tracer of this due to the relatively low attenuation it suffers from dust compared to other emission line tracers , the lack of any strong dependency on metalicity , and because its emission is associated largely with only star forming regions .
it is also a direct measurement of star formation , and measures instantaneous star formation during the last 20myr within the galaxy .
attenuation is still a significant problem however it is variable galaxy - to - galaxy due to its dependence on the amount of dust in a galaxy , and can effect the emission by a factor of three or more . at redshift @xmath15 0.4 ,
the h@xmath1 emission line shifts into the near - infrared ( nir ) and different technologies are required to observe it . in the last decade
there have been a number of ir instruments built for telescopes and with these new instruments , we can now begin to measure this emission to higher redshifts than before .
there have been a few small investigations into h@xmath1 emission at @xmath11 , however most have been hampered by a very high detection limit , or the low number of galaxies studied .
glazebrook et al ( 1999 ) presented the first nir spectroscopy to detect h@xmath1 emission from galaxies at @xmath13 = 1 .
they studied 13 field galaxies drawn from the canada - france redshift survey and compare their derived sfrs with observations of the rest - frame uv emission .
they find that the h@xmath1 sfr density measurements are roughly three times larger than the ones derived using the madau , pozzetti & dickinson ( 1996 ) uv relation .
doherty et al ( 2006 ) followed from works by yan et al .
( 1999 ) , hopkins , connolly & salay ( 2000 ) and tresse et al .
( 2002 ) examining the sfr density at @xmath11 .
doherty et al
. took spectra of 38 galaxies within the _ hubble deep field north _ and stacked their spectra in order to measure the total star formation and hence the sfr density .
they estimate a lower limit for the total sfr in their stacked sample to be @xmath16 312yr@xmath17 but do not measure individual sfrs .
both doherty et al . and
yan et al .
derive dust corrections by comparison with uv measurements .
tresse et al .
( 2002 ) used a mixture of balmer ratios ( h@xmath1/h@xmath18 and h@xmath19/h@xmath18 ) where available to measure a dust correction , in other galaxies they used a fixed attenuation .
most recently , rodrguez - eugenio et al .
( 2007 ) measured 16 star forming galaxies at redshift 0.8 @xmath20 1.0 from the deep2 survey with the camera on the , but had problems removing the sky lines from their data .
garn et al .
( 2010 ) used narrow band imaging to measure h@xmath1 emission from 477 sources in the hizels survey with a sfr detection limit of 2.15yr@xmath17 .
they used other available multiwavelength imaging to calculate the dust attenuation using a number of methods .
sobral et al . (
2010 ) further investigate the sfr - mass relation and specific sfr - mass relation of these galaxies , however focusing on environmental effects .
none of these previously mentioned surveys have however used h@xmath1 spectroscopy to investigate trends and relations found with other sfr tracers , nor do they use a mass selected sample .
we present the first such study using the spectrograph on the to obtain near - infrared multi - object spectroscopy of 42 galaxies from the powir survey ( conselice et al .
2008a ) with stellar masses m@xmath21 and at redshifts @xmath22 .
we measure h@xmath1 luminosities for our sample to calculate their sfrs and to investigate whether the observed trends in global and individual star formation properties are seen when using different sfr tracers . in this paper
we furthermore investigate the number of galaxies with h@xmath1 emission as a function of mass , compare h@xmath1-derived sfrs with other measures of their sfrs and relations between the sfrs , and mass and colour .
this paper is organised as follows .
the data and new observations are summarised in section [ sec : data ] , including different methods of correcting for dust attenuation in section [ sec : corrections ] .
the galaxies where h@xmath1 emission is detected are investigated in section [ sec : results ] and the sfr relations are investigated in section [ sec : starformation ] .
the results , and the astrophysical interpretation of the sfrs are discussed in section [ sec : discussion ] , with a summary of the work in section [ sec : conclusion ] . throughout the paper ,
we adopt the cosmology @xmath23 , @xmath24 and @xmath25kms@xmath17mpc@xmath17 .
the galaxies investigated in this paper are selected from galaxies within the powir survey ( conselice et al .
2008a ) as investigated in conselice et al .
( 2007 , hereafter c07 ) .
the powir survey is a large area ( @xmath161.5deg@xmath26 ) deep nir survey in the _ k _ and _ j _ band covering the goods field north ( giavalisco et al .
2004 ) , the extended groth strip ( egs ; davis et al . 2007 ) and three fields observed by the deep2 team with the deimos spectrograph ( davies et al .
c07 used all the areas covered except the goods - north field .
data in c07 was acquired from the palomar 5-m telescope ( @xmath27 and @xmath28 band imaging ) and used imaging from the _ canada - france - hawaii telescope _
( cfht ; b , r , i ) , the advanced camera for surveys ( acs ) on the ( optical ) , the multiband imaging photometer ( mips ) on the _ spitzer space telescope _ ( 24@xmath14 m ) , the _ very large array _
( 1.4 ghz ) , the _ chandra x - ray observatory _ and spectra from the deep imaging multi - object spectrograph ( deimos ) at the w. m. keck observatory ( @xmath29 ) .
a @xmath30-selected catalogue was created and the different photometry bands matched .
readers are directed to c07 for more information on these observations and the parent sample selection .
the c07 catalogue was used to measure the stellar masses of galaxies to @xmath31 20 21 from sed fitting as in bundy et al .
( 2006 ; for details , see c07 ) .
this measurement uses a chabrier initial mass function ( hereafter imf ; chabrier 2003 ) .
a complete subset was then created with m@xmath10 and a redshift range of @xmath32 0.4 1.4 taken from deep2 spectroscopic redshifts ( davis et al .
this is the dataset from which our sample was constructed .
the targets for were taken from two regions , one purely mass selected from the powir field 3 to have m@xmath10 , the second selected from the egs field .
this second set of targets were selected to be massive ( m@xmath10 ) , have a significant mips flux and no chandra 2 - 8 kev x - ray emission at @xmath33ergs@xmath17@xmath34 .
both sets were selected within the range @xmath35 , giving a total of @xmath167500 possible targets . masks for multi - object spectroscopy were created using pointings that maximised the number of galaxies targeted .
our near infrared ( nir ) spectra were taken during 9 nights over two observing runs on the 4.2-metre using the long - slit intermediate resolution infrared spectrograph ( _ liris _ ) instrument in multi - object spectroscopy mode ( acosta - pulido et al .
the spectra were taken using a 3-point nodding routine and the total integration times range from 2.1 hrs to 6.3 hrs for each of the 7 masks .
each mask targeted between 8 and 14 galaxies , targetting a total of 71 galaxies .
data reduction was done using the dedicated suite of routines lirisdr for iraf , written by jos acosta - pulido and the spectra were flux calibrated using @xmath28-band magnitudes from the powir survey data . in all cases
the spectra covered the wavelength range of the @xmath28-band , and the spectra were scaled to give the same integrated flux measurement within that wavelength range as the @xmath28-band magnitude measurements .
this also corrected for flux lost due to the finite width of the slit ( aperture correction ) . out of the 71 galaxies targeted ,
42 had detected continuum .
there are no galaxies where there is emission at around 6563 but have no detected continuum .
there were 21 galaxies detected in each of the two areas of sky targeted .
one galaxy , 32014349 is affected by an atmospheric absorption feature within 200 of the wavelength of any h@xmath1 emission line and so is discarded leaving 41 galaxies in our sample .
the stellar mass and redshifts of these 41 galaxies are shown in figure [ fig : massz ] ( see also table [ tab : data ] ) .
the 1d spectra were examined by eye and if an emission line was seen at the correct wavelength to be h@xmath1 , the flux was extracted from the continuum subtracted spectra within this region .
even though in most cases where there was substantial @xmath36 emission the h@xmath37 and [ nii]@xmath38 appeared separated , the extraction region was chosen to include both emission lines .
this was to allow the procedure for measuring an upper limit where there was no emission detected to follow as closely as possible the procedure used when there was emission .
the continuum level and noise was taken from two 200 sections of continuum one average full width half maximum ( fwhm , 5 pixels = 31 ) either side of the emission and the root mean squared value of this continuum is used as the error on each pixel .
this average fwhm is used as the extraction width for continuum with no visual signs of emission .
if the h@xmath1 emission is less than the 1.5 times the error and no line has been seen by eye , then these galaxies are classified as non - detections .
this method of measuring the emission line is similar to that used in glazebrook et al ( 1999 ) .
a correction of 30 per cent for [ nii ] is applied ( see for example , gallego et al .
1997 ; tresse et al . 1999 ; garn et al .
2010 ) . in cases where there was no visible emission ,
the continuum noise was used to calculate a 1.5 @xmath39 upper limit to any h@xmath1 emission .
the deep2 spectroscopic redshifts were used to determine the predicted observed h@xmath1 wavelengths and convert fluxes to luminosities .
star formation rates were derived from the h@xmath1 luminosities using the kennicutt ( 1998 ) relation converted to use a chabrier imf ( chabrier 2003 ) : @xmath40 in order to compare our measurements with measurements taken in other wavelengths , we use infrared ( ir ) , uv and radio observations already available of the same sources .
ultraviolet ( uv ) data was taken using the _ galaxy evolution explorer _ ( galex ) and available imaging in the @xmath41 , @xmath42 and @xmath43 filters to match the galaxy s rest - frame emission as close as possible to the galex filters ( 1528 and 2271 ; schiminovich 2007 ) .
sfrs and dust extinction - corrected sfrs were calculated using a composite method using ir calibrated uv - optical colours and broadband sed fitting ( for a description of the method see schiminovich et al .
2007 and seibert et al . 2005 ) .
the uv and ir sfrs are discussed in more detail in section [ sec : corrections ] .
radio data was acquired from a 1.4 ghz catalogue of the egs , ( ivison et al .
2007 ) taken with the _ very large array _ ( vla ) .
they detect 1,123 sources with s@xmath44@xmath14jy to a 5-@xmath39 detection limit which corresponds to rest - frame 1.4 ghz calculated sfr of 275yr@xmath17 at z = 1.0 .
our sample were matched with the radio catalogue , but only reveal 1 match ( from 41 galaxies ) .
the radio catalogue was then matched with the entire set of possible galaxies that could have been observed and are in the same region of sky as those that were observed .
out of these 62 galaxies there are 4 matches ( @xmath45 per cent ) . concluding that our galaxies are not detected at 1.4 ghz
, upper limits can be placed on the total star formation in these galaxies following the methods of condon ( 1992 ) , haarsma et al .
( 2000 ) and condon , cotton & broderic ( 2002 ) . using the detection limit of the nearest galaxy in the sample ( z @xmath16 0.55 ) gives an upper limit of 66yr@xmath17 , the furthest ( z @xmath16 1.23 ) gives @xmath16450yr@xmath17 .
these are considerably higher than the majority of our upper limits derived from the h@xmath1 measurements and thus are not further used . [ cols="^,^,^,^,^,^,^,^,^,^ " , ]
in the scenario of downsizing the most massive galaxies should turn off their star formation at the redshift ranges we probe . in our sample , the most massive galaxies with star formation and with stellar masses m@xmath46are found only at a redshift greater than @xmath13 = 0.93 . the drop - off in the number of star forming galaxies with masses above m@xmath47is sharp , suggesting that the process of turning off star formation happens very quickly , in less than a gyr .
this appears to be a sudden quenching rather than a slow ending of star formation , similarly it is not likely to be a `` running out '' of gas within star forming regions as in this scenario we would expect this to lead to a declining sfr over a longer period of time .
there is only one galaxy that is detected in h@xmath1 and lies within the red sequence of galaxies in our sample , showing that star forming galaxies detected with h@xmath1 at z@xmath161 are mostly blue .
there are only a small number of non - detections ( 4 out of 25 ) that lie definitively within the region of blue galaxies and these could be either non - detections of star forming galaxies or actual non - star forming galaxies . in line with what other studies have found at other wavelengths ,
this study finds a slightly increasing h@xmath1-derived sfr with stellar mass , and a decreasing ssfr with mass relations .
there is some suggestion that the normalisation of the ` galactic main sequence ' ( gms , n07 ) is dependent on the wavelength and method used to measure the sfr , and the h@xmath1-derived sfrs appear to be slightly higher than other measurements .
this could be the result of the sensitivity of measuring sfrs which in this study are limited to greater than a few yr@xmath17 . with a dust correction ,
this becomes a lower limit of sfrs greater than a few tens of yr@xmath17 .
[ oii ] star formation rates can be measured ( uncorrected ) typically down to 0.51yr@xmath17 , which when corrected give sfrs with typically half the value detected in this sample . however , to reconcile our results with the n07 derived gms , there would have to be a significant number of galaxies which have a sfr an order of magnitude below the level of our gms at all masses .
there are two probable causes of our detections having larger average sfrs than the gms measured by n07
. a major contributor will be that some of our galaxies have been selected to have significant 24-@xmath14 m emission , meaning that they are either highly star forming , or have a significant agn component .
two galaxies ( 12024309 & 12024445 ) could have [ nii]/h@xmath1 ratios indicative of agn , but have a sfr@xmath48 excess suggesting otherwise .
these two therefore probably have a slightly larger amount of [ nii ] emission leading to a overestimation of the h@xmath1-derived sfr .
there is little other evidence of any other galaxies in this sample having a dominating agn and so the higher sfrs measured are likely because the galaxies actually have a higher rate of star formation .
these would then lie above ( i.e. with a higher sfr ) any gms relationship found in more complete samples .
the galaxies however lie much above the n07 gms ( even with their estimated raising of the normalisation for higher redshifts ) , significantly beyond their 0.3 dex estimated scatter .
the daddi et al .
( 2007 , d07 ) gms is derived from uv and 24-@xmath14 m emission(as opposed to [ oii ] and 24-@xmath14 m in n07 ) and appears to be at a similar level to our data , but it is not clear whether our data follows their slope .
it is clear then that the choice of sfr indicator can be important .
c07 measured that the 24@xmath14m - derived sfr were considerably higher than the [ oii ] values and it is not unusual that the two methods do not give agreement on a galaxy - by - galaxy basis ( e.g. salim et al .
2007 ) .
the ssfr is a measure of how much of a galaxy s mass has been formed in recent star formation episodes .
the most massive galaxies with m@xmath5have a lower ssfr than the less massive galaxies which have a roughly constant value .
these very massive galaxies have had less of their mass formed in recent star formation and so must contain older stellar populations .
this is a clear indication that the most massive galaxies had their star formation quenched at a higher redshift than lower mass galaxies .
we find a separation between the star forming and non - star forming galaxies in sfr - colour plots ( figure [ fig : sfr_colour ] ) .
galaxies that are towards the red end of our sample ( in @xmath7 ) and have measured star formation , have the same average sfr as the blue galaxies . in the ssfr - colour plots ( figure [ fig : ssfr_colour ] ) ,
the ssfr drops with colour due to the fact that the red star forming galaxies are more massive than the blue galaxies with the same sfr .
this is seen in figure [ fig : colourmass ] and is possibly a selection effect .
half the galaxies in the sample were selected to have a 24-@xmath14 m detection for a galaxy to have enough star formation so that it gives 24-@xmath14 m emission , but not be significant enough to make the galaxy blue , the galaxy must be more massive .
figure [ fig : colourmass ] also shows a considerable difference in the slope of this relation for star - forming and passive galaxies .
the star - forming galaxies are fit by a line with the relation @xmath49 and the galaxies without h@xmath1 detections are fit by the relation @xmath50 a two dimensional kolmogorov
smirnov test ( k - s test ; fasano & franceschini , 1987 ) indicates that these two populations are however not significantly different . colour for galaxies divided into those with detected star formation ( crosses ) and no detected star formation ( triangles ) .
the redder galaxies tend to be more massive than the blue galaxies this is possibly a selection effect due to half our galaxies needing 24@xmath14 m emission for selection and thus have star formation .
there is a lack of massive , blue galaxies which should have been detected .
the solid line is a least - squares fit to the galaxies with h@xmath1 detections and has a slope of @xmath51 .
the dashed line is a least - squares fit to the galaxies without detected h@xmath1 and has a smaller slope : @xmath52,width=321 ] a more detailed look at the sfr and ssfr against colour relations can be obtained by binning the points by stellar mass ( figure [ fig : sfrcolourmass ] ) .
the effects of downsizing are clearly seen in these plots with the highest mass bins having many more passive galaxies than star - forming ones , similarly the lowest mass bin has more star - forming than passive galaxies .
both the lowest and intermediate mass bins in the sfr - colour plot can be described as a number of star - forming galaxies with a range of sfrs but with similar colours , with the passive galaxies ( with similar sfrs ) having a range of @xmath7 colours .
the most massive galaxies are mostly passive , red galaxies . when galaxies turn off their star formation , they also usually start changing colour from blue to red .
this would lead to the galaxies moving downwards and to the right in the sfr - colour diagram .
if the concept of downsizing is correct , then the most massive galaxies shut off their star formation before less massive systems .
we can then use a mass cut to separate a population of galaxies that have shut off their star formation ( the most massive galaxies ) from a population that are yet to shut off their star formation ( the least massive galaxies ) .
figure [ fig : sfrcolourmass ] splits the galaxies into stellar mass bins and shows their position on a sfr - colour diagram . in each stellar mass bin ,
the star forming galaxies have a bluer @xmath7 colour than the passive galaxies and the range of colours that the star forming galaxies have in each stellar mass bin is quite narrow .
passive galaxies have a range of @xmath7 colours .
each stellar mass bin is successively redder , this is due to the relation between stellar mass and colour ( figure [ fig : colourmass ] ) .
as seen in figure [ fig : ssfr_colour ] there is a relationship between the ssfr and colour .
this is partially expected
ssfr traces the effect that star formation has had on the the recent history of the galaxy .
this will also be borne out in the colour of the galaxy , with the more recent the star formation , the more affect it has on a galaxy s colour .
this then explains why the different mass bins are separated along the colour axis of the sfr - colour diagram ( figure [ fig : sfrcolourmass ] ) : the redder systems have lower ssfrs and so for the same sfr they must be more massive than bluer systems .
this relation could be affected by our selection criteria .
we could be missing low - mass passive objects which are red .
however due to their low mass , they would not lie considerably below galaxies already plotted on the ssfr - colour diagram .
the galaxies investigated appear to follow what would be expected in a downsizing senario .
there are significantly fewer galaxies with m@xmath46forming stars and the most massive galaxies are only forming stars at the highest redshifts probed .
more detailed studies into the sfr - m@xmath53 as a function of redshift are not possible due to the small size of our sample .
we have analysed nir spectroscopy of 41 massive ( m@xmath10 ) galaxies taken from the powir survey of which half were chosen to have significant 24-@xmath14 m flux , no x - ray emission and within the range @xmath54 .
we have acquired nir spectroscopic data for these systems and measured sfrs from the h@xmath1 emission line .
the h@xmath1 emission line is the best direct measure of instantaneous star formation as it is coupled directly to the massive stars in formation , is largely independent of metalicity and is less effected by dust attenuation than either the [ oii ] emission line or uv flux .
we investigated several methods of correcting high - redshift galaxies for dust using multi - wavelength methods and chose to ultimately apply a method derived from uv data .
we find that the h@xmath1 sfr measurements are in good agreement with those measured by uv emission , but are slightly higher than those measured from 24-@xmath14 m emission . by not using the correction to the @xmath55 value due to differences associated with gas emission and stellar light ( calzetti et al .
2000 ) our h@xmath1 corrected sfrs are similar to or less than the galaxies 24-@xmath14 m derived sfrs .
the agn contamination of our sample is low
although highly obscured agn may exist within the sample there is no evidence that there are any .
we find that there is a drop in the fraction of galaxies with detections of h@xmath1 for systems with a stellar mass greater than m@xmath56 .
the massive galaxies with detected h@xmath1 are all found at redshifts greater than @xmath57 0.93 .
the fraction of galaxies with detected h@xmath1 also drops for galaxies that are red in u - b colour .
the sharpness of this drop implies that star formation is truncated over a very short period of time rather than dropping steadily over an extended epoch .
we investigated the h@xmath1 sfr as a function of stellar mass and find that sfr stays roughly constant with stellar mass .
likewise , the ssfr drops with increasing mass , especially at m@xmath58 .
this implies that at @xmath32 1 , galaxies with m@xmath59are less likely to be forming stars , and when they are , they are doing so at a much lower rate relative to their mass .
these massive galaxies have a lower ssfr than lower mass galaxies , implying that they have on average a stellar population that formed earlier , and must have had their star formation quenched at a higher redshift than the lower mass galaxies .
we find that the observed slope in the sfr - mass relation appears to be slightly flatter than what is found in previous studies .
the ssfr - mass relation also agrees with the relation predicted by an exponentially decreasing star formation history model with mass dependent formation redshifts .
our measured ssfrs however are slightly higher than the noeske et al .
( 2007b ) model predictions .
we also investigate the sfr as a function of colour and find that the galaxies that are detected in h@xmath1 form stars at similar rates independent of colour .
when split into bins of stellar mass , within each bin we find star forming galaxies that have a narrow range of @xmath7 colours while galaxies not detected in h@xmath1 have a wider range of colours which are mostly redder than the star forming ones .
these mass bins can be viewed as populations of galaxies in different stages of their evolution from star forming to passive , and suggests that galaxies drop in star formation before becoming redder .
we also find that the specific sfr drops with increasing @xmath7 colour ( redness ) for all galaxies ; this is probably due to both being a measure of the effect on the galaxy of recent star formation .
this relation leads to the more massive galaxies ( with lower ssfr ) becoming redder .
evolutionary tracks on the sfr - colour plot are therefore shifted towards redder colours for the most massive galaxies .
we also investigate the structures of our sample which have acs hst imaging .
we find that there is possible relation found between a galaxy s sfr and its asymmetry as measured by the cas parameters ( conselice 2003 ) .
h@xmath1-derived sfrs at @xmath0 appear to be similar to those measured using other tracers .
there is a large body of work studying star formation at these redshifts using these other tracers , and we have shown , using the more reliable h@xmath1 values that these other results are consistent at least up to @xmath60 .
as nir spectroscopic instrument technology continues to evolve , especially on 8 - 10 m telescopes , the growth in the number and quality of h@xmath1 measurements , particularly at high redshift , will allow us to compare these different sfr tracers at higher redshifts and therefore obtain an even more accurate measure of the star formation history of galaxies .
funding for the trip to the obseratorio del roque de los muchachos was provided by the university of nottingham .
we thank nathan bourne for useful and informative discussions .
we acknowledge support from the stfc as well as a leverhulme trust prize to cjc .
the william herschel 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 astrofsica de canarias .
99 acosta - pulido j. a. , ballesteros e. , barreto m. et al . , 2003 ,
ingn , 7 , 15 afonso j. , hopkins a. , mobasher b. , almeida c. , 2003 , apj , 597 , 269 alonso - herrero a. , quillen a. c. , simpson c. , efstathiou a. , ward m. j. , 2001 , aj , 121 , 1369 conselice c. j. , bluck a. f. l. , buitrago f. et al . , 2011 , mnras , 413 , 80 daddi e. , dickinson m. , morrison g. et al .
, 2007a , apj , 670 , 156 ( d07 ) daddi e. , alexander d. m. , dickinson m. et al . , 2007b , apj , 670 , 173 davis m. , faber s. m. , newman j. et al . , 2003 ,
spie , 4834 , 161 davis m. , guhathakurta p. , konidaris n. p. et al . , 2007 , apj , 660 , 1 fontana a. , pozzetti l. , donnarumma i. et al .
, 2004 , a&a , 424 , 23 frster schreiber n. m. , genzel r. , bouch n. et al .
, 2009 , apj , 706 , 1364 gallagher j. d. , hunter d. a. , bushouse , h. , 1989 , aj , 97 , 700 johnson b. d. , schiminovich d. , seibert m. et al .
, 2006 , apj , 644 , l109 johnson b. d. , schiminovich d. , seibert m. et al . , 2007 , apjs , 173 , 377 kauffmann g. , heckman t. m. , white s. d. m. et al . , 2003a , mnras , 341 , 33 kauffmann g. , heckman t. m. , tremonti c. et al . ,
2003b , mnras , 346 , 1055 kennicutt r. c. , 1983 , apj , 272 , 54 , noeske k. g. , weiner b. j. , faber s. m. et al .
, 2007a , apj , 660 , l43 ( n07 ) noeske k. g. , faber s. m. , weiner b. j. et al . 2007b , apj , 660 , l47 osterbrock d. e. 1989 , astrophysics of gaseous nebulae and active galactic nuclei , university science books , mill valley , ca papovich c. , moustakas l. a. , dickinson m. et al . , 2006 , apj , 640 , 92 papovich c. , rudnick g. , le floch e. et al . , 2007 , apj , 668 , 48 prez - gonzlez p. g. , rieke g. h. , villar v. et al . , 2008a , apj , 675 , 234 prez - gonzlez p. g. , trujillo i. , barro g. , gallego j. , zamorano j. , conselice c. j. , 2008b , apj , 687 , 50 salim s. , rich r. m. , charlot s. et al . , 2007 ,
apjs , 173 , 267 schiminovich d. , wyder t. k. , martin d. c. et al . , 2007 , apjs , 173 , 315 seibert m. , martin d. c. , heckman t. m. et al . , 2005 , apj , 619 , l55 silverman j. d. , kovac k. , knobel c. , et al . , 2009 ,
apj , 695 , 171 smail i. , ivison r. j. , blain a. w. , 1997 , apj , 490 , l5 | we present a near - infrared spectroscopic study of a stellar mass selected sample of galaxies at @xmath0 utilising the liris multi - object spectrograph on the william herschel telescope .
we detect continuum , and the h@xmath1 line for our sample , which is one of the better direct tracers of star formation in external galaxies .
we spectroscopically measure the h@xmath1 emission from 41 massive ( m@xmath2 ) galaxies taken from the powir survey with spectroscopic redshifts 0.4 @xmath3 z@xmath4 @xmath3 1.4 .
we correct our h@xmath1 fluxes for dust extinction by using multi - wavelength data , and investigate sfr trends with mass and colour .
we find a drop in the fraction of massive galaxies with m@xmath5which are detected in h@xmath1 emission at @xmath6 .
we furthermore find that the fraction of galaxies with h@xmath1 emission drops steadily and significantly with redder @xmath7 colours at z @xmath8 , and that the ssfr drops with increasing @xmath7 colour for galaxies at all masses . by investigating the sfr - mass relation
we find that the sfr is roughly constant with mass , in possible contrast to previous work , and that the specific star formation rate ( ssfr ) is lower in the most massive galaxies .
the scatter in the sfr vs. mass relationship is very small for those systems with ongoing star formation which suggests that star formation in the most massive galaxies at @xmath0 shuts off rather abruptly over @xmath9gyr , without an obvious gradual decline .
we furthermore investigate the sfr as a function of @xmath7 colour divided into different mass bins , revealing a tracer of the epoch of transition from star forming to passive , as a form of star formation `` downsizing '' .
this suggests that the shut off of star formation occurs before the change in a galaxy s colour .
we find that galaxy stellar mass is the primary driving mechanisms behind the star formation history for these galaxies and discuss several possible mechanisms for regulating this process .
[ firstpage ] galaxies : star formation - galaxies : evolution - galaxies : formation |
investigations of the lyapunov instabilities in many - particle systems are often motivated by the role of chaos in the foundations of statistical physics @xcite , which is still not fully understood . in general , interacting many - particle classical systems are expected to be chaotic .
the largest lyapunov exponents and properties of the entire lyapunov spectra have been calculated numerically for classical many - body systems , such as gases of hard - core particles @xcite , fluids with soft interactions @xcite , and lattice two - dimensional rotators @xcite , and analytically in a few cases @xcite . in the present paper ,
we focus on lattices of interacting classical spins .
classical spins often appear in the theoretical studies as the large - spin limit of quantum spins
. moreover , even when one deals with lattices of spins 1/2 , parallels between classical and quantum spin dynamics still remain .
these parallels have recently received much attention , in particular , in the context of the asymptotic exponential - oscillatory behavior of nuclear spin decays in solids @xcite .
these decays were identified in refs .
@xcite with chaotic eigenmodes in both classical and quantum many - spin systems .
although it appears very likely _ a priori _ that lattices of interacting classical spins exhibit chaotic dynamics , no systematic investigation of the chaotic properties of these lattice was undertaken until our previous work @xcite , which presented a survey of the largest lyapunov exponents for a very large variety of spin lattices and hamiltonian anisotropies .
the principal finding of ref .
@xcite was that all hamiltonians considered , with the exception of the ising case , led to chaotic dynamics as evidenced by the nonzero value of the largest lyapunov exponent .
we also obtained both analytically and numerically the power - law scaling of the largest lyapunov exponent in the vicinity of the integrable ising limit . in the present paper , we complement the findings of ref .
@xcite in several respects .
namely , we compute complete lyapunov spectra for a few selected spin lattices and show their dependence on the lattice size , and also present a more extensive investigation of the lattice size dependence of the largest lyapunov exponent .
finally , we discuss the dependence of the largest lyapunov exponent on the hamiltonian anisotropy with particular emphasis on the difference between bipartite and nonbipartite lattices .
we consider periodically closed spin lattices with the nearest - neighbor ( nn ) interaction hamiltonian of the following kind : @xmath0 where @xmath1 are the three projections of the classical spin vector of unit length on the @xmath2th lattice site ( i.e. @xmath3 ) , and @xmath4 , @xmath5 , @xmath6 are the coupling constants , which we also normalize by condition @xmath7 .
below , we often mention ising , heisenberg and `` anti - heisenberg '' limits of the hamiltonian ( [ h ] ) .
the ising hamiltonian corresponds to @xmath8 , @xmath9 , heisenberg hamiltonian to @xmath10 , and , finally , anti - heisenberg hamiltonian to @xmath11 .
the total number of spins in the lattice is denoted as @xmath12 .
the phase space of such a lattice has dimensionality @xmath13 .
we consider seven lattices shown in fig .
[ fig : lattices ] and labeled as ( l1-l7 ) . lattices ( l1-l5 ) are bipartite , which means that they can be divided into two sublattices such that all the interacting neighbors for a spin on one sublattice belong to the other sublattice
. lattices ( l6,l7 ) are nonbipartite . the equations of motion associated with the hamiltonian ( [ h ] ) can be obtained in the poisson - bracket formalism @xcite : @xmath14 , where the index @xmath15 admits values @xmath16 , @xmath17 or @xmath18 representing the projections @xmath19 , @xmath20 , or @xmath21 , respectively .
the primary poisson brackets are : @xmath22 , where @xmath23 is the kronecker delta and @xmath24 the levi - civita symbol .
the evaluation of @xmath25 involves the following poisson brackets : @xmath26 .
the resulting equations of motion are @xmath27 where @xmath28 is the local field given by the expression @xmath29 here @xmath30 , @xmath31 and @xmath32 are the unit vectors along the respective directions , and @xmath33 implies the summation of the nearest neighbors of the @xmath2-th lattice site . in this work ,
we restrict ourselves to the lyapunov instabilities on the zero energy shell , which corresponds to infinite temperature in the microcanonical sense .
since the dynamics of this system are fully time - reversible , positive and negative lyapunov exponents are expected to form conjugate pairs with equal absolute values . in the general case of different @xmath4 , @xmath5 , and @xmath6
, there should be only two zero lyapunov exponents , corresponding to the energy and the time shift directions . in the case
@xmath34 , of which the anti - heisenberg case is an example , the @xmath21-component of the total spin polarization becomes an integral of motion , and hence one more pair of lyapunov exponents should also become equal to zero . in the heisenberg case , @xmath35 , all three components of the total spin polarization become integrals of motion .
however , these three integrals of motion are not dynamically independent , because one of them can be obtained as the poisson bracket of the two others .
the two independent integrals of motion thus imply two additional pairs of zero lyapunov exponents . in the ising case ,
the @xmath21-component of each spin is an integral of motion .
hence , the system is integrable and all lyapunov exponents are equal to zero .
now we discuss the connections between the lyapunov spectra of different anisotropic hamiltonians .
we first note that infinite - temperature lyapunov spectra are expected to be identical for the hamiltonians with the coupling constants @xmath36 and @xmath37 , because the zero - energy shells in the two cases are identical , while the change of the sign of the coupling constants amounts to the operation of time - reversal and flips the sign of the energy .
we further note that , for bipartite lattices , the infinite temperature lyapunov spectra for the coupling constants @xmath36 and @xmath38 should also be identical . in order to see this , one should make the transformation @xmath39 and @xmath40 for one of the two sublattices forming the bipartite lattice and then examine the resulting equations of motion .
the two observations made above also imply that , for a given bipartite lattice , the lyapunov spectra for the heisenberg and the anti - heisenberg hamiltonians are identical to each other , which means that , in the latter case , the lyapunov spectrum has three pairs of zero lyapunov exponents instead of two expected for the generic case of @xmath41 .
the extra pair of the zero exponents originates from the following two ( not independent ) integrals of motion : @xmath42 or @xmath43 , where @xmath44 is the index taking value 0 for one sublattice and 1 for the other .
the same considerations also imply that any hamiltonian with @xmath45 , or equivalent , on a bipartite lattice can be converted to the axially symmetric hamiltonian with @xmath46 .
therefore , the original hamiltonain has an extra pair of zero lyapunov exponents corresponding to the integral of motion @xmath43 .
small spin clusters may have additional nontrivial integrals of motion .
in particular , any 4-spin periodic chain with the general anisotropic hamiltonian of form ( [ h ] ) is fully integrable .
this is because the first and third spins in this chain rotate in the same local field @xmath47 $ ] , while the second and the fourth spins rotate in another local field @xmath48 $ ] .
therefore , there are two additional integrals of motion , namely : @xmath49 and @xmath50 . finally , one can also check that @xmath51 is also an integral of motion . as a result ,
the number of integrals of motion ( including energy ) becomes 4 , while the dimensionality of the phase space is 8 , i.e. the problem is fully integrable . in the case of the isotropic heisenberg hamiltonian
, a much larger variety of small spin clusters with nontrivial integrals of motion were cataloged in ref .
@xcite .
the equations of motion were integrated using a fourth - order runge - kutta algorithm with a time - step @xmath52 .
this is sufficiently small so that on the time scale of our simulations , energy is conserved with the 6-digit accuracy .
simulations were run for 20000 time units , which was sufficient for accurate convergence of the smaller exponents .
the lyapunov exponents were obtained by using a version of the standard reorthonormalization algorithm @xcite . in order to obtain the first @xmath53 largest lyapunov exponents , we numerically propagate a reference trajectory @xmath54 and @xmath53 initially orthogonal perturbation vectors @xmath55 . at every time step ,
we propagate the perturbations using the linear tangent space map that is obtained by numerically taking derivatives , @xmath56 after each time interval @xmath57 , we hierarchically reorthogonalize the perturbation vectors @xmath55 using the gram - schmidt procedure , and then renormalize their lengths back to the initial values .
the renormalization factors are recorded for the @xmath58th time interval @xmath59 as @xmath60 .
finally , we compute the @xmath2th lyapunov exponent using the formula @xmath61 , where @xmath62 is the number of the renormalization time intervals . as a test of the accuracy of our numerical routine
, we have checked that the lyapunov exponents form conjugate pairs and that the number of zero lyapunov exponents is equal to the number expected from the symmetry of the hamiltonian as discussed in section [ expectations ] .
we have also checked the symmetries between heisenberg and anti - heisenberg lyapunov spectra expected based on the arguments presented in section [ expectations ] . in order to obtain initial conditions at zero total energy ,
we first choose random orientations for all spins .
this produces a total energy close to zero , but with fluctuations of the order of @xmath63 .
then , in order to arrive at zero total energy , we evolve the dissipative equations of motion : @xmath64 depending on the sign in front of the right - hand side , this increases or decreases the total energy associated with the hamiltonian ( [ h ] ) .
once the zero value of the total energy is reached , we additionally assure the randomness of the initial conditions on the energy shell by performing @xmath65 sequential rotations of random spins by random angles around the directions of their respective local fields @xmath28 given by eq.([h ] ) .
these rotations preserve the total energy .
we have computed the full lyapunov spectra for the linear chain ( l1 ) , cubic lattice ( l5 ) and triangular lattice ( l7 ) with the heisenberg hamiltoninan , and also for the triangular lattice ( l7 ) with the anti - heisenberg hamiltonian .
the results are presented in fig .
[ fig : spectraheis ] .
the spectra ( i.e. the lyapunov exponents as a function the exponent s index ) are typically weakly convex , regardless of the lattice . in the case of a cubic lattice with heisenberg interaction , the spectrum is actually very close to linear . as explained in section [ expectations ] , the lyapunov spectra for the bipartite lattices ( l1 ) and ( l5 ) with anti - heisenberg hamiltonian are identical to those already presented in fig .
[ fig : spectraheis ] for the heisenberg hamiltonian .
the effect of the lattice size on the lyapunov spectrum for the linear chain with heisenberg hamiltonian is shown in fig . [
fig : sizespectrum ] .
the size dependence of the spectrum appears to be very weak for @xmath66 and becomes virtually unobservable for chains containing more than 32 spins . comparing the lyapunov spectra of classical spins with the lyapunov spectra of other many - particle systems , we first note , that the spectra presented in fig .
[ fig : spectraheis ] do not exhibit an offset from zero for the smallest positive exponents .
such an offset was observed in gases of particles @xcite and high - dimensional billiards @xcite but not in systems with sufficiently soft interactions @xcite .
the classical spin lattices obviously belong to the latter group .
we further remark on the existence of the delocalized lyapunov - goldstone modes , which were observed in dilute gases @xcite and in some other extended systems @xcite .
if these modes exist in the spin systems , the projections on single spins of the lyapunov vectors corresponding to the smallest nonzero lyapunov exponents should exhibit a sinusoidal dependence on the positions of spins . in the present work
, we did not investigate the properties of the lyapunov vectors systematically .
however , our several attempts to find the lyapunov - goldstone modes for the infinite temperature energy shells did not produce any positive evidence : the projections of the lyapunov vectors for all exponents were strongly localized on the lattice and not sinusoidal .
there is also no indication of a dependence of the smalles non - vanishing exponent on the system length , as is characteristic for the lyapunov - goldstone modes .
the lyapunov - goldstone modes may still exist in classical spin systems at low temperatures but this is a subject beyond the scope of the present paper .
accurate numerical calculations of full lyapunov spectra are very demanding , because the computational cost grows as @xmath67 .
for this reason , the lattices investigated in the preceding subsection were relatively small . in this subsection
, we focus only on the largest lyapunov exponents , @xmath68 .
the cost of computing @xmath68 grows only as @xmath12 , which allows us to investigate much larger lattices and a much greater variety of hamiltonians .
an extensive investigation of this kind was already reported by us in ref .
@xcite . in the present and
the next subsections we present some results and analysis that were not included in ref .
@xcite . in this subsection
, we investigate the dependence of @xmath68 on the lattice size for all seven lattices shown in fig .
[ fig : lattices ] with heisenberg and the anti - heisenberg hamiltonians .
the results are shown in fig .
[ fig : sizedependence ] .
they indicate that , within our numerical accuracy , @xmath68 becomes size - independent for sufficiently large lattices . in particular , on the basis of these results even slow logarithmic growth of @xmath68 with the lattice size
can be excluded .
the saturation of @xmath68 with the lattice size can be explained on the basis of the following consideration .
the exponential growth of the perturbation vector @xmath69 in many - spin phase space with rate @xmath68 implies that the projection @xmath69 on the subspace of each individual spin @xmath70 should also , on average , grow exponentially with the same rate .
the instantaneous growth rate of the perturbations of the coordinates of a given spin can be obtained from the linearized equations of motion , which , in turn can be obtained from equations ( [ eq : eom ] ) and ( [ h ] ) , @xmath71 here , @xmath72 indicates a perturbation in the @xmath2-th spin , while @xmath73 indicates a perturbation in its @xmath19th component . from these equations
, one can see that the instantaneous growth rate is limited from above by the value on the order of the maximum value of the local field @xmath74 , where @xmath75 is the number of nearest neighbors for each lattice site . since this constraint does not depend on the lattice size
, the growth of the perturbation belonging to the largest lyapunov exponent must saturate as the lattice size increases . in principle
, it might also be possible for the largest lyapunov exponent to oscillate with the lattice size , but in a system with exponential decay of spatial correlation , this is extremely unlikely . in ref .
@xcite , we conducted a systematic survey of the dependence of @xmath68 on the hamiltonian anisotropy on the `` interaction sphere '' constrained by the condition @xmath7 .
we have found that the principal parameter controlling this dependence is @xmath76 . in particular , this parameter quantifies the approach to the integrable ising case corresponding to @xmath77 .
the main plot of ref .
@xcite is reproduced in fig .
[ fig : prl ] .
here we focus on the difference between the anisotropy dependence of @xmath68 for bipartite lattices ( l1-l5 ) and nonbipartite lattices ( l6 , l7 ) . as can be seen in fig .
[ fig : prl ] , the bipartite lattices ( l1-l5 ) show a nearly universal dependence @xmath78 , which scales only with the number of interacting neighbors .
the small spread of the sampled values of @xmath68 at a given value of @xmath79 indicates that @xmath68 depends only very little on the ratio of the two coupling constants that have smaller absolute values .
our investigation of the nonbipartite lattices ( l6,l7 ) was , in fact , motivated by the impression that the number of interacting neighbors alone determines the entire anisotropy dependence of @xmath68 .
we wanted to compare @xmath68 for the lattices ( l6 ) and ( l3 ) , where , in the both cases , each site has four nearest neighbors , and for the lattices ( l7 ) and ( l5 ) , each site has six nearest neighbors .
we found , however , that , as seen in fig .
[ fig : prl ] , the nonbipartite lattices ( l6 ) and ( l7 ) exhibit a noticeable fork - like spread of @xmath68 as @xmath79 approaches @xmath80 .
the upper and the lower tips of the fork correspond to the anti - heisenberg and heisenberg hamiltonians , respectively . in the anti - heisenberg case ,
the value of @xmath68 is close to that of a bipartite lattice with the same number of nearest neighbors . in the heisenberg case ,
the value of @xmath68 is closer to the bipartite lattice with one fewer nearest neighbor .
this spread indicates that the knowledge of @xmath79 alone is insufficient to determine @xmath68 . the two - dimensional anisotropy dependence behind this spread is shown in fig .
[ fig : crosssection ] , where we present it for lattices ( l5 ) and ( l7 ) in the form of the color density plots as a function of @xmath4 and @xmath5 . less obvious from fig .
[ fig : prl ] , is the fact that the significant majority of the sampled values of @xmath68 for the lattices ( l6 ) and ( l7 ) agree very well with the dependence @xmath78 for the bipartite lattices ( l3 ) and ( l5 ) , respectively .
the comparison between figs .
[ fig : crosssection](a ) and ( b ) clearly shows that the difference between lattices ( l5 ) and ( l7 ) is only pronounced , when all three interaction constants have the same sign and roughly the same value or , in other words , when they approach the heisenberg limit .
this implies that @xmath68 for the anti - heisenberg hamiltonian has a more typical value than for the heisenberg hamiltonian , and that the expected universality of @xmath78 for the same number of nearest neighbors still roughly holds even for nonbipartite lattices .
it thus appears that the conservation of the total spin , i.e. @xmath81 , in the heisenberg case leads to the reduction of the effective number of the nearest neighbors by roughly one , as far as the value of @xmath68 is concerned .
we suspect that the situation here is similar to the origin of the frustrated low - temperature magnetism for the heisenberg model on nonbipartite lattices .
the interaction energy for a spin pair is minimal when the two spins are antiparallel to each other , but , on a nonbipartite lattice , such an antiparallel configuration can not simultaneously exist for all pairs of interacting spins .
hence the ground state of a nonbipartite lattice is frustrated .
in the case of the lyapunov instabilities , the conservation of the total spin polarization implies that the perturbation vector @xmath82 corresponding to the largest lyapunov exponent does not grow along the direction of the total spin polarization , i.e @xmath83 . at the same time
, @xmath84 grows , on average , exponentially , which means that , for @xmath85 , @xmath86 .
this , in turn , implies that , in the leading order , @xmath87 . as a result ,
when a given projection @xmath88 grows , this growth needs to be compensated by the growth of @xmath89 for other spins in the opposite direction .
however , since the interaction is local , these `` other spins '' can only be the nearest neighbors .
achieving the maximum growth of the perturbation , and hence the largest value of the lyapunov exponent presumably requires that the perturbation vector @xmath69 maximizes the anti - alignment of @xmath88 for the adjacent sites .
for the bipartite lattices , this anti - alignment can , in principle , be made perfect , but for the nonbipartite lattices this is impossible .
such an explanation is consistent with the fact that @xmath68 for lattices ( l6,l7 ) in the heisenberg limit is smaller than in the anti - heisenberg limit .
it seems also to be connected to the fact that the above reduction approximately leads to the value of @xmath68 for a bipartite lattice with roughly one fewer nearest neighbor per site .
we finally remark that the overall small spread of values of @xmath68 for bipartite lattices at a given value of @xmath79 is related to the symmetries described in section [ expectations ] , which imply that the lyapunov exponents are identical for eight combinations of the coupling constants characterized by the same value of @xmath79 , namely : @xmath36 , @xmath38 , @xmath90 , @xmath91 , @xmath37 , @xmath92 , @xmath93 , @xmath94 .
the values of @xmath68 can not change much between these eight points .
we have investigated numerically the lyapunov spectra of systems of many classical spins for a variety of lattices and coupling constants at infinite temperature .
the possibility of varying the coupling constants , from the highly symmetric isotropic heisenberg model , through partially symmetric couplings such as the anti - heisenberg model , to the completely anisotropic integrable case , makes these systems particularly interesting .
we have presented : ( i ) calculations of the lyapunov spectra for selected lattices of interacting classical spins ; ( ii ) investigations of the lattice - size dependence of the lyapunov spectra ; ( iii ) investigations the largest lyapunov exponents for a broader group of coupling constants , lattices and large lattice sizes ; and ( iv ) discussion of the difference between the largest lyapunov exponents for bipartite and nonbipartite lattices .
the computed lyapunov spectra were found to be weakly convex .
we have observed no finite offset of the smallest positive exponents , and , to the extent that we have searched , we have not encountered any evidence of lyapunov - goldstone modes .
both the largest lyapunov exponents and the whole lyapunov spectra were found to become independent of the lattice sizes for sufficiently large lattices .
we have given an analytical argument explaining this finding .
in addition , we have found that the largest lyapunov exponents for bipartite and nonbipartite lattices depend differently on the anisotropy of the coupling .
this is due to the special symmetry of the bipartite lattices with respect to the sign change of the two out of three coupling constants .
a.s.dws work is financially supported by an unga forskare grant from the swedish research council .
b.v.f . acknowledges the hospitality of the kavli institute for theoretical physics at the university of california , santa - barbara , where a part of this paper was written , and support by the national science foundation under grant no .
nsf phy11 - 25915 .
the numerical part of this work was performed at the bwgrid computing cluster at the university of heidelberg .
h. a. posch and r. hirschl .
simulations of billiards and of hard body fluids . in d.
szasz , editor , _ hard ball systems and the lorentz gas _ , volume 101 of _ encyclopedia of mathematical sciences .
_ , page 280 .
springer - verlag , new york , 2000 . | we numerically investigate lyapunov instabilities for one- , two- and three - dimensional lattices of interacting classical spins at infinite temperature .
we obtain the largest lyapunov exponents for a very large variety of nearest - neighbor spin - spin interactions and complete lyapunov spectra in a few selected cases .
we investigate the dependence of the largest lyapunov exponents and whole lyapunov spectra on the lattice size and find that both quickly become size - independent .
finally , we analyze the dependence of the largest lyapunov exponents on the anisotropy of spin - spin interaction with the particular focus on the difference between bipartite and nonbipartite lattices . |
a0620 - 00 was discovered on august 3 , 1975 by the ariel-5 satellite as a bright x - ray nova ( elvis et al . , 1975 ) and designated as v616 mon in the `` general catalogue of variable stars '' ( kholopov et al . , 1998 ) .
the precursor was detected on archival photographic plates at a magnitude of @xmath2 20.0 - 20.5 ( ward et al .
, 1975 , eachus et al . , 1976 ) .
the outburst light curves of the nova in x - ray , optical , infrared and radio wavelengths in 1975 - 76 were gathered by kuulkers ( 1998 ) .
the nova reached 11.@xmath34 in the _ b - _ band ( shugarov , 1976 ) , 30 crab in x - rays and 200 mjy in the 962 mhz radio band .
an outburst in x - rays was observed with satellites ariel-5 , sas-3 and with the x - ray detector `` filin '' on the ussr orbital station salyut-4 .
the object was classified as a soft x - ray transient , and according to white et al .
( 1984 ) , it was in a `` high state '' during most of the x - ray observations .
the spectral changes during the rise to the outburst indicated a change from the `` low state '' to a high state . the different x - ray states ( quiescent , low , intermediate , high , very high ) depend on an increasing mass accretion rate through the accretion disk .
after dramatic fading in the optical brightness in april 1976 ( martynov et al . , 1976 )
, v616 mon passed into a quiescence ( lyuty & shugarov , 1979 ) .
eachus et al .
( 1976 ) found the previous outburst of the nova on archival plates of the harvard college observatory .
the nova reached @xmath4 in november 1917 .
shugarov et al .
( 1976 ) detected the nova at two simeiz plates taken on january 20 , 1918 ( jd 2421461.2 ) and estimated its brightness as @xmath5 .
two pairs of very short ( 0.5 - 5 ms ) flares were detected in quiescence on february 13 and 14 , 1986 by shvartsman et al .
( 1989 ) with the 6-m bta telescope .
the brightness temperature in the flare was at least @xmath6k , indicating non - thermal accretion onto a black hole .
the blue - visual ( 4000 - 6400 fwhm ) passband and @xmath7 photometry and spectroscopy of the object taken by mcclintock & remillard ( 1986 ) showed that the object is a compact binary with the period of 0.3230141(4 ) days .
the photometric binary cycle has a double hump structure as a result of the ellipsoidal shape of the red dwarf companion filling its roche lobe .
the authors derived a mass function @xmath8 = [email protected] m@xmath10 from the radial velocity curve of the red k5-k7 dwarf companion ( oke , 1977 ) and estimated the mass of the compact object to be @xmath11 3.2 m@xmath10 , so it is an excellent black - hole candidate .
mcclintock & remillard ( 1986 ) determined the epoch of minimum light in the binary light curve ( lc ) at jd 244 5477.827 . using radial velocity measurements they established that this is the phase of an inferior conjunction of the compact object ( the black hole is in front of the red dwarf with respect to the observer ) .
the @xmath12 photometry in the years 1991 - 95 confirmed the presence of the double humped light curve with a period of @xmath13 ( leibowitz et al . , 1998 ) . while there were no indications of a short - term variability , the mean brightness of the object varied on a time - scale of hundreds of days .
subsequent @xmath14 photometry in 1999 - 2000 , obtained by gelino et al .
( 2001 ) , has shown that the shape of an ellipsoidal light curve changed with time .
the @xmath15 and infrared @xmath16 photometry spanning the period 1999 - 2007 was published by cantrell et al .
they distinguished three distinct optical states .
light curves in a `` passive '' state can be completely explained by periodic orbital variations and photometric errors .
the `` loop '' and `` active '' states are characterized by the increases of scatter of observations . while a short time scale rms variability in the loop state was 0.05 mag in @xmath15 and 0.047 mag in @xmath17 , in the active state it reached 0.088 mag in @xmath15 and 0.057 mag in @xmath17 .
the loop and active states are brighter than a passive state and were detected mainly in the years 2003 - 2007 .
.journal of observations .
[ cols="<,<,>,^,<,<,<",options="header " , ] telescopes : 50/200-cm in star lesn ; 60/750-cm in crimea ( 60a ) ; 60/750-cm in star lesn ( 60b ) ; 125/2000-cm in crimea ; 250/2000-cm in caucasus .
+ + + + + + cantrell et al . ( 2010 ) modelled ellipsoidal light curves of v616 mon in passive states , which occurred during the years 1988 - 2003 , to determine the following basic parameters of the binary : _ i _
= 50@xmath18 , @xmath19 = [email protected] @xmath20 , @xmath21 = 0.40@xmath9 0.045@xmath20 and its distance _ d _ = [email protected] kpc .
they found that all non - ellipsoidal variability originates in a variable accretion disk component .
the radiation from the disk is responsible for secular changes in brightness , flickering present in non - passive states and periodic modulations in the shape of the lc .
khruzina et al .
( 1988 ) suggested the existence of a hot spot arising on the outer edge of the system .
marsh et al .
( 1994 ) and neilsen et al .
( 2008 ) presented doppler tomography of the accretion disk , revealing an asymmetrical disk with a prominent bright spot associated with the impact of the mass transfer stream on the edge of the disk .
gonzles hernndes et al .
( 2014 ) used spectroscopic data to determine a negative orbital period derivative @xmath22 = [email protected] ms yr@xmath23 from five times of inferior conjunctions of the red dwarf secondary .
comprehensive reviews of close binaries with a black - hole component were published by cherepashchuk ( 1996 , 2013 , 2014 ) .
in the years 2003 - 16 we obtained the time - resolved ccd photometry of v616 mon in @xmath1 passbands and the integral light ( i.e. without filters , denoted by @xmath24 in this paper ) , using the 50125 cm telescopes at the crimean astronomical station of the lomonosov moscow state university and observatory star lesn of the astronomical institute of the slovak academy of sciences . the observations in december 2015 were carried out with the 250 cm telescope , located in the caucasian mountain observatory of the sternberg astronomical institute of the lomonosov moscow state university . in total , we obtained 7660 ccd frames .
the journal of observations is given in table 1 .
the lc of the mean night run observations in different passbands and integral light is shown in fig .
1 . the variability on the scale from days to months can reach 0.35 mag . the standard deviation @xmath25 of the observations taken during every observational season , after removal the mean brightness variations ,
is shown in the upper part of the figure .
it characterizes the activity of the system that reached its maximum in 2012 .
the orbital phase lc can be characterized by a double wave , caused by an ellipsoidal shape of the red dwarf component .
we used the minima times determined by our orbital lcs and the times of inferior conjunction of the red dwarf derived from spectroscopic data ( gonzles hernndez et al . ,
2014 and references therein ) shifted by 0.5 in the orbital phase , to improve the light elements of the binary . for more precise spectroscopic data we applied the weight 10 times larger than for photometric ones .
a period analysis of minima times allowed us to improve the linear ephemeris of the binary to : @xmath26 the primary minimum corresponds to a superior conjunction of the red dwarf .
the average 2003 - 2016 phase lc of v616 mon , after removal of the mean brightness variations , is depicted in fig .
2 . the standard deviation @xmath25 is shown in the upper part of the figure .
we clearly see that the phase lc for the whole time interval exhibits a large scatter , significantly higher than observational errors .
the scatter of the data 0.2 - 0.4 mag , caused by a flickering , can increase to 0.6 mag in orbital phases 0.3 0.4 and 0.8 0.9 .
flickering is due to the gaseous stream - accretion disk interaction in a bright spot , where the gas stream from the l1 point hits the outer edge of the accretion disk .
bisikalo et al .
( 2008 ) and kononov et al .
( 2012 ) performed three - dimensional gas - dynamical simulations for accretion disks and found that the spot at the disk is extended and has the shape of a `` hot line '' , or an arc .
the extended hot line is best visible in the phases 0.3 0.4 and , due to the favorable inclination angle of the binary _ i _ = 50@xmath18 ( cantrell et al . , 2010 ) , it is partly visible also in the phases 0.8 0.9 .
the position of the observer is approximately perpendicular to the hot line in these phases , so the flickering amplitude reaches the maximum .
the lcs of v616 mon at six nights during the years 2003 - 2015 are illustrated in fig .
the observations show a superposition of the orbitally - related wave variation and a rapid flickering variability with a characteristic time scale of about 30 min and the amplitude of 0.2 - 0.4 mag . while the orbital wave variation dominates all lcs , the largest flickering variability was detected in the years 2011 - 14 with its maximum in 2012 .
we plotted also the night run variability of the field star , with approximately the same brightness , located close to v616 mon .
the visible dispersion of the field star brightness shows the actual measurement accuracy of v616 mon . in 2015
the amplitude of flickering decreased , so the system entered a non - active state .
as we have mentioned in the introduction , the last two outbursts of v616 mon occurred in 1917 and 1975 .
if the outburst is triggered by a periodic mechanism , we can expect a new outburst during the next 10 - 20 years . in 2004 - 6 and 2009 - 14 active stages
we observed flickering caused by an increase of the mass transfer from the red dwarf optical component to the accretion disk of the black hole ( shugarov et al . , 2015 ) .
we have detected three types of brightness variations : mean brightness variations with a characteristic time from days to months , orbital lc variations caused by the ellipsoidal shape of the red dwarf component and irregular flickering variations arising in the black hole accretion disk and the bright spot , where the mass transfer stream hits the outer edge of the disk .
v616 mon entered a non - active stage in the year 2015 .
the fast brightness variability was also detected in other x - ray nova - black hole candidates , e.g. in v404 cyg ( pavlenko et al . ,
1996 ; hynes et al .
, 2002 ; shahbaz et al . , 2003 ) and
kv uma ( gonzles hernndez et al . , 2014 ; pavlenko et al . ,
2001 ) . flickering is observed in all types of compact binaries consisting of a late - type star transferring matter onto an accreting white dwarf , a neutron star or a black hole
many statistical properties of this variability can be naturally explained with the fluctuating accretion disk model , where variations in the mass - transfer rate through the disk are modulated on the local viscous timescale and propagate towards the central compact object ( scaringi , 2014 ) .
the radiation produced by accretion is emitted through a release of gravitational potential energy .
most of the radiation of accreting black holes and neutron stars is emitted in the inner parts of the accretion disk in x - rays , while optical radiation is emitted from the outer parts of the accretion disk .
intrinsic brightness fluctuations on timescales of seconds to tens of minutes in mass - exchanging binaries were reviewed by baptista ( 2015 ) . in accreting white dwarf systems most of the radiation
is emitted in the optical .
the flickering originates either in the inner disk region and its boundary layer and/or in the hot spot , where the mass transfer stream hits the outer edge of its accretion disk .
the lcs of classical , symbiotic and dwarf novae with a white dwarf compact object often indicate flickering , e.g. : v2468 cyg ( chochol et al . ,
2012 ) , v959 mon ( shugarov et al . , 2014 ) , v2275 cyg ( ostrova et al . , 2003 ) , dk lac ( katysheva , n.a . & shugarov , s.yu . ) , er uma ( zemko et al . , 2014 ) , v426 oph ( shugarov , 1983 ; hessman , 1988 ; homer et al . , 2004 ) , mv lyr ( pavlenko & shugarov , 1998 ; 1999 ; katysheva et al . , 1999 ;
pavlenko et al .
, 2005 ; scaringi , 2014 ) , sw uma ( pavlenko et al . , 2000 ) , do dra ( szkody et al . , 2002 ) , v795 her ( mironov et al . , 1983 ; imon et al . , 2012 ) , uu aqr ( volkov et al . , 1986 ; volkov & volkova , 2003 ; baptista , 2015 ; khruzina et al . , 2015 ) ,
rs oph ( zamanov et al . , 2015 ) ,
ch cyg ( sokoloski , 2003 ; stoyanov et al . , 2015 , shugarov et al . ,
2015 ) and v407 cyg ( kolotilov et al . , 2003 ; shugarov et al . , 2007 ) .
the authors thank to our referee dr . elena pavlenko for constructive comments and advices .
this work was supported by the slovak research and development agency under the contract no .
apvv-15 - 0458 , the vega grant no .
2/0002/13 , the rfbr grants no
. 14 - 02 - 00825 , no . 15 - 02 - 06178 and the russian president grant nsh-9670.2016.2 .
a.d . acknowledges ( partial ) support from m.v .
lomonosov moscow state university program of development . | v616 mon = a0620 - 00 is a prototype of black hole transient x - ray binaries .
our 2003 - 16 optical photometry of the object during x - ray quiescence , obtained by 50 - 250 cm telescopes in crimea , caucasus mountains and slovakia , consists of @xmath0 7660 ccd frames in johnson - cousins @xmath1 bands and the integral light . during 2003 , 2008 - 9 and 2015
- 16 passive states , the phase light curve of the binary exhibited mainly variations caused by an ellipsoidal shape of the red dwarf component . during 2004 - 6 and 2009 - 14 active states a significant aperiodic broad - band variability ( flickering ) was present , arising in a black hole accretion disk and a bright spot , where the mass transfer stream hits the outer edge of the disk .
long term photometry of our minima times , together with available positions of superior conjunctions of the red dwarf found from spectroscopy , allowed us to refine the orbital period of v616 mon to 0.32301407(5 ) days .
[ intr ] |
[ [ from - the - lindblad - master - equation - to - the - gross - pitaevskii - equation ] ] from the lindblad master equation to the gross - pitaevskii equation + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we begin with a master equation of the lindblad form @xcite @xmath51+i\hbar\hat{\mathcal{l}}\hat{\rho},\ ] ] which describes the time evolution of the density operator of the system under the combined action of the hamiltonian @xmath52 and a local loss term , described by the dissipation operator @xmath53 here , @xmath54 and @xmath55 are the bosonic field operators , obeying the bosonic commutation relation @xmath56=\delta^3(\mathrm{\bf x}-\mathrm{\bf x'}),\ ] ] the interaction strength is denoted with @xmath57 and @xmath58 is an external potential .
the loss term @xmath59 describes the rate of particle losses at the position @xmath60 .
similar master equations for optical lattices with leaky sites have been studied in @xcite .
the time evolution of an operator is then given by the differential equation @xmath61 plugging equation ( 3 ) into ( 7 ) yields after some algebra @xmath62 \hat{\psi}(\mathrm{\bf x } ) \right\rangle - \frac{i\hbar\gamma(\mathrm { \bf x})}{2}\left\langle \hat{\psi}(\mathrm{\bf x } ) \right\rangle.\end{aligned}\ ] ] performing the usual mean - field approximation @xmath63 we arrive at equation ( 1 ) of the main text , i.e. , to the time - dependent gross - pitaevski equation with imaginary potential ( igpe ) .
the comparison between the experiment and the numerical solution of the igpe suggests that the mean - field approximation is well justified for our parameters .
we experimentally determine the value of @xmath19 by monitoring the short - time decay of the ion signal for a fixed current and different electron beam sizes . indeed analysing the solutions of the igpe we have found that such a decay is approximately exponential with a time constant @xmath64 .
the eb size @xmath21 is derived in two different ways .
we first measure @xmath21 by fitting with an integrated gaussian function the electron signal originating from the scan of the beam through the edge of an auxiliary metallic target .
the position of the beam focus and hence the value of @xmath21 at a fixed point can be easily varied by changing the current in its final focussing stage .
a calibration of the beam size as a function of this current is also used to recover the measured @xmath21 when the working position of the microscope is finely adjusted at the actual location of the atoms . for comparison
we also extract @xmath21 directly using the atomic sample as a target , by measuring the visibility of a two - dimensional deep optical lattice imaged by the eb itself .
the lattice that we employed has a spacing of @xmath65 nm and a depth of @xmath66 .
the beam size @xmath21 is obtained by deconvolving the density distribution of the atoms in the lattice by their expected wannier functions distribution .
note that the eb size does not vary appreciably along the radial extension of the cloud ( @xmath67 @xmath27 m ) , as its rayleigh length is always conveniently larger ( @xmath68 @xmath27 m ) .
the two measurements yield compatible results and are also in accordance with the measurement of @xmath69 as a function of the current @xmath18 performed for two different beam sizes : @xmath70 nm and @xmath71 nm , see fig . s1 . from this analysis
we obtain the value of the electron - bec cross - section used in this work : @xmath72 m@xmath73 . as a function of the eb current @xmath18 for two different values of the beam size @xmath35 nm ( gray triangles ) and @xmath74 nm ( red circles ) .
the solid lines are linear fits to the data .
the current @xmath18 is measured by means of a faraday cup , which is integrated in the experimental apparatus.,width=321 ] the main spurious effect is caused by the losses due to the secondary collisions of the produced ions with the trapped atoms .
indeed such ions , produced in the center of the bec , travelling through the atomic cloud can collide with other atoms expelling them from the trap .
this effect can in principle be reduced applying a strong electric field which extracts the ions faster ; but it can not be completely avoided . in order to take into account this loss channel the igpe must be modified adding an ion - atom collision term in the imaginary potential such that : @xmath75 where @xmath76 is the ions production rate and @xmath77 is a gaussian function whose amplitude @xmath78 and extension @xmath79 depend on the extraction geometry of the ions .
we have determined these two parameters fitting the time resolved curves like the one in fig .
1b , finding @xmath80 and @xmath81 .
taking into account this effect is important in our case mainly to estimate our ion detection efficiency , that is @xmath82 , but it has no significant impact on the physics presented . in the main text when we compare the theory with data we have included this effect while pure theoretical graphs report the ideal case .
additional spurious effects that , in principle , must be taken into account are related to the real potentials stemming from the eb . indicating with @xmath83 the radial distance from the center of the eb in the direction perpendicular to its propagation , one can find that the eb generates a magnetic field which has the following expression : @xmath84 the electron microscope machine produces itself a background magnetic field @xmath85 of @xmath86 1 g , directed along the direction of propagation of the eb and is always present at the atoms position .
due to this , the total magnetic field acting on the atoms is @xmath87 .
since @xmath88 is on the order of just 1 - 2 mg , it results that @xmath89 , making any effect due to the magnetic field produced by the eb @xmath88 negligible .
we now consider the electric field induced on the atoms by the flux of electrons , which can be approximated by : @xmath90 where @xmath91 is the mean velocity of the electrons .
this field leads to an effective potential orientated along the direction perpendicular to the beam propagation @xmath92 where @xmath93 is the ground state polarizability of the @xmath94rb atoms . for the beam maximum current we employed in the present experiments , @xmath95 na , the ratio between the maximum modulus of the dissipative and electric potentials is @xmath96 and this latter contribution is again negligible .
finally , we have also taken into account for the effects of the real potentials of the eb in the numerical simulations finding , as expected , no appreciable correction to the calculations performed without .
the numerical simulations made to reproduce the data for the thermal cloud were performed using a molecular dynamics method . we have used a runge - kutta algorithm to simulate the dynamics of @xmath9710@xmath98 atoms at the temperature of 1 @xmath27k in a harmonic potential . for this simulation
the atoms were supposed to be point - like , neglecting any kind of quantum effect like the finite extension due to the debroglie wavelength .
moreover we did not take into account the interactions between the atoms .
the action of the electron beam was simulated introducing @xmath99 electrons per m@xmath73 , gaussian distributed in space along @xmath100 and @xmath101 , and uniformly distributed along @xmath102 , the direction of propagation of the eb .
@xmath41 is the time step of the simulation .
every time the distance @xmath103 between one atom and one electron is smaller than @xmath104 the atom is supposed to collide with the electron leading to the loss of such an atom from the trap .
we have found that this method , despite its simplicity , reproduces quite well our data except for a multiplicative factor , which takes into account the efficiency of the ion detector and the branching - ratio of the collision processes .
we point out that a big advantage originating from the many - body nature of the bec is the possibility to perform continuous measurements on a single wavefunction , giving direct access to the dynamics without the need of ensemble averages . at the same time
the interactions between the atoms play a crucial role in determining the properties of such a wavefunction , through the non - linear term in the igpe , and have a direct impact on the dissipative dynamics . even if experimentally we are not able to change the interactions we can do it when solving the igpe changing the value of @xmath13 .
the results are reported in fig .
the position of the maximum dissipation is shifted to higher values of @xmath18 as the interactions between the atoms are increased .
the normalization is made to compensate the effect of the changing density ( @xmath105 ) .
clearly increasing interactions hinder the onset and the effects of the dzd.,width=321 ] as expecting a change in the interaction strength changes the position of the maximum of dissipation , since it depends on @xmath27 : the stronger is the repulsion between the atoms the more the onset of the dzd is shifted to larger values of @xmath18 and the more its effects are decreased .
moreover an increasing in @xmath13 implies a reduced production of ions , for a given @xmath106 , due to the subsequent change in the density . as shown in fig .
s2 , also a non - interacting bec displays the non monotonic behaviour observed in the experiment . from this
we hence conclude that the observed effects originate from the wave nature of the bec and not from its many - body character . nevertheless , according to fig .
s2 , the mean - field potential stemming from the atomic interactions represents another important controllable parameter that can be used to engineer the response of the system to the environmental action . | we experimentally investigate the action of a localized dissipative potential on a macroscopic matter wave , which we implement by shining an electron beam on an atomic bose - einstein condensate ( bec ) .
we measure the losses induced by the dissipative potential as a function of the dissipation strength observing a paradoxical behavior when the strength of the dissipation exceeds a critical limit : for an increase of the dissipation rate the number of atoms lost from the bec becomes lower .
we repeat the experiment for different parameters of the electron beam and we compare our results with a simple theoretical model , finding excellent agreement . by monitoring the dynamics induced by the dissipative defect we identify the mechanisms which are responsible for the observed paradoxical behavior .
we finally demonstrate the link between our dissipative dynamics and the measurement of the density distribution of the bec allowing for a generalized definition of the zeno effect . due to the high degree of control on every parameter ,
our system is a promising candidate for the engineering of fully governable open quantum systems .
gathering information from a quantum system is never free of cost .
every measurement process provides a coupling between the quantum system and the ( classical ) environment , which leads to non - unitary dynamics , and in some cases to the destruction of essentially quantum effects .
the elusive transition from the quantum to the classical realm must therefore be inherent in the processes that the environment induces on the system . in recent decades
several advances have been made in the study of environmentally induced phenomena like decoherence and decoherence - induced selection of preferred states ( einselection ) @xcite .
more recently environmental action has been used to manipulate qubits in a system of trapped ions @xcite . the knowledge and the mastering of the action of the environment are essential for taming errors in quantum computation schemes @xcite or to engineer decoherence - free subspaces for qubits @xcite , and are also key to understanding the emergence of classicality from the quantum @xcite . in the context of the theory of open quantum systems ,
environmental action gives rise to effective hamiltonians which can contain imaginary terms @xcite .
since these terms actually arise from a collection of an enormous number of degrees of freedom @xcite , however , a complete experimental control over them appears overly challenging . here , we report the engineering of a fully controllable , environmentally induced imaginary potential acting on a quantum system , and present observations of the subsequent induced dynamics .
the _
localized _ imaginary potential is realized by the almost pure dissipative action of an electron beam ( eb ) on an atomic bose - einstein condensate ( bec ) .
we show that such a potential can be used to describe a continuous measurement process that can exhibit a generalized version of the so - called zeno effect .
the combination of the robust and macroscopic many - body quantum behavior of a bec and the high tunability and precision of the eb promotes such a system as a paradigm for governable open quantum systems .
one of the most striking properties of becs is that , despite their many - body nature , they can be described to a good approximation by a mean - field wavefunction obeying the so - called gross - pitaevskii equation ( gpe ) .
this remains valid also when the bec is coupled with the environment .
starting from the lindblad master equation @xmath0 + i\hbar\hat{\mathcal{l}}\hat{\rho}$ ] , where @xmath1 is the density operator of the many - body system , @xmath2 is the hamiltonian operator , and @xmath3 is the dissipation operator such that @xmath4 $ ] , with @xmath5 the local dissipation rate , we can write the equation of motion for the expectation value of the bosonic field operator @xmath6 as @xmath7 , which leads to a time - dependent gpe with an additional imaginary term @xcite : @xmath8 here @xmath9 is the bec wavefunction , obeying the constraint @xmath10 , @xmath11 the trapping potential and @xmath12 , @xmath13 being the s - wave scattering length .
notably our technique allows independent control of the hamiltonian and dissipative terms of this equation .
the ability to describe our open quantum many - body system with such a simple expression is a key asset for understanding and mastering its dynamics .
[ [ experimental - implementation ] ] experimental implementation + + + + + + + + + + + + + + + + + + + + + + + + + + + in our experiment we prepare a pure bec of 75@xmath14 atoms in a single - beam optical trap by means of forced evaporation .
once the evaporation is over , we shine a focussed eb right at the center of the bec .
the eb is produced by a commercial electron microscope mounted inside the vacuum chamber @xcite .
the electron microscope is able to generate a beam of 6 kev electrons with variable beam extensions and currents . when the electrons impact on the bec they collide locally with the atoms , ionizing or exciting them .
those atoms which have undergone an electron collision escape from the trapping potential ( fig .
1a ) .
the eb thus locally dissipates the bec . the ionized atoms , roughly 40@xmath15 of all those scattered , are then directed to an ion detector , where their arrival times are registered . while escaping from the trapping region
the ions can collide with the trapped atoms producing additional losses @xcite .
the total detection efficiency @xmath16 is the product of the branching ratio ( 40% ) and of the combined ion optics and detector efficiency ( 75% ) .
details of the experimental apparatus can be found in @xcite . if the eb is rapidly moved in a controlled pattern , the whole column density profile of the bec
can be reconstructed @xcite . here
we keep the eb fixed in the center of the bec , and monitor the subsequent induced dynamics by looking at the temporal signal from the ion detector ( fig .
1b ) .
by controlling the beam parameters , we can engineer the dissipative term in equation ( [ dgpe ] ) : we write it as @xmath17 , @xmath18 being the eb current , @xmath19 the electron - bec scattering cross - section @xcite , @xmath20 the elementary charge and @xmath21 the standard deviation of the spatial electron distribution , assumed to be gaussian @xcite .
s. points are experimental data averaged over 1800 experimental repetitions , while the solid curve is the numerical simulation ( see text ) .
after 5 ms we typically collect @xmath22 ions . ]
[ [ comparison - between - experimental - results - and - theoretical - expectations ] ] comparison between experimental results and theoretical expectations + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in fig .
2 we report the number of ions collected in the first 5 ms of continuous dissipation as a function of the eb current for three different values of @xmath21 .
notably we observe that the number of ions produced , as a function of the eb current ( i.e. , of the number of electrons sent on the atoms ) , shows a _ non - monotonic _ dependence .
in other words , starting from a critical value of the eb current , the harder we try to dissipate , the less we manage to do it .
this paradoxical behavior is more marked for smaller values of @xmath21 . in the same figure
the data are compared with the results obtained by numerically solving equation ( [ dgpe ] ) , additionally taking into account secondary effects like ion - atom collisions @xcite .
the agreement is very good ( and the same agreement is visible in fig .
1b ) , demonstrating that the description of the eb as a pure dissipative potential is sufficient to capture the observed main features . a detailed description of the dynamics that leads to the curves reported in fig . 2
will be given in the following . from simple textbook calculations , or from more formal analysis like the one made in @xcite
, it is easy to verify that a _ localized _ imaginary potential @xmath23 induces total reflection as the strength of the potential goes to infinity .
hence the effective quantum dissipation vanishes when the localized imaginary potential is either zero or infinity , implying the existence of a maximum of dissipation for some finite value of @xmath23 .
this explains on a qualitative basis the observed non - monotonicity .
the position of the maximum is of special importance , since it sets the parameters which allow one to engineer the most efficient possible absorbing potential . as an example , in @xcite , where the time of arrival of a 1d wave packet is measured by a step - like potential , the maximum dissipation is analytically calculated to be @xmath24 , where @xmath25 is the energy of the wave packet . in our case
the presence of the non - linearity and the less idealized conditions do not allow for an analytic solution , but from the numerical results plotted in fig .
3a we have found @xmath26 , where @xmath27 and @xmath28 are respectively the chemical potential and the healing length of the unperturbed bec . from fig .
3a it also appears that increasing the size of the eb not only moves the position of the maximum dissipation to higher values of @xmath18 , but also increases the number of produced ions for a given current , and reduces or washes out the effect of the reflection . clearly , when @xmath29 , a decrease of the probe size ( an increase of the resolution ) leads to a lower production of ions , making the system more resistant to the environmental action . .
the three panels report the data obtained with @xmath30 and 212 ( 8) nm , from left to right .
each data point is the average over 75 experimental repetitions .
the error bars are mainly due to shot - to - shot fluctuations in the overall ion detection efficiency .
the solid lines are the number of dissipated atoms resulting from the numerical simulations ( see text ) .
please note that in the sum also the initial decay visible in fig .
1b is included .
the scale on the top reports the strength of the imaginary potential @xmath31 corresponding to the measured current.,width=321 ] [ [ comparison - to - classical - systems ] ] comparison to classical systems + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to ascertain to which extent our observations are peculiar to the wave nature of the bec , we have repeated the experiment on a thermal gas of 4@xmath32 atoms at 1@xmath27k with a beam of @xmath33 nm . the results are reported in fig .
3b , where a simple monotonic behavior is observed .
such data are well reproduced by a classical molecular dynamics simulation , which includes the dissipation induced by the eb @xcite .
we then extend the classical simulation to an atomic cloud which has the same density , number of atoms and trapping frequency as our bec . even though this does not represent any real physical system ,
it is instructive to compare the behavior of a quantum system with its hypothetical classical analogue .
this comparison is made in fig .
3c , where the monotonicity of the classical case is confirmed . from this
we can conclude that the observed non - monotonicity is a purely quantum effect stemming from the _ macroscopic _ wave nature of the bec .
moreover it is evident that the effect of the quantum reflection from the imaginary potential leads to a suppression of dissipation in the quantum case , which is already notable for very low currents . solving equation ( [ dgpe ] ) for different values of @xmath21 .
the values of @xmath34 obtained using the approximate expression given in the text are shown as open diamonds over the corresponding curves .
b ) number of ions measured after 5 ms of dissipation for a thermal cloud as a function of the eb current , with @xmath35 nm .
the solid line is the result of the corresponding numerical simulation using the molecular dynamics method .
c ) comparison between the theoretical curves of the number of produced ions as a function of @xmath18 for the bec and the _ corresponding _ ( see text ) classical analogue ( @xmath35 nm).,width=321 ] [ [ dissipative - dynamics ] ] dissipative dynamics + + + + + + + + + + + + + + + + + + + + in order to gain a deeper insight into the dissipation - induced dynamics , we now look in detail at the time - resolved signals coming from the ion detector , reported in fig .
4a .
initially the number of ions produced is well described by the exponential decay @xmath36 , @xmath37 being the effective dissipation rate @xcite . in fig .
4b we show the integral of the signals in the first 5 @xmath27s , together with the simulated values as a function of the eb current . in this phase ,
where no paradoxical behavior is either observed or expected , the eb burns a hole in the bec wavefunction ( see fig .
4d ) , defining a clear border between the space `` inside '' the hole and that `` outside '' . thereafter , the number of ions produced becomes almost constant , signaling the onset of a quasi - stationary dynamics @xcite . in this second phase , the reaction of the quantum system to the external perturbation takes place .
when the strength of the dissipation is increased , the `` outside '' wavefunction passes from a situation of almost total transmission to a situation where reflection takes the leading role .
the non - monotonic dependence on the dissipation strength then becomes apparent ( fig .
4c ) .
this represents the first experimental observation of the so - called back - flow paradox @xcite , i.e. , of the onset of a temporary reflection from a localized perfect absorber .
the curves plotted in fig
. 2 and 3 are then the sum of different contributions like those in fig . 4b and 4c .
[ [ dissipation - as - continuous - measurement ] ] dissipation as continuous measurement + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + finally we demonstrate that the controlled dissipation is equivalent to a local measurement of the bec density , i.e. , of the squared modulus of its wavefunction .
starting from equation ( [ dgpe ] ) and defining @xmath38 , where @xmath39 is the number of atoms , after some algebra we obtain the equation @xmath40 the number of ions produced in a time interval @xmath41 around a certain time @xmath42 is @xmath43 .
hence we can conclude that what we perform is a direct measurement of the bec density @xmath44 in the region illuminated by the eb , as in @xcite . since the seminal trilogy on the time of arrival in quantum mechanics @xcite ,
imaginary potentials have been linked to the action of a measurement apparatus while later refinements @xcite formally demonstrated the equivalence between a pulsed measurement with period @xmath45 and a continuous dissipative potential @xmath23 , provided that @xmath46 .
our findings represent the experimental verification of the equivalence between the action of an imaginary potential and the one of a measurement apparatus on a quantum system .
indeed we show that a continuous measurement of the bec density is nicely reproduced by introducing an imaginary potential in the corresponding schrdinger equation .
furthermore it is well understood theoretically @xcite , and verified experimentally @xcite , that performing continuous measurements strongly modifies the pre - existing dynamics of a quantum system .
this is known as the zeno ( or inverse - zeno ) effect . in general ,
measurements are mainly performed on non - decaying systems and an extension of the standard definition to such systems is needed . in our case
no pre - existing dynamics is present , since in the absence of the eb , the bec is at rest .
we have shown that the action of the continuous dissipative potential , or of the continuous measurement , strongly modifies the output of the measurement itself . in analogy with the standard definition
we define _ dissipation induced zeno dynamics _ ( dzd ) when @xmath47 , @xmath48 being the number of ions produced in the time @xmath42 , and _ dissipation induced simple dynamics _ ( dsd ) when @xmath49 .
these definitions appear to be the natural extension of the standard ones , since the onset of the dzd requires large values of @xmath50 , which corresponds to pulsed measurements with small @xmath45 @xcite . from the definitions it follows that the dsd is observed where the dynamics is dominated by the hamiltonian term of the lindblad master equation , while the onset of the dzd corresponds to a dynamics governed by the dissipative term .
we note that an effect resembling the dzd has been observed also in a system of decaying molecules in 1d @xcite and in an attractive mott - insulator state @xcite .
[ [ conclusion - and - outlook ] ] conclusion and outlook + + + + + + + + + + + + + + + + + + + + + + we have experimentally demonstrated the implementation of an open many - body quantum system whose hamiltonian and dissipative dynamics can be independently and accurately controlled . in the case of extremely strong and localized dissipation
this can lead to the creation of dissipation - resistant states .
the possibility to create such states in a controlled fashion can give new insights for engineering generalized environmental dark states .
these kind of states are of fundamental interest and can possibly have practical applications in quantum computation schemes @xcite . and inasmuch as our technique exploits the demonstrated link between dissipation and measurement , it can be used to address fundamental issues in quantum mechanics , like the definition of the time of arrival @xcite .
the dissipation mechanism studied in the present paper is also particularly suited for lattice systems @xcite , thanks to its localized character and hence to the ability to selectively control the dissipation in a single lattice site . indeed the use of the eb offers the unique possibility to create and study long - living exotic states in optical lattices @xcite and to characterize the interplay between dissipation and interactions @xcite , and so would give access to the engineering of quantum phases in open quantum systems @xcite .
[ [ acknowledgements ] ] acknowledgements + + + + + + + + + + + + + + + + we thank j.r . anglin and a. widera for reading the manuscript and for fruitful discussions .
we are grateful to m. scholl and p. wrtz for their help in the early stage of the experiment .
we also thank v. konotop and s. wimberger for enlightening discussions .
we acknowledge financial support by the dfg within the sfb / trr 49 and grk 792 . g. b. and v. g. are supported by marie curie intra - european fellowships .
r. l. acknowledges support by the mainz graduate school .
w. zurek , _ reviews of modern physics _ * 75 * , 715 ( 2003 ) , and references therein r. omns , _ reviews of modern physics _ * 64 * , 339 ( 1992 ) h. m. wiseman and g. j. milburn , _ quantum measurement and control _ , cambridge university press ( 2010 ) j. t. barreiro , m. mller , p. schindler , d. nigg , t. monz , m. chwalla , m. hennrich , c. f. roos , p. zoller , and r. blatt , _ nature _ * 470*,486 ( 2011 ) p.
w. shor , _ physical review a _ * 52 * , r2493 ( 1995 ) d. g. cory , m. d. price , w. maas , e. knill , r. laflamme , w. h. zurek , t. f. havel , and s. s. somaroo , _ physical review letters _ * 81 * , 2152 ( 1998 ) j. i. cirac and p. zoller , _ physical review letters _ * 74 * , 4091 ( 1995 ) d. a. lidar , i. l. chuang , and k. b. whaley , _ physical review letters _ * 81 * , 2594 ( 1998 ) a. beige , d. braun , b. tregenna , and p. l. knight _ physical review letters _ * 85 * , 1762 ( 2000 ) r. sttzle , m. c. gbel , th .
hrner , e. kierig , i. mourachko , m. k. oberthaler , m. a. efremov , m. v. fedorov , v. p. yakovlev , k. a. h. van leeuwen , and w. p. schleich , _ physical review letters _ * 95 * , 110405 ( 2005 ) see supplementary materials v. a. brazhnyi , v. v. konotop , v. m. perez - garcia , and h. ott , _ physical review letters _ * 102 * , 144101 ( 2009 ) d. witthaut , f. trimborn , h. hennig , g. kordas , t. geisel , and s. wimberger , _ physical review a _ * 83 * , 063608 ( 2011 ) t. gericke , p. wrtz , d. reitz , t. langen , and h. ott , _ nature physics _ * 4 * , 949 ( 2008 ) v. guarrera , p. wrtz , a. ewerbeck , a. vogler , g. barontini , and h. ott , _ physical review letters _ * 107 * , 160403 ( 2011 ) p. wrtz , t. langen , t. gericke , a. koglbauer , and h. ott , _ physical review letters _ * 103 * , 080404 ( 2009 ) g.r .
allcock , _ annals of physics _ ( ny ) * 53 * , 253 ( 1969 ) , g.r .
allcock , _ ibid . _ * 53 * , 286 ( 1969 ) , g.r .
allcock , _ ibid . _ * 53 * , 311 ( 1969 ) d. a. zezyulin , v. v. konotop , g. barontini , h. ott , _ physical review letters _ * 109 * , 020405 ( 2012 ) j.g .
muga , j.p .
palao , c.r .
leavens , _ physics letters a _ * 253 * , 21 ( 1999 ) j. echanobe , a. del campo , and j. g. muga , _ physical review a _ * 77 * , 032112 ( 2008 ) b. misra and e. c. g. sudarshan , _ journal of mathematical physics _ * 18 * , 756 ( 1977 ) a. g. kofman and g. kurizki , _ nature _ * 405 * , 546 ( 2000 ) w. m. itano , d. j. heinzen , j. j. bollinger , and d. j. wineland , _ physical review a _ * 41 * , 2295 ( 1990 ) m. c. fischer , b. gutierrez - medina , and m. g. raizen , _ physical review letters _ * 87 * , 040402 ( 2001 ) e. w. streed , j. mun , m. boyd , g. k. campbell , p. medley , w. ketterle , and d. e. pritchard , _ physical review letters _ * 97 * , 260402 ( 2006 ) n. syassen , d. m. bauer , m. lettner , t. volz , d. dietze , j. j. garcia - ripoll , j. i. cirac , g. rempe , and s. drr , _ science _ * 320 * , 1329 ( 2008 ) m. j. mark , e. haller , k. lauber , j. g. danzl , a. janisch , h. p. bchler , a. j. daley , and h .- c .
ngerl , _ physical review letters _ * 108 * , 215302 ( 2012 ) f. verstraete , m. w. wolf , j.i .
cirac , _ nature physics _ * 5 * , 633 ( 2009 ) p. barmettler and c. kollath , _ phys .
rev .
a _ * 84 * , 041606(r ) ( 2011 ) d. poletti , j .- s .
bernier , a. georges , c. kollath , _ physical review letters _ * 109 * , 045302 ( 2012 ) m. mller , s. diehl , g. pupillo , p. zoller , _ advances in atomic , molecular , and optical physics _
* 61 * , 1 - 80 ( 2012 ) |
0 in this paper we study the interaction of a cosmic string with a ( non - rotating ) black hole .
this problem is interesting for many reasons .
first of all , it is a rare example of gravitational interaction between two extended relativistic objects that allows complete analysis . when a cosmic string comes close to a black hole and is captured by it
, one might expect generation of strong gravitational radiation , which makes the problem potentially interesting for astrophysical applications . besides this
, the study of the interaction of a cosmic string with a black hole has demonstrated some unexpected , physically interesting possibilities , such as the creation of 2d black holes @xcite . for astrophysically interesting cases , the dimensionless parameter @xmath1 , where @xmath2 is the string tension , is small .
for example , for strings formed during a gut phase transition , @xmath3 .
for this reason , in the leading order one can neglect gravitational backreaction effects and consider the string as a test object propagating in a given gravitational background .
the worldsheet of such a string is a minimal surface which gives an extremum to the nambu - goto action @xcite .
stationary configurations of a cosmic string in the field of a rotating charged black hole are described by worldsheets which are tangent to the killing vector generating the time shift .
because of this symmetry , the problem is reduced to finding a geodesic in a three - dimensional space ; the latter can be solved by separation of variables @xcite . in the absence of such a symmetry ,
non - linear partial differential equations describing the string motion in the gravitational field of the black hole do not allow exact solutions . under these conditions
one either needs approximation schemes or numerical solutions . in the present paper we assume that the string is initially far from the non - rotating black hole , straight , and moving with velocity @xmath4 in the direction of the black hole .
the motion of the string is affected by the gravitational field of the black hole .
for a given velocity @xmath4 , the outcome of the interaction depends on the impact parameter @xmath5 .
if the impact parameter is large enough , the string escapes the gravitational attraction of the black hole and is inelastically scattered . for smaller values of @xmath5 the string is captured by the black hole .
the scattering of a cosmic string for very large impact parameters was studied in @xcite .
such a string passes far from the black hole and its motion is modified only slightly . under these conditions one can consider the gravitational interaction as a perturbation and treat it by expanding the solution in powers of the newtonian gravitational potential @xmath6 . the numerical study of string scattering by a black hole was initiated by moss and lonsdale @xcite . in reference
@xcite we used numerical calculations to obtain the dependence of the critical capture impact parameter on velocity @xmath4 , and demonstrated that the results of moss and lonsdale do not correctly reproduce the behaviour of the critical impact parameter at high velocities . in order to obtain more detailed information on the characteristics of scattering and capture of a cosmic string by a black hole we performed further numerical calculations
. in regimes that allow analytical study ,
we compared numerical and analytical results .
the agreement of these results was used as an additional test of our calculations .
we also analysed the role of tension in the string by comparing the relativistic string results with those for a string without pressure ( a so - called `` dust model '' ) .
these results are discussed in the following sections . in section
2 we formulate the scattering problem for the straight string .
section 3 outlines the numerical method that was used and discusses the scattering at shallow impact parameters .
it also contains comparison of these results with the results obtained in the weak - field approximation .
section 4 compares numerical and approximate analytical results for a string moving with ultra - relativistic velocity .
it also discusses coil formation for the scattering in this regime .
( we make the distinction here between coils , which are loop - like configurations in a straight string , and the term loop , as found in the literature , which designates closed , bounded string configurations .
we thank an anonymous referee for suggesting this clarification . )
section 5 discusses near - critical scattering , where the string worldsheet is partially wrapped around the black hole .
section 6 discusses the role played by the tension of the string and makes a comparison with the model of a dust string .
section 7 discusses aspects of the problem connected with string capture .
0 the two - dimensional worldsheet representing a moving string is parametrized as @xmath7 ; @xmath8 ( @xmath9 ) are the spacetime coordinates , and @xmath10 ( @xmath11 ) are the worldsheet coordinates @xmath12 , @xmath13 .
the equations of motion of the string follow from the nambu - goto action @xcite which we write in the polyakov form @xcite @xmath14=-{\mu \over 2}\,\int d\tau d\sigma \sqrt{-h}h^{ab}g_{ab}\ , .\ ] ] we use units in which @xmath15 , and the sign conventions of @xcite . in ( [ 2.1 ] ) @xmath16 is the internal metric with determinant @xmath17 , and @xmath18 is the induced metric on the worldsheet , @xmath19 where @xmath20 is the spacetime metric .
the equations obtained by varying action ( [ 2.1 ] ) with respect to @xmath8 and @xmath16 are of the form @xmath21 and @xmath22 where @xmath23 $ ] . equations ( [ 2.3 ] ) are the dynamical equations for string motion , while equations ( [ 2.4 ] ) are constraints .
consider first the motion of a straight cosmic string in the absence of an external gravitational field , that is when @xmath24 , where @xmath25 is the flat spacetime metric .
in cartesian coordinates ( @xmath26 ) , @xmath27 and @xmath28 , and it is easy to verify that , taking @xmath29 , @xmath30 is a solution of equations ( [ 2.3 ] ) and ( [ 2.4 ] ) .
this solution describes a straight string oriented along the @xmath31-axis which moves in the @xmath32-direction with constant velocity @xmath33 .
initially , at @xmath34 , the string is found at @xmath35 , where @xmath36 is the impact parameter . for definiteness
we choose @xmath37 and @xmath38 , so that @xmath39 .
the two - dimensional surface of the worldsheet of such a string is spanned by vectors @xmath40 and @xmath41 while the vectors @xmath42 and @xmath43 are orthogonal to the worldsheet . vectors @xmath44 ( @xmath45 ) form an orthogonal tetrad .
the study of scattering of an infinitely long cosmic string by a non - rotating black hole is most conveniently carried out in isotropic coordinates , @xmath46 , for which the schwarzschild metric has the form @xmath47 where @xmath48 .
this metric can be written as @xmath49 with @xmath50 and @xmath51 where @xmath52 is the newtonian potential .
the motion of the string in the region with small @xmath53 can be written in the form @xmath54 the straight string starts its motion in the plane @xmath55 .
we call it `` in - plane '' ( see figure 1 ) .
this plane represents the motion of the free string in flat spacetime .
deflection of the string worldsheet from the `` in - plane '' , @xmath56 , can be decomposed in the orthogonal frame as @xmath57 by linearizing string equations ( [ 2.3 ] ) and ( [ 2.4 ] ) one gets the following equations for the first - order perturbations @xmath58 @xcite : @xmath59 where @xmath60 @xmath61 @xmath62 and @xmath63 the solutions of linearized equations ( [ 2.14])([2.16a ] ) were obtained and analyzed in @xcite . here
we reproduce only a solution for @xmath64 which describes the deflection of the string in @xmath65-direction , that is in the spatial direction perpendicular to the string motion .
we assume that a straight string starts its motion with @xmath66 at @xmath67 then @xmath68 \ , , \ ] ] where @xmath69}- \arctan{\left[{(\tau\pm\sigma)\,\tanh\beta}\over{y_0}\right ] } \ , .\ ] ] here @xmath70 .
this expression for @xmath71 can be easily obtained from equation ( 2.22 ) of ref.@xcite if we change @xmath72 in this expression and take the limit @xmath73 .
after this redefinition of @xmath74 , the value of new variable @xmath75 corresponds to the moment of time when the `` distance '' @xmath76 from the string to the black hole is minimal .
the analysis of the late time asymptotic of this solution shows that the string approaches the plane @xmath77 , and the shift in position of the string is given by second order corrections become important ( see @xcite ) .
we do not consider the case of extremely slow motion of strings in this paper and always assume that @xmath78 .
for a discussion of slowly moving strings see @xcite . ]
@xmath79 where @xmath80 is the velocity of the string and @xmath81 .
the late - time solution is represented by a kink and anti - kink , propagating in opposite directions at the speed of light , and leaving behind them the string in a new `` phase '' .
furthermore , it can be shown that each kink has a characteristic width in the external spacetime given by in the expression for the width of the kinks which was absent in @xcite . ]
@xmath82 the above results were obtained under the assumption that the string always propagates in the region where the gravitational field remains weak . in what follows ,
we describe results concerning string scattering in the regime where the gravitational field can not be considered weak .
nevertheless , if the string is not captured it eventually emerges from the strong field region and its subsequent evolution is described again by equations ( [ 2.14 ] ) .
the string s late time features are characterized by the amplitude @xmath83 and width @xmath84 of the kinks ; we refer to these two parameters as the _ scattering data_. as will be seen in the following sections , the breakdown of the weak field approximation manifests itself in two ways : the failure of the analytical expressions to correctly reproduce transient shapes , which develop when the string reaches periastron , and discrepancies in the late - time features of the string , namely the scattering data .
0 in order to study the scattering problem under strong - field conditions ( where the weak - field approximation breaks down ) , a numerical solution to the equations of motion must be found . to deal with an infinite cosmic string numerically ,
a finite computational domain is required , so a method of truncating the string to a reasonably short segment is needed .
such a physical truncation imposes special boundary conditions since it is crucial to reproduce the motion of an infinitely long string with sufficient accuracy .
two methods of truncating a string were developed .
the first places a massive particle at each end of the segment of string and requires that the motion of these particles mimic the motion of the portions of the string lying outside the region of interest .
this is accomplished by letting the mass of the end particles go to infinity .
it should be emphasized that truncating the string and adding infinitely heavy particles at its ends is just a technical trick .
it allows one to consider finite strings . in making the size of the string longer and longer
, we finally reach a size where a further increase in the string s length does not affect the solution within the chosen accuracy .
this may be interpreted as saying that the infinitely massive end particles correctly simulate an infinite string provided the length of the truncated string is chosen sufficiently long ( this issue is discussed in reference @xcite ) .
the second method places the truncation points a reasonable distance away from the region of interest , subject to the condition that the end points remain in a weak gravitational field where a perturbative solution to the equations of motion is applicable .
the analytic weak - field solution is used to prescribe the motion of the boundary points of the string .
each method has distinct advantages , and both were used extensively .
further , the solvers were tested against one another and agreement under identical initial conditions served as an additional validation of the numerical schemes . with the choice of internal metric @xmath85 , the equations of motion have the form @xmath86 the discretization of these equations ( described in detail in reference @xcite ) yields a block - tridiagonal system with which the solution over the entire spatial ( @xmath87 ) grid is obtained by algebraic methods .
the boundary conditions , for which @xmath88 , are imposed by replacing the first and last entries in the block - tridiagonal system by the required expressions . for infinitely massive end particles at the boundaries
the following equations are substituted into the system , @xmath89 the derivation of this boundary condition is given in reference @xcite . for the perturbative boundary conditions , the weak - field solutions of eqn.([2.14 ] )
are substituted into the system ; the solutions are given in reference @xcite .
the technical details of the numerical work are not important , but a few general comments are in order on the practical aspects imposed by the boundary conditions and initial data . under both types of boundary conditions ,
the accuracy of the numerical results may be influenced by the length @xmath90 of the string segment , meaning that if a string segment is too short , the motion of the boundary points induces incorrect behaviour .
the larger the value of @xmath90 , the smaller the effect , but the larger the value of @xmath90 , the greater the number of grid points required to adequately resolve the string .
the choice of string length is therefore a compromise between the need to accurately reproduce the behaviour of an infinite string and the need to use computing resources effectively .
we found that the massive - particle boundary conditions required a string length of at least @xmath91 ( @xmath92 , the schwarzschild radius of the black hole ) to give reasonable results , as discussed in reference @xcite , whereas the perturbative boundary conditions required a much shorter string length , @xmath93 , since the weak - field solutions represent the motion of the string boundary points with sufficient accuracy at these distances .
therefore , the perturbative boundary conditions greatly reduce the number of grid points and time steps needed to resolve the motion of the string near the black hole , and hence speed up execution dramatically .
initial data takes the form of a straight string , minkowski solution with the stipulation that the initial position of the string is sufficiently far away from the black hole to minimize the discrepancy between this approximate initial data and the correct initial configuration in schwarzschild spacetime ; the constraint equations are used as a guide .
the constraint equations are also used to check the accuracy of the calculations during the time evolution of the string .
the evaluation of the discretized form of the constraint equations is carried out periodically during the numerical solution .
statistics are computed for the constraints ( average value over the length of the string and standard deviation ) at the current time step and reported to an output file .
monitoring that the constraints are consistent with zero to several significant digits is done by inspecting this file . if the constraints are not satisfactorily maintained , that is if the average value grows unacceptably large or undergoes sudden changes , the numerical solution is restarted with new parameters ( e.g. finer grid , finer step size ) .
the results for the constraint calculation are influenced significantly by the initial position of the string .
a solution for a string placed initially closer to the black hole requires fewer time steps , but generates a poorer result in the constraint calculation .
typically , the constraints were expected to be consistent with zero from four to six significant figures .
tighter tolerances can also be achieved by increasing the number of grid points ( and significantly increasing solution time ) , but for the purposes of this study , the above values were deemed a reasonable compromise between accuracy and speed .
the perturbative solution to the equations of motion of an infinitely long cosmic string is most conveniently carried out in isotropic coordinates , @xmath46 .
numerically , however , the scattering problem is best studied in eddington - finkelstein in - going coordinates since the christoffel symbols in this coordinate system are regular everywhere away from the origin .
the christoffel symbols associated with this coordinate system , along with their derivatives with respect to the spacetime coordinates , were derived and inserted as analytic expressions in the portions of the code dealing with the initialization of the tridiagonal matrices . also , the coded forms of the analytic expressions for the initial data and for the perturbative boundary conditions were converted to eddington - finkelstein coordinates using standard coordinate transformations .
0 in this section , we compare the numerical solutions of string scattering to those of the weak - field approximation at intermediate velocities ( @xmath94 ) for a range of impact parameters .
our basis for comparison is the scattering data .
figure 2 shows sections of the numerical worldsheet of a string with initial velocity @xmath95 and impact parameter @xmath96 . as shown in table 1
, the maximum amplitude and width of the pulses is in good agreement with the weak - field eqns .
( [ 2.19 ] ) and ( [ 2.20 ] ) .
.scattering data - schwarzschild weak - field .
[ cols= " < , < , < " , ] since the weak - field and ultra - relativistic approximations are accurate only for impact parameters greater than @xmath97 , the boundary of the coil formation is expected to shift due to strong field effects . the solid line shown in figure 5 , based on numerical tests to detect coil formation
, indicates that the boundary is shifted towards lower velocities ; the strong field near the black hole tends to enhance coil formation .
the black hole drives the self - intersection process as its gravitational influence draws points on either side of the string towards the @xmath98 plane .
what happens with the cosmic string forming a cusp which starts growing into a closed coil can not be answered in a simplified model we use .
one of the possibility is that after formation a small coil , as the result of the string interconnection @xcite,@xcite the coil creates a loop . in this scenario ,
the black hole driven self - intersection could leave a line of small loops in the wake of the string .
another option is that loops never form .
instead blobs of higgs particles are ejected continuously as the budding loop cuts from the string without fully forming .
this would result in a trail of fastly decaying higgs particles behind this cusp .
0 as we already mentioned , there are two possible outcomes when a cosmic string interacts with a black hole : either the string is scattered , or it is captured by the black hole . for a given initial velocity @xmath4
the outcome depends on the value of the impact parameter @xmath99 . in other words , in the space of initial data @xmath100 there exists a critical line @xmath101 which separates these two different regimes .
we discuss the critical impact parameter for string capture in section 7 .
this section discusses string scattering where the impact parameter is extremely close to the critical impact parameter for capture and the string comes close to the black hole . in a similar situation for test particle scattering , the particle can execute multiple orbits around the black hole before escaping . some similarities with particle scattering are observed , but there are also important differences that shed light on the process of string capture . figure 8 compares a slice of the string worldsheet through the @xmath102 plane to the motion of a test particle .
a comparison is made of a series of string and particle trajectories with identical velocities ( @xmath103 ) and nearly identical impact parameters .
for the particle trajectories , these range from @xmath104 through @xmath105 .
for the string , the impact parameters are slightly smaller , ranging from @xmath106 through @xmath107 .
the impact parameters at the lower end of each range result in capture , whereas those at the upper end result in scattering .
it is easy to see that , even at ultra - relativistic velocities , tension influences string motion .
although the critical impact parameter for the string is very close to that for a particle , suggesting that tension plays a limited role , the dynamics of the string are , nevertheless , still governed by tension .
this is manifested in the partial winding of the string
while gravity dominates the dynamics , followed by unwinding of the string as a sufficient interval has elapsed for tension to again dominate ; the transition between the two regimes is marked by the cusp - like feature in box b. by comparison , the escaping trajectories of the test particle are far less complicated , showing no folds or cusps .
the detailed motion of the string in the critical regime close to capture is highly sensitive to the initial impact parameter .
figure 8 shows that there are two types of capture trajectories , trajectories such as curve 1 , where the string crosses the horizon directly , and curve 2 , where the string folds back on itself before crossing the horizon .
curve 2 exhibits a coil - like feature , but does not represent a self - intersection since the intersecting points have distinct proper times . escaping trajectories , such as curve 3 , have a folded structure .
there are two critical cases that mark the transition between each of these three generic curves .
the transition between curves of type 1 and type 2 is marked by a structure that develops an increasingly cusp - like shape ( point b in figure 8) as the impact parameter approaches the critical value that marks the transition from type 1 to type 2 curves . in practical terms , this impact parameter is difficult to obtain since it requires a large number of significant digits ( curve 1 and the curve with point b have impact parameters that differ by one part in @xmath108 ) .
the transition between type 2 and type 3 curves is marked by a tangent point , where the string trajectory passes twice through the same point ( point a ) at different times ( again , no self - intersection ) .
furthermore , the transition between type 2 and type 3 curves is associated with the critical impact parameter for capture
. a full @xmath109 rendering of the worldsheet nearest the black hole can be generated , but it is very difficult to interpret and is therefore not shown here .
it is sufficient to note that complicated folds and twists are generated as points on the string cross and recross while executing their partial orbit of the black hole . at late times , however , these folds and twists have dissipated and all that remains are kinks , as shown in figure 9 .
this figure shows a detailed view near the @xmath98 plane ; the kinks evolve and move outward along the string , exiting at the top and bottom of figure 9 and leaving behind the straight portion visible at @xmath110 .
this figure reinforces the idea that scattering at late times is completely understood in terms of the perturbative solutions .
however , it may take a significant amount of time for the simple kink / anti - kink picture to emerge . in the case
studied here , the distortions due to the close encounter persist until the string is about @xmath111 past the black hole .
the numerical results for moderate @xmath112-factors show that the string worldsheet folds back on itself , with no evidence of multiple windings or glory scattering as would be the case for near - critical scattering of particles .
in fact , it seems as if the transition to a `` direct capture '' trajectory ( like curve 1 in figure 8) always occurs before a full turn is achieved .
the fact that multiple windings are absent is a clear indication that the tension in remote parts of the string eventually asserts itself in all cases accessible to the numerical solver .
the question as to whether the string can complete more than one full turn and `` wrap '' the black hole for larger @xmath112-factors is an open one .
0 since a cosmic string is an extended object under tension , motion of a string near a black hole represents the result of the competing influences of tension and gravity . in order to shed light on the role of tension in the dynamics of the string ,
the motion of the string is compared to the motion of an array of test particles initially configured with the same position and initial velocity as the cosmic string .
to do this , consider a family of n test particles arrayed on a line with initial position @xmath113 and initial velocity @xmath114 , where @xmath115 is a position index ( an integer between 1 and n ) that describes the initial location of the test particle on the line .
these particles each satisfy the geodesic equation and constraint , @xmath116 & = & 0\ , .\end{aligned}\ ] ] it is easy to show that , in flat spacetime , @xmath117 satisfies these equations and represents a straight dust string moving with velocity @xmath80 where the discrete @xmath118 mimic the @xmath87 coordinate of the cosmic string .
it should be emphasized that in the general case dust string dynamics is quite different from the dynamics of the nambu - goto string .
for example , without tension a small bump on a dust string simply goes away , while a similar bump on the nambu - goto string produces two small bumps which propagate away from one another with the speed of light . nevertheless , we found it instructive to compare the motion of a straight nambu - goto string in the field of a black hole with the motion of a similar dust string since it clearly demonstrates the role of string tension in capture and scattering processes . in the weak - field and ultra - relativistic cases ,
the role of tension in the string is made apparent by comparing the yz projections of figures 2 and 6 against those of the dust string in figure 10 .
figure 10 shows that a coil always evolves in the dust string worldsheet .
this coil is due to the keplerian nature of the trajectories of each particle on the dust string .
particles lying on one side of the @xmath119 plane are deflected across this plane since their motion is constrained to an orbital plane passing through the black hole . unlike the cosmic string , the y - axis deflection of the dust string is unbounded ( this is not to say that the deflection does not reach an asymptotic angular value )
. close examination of figures 2 , 6 , and 10 reveals that , at the early stages of scattering , the size of the coils in the dust and cosmic strings are virtually identical , reinforcing the idea that tension takes some time to assert itself .
coil formation is a generic feature of the dust string ; this is not the case for the cosmic string , where coil formation is subject to the low - velocity cut - off effect discussed above . in the strong - field regime ( with @xmath120 ) , a comparison to the dust solution is uninformative since the impact parameter is well below the critical value for particles and the portion of the dust string near the equatorial plane is captured by the black hole .
however , it does reinforce the idea that internal tension plays an important role in the dynamics of the string , helping the string avoid capture for impact parameters well below that of the dust string .
0 in reference @xcite we presented a plot of the critical impact parameter for capture as a function of velocity , the capture curve for cosmic strings . in that paper
, we used the massive - particle boundary conditions exclusively and showed that , for string lengths greater than @xmath121 , the capture curve was definitive for velocities greater than @xmath122 .
we are now in a position to revisit this curve using the perturbative boundary conditions in the improved numerical solver , and to merge the numerical findings with the ultra - low velocity results of @xcite to present a complete capture curve .
the revised capture curve shown in figure 11 validates two claims made in the previous paper .
first , that the capture curve for long strings is indeed definitive ( provided focusing of end points is negligible ) .
this is so because the revised capture curve is identical to that of the @xmath123 string for @xmath124 .
second , that there is indeed a minimum impact parameter at intermediate velocities , which occurs at @xmath125 and has a value of @xmath126 .
however , it is important to note that the new solver is also velocity - limited in that , at sufficiently low velocities , second - order perturbative effects become important and invalidate the first - order solutions used to set the boundary conditions
. nevertheless , analytic results @xcite are available for this velocity range , and the numerical results join smoothly with the analytical result for the critical impact parameter for very low velocities @xmath127 and is shown as a dashed line in figure 11 .
this capture curve also is in a good agreement with a recent result by page @xcite for the critical impact parameter , @xmath128\,.\ ] ] this formula reproduces the general shape of the curve shown in figure 11 , without the fine structure that is revealed by the numerical approach .
the error bars in figure 11 are quite small .
this suggests that the capture curve is not a smooth line , but has some fine structure to it ; this is particularly obvious at lower velocities , but this structure is also observed on a smaller scale at higher velocities as well .
this could possibly be explained by the complicated nature of the trajectory of the string for near - critical impact parameters , and its extreme sensitivity to initial data .
however , detailed studies ( carried out subsequent to the drafting of the present paper ) of the capture curve at ultra - relativistic velocities reveal that the structure is damped out to a certain extent by using finer grids , so it is important to note that some ( and perhaps all ) of the `` jitter '' in figure 11 may be numerical in origin .
it should be emphasized that our numerical results demonstrate that in the limit @xmath129 the critical impact parameter reaches the same value , @xmath130 , as the critical impact parameter for capture of ultrarelativistic particles .
this result is inconsistent with the earlier results of moss and lonsdale @xcite .
moreover , the analytical result ( [ n6.1 ] ) also is inconsistent with the results of moss and lonsdale , who quote a low - velocity dependence of @xmath131 obtained from a numerical fit to their capture curve .
0 in this paper , we discussed the gravitational scattering of cosmic strings by schwarzschild black holes .
this paper brought together earlier results for the critical impact parameter for capture , weak - field analytic solutions , and studied the limitations of approximate solutions as a function of velocity and impact parameter by generating numerical solutions to the equations of motion .
earlier findings have been corroborated by these latest results , and numerical , weak - field , and shockwave results are consistent with one another in the regions where they overlap .
the scattering problem was studied in three regimes . in the ultra - relativistic regime
, we showed that strings form coils , provided that the conditions @xmath132 and @xmath133 are satisfied ; this phenomenon cuts off for velocities below @xmath134 .
a detailed comparison of analytic and numerical results showed that the perturbative results break down for @xmath135 ; this is a surprisingly small value and suggests that the weak - field approximation is quite acceptable for all cases except near - critical scattering . where the weak - field approximation breaks down , in the strong - field regime , kinks are produced ( without coils for @xmath136 , with coils for @xmath137 ) , but their amplitude is larger than predicted by approximate solutions . in the case of near - critical scattering , the string worldsheet evolves highly complicated structures .
a detailed investigation of string motion in this case showed that there are two characteristic trajectories for capture , the first where the string enters directly , the second where the string evolves a small coil before crossing the horizon .
this complicated behaviour , and its extreme sensitivity to initial data , may help explain the structure observed in the capture curve ; however , a numerical origin to this structure has not been ruled out .
this paper has also shown the critical role performed by tension in determining the dynamics of the string .
the absence of multiple string windings around the black hole and the complicated folds observed in the string worldsheet at near - critical scattering , even at moderate @xmath112-factors , are consequences of internal tension . :
this work was partly supported by the natural sciences and engineering research council of canada .
one of the authors ( v.f . ) is grateful to the killam trust for its financial support .
the authors are also grateful to don page for numerous discussions .
0 in this appendix we present a solution for an ultra - relativistic straight string scattered by a black hole .
the key observation which leads to a solution of the equations of motion in the ultra - relativistic limit is the following : in the reference frame of the string the black hole moves with @xmath138 and its gravitational field is boosted to the shock wave @xcite . as a result , before and after crossing the null surface @xmath139 representing the black hole , the string obeys the free equations in flat spacetime .
all the information concerning the non - linear interaction with the gravitational field of the black hole can be obtained in the form of `` jump '' conditions on the null surface @xmath139 .
such solutions were studied earlier ( see e.g. reference @xcite ) . to obtain the metric of an ultra - relativistic black hole one starts with the metric ( [ 2.9 ] ) written in the form @xmath140 where @xmath141 and make the boost transformation @xmath142 let @xmath143 so that @xmath144 and , since @xmath145 , @xmath146 using eqns .
( [ n8.3b ] ) the metric ( [ n8.1 ] ) takes the form @xmath147 \\ \nonumber & + & { \gamma^2 \over 2}(1+v)^2 ( \phi+\psi){\left(dx_+ + { 1-v \over 1+v } dx_-\right)}^2\ , .\end{aligned}\ ] ] in the limit @xmath153 and @xmath154 fixed one has @xcite @xmath155 where @xmath156 . from this
it follows that the metric in the aichelburg - sexl form @xcite is obtained @xmath157 where @xmath158 .
the boosted metric in this form represents a gravitational shockwave .
the following components of the christoffel symbols do not vanish : @xmath159 @xmath160 the equation for @xmath65 is @xmath167 the solution obeying the initial conditions ( [ n8.9 ] ) is solved using green s function methods @xcite and has the form @xmath168 \ , .\ ] ] similarly , solving equation @xmath169 for @xmath31 with the initial conditions ( [ n8.9 ] ) one gets @xmath170 \ , .\ ] ] expression for @xmath171 can be easily obtained by using constraint equations ( [ 2.4 ] ) which for the case under consideration take the form ( @xmath172 ) @xmath173 @xmath174 here a dot and prime denote derivatives with respect to @xmath175 and @xmath87 . by integrating these equations one gets @xmath176\ ] ] @xmath177\ , .\ ] ] 9 frolov , v.p .
, hendy , s. , and larsen , a.l .
phys . rev . * d54 * , 5093 - 5102 ( 1996 ) .
nambu , y. , in : _ proceedings of int . conf . on symmetries and quark models_,wayne state unv .
goto , t. , prog .
* 46 * , 1560 .
polyakov , a.m. , phys . lett . *
b103 * , 207 - 210 ( 1981 ) .
frolov , v.p . ,
skarzhinsky , v.d . ,
zelnikov , a.i . , and heinrich , o. phys . lett . *
b224 * , 255 ( 1989 ) . de villiers , j.p . , and frolov , v.p .
phys . rev . * d58 * , 105018(8 ) ( 1998 ) .
page , d.n .
phys . rev . * d58 * , 105026(13 ) ( 1998 ) .
lonsdale , s. and moss , i. nucl .
b298 * , 693 - 700 ( 1988 ) . de villiers , j.p . , and frolov , v.p . int
* d7 * no .
6 ( 1998 ) 957 - 967 .
vilenkin , a. and shellard , e. p. s. , _ cosmic strings and other topological defects . _
( cambridge univ . press ,
cambridge ) ( 1994 ) .
misner , c.w.,thorne , k.s . , and wheeler , j.a .
_ gravitation _ ( w.h .
freeman , san francisco , 1973 ) .
page , d.n .
preprint gr - qc/9902038 ( 1999 ) .
aichelburg , p.c . , and sexl , r.u .
* 2 * , 303 - 312 ( 1971 ) .
amati , d . , and klimcik , c. phys . lett .
* b210 * 92 - 96 ( 1988 ) .
albrecht , a. and york , t. phys . rev . * d38 * , 2958 - 2962 ( 1988 ) .
hayashi , k . , and samura , t. phys .
* d50 * 3666 - 3673 ( 1994 ) . | this paper discusses the gravitational scattering of a straight , infinitely long test cosmic string by a black hole .
we present numerical results that probe the two - dimensional parameter space of impact parameter and initial velocity and compare them to approximate perturbative solutions derived previously .
we analyze string scattering and coil formation in the ultra - relativistic regime and compare these results with analytical results for string scattering by a gravitational shock wave .
special attention is paid to regimes where the string approaches the black hole at near - critical impact parameters .
the dynamics of string scattering in this case are highly sensitive to initial data and transient phenomena arise while portions of the string dwell in the strong gravitational field near the event horizon of the black hole .
the role of string tension is also investigated by comparing the scattering of a cosmic string to the scattering of a tensionless `` dust '' string .
finally , the problem of string capture is revisited in light of these new results , and a capture curve covering the entire velocity range ( @xmath0 ) is given . _
pacs number(s ) : 04.60.+n , 12.25.+e , 97.60.lf , 11.10.gh_ = .8 cm |
percolation theory has attracted a great deal of interest in the last few decades and the activity in the field is still growing @xcite . this is mainly because some aspects of the percolation process such as the geometrical phase transitions occurring in the system have gained a particular impetus due to the introduction of techniques such as monte carlo ( mc ) simulations and series expansions @xcite . however , the problem is far from being exhausted .
in fact , most of the studies are devoted to the percolation of molecules that are irreversibly deposited . in part
this is due to the fact that the deposition ( or irreversible adsorption ) of particles on solid surfaces is a subject of considerable practical importance . in many experiments on adhesion of colloidal particles and proteins on solid substrates ,
the relaxation time scales are much longer than the times of the formation of the deposit . in such processes
, the temperature of the system does not play any relevant role and it is not considered .
however , in numerous systems of both theoretical and practical importance , where the adsorbed particles are in thermodynamic equilibrium , the spatial distribution of the adsorbate might be characterized by using the percolation model @xcite . in these cases ,
the temperature governs the phase in the system and can be an important controlling factor in the percolation process . in the simplest case , where repulsive monomers are adsorbed on a square lattice , the system exhibits a continuous phase transition from a disordered state at high @xmath1 to a doubly - degenerated @xmath2 ordered state at low @xmath1 , with a critical temperature @xmath3 which satisfies @xmath4 @xcite , with @xmath5 , @xmath6 the boltzmann constant and @xmath7 the interaction energy between nearest - neighbour adatoms . on the other hand ,
the same system with attractively interacting particles goes through a first - order phase transition .
it is clear that a percolation study on the spatial configuration of the adlayer should recognize peculiarities related to the phase transitions occurring in the adsorbate .
such a study implies the determination of the critical parameters as a function of the concentration ( surface coverage ) and temperature . to the best of our knowledge
this study has not been done and it is the main purpose of this paper .
the paper is organized as follows . in section [ bd ] the model of adsorption of interacting monomers on a two - dimensional square lattice
is presented .
the analysis of results obtained by using finite - size scaling theory is given in section [ fss ] . in section [ pd ] , the phase diagram is discussed along with the basis of a thermal finite - size scaling study .
finally , conclusions are drawn in section [ c ] .
let us consider that the substrate is represented by a two - dimensional square lattice of @xmath8 equivalent adsorption sites , with periodic boundary conditions . in order to describe the system of @xmath9 monomers
adsorbed on @xmath10 sites at a given temperature @xmath1 , let us introduce the occupation variable @xmath11 which can take the following values : @xmath12 particles can be adsorbed on the substrate with the restriction of at most one adsorbed particle per site and we consider a nearest - neighbour ( @xmath13 ) interaction energy @xmath7 between them . under these considerations ,
the hamiltonian of the system is given by @xmath14 where @xmath15 represents pairs of @xmath13 sites and @xmath16 is the adsorption energy of the sites on the surface .
in addition , we have taken @xmath17 without loss of generality .
for fixed values of surface coverage , @xmath18 , and temperature @xmath1 , the thermodynamic equilibrium is reached in the canonical ensemble by using a standard kawasaki algorithm @xcite .
the procedure is as follows .
an initial arbitrary configuration of @xmath9 adsorbed monomers with the desired surface coverage is generated .
two sites are randomly selected , and their positions are established .
then , an attempt is made to interchange their occupancy state with probability given by the metropolis rule @xcite : @xmath19 where @xmath20 is the difference between the hamiltonian evaluated at the final state and the one computed at the initial state . a monte carlo step ( mcs )
is achieved when @xmath10 pair of sites have been tested to change its occupancy state .
the time ( expressed in units of mcs ) required for equilibration depends on lattice size , temperature and coverage .
typically , @xmath21 mcss suffices for smaller lattices containing up to @xmath22 sites in the whole range of both temperature and coverage .
then , a set of @xmath23 samples in thermal equilibrium is generated by taking configurations separated from each other by @xmath24 mcss in order to avoid possible correlations between the states . in the smallest lattices considered in this paper
, @xmath25 prevents such undesired correlations .
the accuracy of the procedure and the correctness of the algorithm were tested by obtaining the behaviour of different quantities ( for example , the adsorption isotherms ( surface coverage , @xmath26 , versus the normalized chemical potential @xmath27 ) , the specific heat , the order parameter , etc ) and comparing with the corresponding ones derived from the real space renormalization group approach ( rsrg ) @xcite .
the central idea of the percolation theory is based on finding the minimum concentration @xmath26 for which at least a cluster ( a group of occupied sites in such a way that each site has at least one occupied nearest - neighbour site ) extends from one side to the opposite one of the system .
this particular value of the concentration rate is named _ critical concentration _ or _
percolation threshold _ and determines a phase transition in the system . in this paper
, the percolation process will be studied under two different perspectives .
namely , samples will be prepared for fixed temperature ( coverage ) and variable coverage ( temperature ) .
we call this feature _ coverage percolation _ ( _ thermal percolation _ ) . in the random percolation model , a single site
is occupied with probability @xmath28 .
the samples are then generated through irreversible adsorption . in our problem ,
the occupancy state of every site strongly depends on @xmath29 . in both cases , for a precise value of concentration , the percolation threshold of sites , at least one spanning cluster connects the borders of the system ( indeed , there exists a finite probability of finding @xmath30 ( @xmath31 ) spanning clusters @xcite ) .
then , a second - order phase transition appears at such coverage which is characterized by well - defined critical exponents .
it should be emphasized that for @xmath32 the system becomes uncorrelated and formally we have the random percolation model .
a study of the finite - size effects allows us to make a reliable extrapolation to the thermodynamic limit ( @xmath33 ) .
details of this study will be given below .
it is well known that it is a quite difficult matter to analytically determine the value of the percolation threshold for a given lattice @xcite . for some special types of lattices ,
geometrical considerations enable us to derive their percolation thresholds exactly .
thus , exact thresholds for the random percolation problem are known for ( a ) square , triangular and honeycomb lattices and ( b ) triangular and kagome lattice concerning the bond and site problem , respectively . in both cases ,
analytical results are obtained when a monomeric species is considered . for different conditions , i.e. for systems which do not present such a topological advantage
, percolation thresholds have to be estimated numerically by means of computer simulations .
as the scaling theory predicts @xcite , the larger the system size to study , the more accurate the values of the threshold obtained therefrom .
thus , the finite - size scaling theory gives us the basis to achieve the percolation threshold and the critical exponents of a system with a reasonable accuracy .
for this purpose , the probability @xmath34 that a lattice composed of @xmath35 elements ( sites or bonds ) percolates at concentration @xmath26 can be defined @xcite . here , as in @xcite ,
the following definitions can be given according to the meaning of @xmath36 : ( a ) @xmath37 is the probability of finding a rightward ( downward ) percolating cluster ; ( b ) @xmath38 is the probability that we find a cluster which percolates both in a rightward * and * in a downward direction ; ( c ) @xmath39 is the probability of finding either a rightward * or * a downward percolating cluster and ( d ) @xmath40 \equiv \frac{1}{2 } \left[r^{i } _ l(\theta)+r^{u } _
l(\theta ) \right ] $ ] .
the first step for determining the percolation threshold consists in evaluating the effective threshold @xmath41 ( the concentration at which the slope of @xmath42 is the largest ) for a lattice of finite - size @xmath43 . in the mc simulations , @xmath42 is determined for each discrete value of @xmath26 according to the considered finite regular lattice @xcite .
once the temperature is fixed , the next procedure is followed : ( a ) the construction of @xmath44 samples for a given coverage ( according to the scheme presented in section [ bd ] ) and ( b ) the cluster analysis by using the hoshen and kopelman algorithm @xcite . in the last step ,
the number of clusters for each sample , @xmath45 , of size @xmath46 ( a cluster of size @xmath46 is composed by @xmath46 connected elements ) is determined in order to verify whether a percolating island exists .
this spanning cluster could be determined by using the criteria @xmath47 , @xmath48 , @xmath49 or @xmath50 .
@xmath44 runs of two such steps are carried out for obtaining the number @xmath51 of them for which a percolating cluster of the desired criterion @xmath36 is found .
then , @xmath52 is defined and the procedure is repeated for different values both of @xmath26 and lattice sizes , @xmath43 ( @xmath53 and @xmath54 ) . in figure @xmath55 , the probabilities @xmath38 ( squares ) , @xmath39 ( circles ) and @xmath56 ( triangles ) are presented .
three different values of @xmath29 are shown . from a first inspection of the figure ( and from data not shown here for the sake of clarity ) it is observed that ( a ) curves cross each other in a unique universal point , @xmath57 , which depends on the criterion @xmath36 used ; ( b ) those points do not modify their height for the different @xmath29 s .
this finding indicates , as is expected , that the universality class of the phase transition involved in the problem is conserved no matter what the value of @xmath29 is ; ( c ) those points are located at very well - defined values in the @xmath26-axes , determining the critical percolation threshold for each @xmath29 ; ( d ) the standard percolation problem is recovered for @xmath32 giving a critical coverage @xmath58 and ( e ) @xmath59 increases ( decreases ) for positive ( negative ) value of @xmath29 .
a detailed explanation of this point will be given below .
the second step is the extrapolation of @xmath60 towards the limit @xmath33 by using the scaling hypothesis .
thus , the correlation length , @xmath61 , can be expressed as @xmath62 where the critical exponent @xmath63 is analytically shown to be equal to @xmath64 in the case of random percolation @xcite . as @xmath65 the correlation length reaches the linear dimension @xmath43 of the lattice .
thus , we have @xmath66 where @xmath67 is a non - universal constant .
figure @xmath68 shows the extrapolation towards the thermodynamic limit of @xmath60 according to equation ( [ extrapolation ] ) for different values of @xmath29 as indicated .
this figure lends support to the assertion given by equation ( [ extrapolation ] ) : ( a ) all the curves ( different criteria ) are well correlated by a linear function , ( b ) they have a quite similar value for the ordinate in the @xmath69 and ( c ) the fitting determines a different value of the constant @xmath70 depending on the type of criterion used .
it is also important to note that @xmath71 gives an almost perfect horizontal line which is a great advantage of the method because it does not require precise values of critical exponents in the process of estimating percolation thresholds .
the maximum of the differences between @xmath72 and @xmath73 give the error bar for each determination of @xmath59 . the scaling law hypothesis also predicts the collapsing of the curves @xmath74 when they are plotted as a function of a reduced variable @xmath75 : @xmath76 @xmath77 being with the scaling function .
thus , @xmath78 is a universal function with respect to the variable @xmath79 . in figure @xmath80 , as an illustration , we plot @xmath81 as a function of @xmath79 for @xmath32 .
this gives an additional indication for the numerical value of the critical exponent @xmath63 .
as is clearly seen from this analysis , the problem belongs to the same universality class of random percolation regardless of the value of @xmath29 considered .
the same procedure described above can be realized for different values of @xmath29 as is shown in figure @xmath82 just for @xmath32 , @xmath55 and @xmath83 .
thus , for a given value of @xmath29 , all the curves used in the experiment ( for different values of @xmath43 ) collapse into a universal curve according to the theoretical prediction .
however , @xmath78 is not only a function of @xmath26 and @xmath43 but also of @xmath29 . as can be seen
, the collapsing function is different for each value of @xmath29 considered .
this fact determines that the scaling function @xmath78 is not a universal function with respect to the variable @xmath29 ( each value of @xmath29 is represented by using a different type of line as indicated ) . in order to determine the dependence of @xmath78 with @xmath29 , the main features of the collapsing data have to be considered in the range of @xmath29 between @xmath84 and @xmath85 . as can be seen
, the curves become steeper upon increasing the value of @xmath29 .
in fact , the derivative of the universal function @xmath78 with respect to @xmath79 behaves as a gaussian - like function .
thus , we can observe that : 1 .
the derivatives become more pronounced as @xmath29 increases .
it is possible to establish a power law to describe this behavior .
then , + @xmath86 + in a log log scale the points are very well correlated by a linear function , as in equation ( [ sca1 ] ) , the numerical values of the fitting parameters being very similar for the three criteria used here .
the derivatives are narrowed upon increasing @xmath29 .
this behaviour can also be described by a power law according to : + @xmath87 + where @xmath88 is the standard deviation of @xmath89 for each curve .
thus , the standard deviation of each derivative versus @xmath29 when plotted in a log log scale is very well correlated by a linear function ( not shown here ) , with the fitting parameter @xmath90 for @xmath70 , @xmath50 and @xmath49 criteria . according to the above equations
, a metric factor might be included in the scaling function , equation ( [ functionr ] ) , in order to collapse all the curves in figure @xmath82 onto a single one .
following @xcite , in figure @xmath91 we plot the probability @xmath81 as a function of the argument @xmath92 .
as is clearly observed , all the curves collapse onto a single one .
it is remarkable that more than @xmath93 points are included in the collapsing curve .
the metric factor introduced here , @xmath94 , gives an additional indication for the numerical value of the exponent @xmath95 obtained in equation ( [ sca2 ] ) .
by using the scheme discussed above , the critical curve , @xmath59 versus @xmath29 , separating the percolating and non - percolating regions , is presented in figure @xmath96 . in the studied range ,
three regimes can be distinguished : @xmath97 for @xmath98 , @xmath59 remains constant as @xmath29 is decreased ; @xmath99 from @xmath100 up to@xmath101 ( @xmath100 ( @xmath101 ) being the reduced critical temperature for the condensation ( order disorder ) phase transition occurring in the system ) , @xmath59 increases linearly with k and @xmath102 for @xmath103 , @xmath59 remains constant as @xmath29 is increased .
this behaviour can be explained by simple geometrical arguments .
namely , lateral attractive interactions favour the nucleation , which in turn increases the local connectivity and diminishes the percolating fraction of occupied sites .
in contrast , repulsive couplings avoid the occupation of nearest neighbour sites , and consequently , increase the percolation threshold .
in the limit cases , once @xmath104 is reached , the adlayer does not vary significantly as @xmath105 is increased , and @xmath59 reaches its saturation value .
thus , @xmath106 for @xmath98 and @xmath107 for @xmath108 .
as can be seen from the phase diagram , there exists an alternative route for determining the critical curve .
in fact , the surface coverage can be kept constant while the reduced temperature , @xmath29 , is varied .
this procedure does not constitute the ` standard ' technique for calculating percolation features and it will be revised in detail in what follows .
it is important to emphasize that the idea of studying phase transitions upon varying a controlling parameter and at the same time to keep constant the surface coverage is not new .
in fact , this strategy has been used in several works related with either kinetic ( for example , in combination with the well - known zgb model , etc ) @xcite or equilibrium phase transitions @xcite .
however , to the best of our knowledge this study has not been done for the present model despite its simplicity . in figure @xmath109
typical curves of the probabilities @xmath110 ( squares ) , @xmath111 ( circles ) and @xmath112 ( triangles ) are shown for two different values of coverage ( @xmath113 , full symbols and @xmath114 , open symbols ) .
for each case ( fixed values of @xmath26 and a given criterion ) , just three different lattice sizes are shown in the figure .
the curves cross each other in a unique point , @xmath57 , which depends on the chosen criterion .
these points do not change their numerical values regardless of the used coverage .
these findings encourage for considering a finite - size scaling analysis with the temperature as independent variable by following the same rules as in section [ fss ] where the percolation probability was calculated as a function of coverage .
thus , for each curve , @xmath115 is determined by least mean - square fitting .
an extrapolation of @xmath115 towards the limit @xmath33 by using the scaling hypothesis can be argued as @xmath116 where @xmath63 is the critical exponent associated with the correlation length and @xmath117 is a constant for each criterion used .
equation ( [ extrapolation01 ] ) is supported by numerical results as is shown in figure @xmath118 for surface coverage @xmath119 as an illustrative example .
thus , all the curves ( different criteria ) are well correlated by a linear function with a quite similar value for the ordinate in the limit @xmath69 .
it is important to emphasize that @xmath120 is almost independent of the lattice size .
therefore , the crossing points ( figure @xmath109 ) are located at very well - defined values in the @xmath29-axes determining the critical percolation threshold for each coverage . by using standard finite - size scaling ,
it is possible to determine the critical exponent @xmath63 which results to be equal to @xmath121 regardless of the value of @xmath26 considered as both ( a ) in the case of random percolation and ( b ) in section [ fss ] . as a consequence of the above results , it is possible to collapse all the curves in figure @xmath109 onto a single one for each coverage when
@xmath122 is plotted as a function of @xmath123 . as an example , we plot @xmath124 as a function of @xmath125 for the surface coverage @xmath126 in figure @xmath127 , giving an additional indication for the calculated numerical value of the critical exponent @xmath63 . the same procedure described above
was realized for different concentrations .
this fact demonstrates that the scaling function depends not only on the variable @xmath29 but also on the coverage .
thus , for determining the dependence of @xmath78 with @xmath29 , the main features of the collapsing data must be considered : the curves become steeper upon increasing the coverage .
in fact , the derivative of the function @xmath78 with respect to @xmath125 behaves as a gaussian - like function .
thus , we can observe the following facts : the derivatives ( a ) become more pronounced and ( b ) are narrowed as the coverage increases .
the latter is the most important finding where one wishes to find the appropriate parametric factors in order to collapse all the curves , for different values of @xmath26 , into a single universal curve for each criterion .
such behaviour can also be described by a power law according to @xmath128 where @xmath129 is the standard deviation of @xmath130 . according to the above equations
, a metric factor might be included in the scaling function in order to collapse all the curves onto a single one for each criterion . in figure
@xmath131 we plot the probability @xmath132 as a function of the argument @xmath133 .
as is clearly observed , all the curves collapse onto a single one for each used criterion .
this fact allows us to determine the numerical value of the parameter @xmath134 which results to be equal to @xmath135 regardless of the criterion used .
the curves nicely collapse in the close vicinity of @xmath136 ( close to the critical point ) and a tiny deviation is observed as @xmath137 increases . the scaling analysis given above should be rigorously valid only for sufficiently large @xmath43 and from @xmath1 in the asymptotic critical regime .
however , it can be applied to the entire range of @xmath43 and @xmath1 if the data fall in the ` domain of attraction ' of a simple fixed point characterizing only one universality class of critical phenomena .
it is remarkable that more than @xmath93 points are included in the collapsing curve .
the metric factor introduced here , @xmath138 , gives an additional indication for the numerical value of the exponent @xmath134 obtained by using equation ( [ sca4 ] ) .
we presented a model to investigate the process of adsorption of interacting monomers on a square lattice and studied the percolating properties of the adsorbed phase . by using monte carlo simulation and finite - size scaling theory
, we obtained the percolation thresholds for different values of concentration and temperature . from this analysis , a critical curve in the @xmath139 space
was addressed .
the coexistence line , separating the percolating and nonpercolating regions , is characterized by three regimes : ( 1 ) for @xmath140 , @xmath59 remains constant ( @xmath106 ) as @xmath29 is decreased ; ( 2 ) from @xmath100 up to @xmath101 , @xmath59 increases almost linearly with @xmath29 and ( 3 ) for @xmath103 , @xmath59 remains constant ( @xmath141 ) as @xmath29 is increased .
each point in the critical curve was corroborated by following an alternative scheme : the surface coverage can be kept constant while the reduced temperature , @xmath29 , is varied .
this study does not constitute the ` standard ' procedure for calculating percolation properties . the results in this paper show that the new technique seems to be a promising method for describing the percolation behaviour of adlayer at equilibrium or what we called thermal percolation . in all considered cases
, the finite - size scaling study indicates that the model belongs to the universality class of the random percolation model .
this work was supported in part by conicet ( argentina ) , fundacin antorchas ( argentina ) and the universidad nacional de san luis ( argentina ) under project 322000 .
one of the authors ( ajrp ) is grateful to the departamento de qumica , universidad autnoma metropolitana - iztapalapa ( mxico , df ) for its hospitality during the time this manuscript was prepared .
10 l. n. shchur , _ incipient spanning clusters in square and cubic percolation _ in springer proceedings in physics , volume 85 , eds .
landau , s.p .
lewis , and h.b .
schuettler , ( springer verlag , heidelberg , berlin , 2000 ) .
privman v , hohenberg p c and aharony a 1991 universal critical - point amplitude relations ( phase transitions and critical phenomena vol 14 , ed c domb and j l lebowitz ( new york : academic ) , chapter 1 , pp 1134 and 3647 . | in this paper the percolation of monomers on a square lattice is studied as the particles interact with either repulsive or attractive energies . by means of a finite - size scaling analysis ,
the critical exponents and the scaling collapsing of the fraction of percolating lattice are found . a phase diagram separating a percolating from a non - percolating region
is determined .
the main features of the phase diagram are discussed in terms of simple considerations related to the interactions present in the problem .
the influence of the phase transitions occurring in the system is reflected by the phase diagram .
in addition , a scaling treatment maintaining constant the surface coverage and varying the temperature of the system is performed . in all the considered cases ,
the universality class of the model is found to be the same as for the random percolation model .
# 1 # 1 # 1#1 # 1|#1| 2n(n)^2/ n 2x2c@xmath0 pacs : 64.60.ak ; 68.35.rh ; 68.35.fx keywords : percolation , monte carlo simulations , finite size scaling theory , phase transitions . |
the modern cosmology , especially observational cosmology , reminds empirical science in a crisis phase @xcite .
the recent observations of supernovae type ia ( snia ) indicates that the universe is accelerating at the present epoch @xcite .
we accept that the present evolution is well described by the general relativity theory with the robertson - walker type of space symmetry and the source of gravity is perfect fluid then the acceleration of universe expansion can be explained only in the following way .
the universe is filled additionally to nonrelativistic matter with dark energy of unknown origin , which violates the strong energy condition . when the perfect fluid with the energy @xmath2 and pressure @xmath3the source of gravity satisfies the strong energy condition then the explanation of snia observations requires the modification of the einstein equations .
if we postulate the robertson - walker symmetry then there are some propositions of modification of the friedmann first integral .
freese and lewis considered flat cosmological models with the additional term @xmath4 , where @xmath5 and @xmath6 are constants @xcite . the parameters of this model were confronted with the observation of distant type ia supernova @xcite .
both approaches can be tested statistically by searching model parameters which best fits to the snia data . but to distinguish among the models it is necessary to take additional observational constraints .
when we assume the matter density value which is indicated by cosmic microwave background ( cmb ) and galactic counting observations then the most promising model is the dark energy model with the cosmological constant .
the cosmological constant has very long history and is still the source of various problems and troubles @xcite .
the main problem with today face of the cosmological constant is that its value is negligible in comparison to the planck mass . in other words
the @xmath7 cold dark matter ( @xmath7cdm ) model looks like the effective theory which gives us the description of the phenomenon of acceleration without giving any understanding .
another problem is why the energy densities of dark energy and of dust - like matter are of the same order of magnitude at the present epoch @xcite .
this problem is known as a `` cosmic coincidence conundrum '' .
the beginning of the vacuum energy and the dust - like matter is related to different epochs separated by very long interval of time .
one of the possibility of the explanation of this coincidence is the intrinsic feedback between the energy density of dark matter and dark energy modelled by quintessence scalar fields @xcite . in the context of scalar quintessence fields
it was proposed another interesting model with interaction @xcite
. it would be worthy to mention a model with coupling between dark energy and dark cold matter which reproduce power law solutions for energy density @xcite .
this relation was constrained by the cmb observations @xcite and snia observations using statefinder diagnostic parameters ( see @xcite and references therein ) .
we differ in the presented approach that we do not assume any physical mechanism of the energy transfer , which is treated on the phenomenological level .
we argue that luminosity distance versus redshift relation is the very first cosmological test that probes the interactions between dark matter and dark energy .
this interaction is proposed and is checked whether the `` cosmic coincidence conundrum '' is solved .
in particular we are interested in the direction of energy transfer . for this aim
we assume the friedmann equation @xmath8 where @xmath9 is the scale factor , @xmath10 is the curvature index , and a dot means the differentiation with respect to the cosmological time @xmath11 .
the second equation describing the evolution of the model is based on the adiabatic condition @xmath12 which for the friedmann - robertson - walker ( frw ) models with some perfect fluid assumes the form @xmath13 because eq .
( [ eq:2 ] ) has the local character the standard interpretation is that @xmath2 and @xmath3 describe the effective energy density and pressure of multifluid which do not interact each other .
then eq .
( [ eq:2 ] ) describes separately the evolution of each component .
equation ( [ eq:2 ] ) can be rewritten to the form @xmath14 where the hubble function @xmath15 .
let us note that the cosmological constant ( for which @xmath16 ) does not contribute to the conservation condition ( [ eq:3 ] ) as long as it is treated as the noninteracting with the rest matter .
we postulate that apart from dust matter there is dark energy , but both fluids interact now and energy can be transported from dark energy sector to the nonrelativistic matter sector . therefore , eq . ( [ eq:2 ] ) can not be separable for every component of multifluid .
the special case of considered class of model are decaying vacuum cosmologies or @xmath0 models @xcite .
the inspiration for constructing the noninteracting cosmologies is taken from wojciulewitsch @xcite ( in context of dark energy see @xcite ) .
the main aim of the paper is to show that the observation of distant snia offer the possibility of testing the energy transport from the vacuum sector to the nonrelativistic matter sector which includes dark matter .
we show that the measurements of third order term in the expansion of the luminosity distance relation with respect redshift @xmath17 ( jerk ) allows to detect the energy transport .
higher order terms in the expansions ( snap , crackle , etc . )
control the velocity , acceleration of energy transport .
note that statefinder parameters also control third derivatives but they are inadequate if we want to detect the energy transfer directly from observations .
we assume the different interpretation of eq .
( [ eq:2 ] ) rather than its modification .
we construct the general class of the decaying dark energy models with the interaction starting from the friedmann first integral which is independent of the form of pressure of fluid .
we assume for simplicity some two - component fluid with effective pressure and energy @xmath18 where @xmath19 ( @xmath20 ) describes dark energy and @xmath21 is the energy of dust matter .
if we put @xmath22 the special case of the cosmological constant is recovered .
the expression for the conservation condition can be rewritten to the form @xmath23 the first term in eq .
[ eq:5 ] ) describes the net rate of absorption of energy per unit time in unit of comoving volume transfered out of the decaying vacuum fluid to the sector of nonrelativistic fluid .
if we consider @xmath24 the phantom fields are transported .
relation ( [ eq:5 ] ) is usually interpreted without interaction between the sectors .
following wojciulewitsch we postulate that the local energy conservation law ( [ eq:5 ] ) can be written as @xmath25 the function @xmath26 is only a phenomenological description of interaction between two sectors .
of course , the exact model of this interaction should be taken from the particle physics . if @xmath27 the energy is transfered out of the vacuum , while if @xmath28 the energy is transfered in the opposite direction .
integration of eq .
( [ eq:6 ] ) gives @xmath29 where the index `` 0 '' means that the quantities are evaluated today .
it would be useful for our further analysis to represent the friedmann first integral ( [ eq:1 ] ) in the form for a particle moving in the one - dimensional potential @xmath30 because of relation ( [ eq:7 ] ) the potential function is explicitly time dependent and now takes the following form @xmath31\ ] ] where @xmath32 in the concordance @xmath7cdm models both matter and the cosmological constant are treated separately without the interaction , so both functions @xmath33 and @xmath34 ( densities @xmath35 and @xmath36 ) are constant .
the presence of the interaction manifests in the model by appearing the time dependence of the potential function ( [ eq:9 ] ) .
let us note that we postulate the time dependence of @xmath26 through the scale factor , i.e. , @xmath37 , and the potential function becomes only a function of @xmath9 but the exact form of @xmath38 is required .
it is convenient to represent the dynamics of the model in terms of the hubble function @xmath39 if we postulate that @xmath40 , @xmath41 and put @xmath42 then relation ( [ eq:10 ] ) can be used to fit the model parameters @xmath43 to the snia data , where @xmath44 denotes all fluids considered .
differentiation of both sides of eq .
( [ eq:8 ] ) for the potential in the form ( [ eq:9 ] ) gives the expression for acceleration @xmath45\ ] ] where where we substitute derivatives @xmath46 and @xmath47 from ( [ eq:7 ] ) .
let us rewrite equation ( [ eq:11 ] ) to the new form @xmath48\ ] ] where @xmath49 is the deceleration parameter .
to control higher derivatives of the scale factor we introduce the dimensionless parameter @xmath50 and then @xmath51 are the deceleration , jerk , and snap , respectively @xcite . in turn , to control the interaction we introduce the dimensionless transfer parameter @xmath52 by analogy to the matter density parameter @xmath53 . to verify the model we estimate of the parameter @xmath54 at the present epoch from the observation of snia data . from equations ( [ eq:10 ] ) and ( [ eq:12 ] ) we have @xmath55 and @xmath56
of course we have also the constraint condition @xmath57 . after the differentiation of both sides of eq .
( [ eq:11 ] ) we obtain the basic equation relating the jerk to the transfer density parameter @xmath58 and @xmath59 both for strings ( @xmath60 ) and topological defects ( @xmath61 ) relation ( [ eq:17 ] ) does not depend on the density parameter of dark energy @xmath62 .
this relation is obvious for all models . in the special case of the flat model @xmath63
and then we obtain @xmath64 summing ( [ eq:17 ] ) and ( [ eq:16 ] ) for any @xmath65 we obtain @xmath66 in the special case of the flat model formula ( [ eq:20 ] ) reduces to @xmath67 where @xmath68 can be always expressed in terms of @xmath69 and @xmath70 .
note that the relation @xmath71 does not depend on priors on @xmath69 for phantoms .
finally we obtain that the measurements of the jerk @xmath72 at the present epoch probes directly the effects of energy transfer as a consequence of relation ( [ eq:20 ] ) .
while the cubic term in the relation @xmath73 is the first term in the taylor expansion that depends explicitly on the @xmath26 , the higher terms in this expansion are related to the derivatives of @xmath26 .
let us define the parameter @xmath74 for the characterization of variability of @xmath26 as @xmath75 it is obvious that @xmath76 , @xmath77 , . as an illustration that one can control the first derivative of @xmath26 by the measurement of the snap @xmath78
, we prove the existence of some relation obtained by the differentiation of both sides of ( [ eq:21 ] ) . to this aim
we use the following formulas @xmath79 \\ \frac{d\omega_{x}}{dt } & = - h[\nu + \omega_{x}(1 + 3w_{x } - 2q)].\end{aligned}\ ] ] finally we obtain the relation @xmath80 let us briefly comment on the important case of @xmath22 corresponding the decaying cosmological constant @xmath0 cosmologies . of course , they constitute some special case of the considered models @xmath81 the parameter @xmath82 describes the first derivative of @xmath7 and therefore the parameter @xmath54 controls its variability . in turn
the parameter @xmath83 characterizes the convexity of the function @xmath0 @xmath84
let us consider the luminosity distanced versus redshift relation @xmath73 expanded in the taylor series with respect to redshift @xmath17 .
it can be done without knowledge about dynamical equation . for simplicity of presentation of the idea of measurement @xmath54 , @xmath85 ,
we consider the flat universe what is justified by wmap measurements .
then we obtain @xcite @xmath86 \nonumber \\ \label{eq:29 } & + \frac{z^{4}}{24h_{0 } } \left [ 2 + 2q_{0 } - 15q_{0}^{2 } - 15q_{0}^{3 } + 5j_{0 } + 10q_{0}j_{0 } + s_{0 } \right ] + \cdots \end{aligned}\ ] ] we propose to detect the time variation of energy transfer using the parameters @xmath87 .
let us start with estimation of @xmath54 as a first approximation .
we find the current constraints to the plane @xmath88 .
for this aim we mark the shaded region of the @xmath89 confidence level constraint from the recent snia measurements @xcite . because @xmath90 ( or in the general case @xmath91
the detection of the interaction is equivalent to the determination whether the jerk is different from @xmath92 ( or @xmath93 ) . if @xmath94 ( or @xmath95 ) then the energy is transfered from the dark energy sector to the nonrelativistic matter sector . if @xmath96 ( or @xmath97 ) the transport takes place in the opposite direction .
note that the negative curvature ( @xmath98 ) makes the switch of transfer direction to happen for lower value of @xmath72 .
therefore , to find the direction of transfer we should know not only the value of jerk but also the curvature of space .
in the general case for any @xmath68 we have @xmath99 where we use formula ( [ eq:16 ] ) . .
the solid line is the relation @xmath100 fulfilled in the model .
additionally , for this relation the @xmath101 confidence level interval for @xmath102 is drawn.,scaledwidth=80.0% ] we consider , for simplicity , the testing of the interaction for the flat model and the case @xmath103 which corresponds the decaying cosmological constant .
this allows to substitute @xmath104 . on fig .
[ fig:1 ] , from relation ( [ eq:20 ] ) with @xmath22 , the line @xmath105 is drawn when we assume that baryonic matter @xmath106 is equal @xmath107 .
this relation allows us to estimate the interval on @xmath102 and @xmath72 on the @xmath101 confidence level .
we mark the line @xmath100 and the vertical band to denote the interval with the @xmath101 confidence level for @xmath108 which gives @xmath109 . in this interval of @xmath102
the jerk @xmath110 is about @xmath111 .
it is very interesting that present snia observations allowed us to measure the interaction without any special assumptions about physics of the transfer process .
we thus determined the transfer energy parameter @xmath54 and concluded that if we assume that the universe is flat then the energy transfer takes place from the dark energy to dark matter .
in the previous section it was considered that @xmath112 .
now we turn to estimation of the energy transfer parameter @xmath54 using the snia data . it would be useful to consider two situation .
first , the energy transfer is between decaying vacuum and matter sectors , and second , it is between the phantom ( @xmath113 ) and matter sectors .
we would like to answer on two questions . * what is the interval of @xmath102 which rules out the energy energy transfer ( @xmath114 ) on the confidence level @xmath89 ?
* is it possible to tell a scenario with energy transfer and another with variable @xmath68 ? to answer to the first question we test the hypothesis that @xmath115 .
the transfer from decaying vacuum to matter sectors can be ruled out on the confidence level @xmath89 for @xmath116 ( fig . [ fig:2 ] ) , while the transfer from phantom to matter sectors is ruled out for @xmath117 on the same confidence level ( fig . [ fig:3 ] ) .
therefore the transfer between decaying vacuum and matter sector seems to be excluded because the extragalactic observations and cmb observations favor the values of @xmath102 in the obtained interval . on the other hand
these other observations indicate that the transfer between phantom and matter sectors is possible . hence if we have other arguments about phantom existence in the universe and if we accept that @xmath118 as indicated by wmap measurements then the energy transfer is necessary . adopting the same analysis from the previous sections to the case of no energy and variable @xmath119 we obtain analogous formulas in which @xmath120 is replaced by @xmath121 . to answer the second question we analyze the hubble diagram ( fig .
[ fig:4 ] ) .
it is shown that for very distant supernovae ( @xmath122 ) the model with variable @xmath123 predicts the brighter supernovae .
the paper was supported by kbn grant no
. 1 p03d 003 26 .
the author is very grateful dr a. krawiec , dr w. godowski , and t. stachowiak for comments and discussions during the seminar on observational cosmology . | the observations of snia suggest that we live in the acceleration epoch when the densities of the cosmological constant term and matter are almost equal .
this leads to the cosmic coincidence conundrum .
as the explanation for this problem we propose the frw model with dark matter and dark energy which interact each other exchanging energy . we show that the cubic correction to the hubble law , measured by distant supernovae type ia , probes this interaction .
we demonstrate that influences between nonrelativistic matter and vacuum sectors are controlled by third and higher derivatives of the scale factor . as an example we consider flat decaying @xmath0 frw cosmologies .
we point out the possibility of measure of the energy transfer by the cubic and higher corrections to hubble s law .
the statistical analysis of snia data is used as an evidence of energy transfer .
we find that there were the transfer from the dark energy sector to the dark matter one without any assumption about physics governing this process .
we confront this hypothesis about the transfer with snia observations and find that the transfer the phantom and matter sector is admissible for @xmath1 .
we also demonstrate that it is possible to differentiate between the energy transfer model and the variable coefficient equation of state model . |
* problem definition . *
let @xmath0 be a set of @xmath1 points in @xmath2 and @xmath3 . in the largest - empty - convex - subset problem
we want to decide whether there is a set @xmath4 of @xmath5 points in convex position whose convex hull does not contain any other point of @xmath0 .
we show that largest - empty - convex - subset is w[1]-hard with respect to the solution size @xmath5 , under the extra condition that the solution set is _ strictly _ convex , i.e. , the interior of the convex hull of any of its subsets is empty .
this means that ( under standard complexity - theoretic assumptions ) the problem is not fixed - parameter tractable with respect to @xmath5 , i.e. , it does not admit an @xmath6-time algorithm for any computable function @xmath7 and any constant @xmath8 .
see @xcite for basic notions of parameterized complexity theory and @xcite for survey of parameterized complexity results on geometric problems .
largest - empty - convex - subset has been shown to be np - hard in last year s eurocg @xcite .
( in that paper , np - hardness has been shown also for the more general version where the emptiness condition is dropped . )
several interesting questions were also raised such as whether the problem is fixed - parameter tractable with respect to the solution size and whether it admits a polynomial @xmath9-approximation algorithm .
here , we give a negative answer to the first question . note that in the plane , the problem is solvable in polynomial time ; see , for example , @xcite , @xcite . from a combinatorial point of view
, there is a long history of results starting with the famous erds - szekeres theorem @xcite , which states that for every @xmath5 there is a number @xmath10 such that every planar set of @xmath10 points in general position contains @xmath5 points in convex position .
horton @xcite showed that this is not true when the emptiness condition is imposed : there are arbitrarily large sets that do not contain empty @xmath11-gons .
results of this type exist also for higher dimensions ; see @xcite .
we show that largest - empty - convex - subset is w[1]-hard by an _
fpt_-reduction from the w[1]-hard @xmath5-clique problem @xcite : given a graph @xmath12 , e)$ ] and @xmath3 , decide whether @xmath13 contains a clique of size @xmath5 .
we begin with a high - level description of the construction , see fig .
[ fig : high_level_construction ] .
initially , the construction will lie on the plane ; later on , it will be lifted to the elliptic paraboloid with a ( more or less ) standard transform .
the construction is organized as the upper diagonal part of a grid with @xmath5 rows and @xmath5 columns .
the @xmath14th row and column represent a choice for the @xmath14th vertex of a clique in @xmath13 and are made of @xmath15 and @xmath16 gadgets respectively .
there are @xmath1 choices and each choice is represented by a collection of empty convex subsets of points
one subset with a constant number of points from each gadget .
each gadget consists of @xmath17 points within a rectangular region , which are organized in sets ( of pairs ) of collinear points .
there is a constant number of such sets and , since we are looking for strictly convex subsets , only one pair of consecutive points per set can be chosen at any time .
certain choices are rendered invalid by additional points .
neighboring gadgets share the points on their common rectangle edge , see the zoomed - in area in fig [ fig : high_level_construction ] . through these common points ,
the choice of subsets is made consistent among the gadgets .
in particular , the choice in the @xmath14th row is made consistent with the choice in the @xmath14th column via the ` diagonal ' , @xmath18{$\righthalfcup$}}\vskip-.9ex\hbox{$\boxvoid$}}}}$ ] gadget in their intersection corner ; consistency here means that they both correspond to the same choice of a vertex of @xmath13 . on the other hand , in the intersection of the @xmath14th column with the @xmath19th row , for every @xmath20
, there is a ` cross ' , @xmath21 gadget , which ensures that the choice in the column is propagated independently of the choice in the row and vice versa .
finally , the @xmath19th column is ` connected ' to the @xmath14th row , for every @xmath22 , by three additional gadgets .
one of them , the ` star ' , @xmath23 gadget , encodes graph @xmath13 , i.e. , it allows only for combinations of choices ( in the column and the row ) that are consistent with the edges in @xmath13 . locally , every valid subset from a gadget consists of points that are in strictly convex position and whose convex hull is empty . by lifting the whole construction to the paraboloid appropriately , we make sure that this property is true globally , i.e. , for any set constructed from the local choices in a consistant manner . in total ,
the construction consists of a set @xmath0 of @xmath24 points in @xmath2 , such that there exists a set @xmath4 with the desired property and @xmath25 , for some function @xmath26 , if and only if @xmath13 has a clique of size @xmath5 .
it has @xmath1 pairs of points @xmath30 , @xmath31 on the left and right side of its rectangle respectively , which correspond to the vertices of @xmath13 ; pairs corresponding to the same vertex are aligned horizontally .
it also has @xmath32 points inside the rectangle on a vertical line @xmath28 as follows .
for every two pairs @xmath30 and @xmath33 , with @xmath34 , a point is placed such that it is inside the parallelogram @xmath35 formed by the pairs but outside the parallelogram formed by any other two pairs .
effectively , this point cancels a choice of pairs that correspond to different vertices of @xmath13 .
one point is placed above @xmath36 on @xmath28 and close to its boundary ; similarly one point is placed below @xmath37 . due to the strict convexity condition , at most one pair per rectangle side and at most one pair on @xmath28 can be chosen .
thus , there are @xmath1 maximum size empty convex subsets .
each subset contains six points and is formed by three pairs : @xmath30 , @xmath31 , for some @xmath14 , and the pair of points on @xmath28 that are closest to the parallelogram @xmath38 ; this latter pair is formed by the point that cancels @xmath39 and the point that cancels @xmath40 .
the @xmath41{90}{$\boxminusnew$}}$ ] gadget is just a @xmath42-rotated copy of the @xmath27 gadget .
@xmath18{$\righthalfcup$}}\vskip-.9ex\hbox{$\boxvoid$}}}}$ ] * gadget .
* this gadget has basically the same structure as the @xmath27 gadget but propagates information diagonally . for completeness
, it is shown in fig [ fig : diagonal_simple_gadget ] . @xmath43
* this gadget propagates information both horizontally and diagonally , see fig .
[ fig : t_gadget ] .
it has @xmath1 pairs of points @xmath30 , @xmath31 , and @xmath44 , on the left , right , and bottom side of its rectangle respectively .
as before , an @xmath14th pair corresponds to the @xmath14th vertex of @xmath13 .
there will be only @xmath1 valid choices and three pairs per choice , namely , the @xmath14th pair from each side .
this is enforced by the strict convexity condition and by placing , for every @xmath14 , four additional points on two vertical lines @xmath28 and @xmath45 inside the rectangle . these points are placed _
outside _ the convex @xmath29-gon @xmath46 that is formed by the points in the corresponding pairs .
( the points are also outside every other @xmath29-gon for @xmath47 . )
see the example for @xmath48 in fig .
[ fig : t_gadget ] . more specifically , looking at the gadget from top to bottom and from left to right
, a point is placed on the intersection of @xmath28 and the line through the second point of @xmath30 and the second point of @xmath49 ; for the case of @xmath50 , the point is placed just below the @xmath29-gon .
a second point is placed on @xmath28 just above the @xmath29-gon . at the right side ,
two points are placed on @xmath45 as follows .
one point is placed on the intersection of @xmath45 and the line through the second point of @xmath30 and the first point of @xmath51 ; for @xmath50 , the point is placed just above the @xmath29-gon .
a second point is placed on the intersection of @xmath45 and the line through the second point of @xmath31 and the first point of @xmath52 ; for @xmath53 , the point is placed just below the @xmath29-gon .
there are @xmath1 maximum size empty convex subsets with 10 points each .
a subset is formed by the points in the pairs @xmath30 , @xmath31 , @xmath44 , and the four points on @xmath28 and @xmath45 that are closest to the corresponding @xmath29-gon .
it can be thought of as having two inputs ( at the upper and lower left corner ) and two outputs ( at the upper and lower right corner ) .
it propagates the inputs ( choices ) independently from each other , one horizontally and one vertically .
the input subgadgets @xmath41{90}{$\boxtriangleup$}}$ ] and @xmath41{180}{$\boxtriangleup$}}$ ] are respectively connected ( through their left and bottom rectangle sides ) to the @xmath27 and @xmath41{90}{$\boxminusnew$}}$ ] gadgets of the @xmath14th row and @xmath19th column of the global construction ( fig .
[ fig : high_level_construction ] ) .
the output subgadgets @xmath41{270}{$\boxtriangleup$}}$ ] and @xmath57 are similarly connected to a @xmath27 and @xmath41{90}{$\boxminusnew$}}$ ] gadget .
note that the subgadgets @xmath54 and @xmath55 in the middle of the @xmath21 gadget ( fig .
[ fig : high_level_cross_gadget ] ) as well as the subgadgets @xmath41{90}{${\mathrel { \vcenter{\offinterlineskip \hbox{\scalebox{.76}[0.85]{$\righthalfcup$}}\vskip-.9ex\hbox{$\boxvoid$}}}}$}}$ ] and @xmath41{270}{${\mathrel { \vcenter{\offinterlineskip \hbox{\scalebox{.76}[0.85]{$\righthalfcup$}}\vskip-.9ex\hbox{$\boxvoid$}}}}$}}$ ] are _ not _ connected directly to any row or column of the global construction .
roughly speaking , the @xmath21 gadget has the following function : it multiplexes the two inputs , then it mirrors them ( vertically and horizontally ) , and then demultiplexes them .
next , we describe the subgadgets in some more detail .
the @xmath41{90}{$\boxtriangleup$}}$ ] subgadget is shown in fig .
[ fig : horizontal_for_cross_gadget ] .
it is similar to the previously described @xmath27 gadget in fig .
[ fig : horizontal_gadget ] .
it has again @xmath1 pairs of points @xmath30 on the left side .
the difference now is that there are @xmath58 pairs of points @xmath59 , @xmath60 , on the right side of the gadget .
the second index @xmath19 basically encodes the choice coming from the input subgadget @xmath41{180}{$\boxtriangleup$}}$ ] at the lower left corner , which is communicated through the @xmath54 and @xmath41{90}{${\mathrel { \vcenter{\offinterlineskip \hbox{\scalebox{.76}[0.85]{$\righthalfcup$}}\vskip-.9ex\hbox{$\boxvoid$}}}}$}}$ ] subgadgets inbetween . only the @xmath58 pairs @xmath30 and @xmath59 constitute valid choices .
the rest are cancelled by additional points as usually .
together with pairs of canceling points ( which are also chosen as before ) there are exactly @xmath58 empty convex @xmath29-gons , and these are of maximum size .
the input @xmath41{180}{$\boxtriangleup$}}$ ] subgadget propagates information vertically and is defined similarly to the @xmath41{90}{$\boxtriangleup$}}$ ] subgadget .
it has @xmath1 pairs of points @xmath61 on the bottom side and @xmath58 pairs @xmath62 on the top side .
the difference now is that only the @xmath58 pairs @xmath61 and @xmath63 are valid .
the output subgadgets @xmath41{270}{$\boxtriangleup$}}$ ] and @xmath57 are just mirrored images of their input counterparts .
the @xmath41{90}{${\mathrel { \vcenter{\offinterlineskip \hbox{\scalebox{.76}[0.85]{$\righthalfcup$}}\vskip-.9ex\hbox{$\boxvoid$}}}}$}}$ ] and @xmath41{270}{${\mathrel { \vcenter{\offinterlineskip \hbox{\scalebox{.76}[0.85]{$\righthalfcup$}}\vskip-.9ex\hbox{$\boxvoid$}}}}$}}$ ] subgadgets are constructed in the same way as the @xmath18{$\righthalfcup$}}\vskip-.9ex\hbox{$\boxvoid$}}}}$ ] gadget in fig .
[ fig : diagonal_simple_gadget ] but have @xmath58 valid @xmath29-gons , while the @xmath54 and @xmath55 subgadgets are constructed in the same way as the @xmath43 gadget in fig .
[ fig : t_gadget ] and have @xmath58 valid @xmath56-gons . @xmath23
* this gadget encodes the edges of the input graph @xmath13 .
see fig .
[ fig : star_gadget ] .
it is similar to gadget @xmath27 ( fig .
[ fig : horizontal_gadget ] ) and allows only combinations of pairs that correspond to edges of the graph : for every _
non_-edge @xmath64 of @xmath13 , a point is placed inside the parallelogram @xmath35 .
every corner of a gadget rectangle is lifted to the paraboloid with the map @xmath65 .
see fig [ fig : lifting ] .
the images of the corners of each rectangle lie on one distinct plane ( since the corners lie on a circle ) .
the points in a gadget are projected orthogonally on the corresponding plane .
this is an affine map and thus colinearity and convexity within a gadget is preserved .
each gadget now lies on a distinct facet ( a parallelogram ) of a convex polyhedron .
a global valid subset is formed by locally choosing one valid subset from every gadget in a consistent manner . when a largest locally possible subset ( as given above ) can be chosen , the global subset has size @xmath70 .
as we will now prove , such a global subset corresponds to a @xmath5-size clique of @xmath13 .
let @xmath0 be the set of all the points in our construction .
suppose there exists a global valid subset of size @xmath70 .
then , a largest locally possible valid subset must be chosen from every gadget .
consider such a choice of subsets and let @xmath71 be the vertex of @xmath13 corresponding to the choice from the leftmost gadget of the @xmath14th grid row . by construction ,
the subset corresponding to the same vertex @xmath71 must be chosen from every other gadget in this row as well as every gadget in the @xmath14th column .
consider the @xmath19th row , for some @xmath47 . through the @xmath23 gadget that connects the @xmath14th column to the @xmath19th row ,
when @xmath72 , or the @xmath19th column to the @xmath73 row , when @xmath74 , the subset chosen from the @xmath19th row must correspond to a vertex @xmath75 such that @xmath76 is an edge of @xmath13 .
hence @xmath77 is a clique in @xmath13 .
the converse is obvious . | we consider the following problem : given a point set in space find a largest subset that is in convex position and whose convex hull is empty .
we show that the ( decision version of the ) problem is w[1]-hard . |
traders or asset managers willing to sell blocks of shares are increasingly using execution algorithms . amongst the strategies proposed by brokers , the most widely studied from an academic point of view is the implementation shortfall ( is ) strategy . the classical modeling framework for optimal liquidation , developed by almgren and chriss in their seminal papers @xcite , is indeed focused on is orders benchmarked on the arrival price , that is the price at the beginning of the liquidation process . in the case of is orders , the agent faces a trade - off between selling slowly to reduce execution costs and selling rapidly to limit the influence of price fluctuations .
although almost all the literature on optimal liquidation deals with is orders , is algorithms usually account for less volume than vwap ( volume weighted average price ) algorithms
see for instance @xcite .
the aim of traders when they choose vwap orders is to focus on the reduction of execution costs : the order is split into smaller ones and the associated transactions occur on a pre - determined period to obtain a price as close as possible to the average price over this period ( weighted by market volume ) .
vwap is also a neutral and rather fair benchmark to evaluate execution processes .
many agents are willing to trade as close as possible to the vwap as they are benchmarked on the vwap .
+ although vwap orders represent a large part of algorithmic trading , there are only a few papers about vwap orders in the academic literature . the first important paper regarding liquidation with vwap benchmark
was written by konishi @xcite .
he developed a simple model and looked for the best static strategy , that is the best trading curve decided upon at the beginning of the liquidation process ( his goal is to minimize the variance of the slippage with respect to the vwap ) .
he found that the optimal trading curve for vwap liquidation has the same shape as the relative market volume curve when volatility and market volume are uncorrelated .
he also quantified the deviation from the relative market volume curve in the correlated case .
the model was then extended by mcculloch and kazakov @xcite with the addition of a drift in a more constrained framework .
mcculloch and kazakov also developed a dynamic model @xcite in which they conditioned the optimal trajectory with perfect knowledge of the volume by the available information at each period of time .
bouchard and dang proposed in @xcite to use a stochastic target framework to develop vwap algorithms .
recently , carmona and li @xcite also developed a model for vwap trading in which the trader can explicitly choose between market orders and limit orders to buy / sell shares .
related to this literature on vwap trading ( see also @xcite ) , an academic literature appeared on market volume models .
the papers by bialkowski et al .
@xcite or mcculloch @xcite are instances of such papers modeling market volume dynamics and the intraday seasonality of relative volume .
however , all these papers ignore an important point : market impact .
the only paper dealing with vwap trading and involving a form of market impact is the interesting paper by frei and westray @xcite . in this paper ,
the price obtained by the trader is not the market price but a price depending linearly on the desired volume ( as in the early models of almgren and chriss ) .
however , there is no permanent market impact in their model , while it plays an important role in our paper .
+ the aim of our model is to include permanent market impact ( see @xcite for the framework we use ) and any form of execution costs in a model for vwap liquidation .
also , our goal is not to obtain a price as close as possible to the vwap but rather to understand how to provide a guaranteed vwap service while mitigating risk . in other words
, we want to find the optimal strategy if we are given a certain quantity of shares and asked to deliver the vwap over a predefined period .
in addition to the optimal strategy underlying a guaranteed vwap contract , we are interested in the price of such a contract . for that purpose ,
we use indifference pricing in a cara framework , as in @xcite where the author prices a large block of shares .
this price for a guaranteed vwap contract is the minimum premium the trader needs to pay to the broker so that the latter accepts to deliver the vwap to the former .
it depends on the size of the order , on liquidity and market conditions , and on the risk aversion of the broker .
+ in section 2 , we present the general framework of our model .
we introduce the definition of the vwap and the forms of market impact used in the model . the optimization criterion is defined and the price / premium of a guaranteed vwap contract is defined using indifference pricing . in section 3
, we characterize the optimal liquidation strategy when the market volume curve is assumed to be known ( deterministic case ) , along with the premium of the guaranteed vwap contract . in section 4
, we focus on special cases and numerics .
we show that , in the absence of permanent market impact , the optimal trading curve has the same shape as the market volume curve .
we consider also the case of linear permanent market impact and quadratic execution costs that permits to get closed form solutions and to better understand the role played by permanent market impact .
finally , an efficient numerical method is provided to approximate the solution in the general case . in the last section ,
we extend our model to the case of stochastic volumes and we characterize the price of a guaranteed vwap with a pde . +
we consider a filtered probability space @xmath0 corresponding to the available information on the market , namely the market price and market volume of a stock up to the observation time . for @xmath1 ,
we denote @xmath2 the set of @xmath3-valued progressively measurable processes on [ 0,t ] . + we consider a trader who wants to sell @xmath4 shares over the time period @xmath5 $ ] .
his liquidation strategy during this period hereafter denoted @xmath6 is modeled as a stochastic process belonging to the admissible set @xmath7 to model the trader s portfolio , we introduce a process @xmath8 that gives the number of shares remaining in the portfolio at time @xmath9 : @xmath10 in addition to our own volume , we introduce the instantaneous market volume process @xmath11 , assumed to be progressively measurable and nonnegative . the cumulated volume from 0 up to time @xmath9 is denoted @xmath12 .. ] the price process of the stock is defined as a brownian motion with a drift to account for market impact .
permanent market impact is modeled by a function @xmath13 , assumed to be nonincreasing and integrable in @xmath14 : @xmath15 the price received by the trader at time @xmath16 is not @xmath17 because of temporary market impact .
this temporary market impact , also referred to as execution costs , is modeled through the introduction of a function @xmath18 verifying : * @xmath19 , * @xmath20 is an even function , * @xmath21 is increasing on @xmath22 , * @xmath21 is strictly convex , * @xmath21 is asymptotically superlinear , that is : @xmath23 in applications , @xmath21 is often a power function , i.e. @xmath24 with @xmath25 , or a function of the form @xmath26 with @xmath27 .
+ for any @xmath28 , we define the cash process @xmath29 ( hereafter denoted @xmath30 to simplify notations ) by : @xmath31 we are interested in the following optimization problem : @xmath32,\end{aligned}\ ] ] where corresponds to risk neutrality . here
, we choose to consider an expected utility framework with a cara utility function .
this framework boils down to a mean variance setting in the case of gaussian risks . considering a utility function
is more rigorous from an economic viewpoint , especially when it comes to pricing . ]
@xmath33 is the absolute risk aversion parameter , and where @xmath34 stands for the volume weighted average price ( vwap ) on the period @xmath35 $ ] , i.e. : @xmath36 the above definition of the vwap can also be formulated as : @xmath37 where @xmath38 .
this formulation is often used in the literature but it has an important drawback as the above integral is not @xmath39-adapted .
its natural filtration is indeed @xmath40 where : @xmath41 in the above definition of @xmath42 , we did not include our own volume .
an alternative definition , including our trades , is @xmath43 we shall see in the appendix of this article that the results we obtain with our simpler definition can be easily modified to be true in the case of the alternative definition .
the rationale for our optimization criterion is the following .
we consider a stock trader or an asset manager who wants to sell @xmath44 shares at the vwap over the period @xmath5 $ ] .
for that purpose , he signs with an intermediary ( typically a broker ) a guaranteed vwap contract .
he gives the intermediary its @xmath44 shares at time 0 and he receives at time @xmath45 , @xmath44 times the vwap computed over the period @xmath5 $ ] , minus a premium
hereafter denoted @xmath46 to compensate the intermediary for the service and the associated costs , this premium being agreed upon at time @xmath14 .
the problem we address is the problem of the intermediary : he receives @xmath44 shares and sells them over the period @xmath5 $ ] to obtain @xmath47 on its cash account at time @xmath45 .
then , he gives his client @xmath48 in cash .
therefore , if the intermediary has a constant absolute risk aversion @xmath49 , his liquidation strategy is obtained by maximizing @xmath50 = \exp(-\gamma \pi(q_0 ) ) \e \left[-\exp\left(-\gamma ( x_t - q_0 \text{vwap}_t ) \right)\right].\ ] ] since @xmath46 is agreed upon at time @xmath14 , the objective function is simply : @xmath51.\ ] ] now , to decide upon the value @xmath46 , we shall compute the reservation price of the intermediary , that is the price at which the intermediary is indifferent between accepting the contract and refusing it . this approach for pricing ,
also called indifference pricing , leads to the following definition for the premium of a guaranteed vwap : .
however , this is expressed in percentage of the initial price . in practice ,
some guaranteed vwap contracts are also priced in basis points of the realized vwap .
we refer to appendix b for this different definition of the premium and its consequences on our approach .
numerically , the two approaches provide similar figures in basis point of their respective reference price . ]
@xmath52\right).\ ] ] our goal in this paper is twofold : ( i ) we want to solve the optimization problem , that is to find an optimal liquidation strategy @xmath6 , and ( ii ) we want to find the value of @xmath46 .
+ we derive here the key formulas for @xmath53 and @xmath34 : [ lem expression of xt ] for any @xmath54 , we have : @xmath55 where @xmath56 * proof : * + integrating by parts , we have : @xmath57_0^t + \int_0^t q_t ds_t - \int_0^t v_t l\left ( \frac{v_t}{v_t } \right)dt \\ & = q_0 s_0-\int_0^t q_t f\left(\left|q_0-q_t \right| \right ) v_tdt+\int_0^t q_t \sigma dw_t - \int_0^t v_t l\left ( \frac{v_t}{v_t } \right)dt .\end{aligned}\ ] ] we then observe that : @xmath58_0^t+\int_0^t f(q_0-q_t ) v_t dt = \int_0^t f(q_0-q_t ) v_t dt\ ] ] defining in the last integral the change of variables @xmath59 , we have : @xmath60 finally , we obtain : @xmath61 @xmath62 the same calculation can be made for vwap : we have : @xmath63 * proof : * + this is an integration by parts : @xmath64_0^t- \int_0^t\left ( \frac{q_t}{q_t}-1\right)\sigma dw_t + \int_0^t \left(\frac{q_t}{q_t}-1 \right)f(\left| q_0-q_t\right| ) v_t dt\\ & = s_0+\int_0^t \left ( 1-\frac{q_t}{q_t } \right)\sigma dw_t + \left[\left(\frac{q_t}{q_t}-1\right ) f(q_0-q_t ) \right]_0^t- \int_0^t \frac{v_t}{q_t } f(q_0-q_t)dt\\ & = s_0+\int_0^t \left ( 1-\frac{q_t}{q_t } \right)\sigma dw_t- \int_0^t \frac{v_t}{q_t } f(q_0-q_t)dt,\end{aligned}\ ] ] where we recall that @xmath65 .
@xmath62 the slippage @xmath66 is then given by : [ slippage value ] for @xmath54 , we have : @xmath67
in this section we assume that the volume curve is deterministic .
we also assume that @xmath68 is bounded from above and from below : @xmath69 we first study the case where we restrict the set of admissible strategies to the deterministic ones : @xmath70 the problem is then replaced by : @xmath71,\end{aligned}\ ] ] in that framework , we obtain the following lemma : [ lemme 1 section 3 ] for any @xmath72 , @xmath73 is normally distributed with mean @xmath74 and variance @xmath75 * proof : * + the proof is straightforward , considering the formula obtained in corollary [ slippage value ] .
+ using the expression of the laplace transform of a normally distributed variable , we obtain : @xmath76\\ = \sup_{v\in \mathcal{a}_{det } } & -\exp\left(-\gamma \left(-\int_0^{q_0}f(z)dz-\int_0^t v_t l\left(\frac{v_t}{v_t } \right)dt+q_0\int_0^t \frac{v_t}{q_t } f(q_0-q_t ) dt\right . \right.\\ & \left .
\left . - \frac{\gamma}{2}\sigma^2 q_0 ^ 2 \int_0^t \left(\frac{q_t}{q_0}-\left ( 1-\frac{q_t}{q_t}\right ) \right)^2dt \right ) \right).\end{aligned}\ ] ] our optimization problem consequently boils down to : @xmath77 in order to obtain existence and uniqueness , we use the classical framework developed in @xcite .
we denote @xmath78 the set of absolutely continuous arcs @xmath79
\rightarrow \r $ ] such that @xmath80 and @xmath81 and we define the application @xmath82 by : @xmath83 we shall denote @xmath84 for short . + in order to prove the existence of an optimal strategy , we first show the technical lemma : [ lemme technique existence ] @xmath85 verifies the three following assertions : * @xmath85 is convex with respect to the third variable ; * there exists @xmath86 and @xmath87 such that @xmath88 as @xmath89 and @xmath90 , q\in \r , \ell(t , q , v ) \geq
\theta(\left| v\right| ) -c_0;\ ] ] * for all @xmath91 there exists @xmath92 such that @xmath93 for all @xmath94 $ ] , @xmath95 , @xmath96 $ ] , @xmath97 , where @xmath98 is a modulus of continuity .
* proof : * + ( i ) follows from the convexity of @xmath20 and the positivity of @xmath68 .
+ for ( ii ) , we first see that for any @xmath94 $ ] and @xmath99 , we have : @xmath100 where @xmath101 } \left(\frac{q}{q_0 } -a \right)^2.\ ] ] @xmath102 is continuous with @xmath103 and @xmath104 , so that there exists @xmath105 such that @xmath106 on @xmath107 .
+ we next define @xmath108 . by the superlinearity of @xmath20
, we have @xmath88 as @xmath109 .
finally , since @xmath20 is even , we have for all @xmath110 \times \r \times \r$ ] : @xmath111 for ( iii ) , we have for all @xmath112 \times \r\times \r \times \r$ ] : @xmath113 where @xmath114 is a constant uniform in @xmath16 and @xmath115 .
since @xmath116 is nonincreasing on @xmath22 , we have for any @xmath117 : @xmath118 and then : @xmath119 combining and , we obtain : @xmath120 where @xmath121 and @xmath122
. @xmath62 [ existence ] there exists a unique minimizer @xmath123 of @xmath124 in @xmath78 .
moreover we have that @xmath125 for all @xmath126 $ ] .
* proof : * + we divide the proof in three steps .
+ _ step 1 : _ we first show that any strategy can be improved by considering a new one taking values in @xmath127 $ ] . indeed for @xmath128 , we define @xmath129 by @xmath130 .
we have @xmath131 and @xmath132 .
therefore , since @xmath21 is even and increasing on @xmath22 , we have : @xmath133 also , since @xmath134 is odd and nondecreasing , we have for any @xmath126 $ ] : @xmath135 eventually , we have that @xmath136 and then : @xmath137 with strict inequality whenever @xmath138
. + _ step 2 : _ we then show the uniqueness of the minimizer .
let us consider @xmath139 and @xmath140 two minimizers of @xmath124 such that @xmath141 . we know
, using step 1 , that @xmath139 and @xmath140 take values in @xmath127 $ ] .
we next define @xmath95 by @xmath142 . by convexity of @xmath20 , convexity of @xmath143 on @xmath127
$ ] and strict convexity of @xmath144 , we have @xmath145 , which contradicts the optimality of @xmath139 and @xmath140 .
+ _ step 3 : _ the existence of the solution follows from theorem 6.1.2 in @xcite , where we show in lemma [ lemme technique existence ] that @xmath146 verifies the three required conditions .
@xmath62 we shall characterize @xmath147 as the solution of the hamiltonian system associated to the optimization problem . + the function @xmath148 is not convex and so the classical results of @xcite and @xcite can not be applied . since
the optimal solution @xmath123 must verify @xmath149 , we modify the problem and introduce @xmath150 .
we then define the associated function @xmath151 on @xmath35 \times \r \times \r $ ] as an extended real - valued function by : @xmath152 then , using theorem [ existence ] , we clearly see that the problem is identical to the problem : @xmath153 and an optimal strategy for one of the problem , is also an optimal strategy for the other problem . + to solve this problem , we use the technics developed in @xcite and @xcite .
@xmath154 is indeed a convex function , taking values in @xmath155 .
we note that conditions ( b ) , ( c ) and ( d ) enumerated in @xcite are also satisfied , given the assumptions on @xmath156 and the assumptions on @xmath21 .
+ we now introduce the legendre transform of @xmath20 defined by @xmath157 . since @xmath20 is strictly convex , we recall that @xmath158 is a @xmath159 function .
+ using this legendre transform , we have the following characterization of @xmath147 : [ prop : deterministic hamiltonian ] for @xmath160 , we have equivalence between the two following assertions : * @xmath161 ; * there exists @xmath162 such that for all @xmath94 $ ] : @xmath163 * proof * : + the hamiltonian of the system is : @xmath164 the characterization of @xmath147 given in the theorem 6 of @xcite and its corollary is : @xmath165 where @xmath166 stands for the subdifferential . + given the expression for @xmath167 , only @xmath168 may not be a singleton made of a real number , when @xmath169 .
if @xmath170 is finite , then the expression given in the proposition is obtained by straightforward computation . if @xmath171 , then @xmath172 and the expression in the proposition is correct , giving @xmath173 whenever @xmath169 .
@xmath62 this hamiltonian characterization allows to get a regularity result for the optimal strategy if @xmath174 is continuous then @xmath175)$ ] * proof : * + @xmath21 being strictly convex , @xmath176 is @xmath177 and the result is a consequence of the equation @xmath178 .
+ even when @xmath179 is smooth or even constant , @xmath147 may not be @xmath180 .
this remark is important since most studies are using a euler - lagrange equation in the classical sense with the underlying assumption that the strategy is @xmath180 .
we end this section by the optimality of deterministic strategies .
this result was first proved in @xcite in the case of implementation shortfall ( is ) liquidation strategies .
we use the same method here in the case of vwap liquidation strategies .
[ th deterministic strategies are optimal ] assume that @xmath156 is deterministic , then : @xmath181 = \sup_{v \in \mathcal{a } } \e \left [ -\exp\left(-\gamma ( x_t - q_0 \text{vwap}_t ) \right)\right].\ ] ] * proof : * + let us consider @xmath28 .
we have : @xmath182\\ = & -\exp\left(\gamma \int_0^{q_0}f(z)dz \right)\\ & \times \e \left [ \exp \left(\gamma \int_0^t\left ( v_t l \left ( \frac{v_t}{v_t}\right)-q_0 \frac{v_t}{q_t } f(q_0-q_t ) \right)dt \right)\right . \\ & \times \left.\exp\left ( -\gamma \sigma q_0 \int_0^t \left ( \frac{q_t}{q_0}-\frac{q_t - q_t}{q_t } \right)dw_t \right ) \right].\\\end{aligned}\ ] ] we then define a new probability measure @xmath183 by : @xmath184 since @xmath185 , we observe that @xmath186 indeed defines a change of probability .
+ we then have : @xmath187 = -\exp\left(\gamma \int_0^{q_0}f(z)dz \right ) \e^{\q } \left [ \exp \left ( \gamma \mathcal{i}(q ) \right ) \right].\ ] ] now , @xmath188-a.s . , @xmath189 so that @xmath188-a.s . , @xmath190 .
this gives : @xmath191&= -\exp\left(\gamma \int_0^{q_0}f(z)dz \right ) \e^{\q } \left [ \exp \left ( \gamma \mathcal{i}(q ) \right ) \right]\\ & \le -\exp\left(\gamma \int_0^{q_0}f(z)dz \right ) \exp \left(\gamma \mathcal{i}(q^ * ) \right)\\ & = \sup_{v \in \mathcal{a}_{det } } \e \left [ -\exp\left(-\gamma ( x_t - q_0 \text{vwap}_t ) \right)\right].\end{aligned}\ ] ] hence : @xmath192 \le \sup_{v \in \mathcal{a}_{det } } \e \left [ -\exp\left(-\gamma ( x_t - q_0 \text{vwap}_t ) \right)\right],\ ] ] and the result follows since the converse inequality holds .
@xmath62 let us now come to the premium for guaranteed vwap .
the above results show that @xmath193 = -\exp\left(-\gamma \left ( - \int_0^{q_0}f(z)dz - \mathcal{i}(q^ * ) \right)\right).\ ] ] hence , the premium for guaranteed vwap is given by the following theorem : @xmath194
we now turn to specific cases in which closed forms expressions can be obtained .
an interesting case , within the deterministic framework , is the case where there is no permanent market impact ( i.e. @xmath195 ) . in that case indeed , we have that the optimal strategy is to follow the market volume curve .
this is stated in the next proposition : [ noperm ] if @xmath196 then : @xmath197 and @xmath198 * proof : * + although the proof can be made directly using jensen s inequality on @xmath199 , we prove the result using the hamiltonian system of proposition [ prop : deterministic hamiltonian ] .
considering the function @xmath200 defined by @xmath201 and a constant function @xmath202 such that @xmath203 , we have that @xmath204 so that : @xmath205 also , straightforwardly : @xmath206 this proves that @xmath207 , hence the first part of the result .
+ coming to the premium , we have : @xmath208 + the result of the above proposition deserves a few comments .
it states that , in the absence of permanent market impact , the optimal strategy is to have a trading curve that has the same shape as the relative market volume curve .
one consequence is that , in practice , as far as the trading strategy is concerned , when permanent market impact can be ( or is ) ignored , one is interested in the estimation of the relative market volume curve and not in the estimation of the absolute value of the market volume .
this remark is particularly important to understand why the deterministic market volume assumption provides a rather acceptable approximation of the real case for vwap trading .
although the total volume traded over a day is highly variable , we know that the relative market volume curve is stable from one day to the other ( except on witching days ) .
since the cumulated volume @xmath209 only appears in the execution strategy through the ratio @xmath210 , considering it stochastic does not play any role if we have already assumed that the relative market volume curve is deterministic .
the value of @xmath209 is however important to determine the premium of the guaranteed vwap contract . since this premium
must be decided upon at time @xmath211 , the value of @xmath209 must be understood as a forecast at time @xmath14 of the total market volume over the period @xmath5 $ ] . + the above remarks only apply in the absence of permanent market impact but the assumption of a deterministic market volume is acceptable if we deviate slightly through the introduction of a small permanent market impact .
now , our goal is to understand the influence of permanent market impact in this framework , and the nature of the related deviation of the optimal strategy from the relative market volume curve .
+ we now explore , for flat volume curves , the particular case where execution costs are quadratic , i.e. @xmath215 , and where permanent market impact is linear , i.e. @xmath216 , as in the initial almgren - chriss framework .
[ ac ] assume @xmath217 , @xmath218 and @xmath219 .
we have : @xmath220 and @xmath221 where @xmath222.\ ] ] * proof : * + we first see that @xmath223 .
then , using proposition [ prop : deterministic hamiltonian ] , we obtain that : @xmath224 this leads to : @xmath225 therefore , the optimal strategy if of the form : @xmath226 the initial and terminal conditions imply that : @xmath227 we obtain then : @xmath228 @xmath229 since @xmath230 , we obtain : @xmath231.\ ] ] coming to the premium , we have : @xmath232 @xmath62 this result permits to understand the role played by permanent market impact .
we indeed have that @xmath233 and therefore that the liquidation must occur more rapidly with the addition of permanent market impact .
the rationale underlying this point is that the intermediary is going to pay @xmath234 to the client and therefore he has an incentive to sell rapidly so that the price moves down , resulting in a lower vwap .
if @xmath235 is large , the optimal strategy may even be to oversell before buying back the shares so as to reduce the value of the vwap ( see below ) .
+ coming to the premium for a guaranteed vwap , it is straightforward to see that if @xmath236 the premium would be equal to @xmath237 .
therefore , the reduction in the premium due to the use of the optimal strategy @xmath147 is given by @xmath238 + in particular , in the limiting case @xmath239 , corresponding to a risk neutral agent , we obtain the following straightforward formulas for the optimal strategy and the premium : @xmath240 @xmath241 several examples of optimal vwap liquidation are given on figure 1 . we see that taking permanent market impact into account is important since the optimal trading curve may be really different from the simple trading curve obtained in proposition [ noperm ] .
+ , @xmath242 , @xmath243 , @xmath244 , @xmath245 , @xmath246 , @xmath247 day .
plain line : @xmath248 .
dot - dashed line : @xmath249 .
the dashed line corresponds to @xmath250.,scaledwidth=95.0% ] it is important to understand what is at play here .
an agent willing to sell shares at a price close to the vwap over a given period has usually two possibilities .
he may call his favorite broker and ask for an agency vwap order . in that case , the broker will try to sell shares as close as possible to the vwap and the price obtained will be the price for the agent .
in other words , the risk is borne by the agent .
the other possibility is to enter a guaranteed vwap contract . in that case
, the price obtained by the agent will always be the vwap .
our point is that the vwap obtained by the agent in a guaranteed vwap contract is not the same as the vwap obtained on average through agency trades . in a guaranteed vwap contract ,
the counterpart has indeed an incentive to sell more rapidly in order to push down the price and hence push down the vwap .
this is not market price manipulation , as the overall impact of the execution process would be the same independently of the trajectory .
this is however a form of vwap manipulation .
is the agent harmed ? somehow yes , although he gets its benchmark price . nonetheless ,
since the counterpart of the contract makes money by selling more rapidly at the beginning of the execution process , he can redistribute part of it through a reduction of the premium ... + to measure the difference between the naive strategy @xmath251 and the optimal strategy , the best indicator is the premium of a guaranteed vwap contract @xmath46 .
we considered the same cases as on figure 1 , that is @xmath252 , @xmath242 , @xmath243 , @xmath244 , @xmath245 , @xmath246 , @xmath247 day , and two scenarios for the risk aversion parameter @xmath49 .
the results on figure 1 state that if the naive strategy was used , the minimum price of a guaranteed vwap contract would be @xmath253 bps .
however , when optimal strategies are used in the above examples , the intermediary would accept the contract without the payment of a premium , as the theoretical value of the premia are in fact negative .
.premium of a guaranteed vwap contract in the case of a naive strategy and in the case of the optimal strategy .
[ cols="^,^,^",options="header " , ] we treated above special cases for which closed form formulas could be obtained . in general , this is not the case and we present here a general method to approximate the solution of the hamiltonian system .
it is important to notice that the use of the hamiltonian system is preferable to the use of euler - lagrange equation when it comes to numerics since the problem remains of order 1 .
the method we use to approximate the solution @xmath254 of the hamiltonian system on the grid @xmath255 is to apply a newton method on the following nonlinear system of equations : @xmath256 @xmath257 to be more precise , we consider a first couple @xmath258 where : @xmath259 @xmath260 @xmath261 typically , we consider @xmath262 given by @xmath263 .
+ then , to go from @xmath264 to @xmath265 we consider the following method : @xmath266 where @xmath267 solves the linear system : @xmath268 @xmath269 two examples of the use of this method are shown on figure 2 and figure 3 .
the first one corresponds to @xmath270 with @xmath271 and linear permanent market impact .
the second one corresponds to @xmath270 with @xmath271 and nonlinear permanent market impact of the form @xmath272 with @xmath273 . ,
@xmath242 , @xmath243 , @xmath244 , @xmath270 with @xmath274 and @xmath275 , @xmath276 , @xmath247 day .
plain line : @xmath248 .
the dashed line corresponds to @xmath250.,scaledwidth=75.0% ] , @xmath242 , @xmath243 , @xmath244 , @xmath270 with @xmath274 and @xmath275 , @xmath272 with @xmath277 and @xmath278 , @xmath247 day .
plain line : @xmath248 .
the dashed line corresponds to @xmath250.,scaledwidth=75.0% ]
in the previous sections of this article , we focused on the case of a deterministic volume curve .
the reason for this is twofold .
firstly , it permits to understand the role played by permanent market impact in a tractable case .
secondly , it corresponds to the way vwap strategies are often built in practice and we explained above why it was a good approximation . practitioners usually compute relative market volume curves based on historical data and try to follow this curve to get a price as close as possible to the vwap .
+ we now briefly explore the case of stochastic market volume .
the aim of this section is to provide a hamilton - jacobi - bellman pde to characterize the optimal liquidation strategy ( that is no longer deterministic , and hence no longer a trading curve decided upon in advance , at time @xmath14 ) and the premium of a guaranteed vwap contract .
a similar approach , using stochastic optimal control , was adopted by frei and westray @xcite in the mean - variance setup . however , in their paper , the initial filtration is augmented with the knowledge of the final volume and this makes their approach questionable for practical use .
+ in the model we consider , the instantaneous market volume is modeled by a simple stochastic process but it can be generalized to other processes .
in fact , our main goal is to write the hamilton - jacobi - bellman pde characterizing the solution of our problem using as few variables as possible .
if one considers indeed our problem with stochastic market volume in its initial form , 7 variables are necessary to describe the problem : the time @xmath9 , the trader s inventory @xmath200 , the asset market price @xmath279 , the cash account @xmath30 , the instantaneous market volume @xmath280 , the cumulated market volume @xmath281 , and a variable linked to the vwap , namely @xmath282 . using two changes of variables ,
we manage to restrict the number of variables to 5 .
classical numerical ( pde ) methods may fail to approximate the solution of the pde , but some probabilistic methods may be efficient ( see @xcite for robust methods in the case of problems in high dimension ) .
+ coming to the model , we consider that the instantaneous volume is given by @xmath283 , where @xmath102 is @xmath284 , where @xmath285 is a brownian motion independent of @xmath286 , and where @xmath287 .
the dynamics of @xmath68 is then given by the stochastic differential equation : @xmath288 @xmath289 represents obviously the instantaneous market volume , on average , at time @xmath9 . in europe , it is a w - shaped curve with a peak corresponding to the opening of the us market .
+ other dynamics can be considered .
the goal of this last section is not to consider the best possible model for volumes but rather to show how the complexity of the model can be reduced through changes of variables .
+ in order to consider a non - degenerated problem ( one may alternatively use the stochastic target framework ) , we consider a slightly modified problem where @xmath290 is not forced to be equal to @xmath14 . rather than imposing @xmath291
, we consider that , at time @xmath45 , the remaining stocks are not liquidated at price @xmath292 but rather at price @xmath293 , where @xmath294 is chosen positive and high enough to discourage the trader to keep a large position @xmath290 at time @xmath295 .
+ in this slightly modified framework , with computations similar to those of section [ subsect : some easy calculations ] , we obtain the following : @xmath296dw_t\\ & + \int_0^{q_t}f(z)dz - q_t f(q_0-q_t)-k q^2_t.\end{aligned}\ ] ] mathematically , it corresponds to a penalization at time @xmath295 of the form @xmath297 , and we suppose that @xmath298 . + now , we define the set of admissible strategies for all @xmath94 $ ] by : @xmath299 our problem is then the following : @xmath300,\end{aligned}\ ] ] where : @xmath301 to solve this stochastic optimal control problem , we now introduce two processes @xmath302 and @xmath303 : @xmath304 , \
x^{t , x , v}_t = x .\ ] ] @xmath305 , \
\tilde{y}^{t , y , v}_t = y.\ ] ] for any @xmath306 $ ] , we define : @xmath307.\end{aligned}\ ] ] this is the dynamic problem associated to and we observe that corresponds to @xmath308 . we then have the following : @xmath309 is a viscosity solution of : @xmath310 where the operator @xmath311 is defined by : @xmath312 * proof : * + it is straightforward to see that the value function is locally bounded .
the result is then classically obtained by stochastic control technics .
the required dynamic programming principle can indeed be deduced from the apparatus developed in @xcite , and the viscosity subsolution and supersolution are obtained using @xcite .
@xmath62 it is noteworthy that we managed to remove the price @xmath279 from the state variables .
we now remove @xmath313 , using a change of variables . for that purpose ,
we introduce @xmath314 .
then we have : @xmath315 is a viscosity solution of : @xmath316 \times \r^3\times \r_+ \times \r_+^{*}. \\ \tilde{u}(t , y , q , q , v ) = -\exp\left(-\gamma \left ( \frac{y}{q}-h(q ) \right ) \right ) .
\end{array}\right.\ ] ] with @xmath317 defined by : @xmath318 * proof : * + the change of variables @xmath314 being monotonically increasing , the result is obtained by straightforward computation .
@xmath62 now , we are going to prove a technical lemma in order to show that @xmath319 ( or equivalently @xmath315 ) is always negative .
for any @xmath320 \times \r^3\times \r_+ \times \r_+^ { * } $ ] , we have : @xmath321 * proof : * + let us consider @xmath322 \times \r^3\times \r_+ \times \r_+^{*}$ ] and @xmath323 .
we have by jensen s inequality : @xmath324 \le -\exp\left(-\gamma \mathcal{e}^v \right),\ ] ] where @xmath325\\ & = \e \left[x + \frac{y}{q_t } + \sigma q_0 \int_t^t \left ( \frac{q_s}{q_0}-\frac{q_t - q_s}{q_t } \right)dw_s\right.\\ & + \left.\int_t^t \left(-v_s l \left ( \frac{v_s}{v_s } \right ) + \frac{v_s}{q_t } q_0
f(q_0-q_s ) \right ) ds - h(q_t)\right].\\\end{aligned}\ ] ] since @xmath326 , we have @xmath327=0 $ ] . by independence of @xmath280 and @xmath286 , we also have that @xmath328=0 $ ] . therefore : @xmath329 + \e \left[\int_t^t \left(-v_s l \left ( \frac{v_s}{v_s } \right ) + \frac{v_s}{q_t } q_0 f(q_0-q_s ) \right ) ds - h(q_t)\right].\ ] ] since @xmath330 is nonincreasing on @xmath22 , there exists a constant @xmath331 such that : @xmath332 now , since @xmath20 is superlinear , there exists @xmath333 such that : @xmath334 this gives : @xmath335 + \e \left[\int_t^t \left(b - cq_0 |v_s| + \frac{v_s}{q_t } cq_0 \left(1 + ( q_0-q_s)_+\right ) \right ) ds - h(q_t)\right]\\ & \le x + \e \left[\frac{y}{q_t}\right ] + bt + cq_0 + \e \left[\int_t^t \left(- cq_0 |v_s| + \frac{v_s}{q_t } cq_0 ( q_0-q_s)_+ \right ) ds - h(q_t)\right]\\ & \le x + \e \left[\frac{y}{q_t}\right ] + bt + cq_0 + \e \left[cq_0 \left((q_0-q_t)_+ - \frac{q}{q_t } ( q_0-q)_+ \right ) - h(q_t)\right]\\ & + \e \left[\int_t^t \left(- cq_0 |v_s| - c q_0 \frac{q_s}{q_t } v_s \right ) ds\right]\\ & \le x + bt + cq_0 + \e \left[\frac{y}{q_t}\right ] + \e \left [ cq_0 ( q_0-q_t)_+ - h(q_t ) \right].\\\end{aligned}\ ] ] since @xmath336 , we get that @xmath337 < + \infty$ ] .
the assumption on the market volume process @xmath11 then gives that @xmath338 $ ] exists ( and is independent of @xmath6 ) .
putting these inequalities altogether , we get : @xmath339 therefore , @xmath340 @xmath62 now , since @xmath315 is never equal to @xmath14 , we can consider the change of variables @xmath341 .
this new change of variables does not remove another variable but it has two related advantages .
firstly , @xmath342 is in the same unit as the cash account .
hence , it takes values in a range that can be evaluated in advance .
this is particularly important when it comes to numerics .
secondly , the premium @xmath46 for a guaranteed vwap contract is straightforwardly : @xmath343 easy computations lead to the fact that @xmath342 is a viscosity solution of : @xmath344 where @xmath176 is the legendre transform of @xmath21 , and where the nonlinear operator @xmath345 is defined by : @xmath346\\ & + v \partial_q + v \frac{g'(t)}{g(t ) } \partial_v + \frac{\alpha^2 v^2}{2 } \left [ \gamma \left ( \partial_v \right)^2 + \partial_{vv}\right].\end{aligned}\ ] ]
in this article we built a model to find the optimal strategy to liquidate a portfolio in the case of a guaranteed vwap contract .
when there is permanent market impact , we showed that the best strategy is not to replicate the vwap but rather to sell more rapidly to push down the vwap .
also , we use the indifference pricing approach to give a price to guaranteed vwap contracts and we showed that taking into account permanent market impact permits , at least theoretically , to reduce substantially the price of guaranteed vwap contracts . finally , in the case of stochastic volumes , we developed a new model with only 5 variables and not 7 variables as in a naive approach . +
in section 2 , we briefly discussed two alternative definitions of the vwap over @xmath5 $ ] : @xmath349 and @xmath43 in the former case , we exclude our own volume , while in the latter case , closer to the market definition , we include it .
in fact , in both cases , our volume has an influence since the dynamics of the stock price depends on @xmath6 . in this appendix
, we are going to prove that using one or the other definition does not make any difference , up to a change in the function @xmath330 and in @xmath350 .
this is in fact the consequence of the following proposition : for any @xmath54 , @xmath351 * proof : * + let us integrate by parts in the definition of @xmath348 : @xmath352 since we still have that @xmath353 we obtain : @xmath351 @xmath62 the above proposition states that @xmath354 is in fact equal to @xmath66 , had we replaced @xmath330 by @xmath355 and @xmath350 by @xmath356 .
hence , if our volume represents a few percent of the market volume , using one definition of vwap or the other does not really make a difference .
also , when the market volume process is assumed to be deterministic , the simple model we used over the course of this article leads to results that can be used when our volume is taken into account , if we apply the right multiplicative factors .
+ in the case of stochastic volume , instead of changing the values of the volatility and the market impact , we adapt the dynamic problem with the new adapted slippage , and we obtain the same pde with a different terminal condition , which does not induce higher complexity .
we now explore , for a deterministic volume curve , the case where the guaranteed vwap contract is priced in basis point of the vwap . in that case
, the agent has to deliver @xmath357 to his client , for some @xmath358 decided upon at time @xmath211 .
with the notations of section [ sect : general framework ] , we are now facing the maximization problem : @xmath359.\end{aligned}\ ] ] using the same methodology as in sections [ sect : general framework ] and [ sect : deterministic case ] , we obtain easily that , in this framework , the slippage is gaussian .
the following lemma is indeed the equivalent of lemma [ lemme 1 section 3 ] : for @xmath365 , our aim is then to compute : @xmath366 as in section [ sect : deterministic case ] , the maximization problem linked to @xmath367 clearly boils down to the minimization problem : @xmath368 dt.\end{aligned}\ ] ] [ existence lambda ] there exists a unique minimizer @xmath372 of @xmath373 in @xmath78 .
moreover we have that @xmath374 for all @xmath126 $ ] , and we have equivalence between the two following assertions : this result is straightforward with the use of theorem [ existence lambda ] . to compute @xmath360 numerically , we need to compute the values of the function @xmath381
this can be done through a numerical approximation of @xmath382 using the same numerical methods as in section 4 .
r. almgren and n. chriss .
value under liquidation .
risk , 12(12):61 - 63 , 1999 .
r. almgren and n. chriss . optimal execution of portfolio transactions .
journal of risk , 3:5 - 40 , 2001 .
asx , algorithmic trading and market access arrangements , february 2010 .
j. bialkowski , s. darolles and g. le fol , improving vwap strategies : a dynamic volume approach , journal of banking and finance , elsevier , vol .
32(9 ) , pages 1709 - 1722 , september 2008 .
j. bialkowski , s. darolles and g. le fol , how to reduce the risk of executing vwap orders ? - new approach to modeling intraday volume , 2006 preprint b. bouchard , and n.m .
dang , generalized stochastic target problems for pricing and partial hedging under loss constraints - application in optimal book liquidation , finance and stochastics , 2013 , 17(1 ) , 31 - 72 .
b. bouchard and n. touzi , weak dynamic programming principle for viscosity solutions .
_ siam journal on control and optimization _
, 49 , 3 , 948 - 962 , 2011 .
p. cannarsa and c. sinestrari .
semiconcave functions , hamilton - jacobi equations , and optimal control , volume 58 .
birkhauser boston , 2004 .
r. carmona .
indifference pricing : theory and applications .
princeton university press , 2009 .
r. carmona and h. li , dynamic programming and trade execution , ( li . phd thesis ) 2013 .
fleming , and h.m .
soner , controlled markov processes and viscosity solutions , 2nd edition , springer - verlag , 2005 . c. frei , and n. westray , optimal execution of a vwap order : a stochastic control approach , 2013 to appear in mathematical finance .
e. gobet , j .-
lemor , and x. warin , a regression - based monte carlo method to solve backward stochastic differential equations .
15 2172 - 2202 o. gueant , execution and block trade pricing with optimal constant rate of participation , 2012 preprint o. gueant . optimal execution and block trade pricing , a general framework , 2012 preprint .
o. gueant .
permanent market impact can be nonlinear , 2013 preprint .
m. humphery - jenner , optimal vwap trading under noisy conditions , journal of banking and finance , vol .
35 , no . 9 , 2011 .
kakade , m. kearns , y. mansour and l. e. ortiz , competitive algorithms for vwap and limit order trading , proceedings of the acm electronic commerce conference , 2004 .
h. konishi , optimal slice of a vwap trade , journal of financial markets , volume 5 , issue 2 , april 2002 , pages 197 - 221 c .- a .
lehalle , s. laruelle , market microstructure in practice , world scientific , 2014 j. mcculloch , relative volume as a doubly stochastic binomial point process , no 146 , research paper series , 2005 , quantitative finance research centre , university of technology , sydney .
j. mcculloch and v. kazakov , optimal vwap trading strategy and relative volume , no 201 , research paper series , 2007 , quantitative finance research centre , university of technology , sydney .
j. mcculloch and v. kazakov , mean variance optimal vwap trading , 2012 preprint .
d. possamai , g. royer and n. touzi , on the robust superhedging of measurable claims , 2013 preprint .
rockafellar , conjugate convex functions in optimal control and the calculus of variations , j. math .
analysis appl .
32 ( 1970 ) , 174 - 222 .
generalized hamiltonian equations for convex problems of lagrange , pacific j. math .
33 ( 1970 ) , 411 - 427 a. schied , t. schoneborn , and m. tehranchi .
optimal basket liquidation for cara investors is deterministic .
applied mathematical finance , 17(6):471 - 489 , 2010 .
m. schweizer , approximation pricing and the variance - optimal martingale measure , annals of probability 24 , 206 - 236 , 1995 .
a. wranik , a trading system for flexible vwap executions as a design artefact , 2009 , 13th pacific asia conference on information systems ( pacis ) , hyderabad , india . | optimal liquidation using vwap strategies has been considered in the literature , though never in the presence of permanent market impact and only rarely with execution costs . moreover ,
only vwap strategies have been studied and the pricing of guaranteed vwap contracts has never been addressed . in this article , we develop a model to price guaranteed vwap contracts in a general framework for market impact and we highlight the differences between an agency vwap and a guaranteed vwap contract .
numerical methods and applications are also provided .
* key words : * optimal liquidation , vwap strategy , guaranteed vwap contract , optimal control , indifference pricing |
aj acknowledges lukas novotny and mark kasperczyk for helpful discussions .
this work had financial support from cnpq ( grants 460045/2014 - 8 , 407167/2013 - 7 , 303471/2012 - 3 and 307481/2013 - 1 ) .
.[tab : parameters ] parameters used to fit the data from the three materials depicted in fig.[expdiagraph ] , using eq.[eq:10 ] .
@xmath85 is given in mw@xmath64 ( @xmath61 , @xmath86 , @xmath87 and @xmath33 are dimensionless , while @xmath25 is given in mw ) .
temperature @xmath88 is given in k and laser power ( @xmath38 ) in mw .
the constant multiplying @xmath38 is given in k / mw . [ cols="<,^,^,^ " , ] , where @xmath89 stands for steady state , and ( b ) gives the _ sas _ field correlation @xmath36 , according with eqs.[eq:04 ] .
the excitation laser power @xmath38 dependences are plot for different values of the thermal phonon population @xmath37 ( see legend in ( b ) ) .
we set @xmath50 , @xmath90 and @xmath55 for @xmath53 .
@xmath91 in power units .
the stationary phonon population gives nonphysical description for @xmath92 ( more details in the supplemental information ) . ]
intensity ratio in three different solid state systems : three - dimensional diamond ( black data ) , two - dimensional bi - layer graphene ( ab - blg , open red data ) , and two - dimensional twisted bi - layer graphene ( tblg , solid red data ) .
data points are experimental results from refs.@xcite ( diamond ) and @xcite ( graphene - systems ) , and lines are fit to the data using eq.[eq:10 ] and the parameters listed in table[tab : parameters ] . ]
here we provide more details about the theoretical model and include some results not shown in the main text .
we also present a discussion about the analysis of the experimental data from the literature .
to analyze the stokes and anti - stokes populations and their respective correlation functions at zero time delay @xmath34 , we computed the average values @xmath93 which can be obtained by means of the equation @xmath94\rangle+\text{tr}(\hat o\mathcal{l}{\hat\rho}),\ ] ] yielding the following linear system of differential equations : @xmath95 where @xmath96 . initially we will be setting @xmath50 and @xmath51
. the dynamics will depend on how the incident field is implemented . in case of a pulsed laser ,
it is enough to redefine the coupling strength as a time - dependent function @xmath97 .
@xmath98 can , for instance , be modelled as a temporal gaussian profile , such as @xmath99 , where the coherence time is proportional to @xmath100 . in that case
there is nt a steady state and , in order to analyze the behavior of the field population as a function of the laser power @xmath38 , we might consider measures at the excitation time , that is , the time at which the population gets its first maximum , or the time - averaged values of the observables defined in eq .
( [ eq:001 ] ) as @xmath101 .
we first focus on the case of a continuum pumping , i.e. , @xmath102 , on the material . in this case
, the system of differential equations ( [ eq:003 ] ) has a steady state solution due to the environment - induced relaxation . to find them , we write the system in ( [ eq:003 ] ) as a vectorial diffential equation @xmath103 , where @xmath104 is a time - independent square matrix and @xmath105 and set @xmath106 , where @xmath89 stands for steady state . to solve this ,
we invert the matrix @xmath104 , which is only possible if @xmath104 is not singular , i.e. if its determinant @xmath107 is different from zero .
the solutions are given by @xmath108 .
by simple inspection , it is straightforwardly noticed that the stokes , anti - stokes and phonon populations satisfy an equilibrium condition @xmath109 this equation shows that if stokes and anti - stokes are produced at the same rate , the only phonons left are the thermal ones . on the other hand ,
since in general stokes production is favoured over anti - stokes ( as it becomes clear in a few lines ) , whenever these scattering processes take place , the distribution of phonons in the sample is not determined solely by temperature anymore .
one of the central results in the paper is to show the behaviour of the stokes - anti - stokes field intensities in terms of the power of the pumping laser . in the steady state regime
, we can obtain an analytical expression for those intensities , @xmath110 , which are proportional to the average population of the photon modes @xmath111 . for @xmath44- and @xmath45-modes
we find @xmath112 @xmath113 where @xmath114 and @xmath115 .
note that @xmath116 .
the above solutions are well behaved , i.e. , @xmath117 , and are physically meaningful if @xmath118 ; these conditions are fulfilled if @xmath119 .
therefore our description is valid below the laser power upper bound @xmath120 note that for low laser power both @xmath121 increase linearly with @xmath38 . in order to easily identify different regimes for @xmath38 ,
the intensities ratio @xmath7 is a good figure of merit .
if @xmath122 , a condition fulfilled in the analysed experiments , and @xmath123 , the ratio @xmath124 reduces to @xmath125 where @xmath126 .
therefore , the intensities ratio takes the form @xmath127\end{aligned}\ ] ] the figure shows the experimental measurements for the three samples investigated in the paper ( symbols ) .
the lines are the equation ( 7 ) in paper plotted using the fitting parameter of the table .
dashed black line corresponds to eq .
( 7 ) , red line to eq.([eq:009a ] ) only considering the quadratic part .
light - blue line correspond to the full equation eq.([eq:009a ] ) . ] with @xmath128 .
we chose to not use this notation in the paper to keep the equations related to the fundamental constants . here
@xmath58 , with @xmath129 defined as the proportionality constant such that @xmath130 . in figure [ fig:01s ]
we show that the quadratic term in eq.([eq:009a ] ) does not introduce strong corrections .
in fact , all the experiments are done in the limit @xmath131 so it is reasonable that the linear term in this parameter dominates .
the blue - dashed lines corresponds to the full equation ( [ eq:009a ] ) and the black - dashed line only to the linear part of it , both using the fitting parameters of the table presented at the end of the paper .
it is clearly seen that these both lines basically coincide , further justifying the approximation taken in eq .
( 7 ) of the manuscript .
the phonon population can also be analytically obtained from the stationary analysis and it is given by @xmath132 which is only valid if @xmath119 . in the case
@xmath133 , the phonon population reduces to @xmath134 note that if @xmath135 , which activates the _ s - as _ modes , the phonon population is always larger than @xmath37 , therefore the difference @xmath136 .
there are additional effects that allow us to sophisticate our effective description .
for example , we can also consider effects of phase noise caused by elastic scattering of generated phonons with different momenta .
this is done by adding an extra lindbladian term @xmath137 it is easily proven , using eq .
( [ eq:002 ] ) , that phase noise does not modify the equations for the populations in eqs .
[ eq:006a ] and [ eq:007b ] , therefore , the equilibrium condition is preserved .
its main changes are for equations of the expected values of the type @xmath138 and @xmath139 , introducing expectation values of second - order operators , as for instance @xmath140 .
the new system of equations is not trivial and analytical solutions are no longer easy to obtain . in that case , and for the computation of correlation functions @xmath141
( @xmath142 ) , we compute the stationary solution for the density operator @xmath143 , for which @xmath144 .
we must thus solve @xmath145 , which requires an expansion in the fock basis @xmath146 .
this basis increases exponentially and must be truncated for practical implementations .
this is justifiable by the fact that since @xmath147 is always much smaller than one , the steady state photonic population is never large . after numerically obtaining @xmath143
we can compute expected values as @xmath148 , and we are able to analyze changes on the population of the photonic modes and s - as correlation functions in terms of the system parameters .
figure [ fig:02s ] presents how populations depend on the laser power @xmath38 for @xmath149 and different thermal phonon numbers @xmath37 . for @xmath150 and low laser power ,
both photonic modes increase linearly with @xmath38 , with a slope proportional to @xmath151 for @xmath44-modes and @xmath37 for @xmath45-modes . in the limit of extremely low temperature ( @xmath152 ) the as - modes always increase with @xmath40 .
but for a non - zero temperature there are two regimes for the growth of the @xmath45 population , i.e. , @xmath153 for low laser power and @xmath154 for high laser power .
note that for @xmath135 there exists activation of @xmath44- and @xmath45-modes due to the inelastic scattering , however random generation of photonic modes is also expected due to the presence of the thermal bath of phonons .
the interplay between these two processes will affect the production of photon pair @xmath44-@xmath45 , i.e. the @xmath155 event .
panel ( a ) presents stokes @xmath156 , anti - stokes @xmath157 and phonon @xmath158 populations as a function of the laser power @xmath38 in units of @xmath91 .
the different curves correspond to differents temperatures set by the thermal average number @xmath37 ( see legend in panel ( b ) ) . in panel
( b ) we show the ratio between _ as _ and _ s _ mode population as a function of @xmath38 and @xmath37 . for this figure
we set @xmath149 , @xmath159 and a continuum pumping . ]
( a ) stokes ( upper lines ) and anti - stokes ( lower lines ) modes populations as a function of the normalized laser power ( @xmath160 ) for different values of @xmath161 ( different line colors , see legend ) .
( b ) the effect of @xmath161 on @xmath36 for different values of @xmath37 ( see legends ) .
other parameters are same as those in figure [ fig:02s ] . ]
the two regimes are also reflected in the results for cross - correlation function @xmath36 , as shown in fig .
[ fig:03s ] .
note that the presence of phase noise ( @xmath162 ) does not affect relevantly the phenomenology presented in this work ( see fig .
[ fig:03s ] ) . nevertheless , for very large values of @xmath38 the convergence of our numerical fails due to the truncation of the hilbert space , and the results are not conclusive due to the lack of sufficient resources for tackling that regime .
so far our analysis has been done for continuous pumping , which allows for well behaved steady state solutions .
when considering a time - dependent but short - time pumping , the system response is similar in shape to the implemented pulse and an steady state is not expected . in that case , as mentioned before , we computed the long - time averages of the observables of interest .
the results for the photonic populations are shown in fig .
[ fig:04s ] in comparison with the steady state solution for continuum pumping .
stokes ( lines ) and anti - stokes ( lines with squares ) modes populations as a function of the normalized laser power ( @xmath160 ) for different types of incident laser ( see legend ) .
the inset shows the time profile of the implemented lasers .
the parameters are same as those in figure [ fig:02s ] , with @xmath163 . ]
it is clearly seen that the phenomenology studied throughout this paper is preserved , that is , the different regimes of the populations with the increasing laser power @xmath38 is preserved , and only changes in the intensity of the @xmath44- and @xmath45-modes are observed . the results in fig .
[ fig:04s ] allow us to generalize the findings presented along this work .
we finally want to remark that our theory is based on phenomenological constants that will depend on the material under study .
the validity of the approximations made to obtain analytical expressions for the ratio @xmath62 and the phonon population @xmath164 is supported by typical and relevant parameters as @xmath37 , @xmath87 and @xmath33 . in typical experiments , like the one reported in ref .
@xcite , @xmath165 at @xmath166k , @xmath167s ( phonon lifetime ) and @xmath168s ( laser pulse width ) .
the @xmath169 is at least one order of magnitude larger than one , and four orders of magnitud larger that @xmath37 .
the incident field has a power range from 1mw to 1w and wavelength @xmath170 nm .
the number of incident photons is of the order of @xmath171 per pulse .
a fair question one can ask is wether the experimental results for @xmath7 from diamond @xcite and graphene @xcite can not be satisfactorily fit considering just the bose - einstein distribution function . in other words , does the data on the power dependence of @xmath7 from diamond and graphene provide convincing case that anything beyond laser heating has been observed ?
the presence of the _ sas _ correlation in diamond is unquestionable , given the results of the correlation function @xmath36 @xcite .
as discussed in our paper , our theory provides perfect description for the excitation power dependence of @xmath43 , @xmath42 , @xmath7 and @xmath36 .
the ab - blg is fit in our work considering that the @xmath7 is ruled by thermal effects and , although relatively good fitting can be obtained with a small contribution from the _ sas _ phenomenon , we raise no controversy in this case .
we now discuss results for the tblg sample @xcite , which according to our analysis , is dominated by the _ sas _ phenomenon .
figure[blgallt ] is a plot of what would be the sample temperature @xmath88 , if extracted directly from the @xmath7 experimental data and using the bose - einstein phonon distribution function , which from eq.(1 ) in the paper is given by : @xmath172 } , \ ] ] where @xmath173 is the phonon energy , @xmath6 is the boltzmann constant , and @xmath1 is the proportionality constant that accounts for all the optical properties of the setup and sample .
@xmath1 is chosen such that @xmath174k as @xmath175 .
for tblg we obtain @xmath176 and for the ab - blg we obtain @xmath177 . the fact that @xmath178 is expected , since tblg was engineered for specific resonance behavior
. however , the fact that @xmath88 is generally larger for tblg , as compared to ab - blg ( see data in fig.[blgallt ] ) is not expected .
first , the tblg was engineered for resonance with the anti - stokes photon emission .
therefore it is expected that this sample would have a stronger cooling channel , and it would actually exhibit lower temperatures than the ab - blg .
second , this expectation is proven to be correct by the lower shift in phonon frequency observed for tblg , as shown in ref.@xcite .
therefore , the tblg * can not * be hotter than the ab - blg , and the result obtained with eq.[eq:15 ] is proven to be inconsistent . three different fittings for the @xmath7 data from tblg @xcite .
the red points are the @xmath7 data .
the red line is the same fitting as in the paper .
the black line is a fit to the data using eq.(7 ) from the paper , with @xmath179 , @xmath180mw@xmath64 and @xmath166k .
the blue line is a fit to the data using eq.(7 ) from the paper , with @xmath179 , @xmath181 and @xmath182k . ]
one can also argue on how sensitive the data fitting is with respect to the parameters choice .
parameter values outside the errors given in table i of the paper do not fit the data properly .
figure[tblg - allfits ] shows three possible fittings of the tblg data .
the red line is the best fit , as given in the paper .
the black and blue lines were obtained by considering that the @xmath7 is ruled by pure _ sas _ phenomena and by pure thermal phenomena , respectively .
notice that these two fitting options do not fit the data properly .
the worse fitting is obtained using the pure thermal phenomena , demonstrating the predominance of the _ sas _ phenomena .
figure[tvssas ] presents the separate contribution from the thermal phonons ( dashed blue lines ) and from the _ sas _ phenomena ( dotted black lines ) to the @xmath7 data fitting ( solid black lines ) for diamond ( a ) , tblg ( b ) and ab - blg ( c ) . as discussed in the paper , for diamond and tblg ,
the _ sas _
phenomena dominates until large laser powers , when the thermal phonons overcome . for ab - blg , the _
phenomena is minor .
it remains to be discussed the effect of @xmath183 , which is important when resonance effects are in place , i.e. when either stokes or anti - stokes has a stronger coupling to the pumping field . considering @xmath184 and setting @xmath185 and @xmath186 , once again as long as @xmath187 , eq.(7 ) in the paper changes to : @xmath188 for @xmath189 , this expression is well approximated by @xmath190 which approximately reduces to : @xmath191 in non - resonant scattering , @xmath199 is expected due to time - reversal symmetry .
this is the case for diamond @xcite . if resonance with electronic states exists , but there is not much difference between the resonance with the _ s _ and _ as _ photon emission , then @xmath200
this is the case of ab - blg , where the density of states does not vary too much between the _ s _ and _ as _ photon energies @xcite .
however , in the case of tblg , the electronic structure is engineered to exhibit specific resonance with the _ as _ photon emission @xcite , i.e. @xmath201 .
therefore , strictly speaking , eq.[eq:13b ] from this supplemental information should be used to fit the data , rather than eq.7 from the paper .
the value of @xmath202 is easily obtained from the data if we consider that this parameter rules the enhancement of @xmath61 from tblg as compared to ab - blg . from table
i in the paper and eq.[eq:13a ] from this supplemental information we obtain @xmath203 .
we now add @xmath204 in eq.[eq:13a ] from this supplemental information and fit the tblg data with @xmath205 , which is consistently the value for ab - blg .
this means that the effect of resonance in tblg is now accounted for with @xmath202 .
contribution from the thermal phonons and from the _ sas _ phenomena in the fitting of @xmath7 from ( a ) diamond , ( b ) tblg and ( c ) ab - blg .
the symbols are the experimental data .
the solid black lines are the best fits using eq.(7 ) from the paper , with the parameters shown in table i. the dashed blue and the dotted black lines give the separate contribution from the thermal phonons and from the _ sas _ phenomena , respectively . ] the data fitting gives @xmath206mw@xmath64 and ( @xmath67 } = 295 + ( 15\pm 1)p_l{\rm [ mw]}$ ] ) .
therefore , considering the possibility of @xmath183 in our analysis actually enhances the importance of the _ sas _ phenomena in tblg by almost one order of magnitude ( compare with the value for @xmath207 in table i ) .
furthermore , the dependence on temperature decreases to half of the value obtained for ab - blg , which is ( @xmath67 } = 295 + 30p_l{\rm [ mw]}$ ] ) , consistent with the g band frequency dependence observed in fig.7 of ref.@xcite .
although this picture is fully consistent , we chose to keep the simpler equation in the paper since the general picture does not change , and the discussion is simpler . | the stokes and anti - stokes components in the inelastic scattering of light are related to phonon statistics and have been broadly used to measure temperature and phonon lifetimes in different materials .
however , correlation between the components are expected to change the stokes / anti - stokes intensity ratio , imposing corrections to the broadly used bose - einstein statistics . here
the excitation power dependence of these scattering processes is theoretically described by an effective hamiltonian that includes correlation between the stokes and the anti - stokes events .
the model is used to fit available experimental results in three - dimensional diamond and two - dimensional graphene , showing that the phenomenon can significantly increase in the low - dimensional system under specific resonance conditions . by setting the scientific basis for the stokes - anti - stokes correlated phenomenon ,
the use of the bose - einstein population function for reasoning the inelastic scattering is generalized , providing a model to predict the conversion of optical phonons into heat or light , according to coupling constants and decay rates .
the model applies to inelastic scattering in general .
* keywords * : light scattering , diamond , graphene , stokes - anti - stokes correlation the inelastic scattering of light by matter @xcite exhibits two components : the _ stokes _ ( _ s _ ) , where an incident photon is converted into a phonon and a red - shifted _ s _ photon , and the _ anti - stokes _ ( _ as _ ) , where one incident photon and one existing phonon are annihilated , generating a blue - shifted _ as _ photon .
the stokes / anti - stokes intensity ratio is a signature of the quantum character of lattice vibrations , considered to be defined by the bose - einstein distribution @xmath0 where @xmath1 depends on the optical setup , and @xmath2 is the effective phonon population @xcite .
@xmath3 for phonons with energies @xmath4 larger than the thermal energy @xmath5 ( @xmath6 is the boltzmann constant ) .
the @xmath7 intensity ratio is used to measure phonon lifetimes @xcite , local effective temperatures @xcite and optical resonances @xcite , and except for very specific resonance conditions , the @xmath7 ratio should approach the unit only if the temperature is high enough to activate a very large phonon population .
however , klyshko @xcite has proposed a correlated process , called here the _ stokes - anti - stokes _ ( _ sas _ ) event , where a phonon created by the stokes process is subsequently annihilated in the anti - stokes process .
if the _ sas _
event is significant , the picture described by the bose - einstein phonon distribution is not complete , and eq.[eq - boseeinstein ] has to be generalized .
evidences for the _ sas _ process are accumulating in materials science @xcite , generating interests in quantum optics @xcite .
lee _
et al .
_ @xcite demonstrated the quantum nature of the _ sas _ correlation in diamond by measuring a non - classical _ sas _ field correlation function @xmath8 .
klyshko @xcite pointed that the correlated character of this _ s _ and _ as _ photons can be continuously varied from purely quantum to purely classical and kasperczyk _ et al . _
@xcite explored this transition in character by changing the photon and phonon reservoirs through changes in the excitation laser power . while these experiments are usually performed with ultra - fast pulsed lasers to enhance the response of the non - linear _ sas _ event , jorio _ et al .
_ @xcite have provided evidences for the observation of dominant _ sas _ event using a few miliwatts continuum wave ( cw ) laser , i.e. achievable even with a simple laser pointer .
this result was obtained in twisted - bilayer graphene ( tblg ) , a two - dimensional system specially engineered to exhibit resonance with the _ as _ photon emission @xcite .
since phonons have a significant lifetime , it has been proposed that these systems can work as a solid state quantum memory , storing information between the write ( _ stokes _ ) and read ( _ anti - stokes _ ) processes @xcite . in diamond and graphene @xmath9 , and
the quantum memory would be able to work at room temperature .
here the @xmath7-based phonon population analysis is generalized by proposing an effective hamiltonian that explicitly considers the stokes and anti - stokes fields correlation , on a simple formalism that is able to fit the experimental results from diamond and graphene .
the hamiltonian is given by @xmath10 where @xmath11 ( @xmath12 ) , @xmath13 ( @xmath14 ) and @xmath15 ( @xmath16 ) represent the annihilation ( creation ) operator of stokes ( _ s _ ) , anti - stokes ( _ as _ ) , phonon ( _ c _ ) and incident ( _ a _ ) fields respectively .
@xmath17 and @xmath18 are the coupling constants .
the @xmath17 coupling term describes the creation of an _ s _ photon and a phonon through the absorption of an incident photon , and the @xmath18 term describes the creation of an _ as _ photon through the absorption of an incident photon and a phonon . @xmath19 and @xmath20 are the incident and the phonon fields frequencies , respectively , then the stokes and anti - stokes modes have energies given by @xmath21 . typically , the coherent sources used in experiments of this type can be considered to have large enough number of photons , which allows us to replace @xmath22 , with @xmath23 being the mean number of incident photons . the laser power is given by @xmath24 , @xmath25 being a constant with power units , depending on the laser frequency .
the hamiltonian model above is valid within the coherence time of the pumping laser , whether continuum or pulsed .
hamiltonian ( [ eq:01 ] ) accounts for the interaction at the material but it does not properly describe the dissipation of the created excitations due to the presence of the photonic and phononic reservoirs .
such dynamics can be computed by means of the markovian master equation @xcite for the overall reduced density operator of the three fields , introducing decay rates of phonons and scattered photons .
the master equation for the density operator @xmath26 , in the lindblad form , reads @xmath27+\mathcal{l}(\hat\rho)\,\ ] ] where @xmath28 indicates the tracing out of the reservoir degrees of freedom .
the lindbladian term in our model is @xmath29 , with : @xmath30 with @xmath31 , @xmath32 and @xmath33 being the decay rates ( proportional to the inverse of the coherence time ) of the respective stokes , anti - stokes and phonon fields . to analyze the stokes and anti - stokes field intensities and their respective correlation function at zero delay @xmath34 ,
we compute the average values @xmath35 the power dependence for the intensity ratio @xmath7 and the _ sas _ field correlation @xmath36 are described in panels ( a ) and ( b ) of fig.[expdiamond ] , respectively , for different values of the thermal phonon population @xmath37 . while the stokes field is found to be proportional to the excitation laser power @xmath38 ,
the anti - stokes field exhibits two different regimes . for lower power , thermal phonons
dominate the process and @xmath39 is proportional to @xmath38 ( assuming there is no laser induced heating ) . for higher powers , the _ sas _
phenomenon dominates , and @xmath39 is proportional to @xmath40 ( see supplementary information for details ) .
this rationale explains the results in fig.[expdiamond](a ) , @xmath7 being constant for low power , and for high power , @xmath41 . in fig.[expdiamond](b ) , the _ sas _
correlation function @xmath36 is shown to be proportional to the inverse laser power , and it goes to a @xmath37-dependent constant value for small values of @xmath38 . the behaviors of @xmath42 , @xmath43 and @xmath36 observed by kasperczyk _ et al .
_ @xcite are , therefore , perfectly described by the steady - state number of @xmath44 , @xmath45 photons @xmath46 , @xmath47 and @xmath36 in eqs.[eq:04 ] , calculated by considering the model in eqs.[eq:01]-[eq:04 ] . here
, @xmath48 is the steady state number of @xmath44,@xmath45 photons . a rationale for @xmath7 that takes into account the _ sas _ phenomenon
can then be developed based on the theory for @xmath49 . setting @xmath50 , @xmath51 , @xmath52 , and considering the limit of @xmath53 ,
the anti - stokes / stokes intensity ratio is given by ( see supplemental information ) @xmath54 where @xmath55 for @xmath53 , given in power units .
in the experimental range of interest , eq.[eq:10first ] can be approximated by @xmath56\ , , \ ] ] where the fundamental constants @xmath57 measure the importance of the _ sas _ phenomena , and @xmath58 , with @xmath59 defined as the proportionality constant such that @xmath60 .
the constant @xmath61 depends on the optical parameters , for instance , absorption coefficients , inelastic scattering cross sections and the dimension ( geometry ) of the sample under study @xcite .
figure[expdiagraph ] shows a comparison of the experimental @xmath7 results from the three - dimensional diamond ( black data ) @xcite , from a two - dimensional bi - layer graphene , where the layers are superposed in the so - called ab - staking configuration ( ab - blg , open red data ) @xcite , and finally from the two - dimensional twisted bi - layer graphene ( tblg , solid red data ) , which was engineered to exhibit electronic resonance with the @xmath45 photon emission @xcite .
the three different results observed for the three different samples in fig.[expdiagraph ] can all be fit with eq.[eq:10 ] , giving the parameters provided in table[tab : parameters ] ( see supplementary information for fitting details ) .
analysis of the data shows that the increase in @xmath62 for diamond is dominated by the _ sas _
process ( @xmath63mw@xmath64 ) . actually , the diamond data can be satisfactorily fit by imposing a constant temperature ( @xmath65k ) , with a larger _ sas _ contribution ( @xmath66mw@xmath64 ) , consistent with the fact that , for measurements with a cw laser , within the same excitation power range , the @xmath7 power dependence in diamond exhibits a constant behavior ( see data in the supplementary information ) . for the ab - stacked bilayer graphene ( ab - blg ) ,
the @xmath62 power dependence is dominated by heating , with a linear dependence of the effective phonon temperature on laser power ( @xmath67 } = 295 + 30p_l{\rm [ mw]}$ ] ) . finally , the increase in @xmath62 for twisted - bilayer graphene ( tblg ) is dominated by the _ sas _
process ( @xmath68mw@xmath64 ) , with a contribution from laser induced heating ( @xmath67 } = 295 + 19p_l{\rm [ mw]}$ ] ) that is consistently smaller than that from ab - blg .
notice that the constant found to rule the temperature increase in tblg ( @xmath69k / mw ) is roughly half of the value obtained for the ab - blg ( @xmath70k / mw ) , consistent with the g band frequency changes measured in tblg and ab - blg@xcite . due to resonance with the _ as _ photon emission in tblg , the assumption @xmath71 should be discussed . however , considering @xmath72 does not change the overall picture .
more details can be found in the supplemental information , including a comparative analysis of the thermal versus _
sas _
contributions to @xmath62 for the three samples . comparing the graphene samples , while in ab - blg the generated phonons are mostly converted into heat , in tblg they are mostly converted into light ( _ as _ photons ) . but the most striking result is the much larger efficiency of the _ sas _ process in tblg , with a @xmath73 value that is two orders of magnitude larger than in bulk diamond . besides these two orders of magnitude difference , the @xmath40 dependence for @xmath43 in diamond can only be observed with the use of femtosecond lasers , where the pulse intensity is about @xmath74 times larger than the intensity of continuous wave laser radiation of the same average power .
furthermore , in bulk diamond the number of active atoms in the focal volume is @xmath75 times larger than in the two - dimensional graphene system . finally , the ratio @xmath62 at the highest power in tblg reaches values of @xmath760.5 , while in diamond @xmath77 was reached .
putting all numbers together , correlated _ sas _
generation per involved atom in the graphene - based system is roughly @xmath78 times more efficient than in bulk diamond .
qualitatively , the reason for the striking efficiency of correlated _ sas _ raman scattering in the twisted bilayer graphene system might be related to a reduction in phase - space in the scattering event . in real space ,
the confined two - dimensional structure , which enhances electron - hole interactions in low - dimensional structures @xcite , leads to a large overlap between the photon and phonon wavefunctions . in reciprocal space ,
the logarithmically diverging two - dimensional van - hove singularity at the hexagonal saddle ( m ) point @xcite strongly enhances the generation and recombination of electron - hole pairs at the _ as _ photon energy @xcite .
the _ sas _
event is also expected to play a role in other low dimensional materials @xcite , where phase - space reduction becomes important .
raman spectroscopy is established as an important tool to study and characterize nanostructures @xcite . by setting the scientific basis for the _ sas _ phenomenon , the use of the bose - einstein population function for reasoning the inelastic scattering is generalized , providing a model to predict the conversion of optical phonons into heat or light , according to coupling constants and decay rates .
it is surprising that the correlation between the _ stokes _ and the _ anti - stokes _ components can seriously affect @xmath7 , being measurable indirectly from the intensity ratio .
this aspect made it possible to measure the importance of the _ sas _ correlation in graphene , where a direct @xmath36 measurement is not possible due to ultrafast luminescence @xcite .
the stationary phonon population can then be analytically obtained from the steady state solution of eq.([eq:02 ] ) and , in the limit of @xmath53 , it is given by @xmath79.\end{aligned}\ ] ] the phonon population has two contributions , one thermal and the other related to the _ sas _ phenomenon , each contribution dominating at different excitation power ranges ( see a plot of @xmath80 in fig.1s of the supplementary information ) .
the coupling constants and the decay rates governing the phenomena can themselves be dependent on temperature , through anharmonicities or electron - phonon scattering . on the other way around , the _ sas _
event is an important decay channel that has to be explicitly considered when using raman spectroscopy to extract structural and transport properties related to phonon scattering in low - dimensional structures @xcite . more generally , since the _ stokes _ and _ anti - stokes _ components are common to inelastic scattering processes in general , the model is not limited to the raman scattering by phonons .
it is generic , for instance applying also to the scattering of he atoms on surfaces , to the inelastic neutron scattering , etc .
experimental and theoretical work is needed to address the physics of the @xmath81 , @xmath82 and @xmath33 parameters when applied to different materials and scattering phenomena .
99 c. v. raman and k. s. krishnan , a new type of secondary radiation .
_ nature _ ( london ) * 121 , * 501 ( 1928 ) . a. jorio , m. s. dresselhaus , r. saito , _ raman spectroscopy in graphene related systems _
( wiley - vch , 2011 ) .
d. song _ et al .
_ direct measurement of the lifetime of optical phonons in single - walled carbon nanotubes .
_ phys .
rev .
lett . _ * 100 , * 225503 ( 2008 ) .
c. faugeras _ et al .
_ thermal conductivity of graphene in corbino membrane geometry .
_ acs nano _ * 4 * , 1889 ( 2010 ) .
s. berciaud _
et al . _
electron and optical phonon temperatures in electrically biased graphene .
_ phys .
rev .
lett .
_ * 104 , * 227401 ( 2010 )
. a. g. souza filho _
et al . _
electronic transition energy @xmath83 for an isolated ( @xmath84 ) single - wall carbon nanotube obtained by anti - stokes / stokes resonant raman intensity ratio .
_ phys .
rev .
b _ * 63 , * 241404(r ) ( 2001 ) . d. n. klyshko , correlation between the stokes and anti - stokes components in inelastic scattering of light . _
sov .
j. quantum eletron . _
* 7 * , 755 ( 1977 ) .
d. n. klyshko , _ photons and nonlinear optics _ ( gordon and breach science publishers , new york , 1988 ) .
k. c. lee , _ et al . _ macroscopic non - classical states and terahertz quantum processing in room - temperature diamond .
_ nature photon . _ * 6 , * 41 ( 2012 ) .
a. jorio _
et al . _ optical - phonon resonances with saddle - point excitons in twisted - bilayer graphene . _
nano lett .
_ * 14 , * 5687 ( 2014 ) m. kasperczyk , a. jorio , e. neu , p. maletinksy , l. novotny , stokes - anti - stokes correlations in raman scattering from diamond membranes . _
. letters _ * 40*(10 ) , 2393 ( 2015 ) h. carmichael , _ an open systems approach to quantum optics _ ( springer - verlag , berlin , 1993 ) . h. w. lo and a. compeen , raman measurement of lattice temperature during pulsed laser heating of silicon .
_ phys .
rev .
lett .
_ * 44 , * 1604 ( 1980 )
. k. f. mak , j. shan , t. f. heinz , seeing many - body effects in single- and few - layer graphene : observation of two - dimensional saddle - point excitons .
_ phys .
rev .
lett . _ * 106 , * 046401 ( 2011 ) . c. d. spataru , s. ismail - beigi , l. x. benedict , s. g. louie , excitonic effects and optical spectra of single - walled carbon nanotubes .
_ phys .
rev .
lett . _ * 92 , * 077402 ( 2004 ) .
l. van - hove , the occurrence of singularities in the elastic frequency distribution of a crystal .
_ phys .
rev .
b _ * 89(6 ) , * 1189 ( 1953 ) . a. jorio and l. g. canado , raman spectroscopy of twisted bilayer graphene .
_ solid state comm . _ * 175 - 176 , * 3 ( 2013 ) .
m. steiner , h. qian , a. hartschuh , a. j. meixner , controlling nonequilibrium phonon populations in single - walled carbon nanotubes .
_ nano lett . _ * 7(8 ) , * 2239 ( 2007 ) . c. h. lui , k. f. mak , j. shan , t. f. heinz , ultrafast photoluminescence from graphene .
_ phys .
rev .
lett . _ * 105 , * 127404 ( 2010 ) .
t. li _
et al . _
femtosecond population inversion and stimulated emission of dense dirac fermions in graphene .
_ phys .
rev .
lett . _ * 108 , * 167401 ( 2012 ) . n. bonini , m. lazzeri , n. marzari , f. mauri , phonon anharmonicities in graphite and graphene .
_ phys .
rev .
lett . _ * 99 , * 176802 ( 2007 ) .
m. steiner _
et al .
_ phonon populations and electrical power dissipation in carbon nanotube transistors . _ nature nanotech . _ * 4 , * 320 ( 2009 ) .
a. barreiro , m. lazzeri , j. moser , f. mauri , a. bachtold , transport properties of graphene in the high - current limit .
_ phys .
rev .
lett . _ * 103 , * 076601 ( 2009 ) . |
the properties of dense and/or highly asymmetric nuclear matter , where _ asymmetric _ may refer to isospin or spin asymmetries , are of great current interest in nuclear physics and astrophysics .
this topic is broad - scoped since it reaches out to exotic systems on the nuclear chart as well as , on a dramatically different scale , exotic objects in the universe such as compact stars . in this paper
, we investigate the bulk and single - particle properties of spin - polarized neutron matter .
the study of the magnetic properties of dense matter is of considerable interest in conjunction with the physics of pulsars , which are believed to be rapidly rotating neutron stars with strong surface magnetic fields .
the polarizability of nuclear matter can have strong effects on neutrino diffusion and , in turn , variations of the neutrino mean free path due to changes in the magnetic susceptibility of neutron matter can impact the physics of supernovae and proton - neutron stars .
the magnetic properties of neutron / nuclear matter have been studied extensively since a long time by many authors and with a variety of theoretical methods @xcite .
nevertheless , conclusions about the possibility of a phase transition to a ferromagnetic state at some critical density are still contradictory .
for instance , calculations based on skyrme - type interactions @xcite predict that such instabilities will occur with increasing density .
in particular , currently used skyrme forces show a ferromagnetic transition for neutron matter at densities between 1.1@xmath0 and 3.5@xmath0 @xcite . on the other hand , more recent predictions based on monte carlo simulations @xcite and the brueckner - hartree - fock ( bhf ) approach with realistic nucleon - nucleon ( nn ) interactions @xcite exclude these instabilities , at least at densities up to several times normal nuclear density .
similarly , no evidence of a transition to a ferromagnetic state was found in older calculations based on the brueckner - hartree - fock approach with the reid hard - core potential as well as a non - local separable potentials @xcite .
relativistic calculations based on effective meson - nucleon lagrangians @xcite predict the ferromagnetic transition to take place at several times nuclear matter density , with its onset being crucially determined by the inclusion of the isovector mesons .
clearly , the existence of such phase transition depends sensitively on the modeling of the spin - dependent part of the nuclear force and its behavior in the medium .
thus , this unsettled issue goes to the very core of nuclear physics .
our calculation is microscopic and treats the nucleons relativistically .
a parameter - free and internally consistent approach is important if we are to interpret our conclusions in terms of the underlying nuclear force .
this is precisely our focus , namely to understand the in - medium behavior of specific components of the nuclear force ( in this case , the spin dependence ) .
different nn potentials can have comparable quality as seen from their global description of nn data and yet differ in specific features .
thus , it will be interesting to explore how , for a given many - body approach , predictions for spin - polarized neutron matter depend upon specific features of the nn potential .
second , it will be insightful to compare with predictions based on a realistic nn potential and the bhf method @xcite , especially at the higher densities , where the repulsive dirac effect can have a dramatic impact on the short - range nature of the force .
this work is organized in the following way : after the introductory notes in this section , we briefly review our theoretical framework ( section 2 ) ; our results are presented and discussed in section 3 ; we conclude in section 4 with a short summary and outlook .
the starting point of any microscopic calculation of nuclear structure or reactions is a realistic free - space nn interaction .
a realistic and quantitative model for the nuclear force with reasonable theoretical foundations is the one - boson - exchange ( obe ) model @xcite .
unless otherwise specified , our standard framework consists of the bonn b potential together with the dirac - brueckner - hartree - fock ( dbhf ) approach to nuclear matter . a detailed description of our application of the dbhf method to nuclear , neutron , and asymmetric matter can be found in our earlier works @xcite .
similarly to what we have done to describe isospin asymmetries of nuclear matter , the single - particle potential is the solution of a set of coupled equations @xmath1 @xmath2 where @xmath3 and @xmath4 refer to `` up '' and `` down '' polarizations , respectively , and where each @xmath5 term contains the appropriate ( spin - dependent ) part of the interaction , @xmath6 .
more specifically , @xmath7 where the second summation indicates integration over the two fermi seas of spin - up and spin - down neutrons , and @xmath8 the notation @xmath9 is used for the clebsh - gordon coefficients .
clearly , the need to separate the interaction by individual spin components brings along angular dependence , with the result that the single - particle potential depends also on the direction of the momentum .
notice that the @xmath10-matrix equation is solved using partial wave decomposition and the matrix elements are then summed as in eq .
( 4 ) to provide the new matrix elements in the uncoupled - spin representation needed for eq . ( 3 ) . the three - dimensional integration in eq .
( 3 ) is performed in terms of the spherical coordinates of @xmath11 , @xmath12 , with the final result depending upon both magnitude and direction of @xmath13 . on the other hand ,
the scattering equation is solved using relative and center - of - mass coordinates , @xmath14 and @xmath15 .
these are easily related to the momenta of the two particles in the nuclear matter rest frame through the standard definitions @xmath16 , and @xmath17 .
( the latter displays the dependence of the argument of the spherical harmonics upon @xmath13 and @xmath11 . )
solving the @xmath10-matrix equation requires knowledge of the single - particle potential , which in turn requires knowledge of the interaction .
hence , eqs .
( 1 - 2 ) together with the @xmath10-matrix equation constitute a self - consistency problem , which is handled , technically , exactly the same way as previously done for the case of isospin asymmetry @xcite . the pauli operator for scattering of two particles with unequal fermi momenta , contained in the kernel of the @xmath10-matrix equation , is also defined in perfect analogy with the isospin - asymmetric one @xcite , @xmath18 notice that , although a full three - dimensional integration is performed in eq .
( 3 ) , the usual angle - average procedure is applied to the pauli operator ( when expressed in terms of @xmath14 and @xmath15 ) and to the two - particle propagator in the kernel of the @xmath10-matrix equation .
once a self - consistent solution is obtained for the single - particle spectrum , the average potential energy for particles with spin polarization @xmath19 is obtained as @xmath20 the average potential energy per particle is then @xmath21 the kinetic energy ( or , rather , the free - particle operator in the dirac equation when using the dbhf framework ) , is also averaged over magnitude and direction of the momentum . in particular , we calculate the average free - particle energy for spin - up(down ) neutrons as @xmath22 where @xmath23 is the average over the magnitude of the momentum .
notice that the angular dependence comes in through the effective masses , which , being part of the parametrization of the single - particle potential , are themselves direction dependent ( and of course different for spin - up or spin - down neutrons ) .
finally @xmath24 and the average energy per neutron is @xmath25 as in the case of isospin asymmetry , it can be expected that the dependence of the average energy per particle upon the degree of polarization @xcite will follow the law @xmath26 where @xmath27 is the spin asymmetry , defined by @xmath28 .
a negative value of @xmath29 would signify that a polarized system is more stable than unpolarized neutron matter . from the energy shift , @xmath30 the magnetic susceptibility can be easily calculated . if the parabolic dependence is assumed
, then one can write the magnetic susceptibility as @xcite @xmath31 where @xmath32 is the neutron magnetic moment .
the magnetic susceptibility in often expressed in units of @xmath33 , the magnetic susceptibility of a free fermi gas , @xmath34 where @xmath35 denotes the average fermi momentum which is related to the total density by @xmath36 the fermi momenta for up and down neutrons are @xmath37 .
the angle is defined relative to the polarization axis . ] ( from lowest to highest curve ) .
predictions obtained with our dbhf calculation . ] .
predictions are obtained with the dbhf model . ] for the most general case , it will be necessary to combine isospin and spin asymmetry . with twice as many degrees of freedom , the coupled self - consistency problem schematically displayed in eqs.(1 - 2 ) is numerically more involved but straightforward .
this is left to a future work .
we begin by showing the angular and momentum dependence of the single - neutron potential , see figs . 1 - 2 .
the angular dependence is rather mild , especially at the lowest momenta . as can be reasonably expected , it becomes stronger at larger values of the asymmetry , see fig .
2 . in fig . 3
, the asymmetry dependence is displayed for fixed density and momentum ( here the angular dependence is averaged out ) . as the density of @xmath3 particles goes up , the total density remaining constant , the most likely kind of interaction for @xmath3 neutrons is of the @xmath38 type .
similarly , the largest contribution to the @xmath4-particle potential is of the @xmath39 type , see eqs .
( 1 - 2 ) , with the latter being apparently more attractive , as can be inferred by the spin splitting of the potential shown in fig .
before we move on to discuss this issue in greater details , we also show the effective masses of @xmath3 and @xmath4 neutrons , see fig . 4 , and observe that they display a qualitatively similar behavior as the one of the corresponding single - particle potentials .
the average energy per particle at various densities and as a function of the asymmetry parameter is shown in the third frame of fig .
the first two frames display the contribution from the average potential energy and the average kinetic energy , respectively . the parabolic dependence on @xmath27 , or linear on @xmath40 ,
is obviously verified . in fig .
6 we show the corresponding predictions obtained with the conventional brueckner - harteee - fock approach
. this comparison may be quite insightful , as we further discuss next .
we notice that the dirac energies are overall more repulsive , but the parabolas predicted with the bhf prescription appear to become steeper , relative to each other , as density grows .
the energy difference between the totally polarized state and the unpolarized one for both the relativistic and the non - relativistic calculation is shown in fig .
although initially higher , the growth of the dbhf curve shows a tendency to slow down and the two sets of predictions cross over just above 3@xmath0 . before leaving this detour into the non - relativistic model ,
we observe that the predictions shown in fig . 6 are reasonably consistent with those from previous studies which used the brueckner - hartree - fock approach and the nijmegen ii and reid93 nn potentials @xcite . in fact , comparison with that work allows us to make some useful observations concerning the choice of a particular nn potential , for a similar many - body approach ( in this case , bhf ) .
we must keep in mind that off - shell differences exist among nn potentials ( even if nearly equivalent in their fit of nn scattering data ) and those will impact the @xmath10-matrix ( which , unlike the @xmath41-matrix , is not constrained by the two - body data ) .
furthermore , off - shell differences will have a larger impact at high fermi momenta , where the higher momentum components of the nn potential , ( usually also the most model dependent ) , play a larger role in the calculation .
accordingly , the best agreement between our bhf predictions and those of ref .
@xcite is seen at low to moderate densities .
furthermore , as far as differences based on the choice of the nn potential are concerned , we would expect them to be more pronounced for nuclear matter than for pure neutron matter , since the largest variations among modern realistic potentials are typically found in the strength of the tensor force , which is stronger in t=0 partial waves ( obviously absent in the @xmath42 system ) .
this point will be explored in a later investigation . in the remainder of this paper , we will focus on the dbhf model , which is our standard operational approach . to further explore the possibility of a ferromagnetic transition
, we have extended the dbhf calculation to densities as high as 10@xmath0 .
the same method as described in ref .
@xcite is applied to obtain the energy per particle where a self - consistent solution can not be obtained ( see section iii of ref .
@xcite for details ) .
the ( angle - averaged ) neutron effective masses for both the unpolarized and the fully polarized case are shown in fig .
8 as a function of density .
dbhf predictions for the average energy per particle are shown in fig . 9 at densities ranging from @xmath43=0.5@xmath0 to 10@xmath0 .
what we observe is best seen through the spin - symmetry energy , which we calculate from eq .
( 12 ) and show in fig .
we see that at high density the energy shift between polarized and unpolarized matter continues to grow , but at a smaller rate , and eventually appear to saturate .
similar observations already made in conjunction with isospin asymmetry were explained in terms of stronger short - range repulsion in the dirac model @xcite . it must be kept in mind that some large contributions , such as the one from the @xmath44 state , are not allowed in the fully polarized case .
now , if such contributions ( typically attractive at normal densities ) become more and more repulsive with density ( due to the increasing importance of short - range repulsive effects ) , their absence will amount to less repulsive energies at high density . on the other hand ,
if large and attractive singlet partial waves remain attractive up to high densities , their suppression ( demanded in the totally polarized case ) will effectively amount to increased repulsion . in conclusion , although the curvature of the spin - symmetry energy may suggest that ferromagnetic instabilities are in principle possible within the dirac model , inspection of fig .
10 reveals that such transition does not take place at least up to 10@xmath0
. clearly it would not be appropriate to explore even higher densities without additional considerations , such as transition to a quark phase .
in fact , even on the high side of the densities considered here , softening of the equation of state from additional degrees of freedom not included in the present model may be necessary in order to draw a more definite conclusion . finally , in fig .
11 we show the ratio @xmath45 , whose behavior is directly related to the spin - symmetry energy , see eq .
( 13 ) . clearly
, similar observations apply to both fig .
11 and fig . 10 .
( the magnetic susceptibility would show an infinite discontinuity , corresponding to a sign change of @xmath29 , in case of a ferromagnetic instability . )
we have calculated bulk and single - particle properties of spin - polarized neutron matter .
the eoss we obtain with the dbhf model are generally rather repulsive at the larger densities .
the energy of the unpolarized system ( where all @xmath42 partial waves are allowed ) , grows rapidly at high density with the result that the energy difference between totally polarized and unpolarized neutron matter tends to slow down with density .
this may be interpreted as a _ precursor _ of spin - separation instabilities , although no such transition is actually seen up to 10@xmath0 .
our analysis allowed us to locate the origin of this behavior in the contributions to the energy from specific partial waves and their behavior in the medium , particularly increased repulsion in the singlet states . in future work
, the impact of further extensions will be considered , such as : examining the effects of contributions that soften the eos ( especially at high density ) ; extending our framework to incorporate both spin- and isospin - asymmetries ; examining the temperature dependence of our observations for spin- and isospin - asymmetries of neutron and nuclear matter .
the authors acknowledge financial support from the u.s .
department of energy under grant number de - fg02 - 03er41270 . | the relation between energy and density ( known as the nuclear equation of state ) plays a major role in a variety of nuclear and astrophysical systems .
spin and isospin asymmetries can have a dramatic impact on the equation of state and possibly alter its stability conditions .
an example is the possible manifestation of ferromagnetic instabilities , which would indicate the existence , at a certain density , of a spin - polarized state with lower energy than the unpolarized one .
this issue is being discussed extensively in the literature and the conclusions are presently very model dependent . we will report and discuss our recent progress in the study of spin - polarized neutron matter .
the approach we take is microscopic and relativistic .
the calculated neutron matter properties are derived from realistic nucleon - nucleon interactions .
this makes it possible to understand the properties of the equation of state in terms of specific features of the nuclear force model . |
the lotka - volterra system has been widely investigated in the last years .
this system , studied by may and leonard @xcite , models the evolution of competition between three species . among the studied topics related with the lotka - volterra system , we recall a few of them together with a partial list of references , namely : integrals and invariant manifolds ( @xcite , @xcite ) , stability ( @xcite , @xcite ) , analytic behavior @xcite , nonlinear analysis @xcite , and many others .
in this paper we consider a special case of the lotka - volterra system , recently introduced in @xcite .
we write the system as a hamiltonian system of poisson type in order to analyze the system from the poisson dynamics point of view .
more exactly , in the second section of this paper , we prepare the framework of our study by writing the lotka - volterra system as a hamilton - poisson system , and also find a @xmath0 parameterized family of hamilton - poisson realizations .
as consequence of the hamiltonian setting we obtain two new first integrals of the lotka - volterra system that generates the first integrals of this system found in @xcite . in the third section of the paper
we determine the equilibria of the lotka - volterra system and then analyze their lyapunov stability .
the fourth section is dedicated to the study of the poincar compactification of the lotka - volterra system .
more exactly , we integrate explicitly the poincar compactification of the lotka - volterra system . in the fifth section of the article we present some convexity properties of the image of the energy - casimir mapping and define some naturally associated semialgebraic splittings of the image .
more precisely , we discuss the relation between the image through the energy - casimir mapping of the families of equilibria of the lotka - volterra system and the canonical whitney stratifications of the semialgebraic splittings of the image of the energy - casimir mapping . in the sixth part of the paper
we give a topological classification of the fibers of the energy - casimir mapping , classification that follows naturally from the stratifications introduced in the above section .
note that in our approach we consider fibers over the regular and also over the singular values of the energy - casimir mapping . in the last part of the article we give two lax formulations of the system . for details on poisson geometry and hamiltonian dynamics
see e.g. @xcite , @xcite , @xcite , @xcite .
the lotka - volterra system we consider for our study , is governed by the equations : @xmath1 note that the above system is the lotka - volterra system studied in @xcite in the case @xmath2 .
in @xcite it is shown that this system admits the following polynomial conservation laws : @xmath3 @xmath4 using hamiltonian setting of the problem , we provide two degree - two polynomial conservation laws of the system which generates @xmath5 and @xmath6 .
these conservation laws will be represented by the hamiltonian and respectively a casimir function of the poisson configuration manifold of the system . as the purpose of this paper is to study the above system from the poisson dynamics point of view , the first step in this approach is to give a hamilton - poisson realization of the system .
[ t22 ] the dynamics has the following hamilton - poisson realization : @xmath7 where , @xmath8\ ] ] is the poisson structure generated by the smooth function @xmath9 , and the hamiltonian @xmath10 is given by @xmath11 .
note that , by poisson structure generated by the smooth function @xmath12 , we mean the poisson structure generated by the poisson bracket @xmath13 , for any smooth functions @xmath14 .
indeed , we have successively : @xmath15\cdot \left [ { \begin{array}{*{20}c } { y - z } \\ { x } \\ { -x } \\ \end{array } } \right]= \left [ { \begin{array}{*{20}l } { -x(x - y - z ) } \\ { -y(-x+y - z ) } \\ { -z(-x - y+z ) } \\ \end{array } } \right]= \left [ { \begin{array}{*{20}r } { \dot x } \\ { \dot y } \\ { \dot z } \\ \end{array } } \right],\ ] ] as required . since the signature of the quadratic form generated by @xmath16 is @xmath17 , the triple @xmath18 it is isomorphic with a lie - poisson realization of the lotka - volterra system on the dual of the semidirect product between the lie algebra @xmath19 and @xmath20 .
by definition we have that the center of the poisson algebra @xmath21 is generated by the casimir invariant @xmath16 .
the conservation laws @xmath5 and @xmath6 found in @xcite , can be written in terms of the casimir @xmath12 and respectively the hamiltonian @xmath22 as follows : @xmath23 @xmath24.\ ] ] next proposition gives others hamilton - poisson realizations of the lotka - volterra system .
the dynamics admits a family of hamilton - poisson realizations parameterized by the group @xmath25 .
more exactly , @xmath26 is a hamilton - poisson realization of the dynamics where the bracket @xmath27 is defined by @xmath28 for any @xmath14 , and the functions @xmath29 and @xmath30 are given by : @xmath31 @xmath32 respectively , the matrix of coefficients @xmath33 is @xmath34\in sl(2,{{\mathbb r}}).$ ] the conclusion follows directly taking into account that the matrix formulation of the poisson bracket @xmath27 is given in coordinates by : @xmath35.\ ] ]
in this short section we analyze the stability properties of the equilibrium states of the lotka - volterra system .
the equilibrium states of the system are given as the union of the following three families : @xmath36 @xmath37 @xmath38 figure [ figech ] presents the above defined families of equilibrium states of the lotka - volterra system . in the following theorem
we describe the stability properties of the equilibrium states of the system .
[ t33 ] all the equilibrium states of the lotka - volterra system are unstable .
the conclusion follows from the fact that the characteristic polynomial associated with the linear part of the system evaluated at an arbitrary equilibrium state , is the same for any of the families @xmath39 , and is given by : @xmath40 for @xmath41 , we get the origin @xmath42 which is also unstable since in any arbitrary small open neighborhood around , there exists unstable equilibrium states .
in this section we integrate explicitly the poincar compactification of the lotka - volterra system , and consequently the lotka - volterra system ( on the sphere ) at infinity . recall that using the poincar compactification of @xmath43 , the infinity of @xmath43 is represented by the sphere @xmath44 - the equator of the unit sphere @xmath45 in @xmath46 . for details regarding the poincar compactification of polynomial vector fields in @xmath43
see @xcite .
fixing the notations in accordance with the results stated in @xcite we write the lotka - volterra system as @xmath47 with @xmath48 , @xmath49 , @xmath50 .
let us now study the poincar compactification of the lotka - volterra system in the local charts @xmath51 and @xmath52 , @xmath53 , of the manifold @xmath45 .
the poincar compactification ( @xmath54 in the notations from @xcite ) of the lotka - volterra system is the same for each of the local charts @xmath55 , @xmath56 and respectively @xmath57 and is given in the corresponding local coordinates by @xmath58 regarding the poincar compactification of the lotka - volterra system in the local charts @xmath59 , @xmath60 and respectively @xmath61 , as a property of the compactification procedure , the compactified vector field @xmath54 in the local chart @xmath52 coincides with the vector field @xmath54 in @xmath51 multiplied by the factor @xmath62 , for each @xmath53 .
hence , for each @xmath53 , the flow of the system on the local chart @xmath52 is the same as the flow on the local chart @xmath51 reversing the time .
the system is integrable with the solution given by @xmath63 where @xmath64 are arbitrary real constants . to analyze the lotka - volterra system on the sphere @xmath44 at infinity , note that the points on the sphere at infinity are characterized by @xmath65 .
as the plane @xmath66 is invariant under the flow of the system , the compactified lotka - volterra system on the local charts @xmath51 ( @xmath53 ) on the infinity sphere reduces to @xmath67 the system is integrable and the solution is given by @xmath68 where @xmath69 are arbitrary real constants .
consequently , the phase portrait on the local charts @xmath51 ( @xmath53 ) on the infinity sphere is given in figure [ sphe ] .
the aim of this section is to study the image of the energy - casimir mapping @xmath70 , associated with the hamilton - poisson realization of the lotka - volterra system .
we consider convexity properties of the image of @xmath71 , as well as a semialgebraic splitting of the image that agree with the topology of the symplectic leaves of the poisson manifold @xmath72 .
recall that by a semialgebraic splitting , we mean a splitting consisting of semialgebraic manifolds , namely manifolds that are described in coordinates by a set of polynomial inequalities and equalities . for details on semialgebraic manifolds and their geometry
see e.g. @xcite .
all these will be used later on , in order to obtain a topological classification of the orbits of .
recall first that the energy - casimir mapping , @xmath73 is given by : @xmath74 where @xmath75 are the hamiltonian of the system , and respectively the casimir of the poisson manifold @xmath72 , both of them as considered in theorem [ t22 ] .
next proposition explicitly gives the semialgebraic splitting of the image of the energy - casimir map @xmath76 .
[ p51 ] the image of the energy - casimir map - @xmath77 - admits the following splitting : @xmath78 , where the subsets @xmath79 , @xmath80 , @xmath81 are splitting further on a union of semialgebraic manifolds , as follows : @xmath82 the conclusion follows directly by simple algebraic computation using the definition of the energy - casimir mapping . the superscripts used to denote the sets @xmath83 , are in agreement with the topology of the symplectic leaves of the poisson manifold @xmath72 , namely : * for @xmath84 , @xmath85 is a hyperbolic cylinder . * for @xmath86 , @xmath85 is a union of two intersecting planes .
the connection between the semialgebraic splittings of the image @xmath87 given by proposition [ p51 ] , and the equilibrium states of the lotka - volterra system , is given in the following remark .
[ strt ] the semialgebraic splitting of the sets @xmath83 is described in terms of the image of equilibria of the lotka - volterra system through the map @xmath71 as follows : * @xmath88 * @xmath89 * @xmath90 all the stratification results can be gathered as shown in figure [ fig3 ] .
note that for simplicity we adopted the notation @xmath91 .
@xmath79 & @xmath80 + & + + + + + as a convex set , the image of the energy - casimir map is convexly generated by the images of the equilibrium states of the lotka - volterra system , namely : @xmath92
in this section we describe the topology of the fibers of @xmath71 , considering for our study fibers over regular values of @xmath71 as well as fibers over the singular values .
it will remain an open question how these fibers fit all together in a more abstract fashion , such as bundle structures in the symplectic arnold - liouville integrable regular case . according to the stratifications from the previous section
, the topology of the fibers of @xmath71 can be described as in tables [ tab1 ] , [ tab2 ] , [ tab3 ] : @xmath93 & @xmath94 & @xmath95 & @xmath96 + @xmath97 & & & + @xmath98 & @xmath99 & @xmath100 & @xmath99 + dynamical & union of & union of 8 & union of + description & 4 orbits & orbits and two & 4 orbits + & & equilibrium points & + @xmath101 & @xmath102 & @xmath103 & @xmath104 + @xmath97 & & & + @xmath98 & @xmath105 & @xmath106 & @xmath107 + dynamical & union of 8 & union of 8 & union of + description & orbits and two & orbits and one & 4 orbits + & equilibrium points & equilibrium point & + @xmath108 & @xmath109 & @xmath110 & @xmath111 + @xmath97 & & & + @xmath98 & @xmath99 & @xmath100 & @xmath99 + dynamical & union of & union of 8 & union of + description & 4 orbits & orbits and two & 4 orbits + & & equilibrium points & + the conclusion follows by simple computations according to the topology of the solution set of the system : @xmath112 where @xmath113 belongs to the semialgebraic manifolds introduced in the above section .
a presentation that puts together the topological classification of the fibers of @xmath71 and the topological classification of the symplectic leaves of the poisson manifold @xmath72 , is given in figure [ fig6 ] .
@xmath114 & @xmath115 + & + + + + +
in this section we present a lax formulation of the lotka - volterra system .
let us first note that as the system restricted to a regular symplectic leaf , give rise to a sypmlectic hamiltonian system that is completely integrable in the sense of liouville and consequently it has a lax formulation .
the lotka - volterra system can be written in the lax form @xmath116 $ ] , where the matrices @xmath117 and respectively @xmath118 are given by : @xmath119 , \
b= \left[{\begin{array}{*{20}c } 0 & { iz } & { i(x - y ) } \\ { -iz } & 0 & 0 \\ { -i(x - y ) } & 0 & 0 \\ \end{array } } \right].\ ] ] r.m .
tudoran + the west university of timioara + faculty of mathematics and c.s .
, department of mathematics , + b - dl .
vasile parvan , no .
4 , + 300223-timioara , romania . +
e - mail : [email protected] + supported by cncsis - uefiscdi , project number pn ii - idei code 1081/2008 no . | in this paper we present some relevant dynamical properties of a 3d lotka - volterra system from the poisson dynamics point of view .
[ 1][proof]*#1 . * ' '' '' [ section ] [ theorem]definition [ theorem]lemma [ theorem]remark [ theorem]proposition [ theorem]corollary [ theorem]example * ams 2000 * : 70h05 ; 37j25 ; 37j35 .
* keywords * : hamiltonian dynamics , lotka - volterra system , stability of equilibria , poincar compactification , energy - casimir mapping . |
the accuracy of the monte carlo trigger efficiency has been checked from the two - track trigger , which requires at least two cdc tracks with an opening angle larger than @xmath17 , two or more tof hits as well as the ecl timing signal , using experimental events passing the high energy trigger based on a 1 gev threshold for an ecl total energy sum @xcite .
the trigger efficiency depends on the average transverse momentum of the two tracks in the laboratory frame , @xmath18^{1/2 } \cdot ( 1-|\cos{\theta^*}|^2)^{1/2}$ ] , where the latter is the transverse momentum of @xmath19 in the @xmath16 c.m .
we determine the trigger efficiency as a function of @xmath20 , since each of the two - dimensional bins in @xmath21 and @xmath22 , where the number of events is measured , is associated with a @xmath20 value using the relation above . from the data ,
the trigger efficiency is @xmath23 at @xmath24 gev/@xmath25 and @xmath26 at @xmath27 gev/@xmath25 .
corrections for the monte carlo trigger efficiency are implemented according to the data , with a systematic error within @xmath28 .
the accuracy of the pid efficiencies has been checked by comparing monte carlo estimates to those based on data .
the efficiency associated with each of the four pid conditions ( section 2 ) is studied , using events passing all selection criteria except the condition in question .
the overall pid efficiency is @xmath29 , @xmath30 and down to @xmath31 at @xmath32 , 3 and 4 gev respectively , with a systematic error less than @xmath33 in the whole @xmath21 range .
the fake rate is @xmath34 - @xmath35 for @xmath36 - 4 gev , respectively .
monte carlo studies indicate that the pid requirements are very efficient in rejecting electrons and other relativistic particles , so that events from @xmath37 and @xmath2 do not survive the selection , leaving those from @xmath38 as the main residual background . from monte carlo simulation and the measured cross sections for @xmath38 @xcite
, the fraction of data that can be attributed to residual @xmath39 background , @xmath40 , is evaluated .
based on monte carlo studies , the dependence of @xmath40 on @xmath22 is negligible and @xmath41 , @xmath42 and @xmath43 at @xmath44 , 3.6 and 4.0 gev , respectively . for @xmath45 gev
, @xmath40 is negligible .
the values of @xmath40 have been checked in the data , using events passing all selection criteria except that on the normalized likelihood , @xmath46 .
the number of signal events that would pass all selection criteria is estimated from the @xmath47 distribution of one of the two tracks , after requiring the other to satisfy @xmath46 .
the values of @xmath40 inferred in this way are in good agreement with those above . the contribution from this source of background
is subtracted from the data , using the expected @xmath40 from the monte carlo studies . the systematic uncertainty due to this source is @xmath48 or less in the whole @xmath21 range .
possible non - exclusive backgrounds ( @xmath49 ) , most of them from @xmath50 events , have been searched for in the data .
monte carlo studies show that a high purity sample of such background can be derived from events with larger @xmath51 and smaller @xmath52 , the latter being the transverse momentum balance of the three particles . by comparing monte carlo and data distributions of these parameters
we obtain the fraction of the data attributed to this background type , @xmath53 .
we find that the dependence of @xmath53 on @xmath22 is negligible , and it ranges from @xmath54 to @xmath55 for @xmath21 from 2 to 4 gev , respectively .
corrections are made using the @xmath56 obtained above , and in total @xmath57 of the selected data are subtracted .
the systematic error from this source is 2 - @xmath58 for @xmath21 from 2 to 4 gev , respectively .
the fraction of the data attributed to @xmath50 events is also obtained as a function of @xmath51 . before the correction , a comparison of the @xmath51 distribution between data and monte carlo exhibits a total difference of @xmath59 between the two samples , while it is reduced to less than @xmath60 for any @xmath21 range after the correction ( fig .
since the residual excess in the @xmath51 distribution could be attributed to residual non - exclusive backgrounds and a broader nature of the signal distribution than the monte carlo , the systematic uncertainty due to other possible non - exclusive backgrounds is limited to @xmath60 after the correction .
distributions for the data before ( left column ) and after ( right column ) the subtraction of residual non - exclusive backgrounds ( @xmath50 ) .
the monte carlo distributions are scaled with the first bin normalized to the data.,width=423,height=423 ]
to the data in the range of @xmath61 - 2.9 gev and 3.2 - 4.0 gev , with ( a ) @xmath62 floating ; ( b ) @xmath63 and @xmath64 .
the error bars include statistical and systematic errors .
the @xmath65 values for each fit are indicated in the figure.,title="fig:",width=257,height=257 ] to the data in the range of @xmath61 - 2.9 gev and 3.2 - 4.0 gev , with ( a ) @xmath62 floating ; ( b ) @xmath63 and @xmath64 .
the error bars include statistical and systematic errors .
the @xmath65 values for each fit are indicated in the figure.,title="fig:",width=257,height=257 ] from the asymptotic qcd prediction of eq.(1 ) and after integration over @xmath66 , the cross section for @xmath0 is proportional to @xmath67 for asymptotically large @xmath21 .
all models based on this framework behave asymptotically as @xmath68 . for the diquark scenario ,
two curves are provided @xcite : from the complete diquark model and from the same model with only helicity conserved amplitudes , where @xmath69 and @xmath70 are in opposite helicity states .
the scale of the diquark predictions matches the data for @xmath61 - 3.0 gev , but the deviation becomes larger as @xmath21 increases ( fig . 4 ) . at higher energies
, the data fall below the diquark predictions and exhibit a gradual approach to the three - quark model predictions @xcite . at medium energies between 2.5 and 4.0 gev , a steeper fall of the total cross section in @xmath21 is observed . if we fit the data with a power law @xmath71 with @xmath62 floating ( fig .
8(a ) ) , taking into account both statistical and systematic uncertainties as well as possible correlations between the latter , we obtain @xmath72 and @xmath73 in the range of @xmath61 - 2.9 gev and 3.2 - 4.0 gev , respectively ( the charmonium region between 2.9 and 3.2 gev is excluded ) . for completeness , we also show in fig .
8(b ) the results of the fits with @xmath62 fixed at 10 and 15 .
although for both ranges a good fit to the data can be obtained at @xmath64 , a smaller power , @xmath63 , describes the data above 3.2 gev reasonably well .
this may imply that lower power terms become dominant at higher enegies , which is an indication for the transition to the asymptotic predictions .
the angular differential cross section in @xmath22 is another observable most important to the study of the nature of the interactions involved in the process @xmath0 .
all existing models based on the constituent scattering picture @xcite , as expected , predict an ascending trend , which is in agreement with the data for @xmath74 gev .
this is due to the factor @xmath75 contained in the hard scattering amplitudes .
the same trend is obtained from naive qed @xcite estimates : @xmath76 , in the massless limit .
a simplified picture with diquarks would follow the naive qed expectation above @xcite , if all quark masses are neglected and only scalar diquarks are considered . in fig .
9 we plot the differential cross section normalized to that averaged within @xmath77 , and compare various predictions to the data .
we observe that the data rise more sharply in @xmath22 at higher energy ( see also fig . 5 ) . in comparison ,
all current models predict a flatter trend in @xmath22 . , normalized to that averaged within @xmath77 , for the two higher ranges of @xmath21 .
the error bars are statistical only .
theoretical predictions are from @xcite ( diquark ) , @xcite ( three - quark ) and @xcite ( handbag).,title="fig:",width=257,height=257 ] , normalized to that averaged within @xmath77 , for the two higher ranges of @xmath21 .
the error bars are statistical only .
theoretical predictions are from @xcite ( diquark ) , @xcite ( three - quark ) and @xcite ( handbag).,title="fig:",width=257,height=257 ] the deviation of the leading term qcd calculations @xcite from the data at @xmath61 - 4.0 gev implies that power corrections are still significant at these intermediate energies .
it is not surprising since the very threshold of @xmath3 production corresponds to @xmath78 gev .
however , the diquark and handbag models @xcite were developed in order to describe the intermediate energy region at the price of introducing model form factors , _
etc._. the disagreement of the data at @xmath61 - 4.0 gev with their predictions ( see fig . 4 and 9 ) obviously necessitates their improvement .
the descending trend of the differential cross section in @xmath22 observed at low energies ( @xmath79 gev ) can not be understood within the hard scattering picture ( fig .
in a recent study based on non - perturbative qcd sum rules @xcite , this trend was proposed as a general feature for hadron pair production from two - photon collisions .
the behavior is very natural if low partial waves are involved . in ref . @xcite it was shown that even a simple model based on pole- and resonance - dynamics can reproduce this behavior .
using the belle detector at the high - luminosity kekb collider , the cross sections for @xmath0 have been measured for @xmath21 from 2.025 to 4.0 gev and @xmath4 , with systematic uncertainties ranging from 7% to 14% .
these results represent a great improvement in precision compared to all previous measurements and allow more accurate tests of various theoretical models .
we also observed the production of @xmath5 and determined the product of the two - photon width of the @xmath6 and its branching ratio to @xmath3 .
fitting to a power law @xmath71 shows that the best fit value of @xmath62 decreases as energy increases , and @xmath63 can not be rejected at energies above 3.2 gev , implying the gradual transition to the expectation from asymptotic predictions .
the ascending trend for the differential cross section in @xmath22 predicted by the hard scattering picture is in agreement with the data for @xmath74 gev ; however , the data rise more sharply in @xmath22 as @xmath21 increases . the descending trend in @xmath22 at lower energies @xmath80 gev can be reproduced by non - perturbative approaches @xcite .
the descending trend of the differential cross section in @xmath22 changes to an ascending one with the increase of energy , which could be an indication for the transition from a soft resonance regime to the beginning of a hard regime .
existing models suggested for the intermediate energies @xcite can not provide satisfactory description of the observed energy and angular dependence in the studied energy range .
we thank the kekb group for the excellent operation of the accelerator , the kek cryogenics group for the efficient operation of the solenoid , and the kek computer group and the nii for valuable computing and super - sinet network support .
we thank c. f. berger , v. l. chernyak , h .-
li and k. odagiri for fruitful discussions .
we acknowledge support from mext and jsps ( japan ) ; arc and dest ( australia ) ; nsfc ( contract no . 10175071 , china )
; dst ( india ) ; the bk21 program of moehrd and the chep src program of kosef ( korea ) ; kbn ( contract no .
2p03b 01324 , poland ) ; mist ( russia ) ; mess ( slovenia ) ; snsf ( switzerland ) ; nsc ( under grant no .
nsc-92 - 2112-m-008 - 028 ) and moe ( taiwan ) ; and doe ( usa ) .
99 v. l. chernyak and a. r. zhitnitsky , jetp lett . 25 ( 1977 ) 510 ( pisma v
zhetf 25 ( 1977 ) 544 ) ; v. l. chernyak , v. g. serbo , a. r. zhitnitsky , jetp lett . 26
( 1977 ) 594 ( pisma v zhetf 26 ( 1977 ) 760 ) .
g. p. lepage and s. j. brodsky , phys .
b 87 ( 1979 ) 359 ; g. p. lepage and s. j. brodsky , phys . rev .
d 22 ( 1980 ) 2157 . v. l. chernyak and a. r. zhitnitsky , phys .
( 1984 ) 173 .
g. r. farrar , e. maina and f. neri , nucl .
b 259 ( 1985 ) 702 .
g. r. farrar et al . , nucl .
b 311 ( 1989 ) 585 .
v. l. chernyak and i. r. zhitnitsky , nucl .
b 246 ( 1984 ) 52 .
m. althoff et al .
( tasso collaboration ) , phys .
b 130 ( 1983 ) 449 ; w. bartel et al .
( jade collaboration ) , phys .
b 174 ( 1986 ) 350 ; h. aihara et al .
( tpc / two - gamma collaboration ) , phys . rev .
d 36 ( 1987 ) 3506 ; h. albrecht et al .
( argus collaboration ) , z. phys .
c 42 ( 1989 ) 543 . m. artuso et al .
( cleo collaboration ) , phys .
d 50 ( 1994 ) 5484 .
h. hamasaki et al .
( venus collaboration ) , phys .
b 407 ( 1997 ) 185 .
g. abbiendi et al .
( opal collaboration ) , eur .
j. c 28 ( 2003 ) 45 .
p. achard et al .
( l3 collaboration ) , phys .
b 571 ( 2003 ) 11 .
m. anselmino , f. caruso , p. kroll and w. schweiger , int . j. mod
. phys . a 4 ( 1989 ) 5213 .
p. kroll , th .
pilsner , m. schrmann and w. schweiger , phys .
b 316 ( 1993 ) 546 . c. f. berger and w. schweiger , eur . phys .
j. c 28 ( 2003 ) 249 .
m. diehl , p. kroll and c. vogt , eur .
j. c 26 ( 2003 ) 567 .
k. odagiri , nucl .
a 748 ( 2005 ) 168 .
m. ambrogiani et al .
( fermilab e835 collaboration ) , phys . rev .
d 62 ( 2000 ) 052002 .
a. abashian et al .
( belle collaboration ) , nucl .
instr . meth . a 479 ( 2002 ) 117 .
s. kurokawa and e. kikutani , nucl .
instr . meth . a 499 ( 2003 ) 1 .
s. eidelman et al .
( particle data group ) , phys .
b 592 ( 2004 ) 1
. v. m. budnev , i. f. ginzburg , g. v. meledin and v. g. serbo , phys . rep . 15
( 1974 ) 182 .
s. uehara , kek report 96 - 11 ( 1996 ) .
f. a. berends , p. h. daverveldt and r. kleiss , comp . , phys .
40 ( 1986 ) 285 .
egpc v207 , l3 program library , 1999 , converted for the belle configuration by s. hou , 2000 .
r. brun et al . , cern dd / ee/84 - 1 ( 1987 ) .
j. h. boyer et al . (
mark ii collaboration ) , phys .
d 42 ( 1990 ) 1350 ; h. j. behrend et al .
( cello collaboration ) , z. phys .
c 56 ( 1992 ) 381 ; f. yabuki et al .
( venus collaboration ) , jour .
64 ( 1995 ) 435
. h. nakazawa et al .
( belle collaboration ) , hep - ex/0412058 .
k. abe et al .
( belle collaboration ) , phys .
b 540 ( 2002 ) 33 . c. baglin et al . , phys . lett .
b 187 ( 1987 ) 191 ; t. a. armstrong et al . ,
d 52 ( 1995 ) 4839 ; m. ambrogiani et al . ,
b 566 ( 2003 ) 45 . b. g. cheon et al . , nucl .
instr . meth . a 494 ( 2002 ) 548
. k. abe et al .
( belle collaboration ) , eur .
j. c 32 ( 2004 ) 323
. f. m. renard , basics of electron positron collisions , editions frontires , 1981 .
hsieh and h .- n .
li , phys . rev .
d 70 ( 2004 ) 056002 . | a high precision study of the process @xmath0 has been performed using a data sample of 89 fb@xmath1 collected with the belle detector at the kekb @xmath2 collider .
the cross section of @xmath3 production has been measured at two - photon center - of - mass ( c.m . )
energies between 2.025 and 4.0 gev and in the c.m .
angular range of @xmath4 .
production of @xmath5 is observed and the product of the two - photon width of the @xmath6 and its branching ratio to @xmath3 is determined .
+ pacs numbers : 12.38.qk , 13.60.rj , 14.40.gx [ cols= " < " , ] the total @xmath6 yield , @xmath7 , can be converted to the product of the two - photon width of @xmath6 and the branching fraction of @xmath8 : @xmath9 ev , using the luminosity function @xmath10 determined at the energy of the @xmath6 mass ( @xmath11 ) and the efficiency @xmath12 from monte carlo . for the systematic error ,
effects from the uncertainties of the continuum background shape and the signal width are taken into account , in addition to all other sources listed in table 3 .
the above result gives @xmath13 kev , where the last error comes from the branching fraction @xmath14 uncertainty @xcite .
since observations of the @xmath6 in the @xmath3 channel are scarce and suffer from low statistics , current measurements for @xmath14 available in ref .
@xcite are not very consistent with each other .
our result is the first measurement of @xmath15 in two - photon collisions and , together with the observation of the @xmath6 in @xmath3 collisions in its @xmath16 decay mode @xcite , will help to decrease the errors on both the @xmath6 two - photon width and branching fraction . |
kippenhahn , kohl & weigert ( 1967 ) were the first who followed the formation of helium white dwarfs ( wd ) of low mass in a binary system .
the evolution of a helium wd of 0.26@xmath4 ( remnant ) was investigated by kippenhahn , thomas & weigert ( 1968 ) who found that a hydrogen flash can be initiated near the base of the hydrogen rich envelope .
the energy of the flash is sufficient to cause the envelope to expand to giant dimensions and hence it may be possible that another short term roche lobe filling can occur . in webbink ( 1975 ) , models of a helium white dwarf were constructed by formally evolving a model from the homogeneous zero age main sequence with the reduction of the mass of the hydrogen rich envelope . when the mass of the envelope is less than some critical value , the model contracts adopting white dwarf dimensions .
webbink found that thermal flashes do not occur for wds less massive than 0.2@xmath4 .
alberts et al .
( 1996 ) have confirmed webbink s finding that low
mass white dwarfs do not show thermal flashes and the cooling age for wds of mass @[email protected]@xmath7 can be considerably underestimated if using the traditional wd cooling curves which were constructed for @[email protected]@xmath7 ( iben & tutukov 1986 , it 86 ) .
recently , hansen & phinney ( 1998a hp98 ) and benvenuto & althaus ( 1998 ba98 ) investigated the effect of different mass of the hydrogen layer ( @xmath10 ) on the cooling evolution of @xmath11 helium wds . in both calculations ( ba98 and hp98 ) the mass of the hydrogen envelope left on the top of white dwarf has been taken as free parameter .
ba98 found that thick envelopes appreciably modify the radii and surface gravities of no h models , especially in the case of low
mass helium white dwarfs .
driebe et al .
( 1998 dsbh98 ) present a grid of evolutionary tracks for low - mass white dwarfs with helium cores in the mass range from 0.179 to 0.414@xmath4 .
the tracks are based on a 1@xmath4 model sequence extending from the pre
main sequence stage up to the tip of red giant branch . applying large mass loss rates forced the models to move off the giant branch and evolve across the hertzsprung
russell diagram and down the cooling branch .
they found that hydrogen flashes take place only for two model sequences , 0.234@xmath4 and 0.259@xmath12 , and for very low mass wds the hydrogen shell burning remains dominant even down to effective temperatures well below 10 000 k. according to our previous calculations ( ergma , sarna & antipova , 1998 ) we find that for a low mass white dwarf with a helium core , which was formed during low mass binary evolution ( after detachment from the roche lobe ) , the hydrogen layer left on the top of the helium core is much thicker ( @xmath13 with @xmath14 ranging from 0.3 to 0.52 ) than used in cooling calculation by hp98 and ba98 .
also in dsbh98 ( see their table 1 ) , for the two lowest total remnant masses the envelope mass value is smaller that obtained in our calculations .
low mass helium white dwarfs are present in millisecond binary pulsars and double degenerate systems .
this gives a unique opportunity to test the cooling age of the wd in a binary and , especially in the case of millisecond binary pulsars , allows for age determinations for neutron stars that are independent of their rotational history .
the evolutionary sequences we have calculated are comprised of three main phases : @xmath15 detached evolution lasting until the companion fills its roche lobe on the time scale @xmath16 ; @xmath17 semi detached evolution ( non conservative in our calculations ) on the time scale @xmath18 ; @xmath19=@xmath20 ; @xmath17 a cooling phase of the wd on the time
scale @xmath21 ( the final phase during which a system with a ms pulsar + low mass helium wd is left behind ) .
the total evolutionary time is @xmath22 .
the duration of the detached phase is somewhat uncertain ; it may be determined either by the nuclear time
scale or by the much shorter time
scale of the orbital angular momentum loss owing to the magnetized stellar wind . in our calculations
we assume that the semi detached evolution of a binary system is non conservative , i.e. the total mass and angular momentum of the system are not conserved .
we can express the total orbital angular momentum ( j ) of a binary system as @xmath23 where the terms on the right hand side are due to : stellar mass angular momentum loss from the system , magnetic stellar wind braking , and gravitational wave radiation .
the formalism which we have adopted is described in muslimov & sarna ( 1993 ) .
we introduce the parameter @xmath24 characterizing the loss of mass from the binary system and defined by the relations , @xmath25 where @xmath26 is the mass loss rate from the system , @xmath27 is the rate of mass loss from the donor ( secondary ) star and @xmath28 is the accretion rate onto the neutron star ( primary ) .
the matter leaving the system will carry off its intrinsic angular momentum in agreement with formula @xmath29 where @xmath30 and @xmath31 are the masses of the neutron star and donor star , respectively and m=@xmath30+@xmath31 . here
we have introduced the additional parameter @xmath32 , which describes the efficiency of the orbital angular momentum loss from the system due to a stellar wind ( tout & hall 1991 ) . in our calculations we have @xmath32=1 and @xmath24=1 ; we calculate the fully non conservative case , although additional calculations with @xmath24 = 0.9 and 0.5 ( with @xmath32=1 ) give similar results . a similar result to ours
was found by tauris ( 1996 ) , who showed that the change in orbital separation due to mass transfer in lmxb ( low mass x - ray binaries ) as a function of the fraction of exchanged matter @xmath24 which is lost from system is small ( for 0.5@xmath33 ) . to understand whether the system evolution is conservative or non conservative
is not easy in the case of a rapidly rotating neutron star ; no easy solution can be found .
we propose as one possibility a factor which may help us to distinguish between the two cases the surface magnetic field of the neutron star and its evolution during the accretion .
we also assume that the donor star , possessing a convective envelope , experiences magnetic braking ( mestel 1968 ; mestel & spruit 1987 ; muslimov & sarna 1995 ) , and , as a consequence , the system loses its orbital angular momentum . for a magnetic stellar wind we used the formula for the orbital angular momentum loss @xmath34 where @xmath35 and @xmath36 are the separation of the components and the radius of the donor star in solar units . for systems with very short orbital periods , during the final stages of their evolution we also take into account the loss of orbital angular momentum due to emission of gravitational radiation ( landau & lifshitz 1971 ) :
@xmath37 the mass and accompanying orbital angular momentum loss from these system are poorly understood problems in the evolution of binary stars .
as is well known , the variation of the angular momentum depends critically on the assumed model ( ergma et al .
1998 ) . in the case of binary systems with ms pulsar typically
two different models concerning the mass ejection and angular momentum loss can be adopted .
the first is that the amount of angular momentum lost per 1 gram of ejected matter is equal to the average orbital angular momentum of 1 gram of the binary .
the second is that the matter that flows from the companion star onto the neutron star ( after accretion ) is ejected isotropically with the specific angular momentum of the neutron star . in this paper , for our non conservative approach we have adopted the first model .
this affects significantly our results on the semi detached evolution ( see fig . 2 in ergma et al .
1998 ) , but very little changes the cooling time scale of the helium white dwarf . in all cases
we have included the effect of illumination of the donor star by the millisecond pulsar . in our calculations
we assume that illumination of the component by the hard ( x ray and @xmath38ray ) radiation from the millisecond pulsar leads to additional heating of its photosphere ( muslimov & sarna 1993 ) .
the effective temperature @xmath39 of the companion during the illumination stage is determined from the relation @xmath40 where @xmath41 is the intrinsic luminosity corresponding to the radiation flux coming from the stellar interior and @xmath42 is the stefan boltzmann constant .
@xmath43 is the millisecond pulsar radiation that heats the photosphere , which is determined by @xmath44 and @xmath45 is `` rotational luminosity '' of the neutron star due to magneto
dipole radiation ( plus a wind of relativistic particles ) @xmath46 where @xmath47 is the neutron star radius , b is the value of the magnetic field strength at the neutron star and @xmath48 is the pulsar period .
@xmath49 is the factor characterizing the efficiency of transformation of irradiation flux into thermal energy ( in our case we take @xmath50 ) .
note that in our calculations the effect of irradiation is formally treated by means of modification of the outer boundary condition , according to relation ( 6 ) .
in this paper we do not follow the magnetic field and pulsar period ( @xmath51 ) evolution , as we did in our earlier papers ( muslimov & sarna 1993 , ergma & sarna 1996 ) .
we were mainly interested in finding initial models for low
mass helium white dwarfs and in investigating the initial cooling phase of these low mass helium white dwarfs . from earlier calculations
we know that if the magnetic field strength is greater than about @xmath52 g , the neutron star spins up to tenths and hundreds of milliseconds , rather than several milliseconds .
this leads to a situation where the pressure of the magneto
dipole radiation is insufficient to eject matter from the system . also from our previous calculations
( see for example ergma & sarna 1996 ) we find that after accretion of a maximum of about 0.2@xmath1 , the neutron star has spun up to millisecond periods if b@xmath53 g .
therefore in this paper we accept that after accretion of 0.2@xmath1 the neutron star spins up to about 2 ms .
after spin up the pulsar irradiation is strong enough to prevent accretion , and at this moment we include non conservative mass loss from the system as described above . during the initial high mass accretion phase ( @xmath54 , @xmath55yrs )
the system may be observed as a bright low mass x
ray binary ( lmxb ) .
it is necessary to point out that majority of lmxbs for which orbital period determinations are available ( 21 systems out of 24 according to van paradijs catalogue 1995 ) , have orbital period of less than one day .
these systems therefore can not be the progenitors of the majority of low
mass helium white dwarf + millisecond pulsar binary systems .
a lack of lmxb systems with orbital period between 1 3 days does not allow us to make a direct comparison between the observational data and the results of our calculations .
the models of the stars filling their roche lobes were computed using a standard stellar evolution code based on the henyey type code of paczyski ( 1970 ) , which has been adapted to low mass stars .
the henyey method involves iteratively improving a trail solution for the whole star . during each iteration
, corrections to all variables at all mesh points in the star are evaluated using the newton
raphson method for linearised algebraic equations ( see for example hansen & kawaler 1994 ) .
the henyey method extended to calculate stellar evolution with mass loss , as adopted here , is well explain by zikowski ( 1970 ) .
we note here that our code makes use of the stationary envelope technique , which was developed early on in the life of our code in order to save disc space ( paczyski 1969 ) .
this method makes the assumption that the surface 0.5 5% ( by mass ) of the star is not significantly affected by nuclear processes , such that it can be treated to a good approximation as homogeneous region ( in composition ) throughout the whole evolutionary calculation . during the cooling phase
we assume that the static envelope is the surface 0.5% of the star .
this assumption is valid during the flashes because the time scale is longer than thermal time
scale of the envelope .
we tested the possibility that the algorithm for redistributing meshpoints introduces numerical diffusion into the composition profile .
we find that if such numerical diffusion is real , it has only a marginal influence on the hydrogen profile .
we would also like to note that in the heat equation we neglect the derivative with respect to molecular weight , since its effect is small .
convection is treated with the mixing
length algorithm proposed by paczyski ( 1969 ) .
we solve the problem of radiative transport by employing the opacity tables of iglesias & rogers ( 1996 ) .
where the iglesias & rogers ( 1996 ) tables are incomplete , we have filled the gaps using the opacity tables of huebner et al .
( 1977 ) . for temperatures less than 6000
k we use the opacities given by alexander & ferguston ( 1994 ) and alexander ( private communication ) . the contribution from conduction present in the opacity tables of huebner et al .
( 1977 ) has been included by us in the other tables , since they do nt include it ( haensel , private communication ) .
the equation of state ( eos ) includes radiation and gas pressure , which is composed of the ion and electron pressure .
contribution to the eos owing to the non ideal effects of coulomb interaction and pressure ionization which influence the eos , as discussed by pols et al .
( 1995 ) , have not been included in our program , and for this reason we stopped our cooling calculations before these effects become important . during the initial phase of cooling , the physical conditions in the hot white dwarfs are such that these effects are usually small .
we perform our evolutionary calculations for binary systems initially consisting of a 1.4 @xmath56 neutron star ( ns ) and a slightly evolved companion ( subgiant ) of two masses , 1 and 1.5 , and four chemical compositions ( z : 0.003 , 0.01 , 0.02 , 0.03 ) .
we have produced ( table 1 ) a number of evolutionary tracks corresponding to the different possible values of the initial orbital period ( ranging from 0.7 to 3.0 days ) at the beginning of mass transfer phase .
in table 1 we list the characteristic of the cooling phase of the wd , @xmath57 , and the maximum possible evolution time of a system , @xmath58 , which is a sum of times of detached ( determined by nuclear evolution ) , semi detached , and cooling phases .
the cooling is the last phase of evolution of the wd , and in our calculations starts at the end of rlof .
the cooling time , @xmath57 , is limited to an initial cooling stage during which the wd cools until its central temperature has decreased by 50 @xmath59 of its maximum value . from table 1
it is clearly seen that to produce short orbital period systems in a time
scale shorter than hubble time it is necessary either to have low z or a more massive secondary . in our calculations
the donor star fills its roche lobe while it is evolving through the hertzsprung gap , and therefore it transfers mass on its companion in a thermal time scale .
figure 1 show the evolutionary cooling sequences for models 20 and 22 ( more details in table 1 ) .
model 20 presents the case with stable hydrogen burning .
model 22 shows the case when the thermal instability of the hydrogen
burning shell occurs .
the first flash is not strong enough to allow the star to overflow its roche lobe , but during the second flash the radius of the secondary increases to fill its roche lobe and short time roche lobe overflow ( rlof ) occurs . in table 2
we present the mass radius relationship for wds from our calculations , dsbh98 , the wood models , and the hamada & salpeter ( 1961 ) zero temperature helium wd models calculated for a surface temperature of 8500 k ( as in van kerkwijk , bergeron & kulkarini 1996 for psr1012 + 5307 ) .
comparison of the numbers demonstrate that for wd masses of @xmath60 0.25@xmath1 , the results of our calculations differ significantly from a simple extrapolation obtained from the cooling curves ( wood 1990 ) performed for carbon wds with the thick hydrogen envelopes .
in addition comparing the cooling time scales of hp98 and ba98 with those of webbink and our models , shows differences of an order of magnitude ( table 3 ) for wd masses of @xmath60 0.25@xmath1 .
the problem of unstable hydrogen shell burning in low mass helium wds was first discussed in the literature more than 30 years ago ( kippenhahn , thomas & weigert 1968 ) .
recently , alberts et al .
( 1996 ) have claimed that they do not see any thermal flashes that result from thermally unstable shell burning , as reported in papers it86 and kippenhahn , thomas & weigert ( 1968 ) .
webbink ( 1975 ) found that in none of his model sequences , such a severe thermal runaway as described by kippenhahn et al .
( 1968 ) was found , although mild flashes for [email protected] @xmath7 did take place .
alberts et al . found that even reducing the time step to 50100 years would not lead to thermally unstable shell burning for @[email protected] @xmath4 . in dsbh98 ,
thermal instabilities of the hydrogen
burning shell occurs in their two models , 0.234@xmath4 and 0.259@xmath4 .
they concluded that hydrogen flashes take place only in the mass interval 0.21@xmath6 @xmath61 . according to our computations , low
mass helium wds with masses more than 0.183@xmath4 ( z=0.03 ) , 0.192@xmath4 ( z=0.02 ) , 0.198@xmath4 ( z=0.01 ) and 0.213@xmath4 , ( z=0.003 ) may experience up to several hydrogen flashes before they enter the cooling stage . in table 4
we present several characteristics for the computed flashes .
we discussion two kinds of flashes : in the first case ( in table 4 shown as `` 1 '' ) , during the flash the secondary does not fill its roche lobe i.e. the mass of the white dwarf does not change , and in the second case ( `` 2 '' ) , during the unstable hydrogen burning phase the secondary fills its roche lobe and the system again enters into a very short duration accretion phase ( see table 4 ) .
we introduce four time scales to describe the flash behaviour : ( i ) the flash rise time
scale @xmath62 , which is the time for the luminosity to increase from minimum to maximum value ( typically this value is between few @xmath63 to few @xmath64 yrs third column in the table 4 ) ; ( ii ) the flash decay time
scale @xmath65 , which is the time for the luminosity to decrease to the initial value ( typically from few hundred thousand to few tenth million years ) ; ( iii ) @xmath66 t is the recurrence time between two successive flashes ( iv ) @xmath67 is the duration of the accretion phase when the secondary fills its roche lobe during hydrogen shell flash . for all sequences with several unstable hydrogen shell burning stages ( usually for case `` 1 '' ) ,
the first flash is the weakest . in the majority of cases
when the flash forces the star to fill its roche lobe , only one flash takes place . for four cases we found two successive flashes with roche lobe overflow ( models 17 , 23 , 24 , 31 ) , and for another two cases ( models 47 , 53 ) to the first flash is not powerful enough to force the secondary fill its roche lobe , but during the second flash it is
how does the hydrogen flash burning influence the cooling time scale ?
in fig.2 , the luminosity and nuclear energy production rates versus cooling time for models 20 and 22 are shown .
model 20 shows stationary hydrogen burning and model 22 , hydrogen flash burning .
although before flash model 22 was more luminous than model 20 , later the situation is reversed .
after the flash , the burning mass of the hydrogen rich envelope in model 22 has decreased to 0.0116@xmath68 , whereas the mass of the hydrogen envelope in model 20 , in which stationary hydrogen burning occurs , is almost twice as large ( 0.0241@xmath68 ) .
if we look at how the maximum nuclear energy rate behaves with cooling time , we can see that after the flash in model 22 , the maximum energy production rate is less than in model 20 ( stationary hydrogen burning ) . in fig .
3 we present the behaviour of log@xmath69 , log @xmath70 and log l/@xmath71 , and in fig.4 log @xmath72 , @xmath73 and @xmath74 as a function of cooling time for model 7 . before the helium white dwarf enters the final cooling phase , four unstable hydrogen flash burnings occur .
the same parameters for model 17 ( with rlof ) are shown in figs . 5 and 6 . to investigate in more detail how the flashes develop , we show in fig .
7 the evolution of the white dwarf radius ( upper panel ) , nuclear energy generation rate ( upper middle panel ) , maximum shell temperature and central temperature ( lower middle panel ) and the surface luminosity ( lower panel ) as a function of computed model number . in fig . 7 , as vertical dashed lines we marked several time scales which characterize the flash behaviour ( for numbers see table 4 ) .
@xmath62 and @xmath75 describe the rise and decay times ; the first characterizes the nuclear shell burning time scale ( @xmath76 ) , the second the kelvin helmholtz ( thermal ) envelope time scale modified by nuclear shell burning ( @xmath77 ) . the accretion time ( @xmath78 ) is described by the square of the kelvin helmholtz time
the radiative diffusion time is defined as the kelvin
helmholtz time
scale of the extended envelope above the shell ( @xmath79 ) .
the shape of the first flash on fig .
7 shows some characteristic changes which are connected with physical processes in the stellar interior . at the beginning of the flash
the luminosity increases due to the more effective hydrogen burning in the shell source . after reaching a local maximum
, the luminosity then decreases while the nuclear energy generation rate is still increasing rapidly .
this decrease of the surface luminosity is due to a temperature inversing forming below the hydrogen shell .
the energy generated in the hydrogen shell splits into two fluxes ; coming outwards and going inwards .
the helium core is heated effectively by the shell nuclear source the central temperature increases by 2% . on fig .
8 the evolution of the luminosity and temperature profiles during the @xmath80 and @xmath75 phases are shown .
we clearly see how the inversion profile evolves and how the luminosity wave moves into the surface .
the nuclear energy generation rate in the shell has a maximum value far away from maximum surface luminosity .
this is because the luminosity front is moving towards the stellar surface in a time
scale described by radiative diffusion ( @xmath81 ) . after reaching a maximum value ,
the luminosity starts to decrease and the energy generation rate also declines in the hydrogen shell over a time - scale @xmath75 ( for a contracting envelope ) the luminosity decreases to the minimum value . during the first flash
, the stellar radius does not fill the inner roche lobe . in the second and third flashes we have short episodes of super eddington mass transfer ( see table 4 ) . during the rlof phase , the orbital period slightly increases and the subgiant companion evolves quickly from spectral type f0 to a0 .
as already pointed out , for several cases the secondary fills its roche lobe and the system enters an accretion phase . during rlof , the mass accretion rate is about three orders of magnitude greater than the eddington limit ( fig .
all the accreted matter will be lost from the system ( @xmath82 ) .
the accretion phase is very short , usually less than 1000 years ( ranging from 160 to 2500 yrs see table 4 ) . during the short super eddington accretion phase the system is a very bright x ray source , with orbital period between 2 to 8 days .
we notice that during the flash the evolutionary time step strongly decreases and may be as short as several years .
dsbh98 modelled single star evolution and produced white dwarfs with various masses by applying large mass loss rates at appropriate positions in the red giant branch to force the models to move off the giant branch . to show how binarity influences the final fate of the white dwarf cooling ,
we have computed extra sequences ( 1.0 + 1.4 @xmath7 , z=0.02 , @xmath83=2.0 days ) where we did not take into account that the star is in a binary system e.g. during hydrogen shell flash we do not allow rlof . in complete binary model calculation , only one shell flash occurs accompanied with rlof , whereas for the single star model calculation , four hydrogen shell flashes take place . due to rlof ,
the duration of the flash phase is 2.7@xmath2yrs ; if we do not include binarity the duration of the flash phase is 1.8@xmath84yrs .
however , the cooling time for helium white dwarfs less massive than 0.2@xmath1 is not significantly changed .
this is because the duration of flash phase is very short in comparison to the normal cooling phase ( towards the white dwarf region ) .
however , the effect of binarity will be important for the cooling history of more massive helium white dwarfs . in fig .
10 both cases of evolution on the hertzsprung russell diagram are shown on the left panel roche lobe overflow is not allowed , on right panel rlof takes place .
below we discuss the observational data for several systems for which results of our calculations may be applied , by taking into account the orbital parameters of the system , the pulsar spin down time , and the white dwarf cooling timescale .
timing information for this millisecond binary system : @xmath85=5.757 ms , @xmath86=5.741 days , @xmath87 ( intrinsic characteristic age of pulsar ) = 4.4 4.91 gyrs , mass function f(m)= 1.239@xmath88@xmath7 ( johnston et al . 1993 ; bell et al . 1995 ) .
hansen & phinney ( 1998b ) have discussed the evolutionary stage of this system using their own cooling models described in hp98 .
they found consistent solution for all masses in the range 0.15 0.375@xmath4 with thick ( in the terminology of hp98 ) hydrogen envelopes of 3@xmath89 .
timing measurements by sandhu et al .
( 1997 ) have detected a rate of change in the projected orbital separation @xmath90 , which they interpret as a change in @xmath91 and they calculate for an upper limit for @xmath91@xmath60 @xmath92 and new lower limit to the mass of the companion of m@xmath93 0.22@xmath4 .
our calculations also allow us to produce the orbital parameters and secondary mass for the psr j04374715 system and fit its cooling age ( 2.55.3 gyrs , hansen & phinney , 1998b ) , and we find that the secondary fills its roche lobe when the orbital period @xmath83 is @xmath93 2.5 days ( tables 1 , 4 ) . from our cooling tracks for a binary orbital period of 5.741 days , the mass of the companion is 0.21@xmath94 and its cooling age 1.262.25 gyrs ( for a population i chemical composition ) .
these cooling models usually have one strong ( with rlof ) hydrogen shell flash , after which the helium wd enters the normal cooling phase .
lorimer et al .
( 1995 ) determined a characteristic age of the radio pulsar to be 7 gyr , which could be even larger if the pulsar has a significant transverse velocity ( hansen & phinney 1998b ) . using the it86 cooling sequences , they estimated the companion to be at most 0.3 gyr old .
hp98 models yield the following results for this system : the companion mass lies in the range 0.130.21@xmath4 and the wd age is @xmath60 0.6 gyr , the neutron star mass in the range 1.32.1@xmath4 .
alberts et al .
( 1996 ) were the first to show that the cooling timescale of a low mass wd can be substantially larger if there are no thermal flashes which lead to rlof and a reduction of the hydrogen envelope mass .
our and dsbh98 calculations confirmed their results that for low
mass helium wds ( @xmath60 0.2 @xmath7 ) , indeed stationary hydrogen burning plays important role . to produce short ( less that one day ) orbital period systems with a low mass helium wd and a millisecond pulsar
it is necessary that the secondary fills its roche lobe between @xmath95 and @xmath96 ( ergma , sarna & antipova , 1998 ) .
if the initial orbital period @xmath97 ( at rlof ) is less than @xmath95 , the binary system evolves towards short orbital periods .
@xmath96 is another critical orbital period value . if @xmath96 @xmath60 @xmath98 @xmath60 @xmath95 , then a short orbital period ( @xmath60 1 day ) millisecond binary pulsar with low
mass helium white dwarf may form .
so the initial conditions of the formation of such systems are rather important .
we calculated one extra sequence to produce a binary system with orbital parameters similar to psr j1012 + 5307 .
initial system : 1 + 1.4 @xmath7 , @xmath97(rlof ) = 1.35 days , z=0.01 .
final system : @xmath99=0.168@xmath4 , @xmath100=0.605 days , @xmath73=0.041@xmath4 . in fig .
11 in the effective temperature and gravity diagram we show the cooling history of this white dwarf after detachment of the roche lobe .
the two horizontal regions are the gravity values inferred by van kerkwijk et al .
( 1996 ) ( lower ) and callanan et al .
( 1998 ) ( upper ) .
our results are consistent with the callanan et al .
( 1998 ) estimates .
it is necessary to mention that after detachment from its roche lobe , the outer envelope is rather helium rich .
bergeron et al . (
1991 ) have shown that a small amount of helium in a hydrogen dominated envelope can mimic the effect of a larger gravity .
the results of our evolutionary calculations differ from those of iben & tutukov ( 1986 ) and driebe et al .
( 1998 ) because of the different formation scenarios for low mass helium wds . in it86
s calculations a donor star fills its roche lobe while it is on the red giant branch ( i.e. has a thick convective envelope ) with a well developed helium core and a thin hydrogen burning layer .
they proposed that the mass transfer time scale is so short that the companion will not be able to accrete the transferred matter and will itself expand and overflow its roche lobe .
the final output is the formation of a common envelope and the result of this evolution is a close binary with a helium wd of mass @xmath101 having a rather thin ( @xmath102 ) hydrogen rich ( x=0.5 ) envelope .
dsbh98 did not calculate the mass exchange phases during the red giant branch evolution in detail but they also simulated the mass exchange episode by subjecting a red giant branch model to a sufficiently large mass loss rate . in both cases ( it86 and
dsbh98 ) mass loss starts when the star ( with a well developed helium core ) is on the red giant branch . in our calculations
the roche lobe overflow starts when the secondary has either almost exhausted hydrogen in the center of the star or has a very small helium core with a thick hydrogen burning layer . during the semi
detached evolution the mass of the helium core increases from almost nothing to final value ( for more detail about evolution of such systems , see ergma , sarna & antipova , 1998 ) .
this is the reason that a much thicker ( @xmath103 \times 10^{-2}~m_{\odot } $ ] , with x ranging from 0.30 to 0.52 ) hydrogen rich envelope is left on the donor star at the moment it shrinks within the roche lobe .
the second important point where our results differ from that of dsbh98 is that in our calculations we can produce ( after the secondary detaches from its roche lobe ) final millisecond binary pulsar parameters which we compare with observational data ( orbital period , spin period of ms pulsar , mass of the companion ) .
it was shown by joss , rappaport & lewis ( 1987 ) and more recently by rappaport et al .
( 1995 ) that the evolution of a binary system initially comprising of a neutron star and a low mass giant will end up as a wide binary containing a radio pulsar and a white dwarf in a nearly circular orbit .
the relation between the white dwarf mass and orbital period ( see eq .
( 6 ) in rappaport et al .
1995 ) shows that if the secondary fills its roche lobe while on the red giant branch , then for @xmath104 the final orbital period would be @xmath93 5 days , which is far from observed orbital period of the binary pulsar psr j 1012 + 5307 ( @xmath86=0.6 days ) .
alberts et al .
( 1996 ) , dsbh98 , and the results of our calculations demonstrate clearly that especially for low mass helium wds ( @xmath60 0.2@xmath4 ) stationary hydrogen burning remains an important , if not the main , energy source . hp98 and ba98 did consider nuclear burning but found it to be of little importance since their artificially chosen hydrogen envelope mass was less than some critical value , disallowing significant hydrogen burning . if we compare now the cooling curves of hp98 , dsh98 with ours then there is one very important difference ; they did not model the evolution of the helium wd progenitor and all their cooling models ( see for example figs
. 11 , 12 in hp98 ) start with a high @xmath105 . in our models ,
cooling of the helium wd starts after detachment of the secondary from its roche lobe ( dsbh98 mimic this situation with mass loss from the star ) .
this time , the secondary ( proto white dwarf ) has rather low effective temperature ( see for example fig.1 ) . during the evolution with l approximately constant , the effective temperature increases to a maximum value , after which it decreases while still having a active hydrogen shell burning source . the evolutionary time needed for the proto
white dwarf to travel from the minimum @xmath105 ( after detachment from roche lobe ) to maximum @xmath105 depends strongly on mass of the wd ( for a smaller mass a longer evolutionary time
scale ) .
so for low
mass helium wds the evolutionary prehistory plays a very important role in cooling history of the white dwarf .
we have performed comprehensive evolutionary calculations to produce a close binary system consisting of a ns and a low mass helium wd .
we argue that the presence of a thick hydrogen layer changes dramatically the cooling time scale of the helium white dwarf ( @xmath106 ) , compared to the previous calculations ( hp98 , ba98 ) where the mass of the hydrogen envelope was chosen as free parameter and was usually one order of magnitude less than that obtained from real binary evolution computations .
also , we have demonstrated that using new cooling tracks we can consistently explain the evolutionary status of the binary pulsar psr j1012 + 53 .
tables with cooling curves are available on * http://www.camk.edu.pl/ sarna/*.
we would like to thank dr .
katrina m. exter for help in improving the form and text of the paper .
we would like to thank our referee dr . peter eggleton for very useful referee opinion . at warsaw , this work is supported through grants 2p03d01407 and 2p03d00516 of the polish national committee for scientific research . also , j.a . and
e.e . acknowledges support through estonian sf grant 2446 .
clrrrrrlrrcc + + + model & @xmath107 & & @xmath108 & & @xmath109 & & & @xmath110 & @xmath111 & @xmath112 & @xmath113 + & [ days ] & [ @xmath7 ] & & & & [ days ] & [ @xmath7 ] & & & [ @xmath71 ] & + + + + + 1 & 1.02 & 1.0 & 10.004 & 10.197 & 0.38 & 0.421 & 0.172 & 0.112 & 0.150 & -2.299 & 3.920 + 2 & 1.05 & 1.0 & 9.948 & 10.159 & 0.39 & 0.554 & 0.175 & 0.115 & 0.152 & -2.232 & 3.936 + 3 & 1.10 & 1.0 & 9.867 & 10.108 & 0.39 & 0.708 & 0.178 & 0.119 & 0.158 & -2.139 & 3.959 + 4 & 1.30 & 1.0 & 9.685 & 10.008 & 0.40 & 1.180 & 0.187 & 0.134 & 0.169 & -1.944 & 4.008 + 5 & 1.50 & 1.0 & 9.582 & 9.958 & 0.40 & 1.584 & 0.192 & 0.143 & 0.177 & -1.843 & 4.035 + 6 & 2.00 & 1.0 & 9.399 & 9.885 & 0.41 & 2.614 & 0.203 & 0.160 & 0.189 & -1.673 & 4.080 + 7 & 2.50 h & 1.0 & 9.211 & 9.831 & 0.43 & 4.275 & 0.213 & 0.176 & 0.201 & -1.499 & 4.125 + 8 & 3.00 h & 1.0 & 9.107 & 9.809 & 0.44 & 5.498 & 0.219 & 0.185 & 0.207 & -1.403 & 4.142 + + 9 & 0.70 & 1.5 & 9.479 & 9.645 & 0.43 & 1.591 & 0.191 & 0.137 & 0.175 & -1.659 & 4.066 + 10 & 0.80 & 1.5 & 9.122 & 9.474 & 0.44 & 2.092 & 0.199 & 0.146 & 0.184 & -1.364 & 4.129 + 11 & 0.90 & 1.5 & 9.061 & 9.392 & 0.44 & 2.450 & 0.204 & 0.154 & 0.190 & -1.201 & 4.165 + 12 & 1.20 h * & 1.5 & 8.845 & 9.304 & 0.46 & 3.409 & 0.213/ & 0.169 & 0.200 & -0.986 & 4.210 + & & & & & & & 0.212 & & & & + 13 & 1.50 h * & 1.5 & 8.788 & 9.286 & 0.48 & 4.280 & 0.217/ & 0.178 & 0.206 & -0.929 & 4.224 + & & & & & & & 0.217 & & & & + 14 & 1.80 h * & 1.5 & 8.753 & 9.277 & 0.49 & 5.105 & 0.221/ & 0.185 & 0.210 & -0.925 & 4.230 + & & & & & & & 0.221 & & & & + 15 & 2.10 h * & 1.5 & 8.766 & 9.283 & 0.49 & 5.866 & 0.225/ & 0.192 & 0.214 & -0.957 & 4.229 + & & & & & & & 0.224 & & & & + 16 & 2.50 h * & 1.5 & 8.742 & 9.278 & 0.48 & 6.831 & 0.229/ & 0.197 & 0.219 & -0.976 & 4.232 + & & & & & & & 0.228 & & & & + 17 & 3.00 h * & 1.5 & 8.665 & 9.259 & 0.49 & 7.888 & 0.232/ & 0.203 & 0.223 & -1.010 & 4.231 + & & & & & & & 0.231 & & & & + + + + + 18 & 1.30 & 1.0 & 10.212 & 10.388 & 0.38 & 0.366 & 0.163 & 0.120 & 0.143 & -2.601 & 3.847 + 19 & 1.35 & 1.0 & 9.907 & 10.316 & 0.39 & 0.605 & 0.168 & 0.127 & 0.150 & -2.477 & 3.877 + 20 & 1.45 & 1.0 & 9.886 & 10.193 & 0.40 & 1.092 & 0.177 & 0.139 & 0.161 & -2.231 & 3.934 + 21 & 1.65 & 1.0 & 9.661 & 10.094 & 0.42 & 1.945 & 0.188 & 0.154 & 0.175 & -1.995 & 3.993 + 22 & 2.00 h * & 1.0 & 9.490 & 10.037 & 0.43 & 2.936 & 0.197/ & 0.166 & 0.185 & -1.829 & 4.035 + & & & & & & & 0.196 & & & & + 23 & 2.50 h * & 1.0 & 9.165 & 9.967 & 0.45 & 4.272 & 0.205/ & 0.173 & 0.194 & -1.606 & 4.085 + & & & & & & & 0.203 & & & & + 24 & 3.00 h * & 1.0 & 9.152 & 9.965 & 0.45 & 5.546 & 0.211/ & 0.184 & 0.201 & -1.546 & 4.104 + & & & & & & & 0.209 & & & & + + 25 & 0.90 & 1.5 & 9.902 & 10.007 & 0.44 & 1.075 & 0.174 & 0.132 & 0.156 & -2.168 & 3.937 + 26 & 1.05 & 1.5 & 9.650 & 9.810 & 0.47 & 1.855 & 0.186 & 0.148 & 0.172 & -1.872 & 4.008 + 27 & 1.10 & 1.5 & 9.596 & 9.772 & 0.46 & 2.032 & 0.188 & 0.152 & 0.175 & -1.832 & 4.020 + 28 & 1.20 & 1.5 & 9.504 & 9.710 & 0.47 & 2.378 & 0.192 & 0.157 & 0.180 & -1.741 & 4.042 + 29 & 1.50
h * & 1.5 & 9.368 & 9.629 & 0.47 & 3.152 & 0.200/ & 0.169 & 0.188 & -1.645 & 4.069 + & & & & & & & 0.199 & & & & + 30 & 2.00 h * & 1.5 & 9.273 & 9.578 & 0.48 & 4.153 & 0.206/ & 0.178 & 0.195 & -1.572 & 4.091 + & & & & & & & 0.205 & & & & + 31 & 2.50 h * & 1.5 & 9.111 & 9.505 & 0.49 & 5.091 & 0.211/ & 0.185 & 0.201 & -1.475 & 4.114 + & & & & & & & 0.209 & & & & + 32 & 3.00 h * & 1.5 & 9.091 & 9.501 & 0.50 & 7.896 & 0.221/ & 0.197 & 0.213 & -1.455 & 4.130 + & & & & & & & 0.221 & & & & + clrrrrrlrrcc + + + model & @xmath107 & & @xmath108 & & @xmath109 & & & @xmath110 & @xmath111 & @xmath112 & @xmath113 + & [ days ] & [ @xmath7 ] & & & & [ days ] & [ @xmath7 ] & & & [ @xmath71 ] & + + + + + 33 & 1.20 & 1.0 & 10.237 & 10.467 & 0.37 & 0.416 & 0.162 & 0.128 & 0.145 & -2.673 & 3.830 + 34 & 1.50 & 1.0 & 9.850 & 10.277 & 0.41 & 1.489 & 0.179 & 0.149 & 0.166 & -2.245 & 3.929 + 35 & 2.00 h * & 1.0 & 9.582 & 10.193 & 0.43 & 2.912 & 0.192/ & 0.164 & 0.181 & -1.978 & 3.995 + & & & & & & & 0.191 & & & & + 36 & 2.50 h * & 1.0 & 9.420 & 10.159 & 0.45 & 4.242 & 0.200/ & 0.175 & 0.190 & -1.819 & 4.033 + & & & & & & & 0.199 & & & & + 37 & 3.00 h * & 1.0 & 9.299 & 10.139 & 0.47 & 5.551 & 0.206/ & 0.182 & 0.196 & -1.698 & 4.062 + & & & & & & & 0.205 & & & & + + 38 & 1.20 & 1.5 & 10.168 & 10.255 & 0.44 & 0.736 & 0.170 & 0.141 & 0.155 & -2.570 & 3.854
+ 39 & 1.50 & 1.5 & 9.835 & 9.987 & 0.44 & 1.737 & 0.183 & 0.154 & 0.171 & -2.198 & 3.939 + 40 & 2.00 h * & 1.5 & 9.448 & 9.747 & 0.48 & 4.230 & 0.203/ & 0.179 & 0.193 & -1.831 & 4.032 + & & & & & & & 0.202 & & & & + 41 & 2.50 h * & 1.5 & 9.295 & 9.677 & 0.49 & 5.910 & 0.210/ & 0.188 & 0.202 & -1.690 & 4.067 + & & & & & & & 0.209 & & & & + 42 & 3.00 h * & 1.5 & 9.074 & 9.599 & 0.52 & 7.686 & 0.216/ & 0.196 & 0.209 & -1.537 & 4.100 + & & & & & & & 0.215 & & & & + + + + + 43 & 1.15 & 1.0 & 10.287 & 10.553 & 0.37 & 0.305 & 0.160 & 0.130 & 0.143 & -2.753 & 3.809 + 44 & 1.30 & 1.0 & 10.104 & 10.462 & 0.38 & 0.882 & 0.169 & 0.140 & 0.156 & -2.562 & 3.856 + 45 & 1.50 & 1.0 & 9.909 & 10.384 & 0.40 & 1.488 & 0.177 & 0.149 & 0.166 & -2.343 & 3.906 + 46 & 1.65 & 1.0 & 9.809 & 10.353 & 0.41 & 1.884 & 0.182 & 0.156 & 0.171 & -2.242 & 3.930 + 47 & 1.80
h * & 1.0 & 9.693 & 10.323 & 0.41 & 2.303 & 0.185/ & 0.160 & 0.175 & -2.149 & 3.950 + & & & & & & & 0.184 & & & & + 48 & 2.50 h * & 1.0 & 9.474 & 10.281 & 0.45 & 4.222 & 0.197/ & 0.175 & 0.188 & -1.912 & 4.008 + & & & & & & & 0.196 & & & & + 49 & 3.00 h * & 1.0 & 9.366 & 10.265 & 0.47 & 5.541 & 0.203/ & 0.181 & 0.195 & -1.798 & 4.035 + & & & & & & & 0.202 & & & & + + 50 & 1.35 & 1.5 & 10.264 & 10.353 & 0.42 & 0.497 & 0.165 & 0.137 & 0.150 & -2.694 & 3.822 + 51 & 1.50 & 1.5 & 10.045 & 10.173 & 0.42 & 1.190 & 0.174 & 0.146 & 0.161 & -2.464 & 3.876 + 52 & 1.70 & 1.5 & 9.829 & 10.015 & 0.42 & 1.968 & 0.184 & 0.156 & 0.173 & -2.246 & 3.929 + 53 & 1.80 h * & 1.5 & 9.580 & 9.873 & 0.46 & 3.380 & 0.196/ & 0.174 & 0.187 & -2.020 & 3.984 + & & & & & & & 0.194 & & & & + 54 & 2.50 h * & 1.5 & 9.353 & 9.772 & 0.49 & 5.850 & 0.208/ & 0.188 & 0.200 & -1.801 & 4.039 + & & & & & & & 0.207 & & & & + 55 & 3.00 h * & 1.5 & 9.151 & 9.705 & 0.51 & 7.671 & 0.214/ & 0.195 & 0.207 & -1.651 & 4.073 + & & & & & & & 0.213 & & & & + listed are : + @xmath116 is initial orbital period of the system ( at the beginning of mass transfer ) + @xmath117 is the mass of the progenitor of white dwarf + @xmath57 is duration of the cooling phase of a white dwarf starting at the end of rlof + @xmath118 is total evolution time + @xmath119 is the final surface hydrogen content + @xmath120 is final orbital period at the moment of shrinking of the donor within its roche lobe + @xmath121 is final wd mass + @xmath122 is the mass of the helium core at the moment of shrinking of the donor within its roche lobe + @xmath123 is the final mass of helium core after the central temperature has decreased by 50% of its maximum value + @xmath124 is the final luminosity + @xmath125 is the final effective temperature .
+ h - hydrogen flashes without rlof + h * - hydrogen flashes with rlof + lcccccc + + @xmath126 & @xmath127 & @xmath128 & @xmath128 & log @xmath129 & @xmath128 & log @xmath130 + + 0.155 & 0.0218 & 2.100 & - & 6.31 & 1.351 & 6.69 + 0.180 & 0.0208 & 1.594 & 1.687 & 6.65 & 1.300 & 6.82 + 0.206 & 0.0198 & 1.469 & 1.476 & 6.83 & 1.236 & 7.00 + 0.296@xmath131 & 0.0173 & 1.224 & 1.220 & 7.26 & 1.111 & 7.36 + the first two columns present the zero temperature m
r relation for a helium wd obtained by hamada & salpeter ( 1961 ) .
the third and fifth columns display our calculations of the stellar radius and gravity , while fourth and fifth the dsbh98 calculations , respectively .
the last two columns illustrate the same quantities taken from the cooling tracks produced by wood ( 1990 ) for carbon wds with thick hydrogen envelopes .
the stellar radius is calculated at t = 8500 k and is normalized by the zero temperature radius . + * the last two values in this row are taken from it86 .
ccrrrrccrrr + + + model & case & & & & & @xmath134 & @xmath135 & @xmath136 & @xmath137 & + & & [ yrs ] & [ yrs ] & [ yrs ] & [ yrs ] & & [ l=@xmath138 & & & + + + 7 & 1 & 6.377 & & 6.349 & 6.595 & 0.526 & 4.349 & & & + & 1 & 6.852 & & 6.599 & 6.450 & 1.525 & 4.117 & & & + & 1 & 7.095 & & 6.609 & 6.584 & 1.603 & 4.073 & & & + & 1 & 7.346 & & 6.598 & & 1.693 & 4.011 & & & + + 8 & 1 & 6.453 & & 6.509 & 6.520 & 1.571 & 4.147 & & & + & 1 & 6.906 & & 6.580 & 6.553 & 1.658 & 4.084 & & & + & 1 & 7.183 & & 6.596 & & 1.740 & 4.037 & & & + + 12 & 1 & 6.634 & & 6.551 & 6.484 & 0.912 & 4.330 & & & + & 1 & 6.907 & & 6.608 & 6.449 & 1.548 & 4.084 & & & + & 1 & 7.139 & & 6.601 & 6.649 & 1.619 & 4.044 & & & + & 2 & 7.367 & 2.673 & 6.456 & & 1.703 & 4.015 & 0.0129 & 0.0123 & 1.7 + + 13 & 1 & 6.717 & & 6.590 & 6.474 & 1.577 & 4.099 & & & + & 1 & 6.971 & & 6.612 & 6.574 & 1.646 & 4.053 & & & + & 2 & 7.215 & 2.192 & 6.571 & & 1.727 & 3.979 & 0.0119 & 0.0118 & 0.2 + + 14 & 1 & 6.372 & & 6.519 & 6.473 & 1.588 & 4.119 & & & + & 1 & 6.861 & & 6.585 & 6.475 & 1.670 & 4.078 & & & + & 1 & 7.141 & & 6.588 & 7.603 & 1.751 & 3.994 & & & + & 2 & 6.435 & 2.593 & 6.258 & & 1.843 & 3.982 & 0.0113 & 0.0105 & 3.5 + + 15 & 1 & 6.538 & & 6.558 & 6.443 & 1.669 & 4.099 & & & + & 1 & 6.975 & & 6.567 & 6.516 & 1.748 & 4.038 & & & + & 2 & 7.283 & 2.433 & 6.450 & & 1.842 & 3.963 & 0.0108 & 0.0103 & 1.2 + + 16 & 1 & 6.643 & & 6.551 & 6.462 & 1.757 & 4.082 & & & + & 1 & 7.091 & & 6.553 & 6.622 & 1.846 & 4.016 & & & + & 2 & 7.501 & 2.513 & 6.172 & & 1.965 & 3.983 & 0.0100 & 0.0095 & 4.4 + + 17 & 1 & 5.774 & & 6.646 & 6.510 & 1.829 & 4.047 & & & + & 2 & 7.202 & 2.272 & 6.477 & 6.723 & 1.933 & 3.940 & 0.0120 & 0.0094 & 0.6 + & 2 & 7.688 & 2.389 & 5.931 & & 2.094 & 3.983 & 0.0094 & 0.0082 & 8.2 + + 22 & 1 & 6.495 & & 6.513 & 6.674 & 0.312 & 4.253 & & & + & 2 & 7.029 & 3.154 & 6.113 & & 1.311 & 3.938 & 0.0121 & 0.0116 & 4.7 + + 23 & 2 & 6.998 & 2.994 & 6.124 & 6.884 & 1.459 & 3.922 & 0.0115 & 0.0105 & 4.9 + & 2 & 7.902 & 2.806 & 5.915 & & 1.679 & 3.975 & 0.0105 & 0.0088 & 14.1 + + 24 & 2 & 6.823 & 2.914 & 6.171 & 6.801 & 1.546 & 3.914 & 0.0103 & 0.0097 & 3.9 + & 2 & 7.623 & 2.769 & 5.966 & & 1.695 & 3.939 & 0.0097 & 0.0086 & 10.4 + + 29 & 1 & 6.662 & & 6.583 & 6.634 & 0.371 & 4.246 & & & + & 2 & 6.971 & 3.177 & 5.984 & & 1.307 & 3.293 & 0.0120 & 0.0108 & 6.2 + + 30 & 2 & 6.880 & 3.066 & 5.967 & & 1.412 & 3.918 & 0.0112 & 0.0100 & 5.7 + + 31 & 2 & 6.652 & 2.994 & 6.011 & 6.793 & 1.470 & 3.906 & 0.0103 & 0.0095 & 4.3 + & 2 & 7.666 & 2.843 & 5.899 & & 1.641 & 3.955 & 0.0095 & 0.0083 & 11.6 + + 32 & 2 & 7.005 & 2.779 & 5.825 & & 1.682 & 3.900 & 0.0090 & 0.0078 & 7.9 + ccrrrrccrrr + + + model & case & & & & & @xmath134 & @xmath135 & @xmath136 & @xmath137 & + & & [ yrs ] & [ yrs ] & [ yrs ] & [ yrs ] & & [ l=@xmath138 & & & + + + 35 & 1 & 6.510 & & 6.425 & 6.799 & -0.068 & 4.205 & & & + & 2 & 6.864 & 3.290 & 5.969 & & 1.173 & 3.911 & 0.0111 & 0.0102 & 7.4 + + 36 & 2 & 6.720 & 3.158 & 5.850 & & 1.313 & 3.897 & 0.0102 & 0.0089 & 8.6 + + 37 & 2 & 6.885 & 2.999 & 5.810 & & 1.419 & 3.877 & 0.0097 & 0.0082 & 10.5 + + 40 & 2 & 6.622 & 3.168 & 5.549 & & 1.275 & 3.883 & 0.0098 & 0.0084 & 9.2 + + 41 & 2 & 6.876 & 2.981 & 5.719 & & 1.427 & 3.875 & 0.0087 & 0.0076 & 10.2 + + 42 & 2 & 7.362 & 2.744 & 5.737 & & 1.615 & 3.876 & 0.0079 & 0.0063 & 11.6 + + 47 & 1 & 7.373 & & 6.594 & 6.966 & 0.208 & 4.151 & & & + & 2 & 6.095 & 3.388 & 5.938 & & 1.072 & 3.926 & 0.0109 & 0.0095 & 12.7 + + 48 & 2 & 6.809 & 3.200 & 5.820 & & 1.239 & 3.864 & 0.0094 & 0.0080 & 11.1 + + 49 & 2 & 6.871 & 3.081 & 5.836 & & 1.317 & 3.863 & 0.0089 & 0.0074 & 9.9 + + 53 & 1 & 6.471 & & 6.690 & 6.819 & 0.190 & 4.170 & & & + & 2 & 6.920 & 3.239 & 6.785 & & 1.160 & 3.889 & 0.0094 & 0.0078 & 12.4 + + 54 & 2 & 6.927 & 2.962 & 5.722 & & 1.355 & 3.855 & 0.0081 & 0.0066 & 11.1 + + 55 & 2 & 7.180 & 2.785 & 5.734 & & 1.521 & 3.849 & 0.0074 & 0.0058 & 11.8 + listed are : + number of model ( table 1 ) + number of case ( 1 or 2 ) + @xmath62 and @xmath65 are the rise and decay times responsively + @xmath139 and @xmath140 are recurrence time between two successful flashes and duration of accretion phase during the flash + @xmath105 is effective temperature when the luminosity has its maximum value @xmath141 + @xmath136 and @xmath137 are the envelope masses before and after flash + @xmath142 is accreted mass + @xmath136-@xmath137= @xmath143+@xmath142 where @xmath143 is the increase of the helium core mass during the flash . | we present a detailed calculation of the evolution of low mass ( @xmath0 ) helium white dwarfs .
these white dwarfs ( the optical companions to binary millisecond pulsars ) are formed via long term , low mass binary evolution .
after detachment from the roche lobe , the hot helium cores have a rather thick hydrogen layer with mass between 0.01 to 0.06@xmath1 . due to mixing between the core and outer envelope ,
the surface hydrogen content is 0.5 to 0.35 , depending on the initial value of the heavy element ( z ) and the initial secondary mass .
we found that the majority of our computed models experience one or two hydrogen shell flashes .
we found that the mass of the helium dwarf in which the hydrogen shell flash occurs depends on the chemical composition .
the minimum helium white dwarf mass in which a hydrogen flash takes place is 0.213@xmath1 ( z=0.003 ) , 0.198@xmath1 ( z=0.01 ) , 0.192@xmath1 ( z=0.02 ) or 0.183@xmath1 ( z=0.03 ) .
the duration of the flashes ( independent of chemical composition ) is between few @xmath2 years to few @xmath3 years . in several flashes the white dwarf radius will increase so much that it forces the model to fill its roche lobe again .
our calculations show that cooling history of the helium white dwarf depends dramatically on the thickness of the hydrogen layer .
we show that the transition from a cooling white dwarf with a temporary stable hydrogen burning shell to a cooling white dwarf in which almost all residual hydrogen is lost in a few thermal flashes ( via roche lobe overflow ) occurs between 0.1830.213@xmath1 ( depending on the heavy element value ) .
epsf -0.5 in binaries : close binaries : general stars : mass loss evolution stars : millisecond binary pulsars pulsars : individual : psr j0437 + 4715 pulsars : individual : psr j1012 + 5307 |
for the dft calculations , we have employed the all - electron flapw band method implemented in wien2k @xcite .
we used 17 @xmath60 17 @xmath60 17 * @xmath61 * -point mesh in the full brillouin zone .
the muffin - tin radii @xmath62 were set to 2.50 and 2.30 a.u . for sm and s , respectively , and the product of @xmath62 and the maximum reciprocal lattice vector @xmath63 was chosen as @xmath64 .
we used the maximum @xmath65 value of 10 for the waves inside the atomic spheres and the largest reciprocal lattice vector @xmath66 of 12 in the charge fourier expansion .
figure [ black - dft ] shows the band structures and densities of states ( doss ) of a black phase of sms ( b - sms ) obtained by the dft schemes .
figure [ black - dft](a ) shows symmetry decomposed band structure of b - sms in the gga + soc scheme ( soc : spin - orbit coupling ) .
since sm 4@xmath4-electrons feel much larger soc than the cubic crystal field , the soc bases incorporating the cubic crystal field should be utilized @xcite .
the cubic crystal field splits @xmath14 states at @xmath19 into lower @xmath67 doublet and higher @xmath68 quartet , as in smb@xmath2 @xcite .
sm 4@xmath4 bands are dominant near the fermi level ( @xmath6 ) , and @xmath14 and @xmath15 states are located mainly below and above @xmath6 , respectively .
the splitting between @xmath14 and @xmath15 states is about 0.75 ev , which is related to the strength of the soc of 4@xmath4 electrons in sm ion .
in fact , most of @xmath15 states are to be shifted up in energy due to the strong correlation effect of 4@xmath4-electrons .
sm @xmath17 @xmath33 characters are also shown in order to visualize the hybridization between the @xmath67 doublet and @xmath33 band . since the lobes of @xmath69 and @xmath33 orbitals of sm
@xmath17 electrons are along and away from the anion sulfur ions , respectively , the energy of @xmath33 band is to be lower than that of @xmath69 band .
this feature in sms is contrary to that in smb@xmath2 , for which the @xmath69 band is lower in energy than the @xmath33 band because the lobes of @xmath33 orbitals are toward anion boron ions @xcite .
therefore , in smb@xmath2 , the @xmath4-@xmath30 hybridization occurs between the @xmath67 doublet and @xmath69 band , while , in sms , it occurs between the @xmath67 doublet and @xmath33 band . at @xmath21
, the @xmath33 band is located below @xmath14 bands , and so the order of parities is changed .
the valence state of sm in the gga + soc scheme is estimated to be 2.56 + , which is far from the semiconducting black phase of the valence state of sm , 2 + .
furthermore , the band gap is not obtained which is contrary to the experiment .
figure [ black - dft](b ) presents the symmetry - projected partial doss ( pdoss ) of b - sms in the gga + soc scheme .
sm @xmath14 dos is projected into the relativistic double group bases and sm @xmath15 and @xmath17 @xmath33 pdoss are also provided .
it is seen that the dos at @xmath6 is not zero , implying that the system is not an insulator .
rather it is semimetallic , as shown in fig .
[ black - dft](a ) . figure [ black - dft](c ) and ( d ) show electronic structures of b - sms in the gga + soc + @xmath7 scheme ( @xmath7 : coulomb correlation of sm @xmath0 electrons ) .
when @xmath70 ev , it gives an insulating phase with the band gap of 87 mev , which is quite comparable to the experimental band gap of @xmath30.15 ev @xcite .
once the semiconducting phase of sms is obtained , the parity inversion does not happen , which results in the trivial @xmath71 topological number .
the valence state of sm is estimated to be 2.23 + , which is close to 2 + . however , the band symmetries at high symmetry * @xmath61 * -points are different from those in the gga + soc scheme , which turns out to be wrong in view of the dmft results ( see fig .
2(c ) in the main text ) . figure [ black - dft ] ( e ) and ( f ) show symmetry decomposed band structure and pdos of b - sms in the gga + soc with 10 times enhanced soc of sm @xmath0-electron ( 10@xmath59-soc scheme ) . due to the artificially large soc strength ,
the @xmath15 bands are shifted up to the higher energy side .
in contrast to the gga + soc + @xmath7 scheme , the band symmetries at high symmetry * @xmath61 * -point in the 10@xmath59-soc scheme are consistent with those of the gga + soc scheme . sm @xmath17
@xmath33 band hybridizes with @xmath67 doublet , resulting in the hybridization gap near @xmath21 .
the valence state of sm is estimated to be 2.27 + .
thus the dft schemes can not describe the ground state electronic structure of b - sms properly .
also , according to the angle - resolved photoemission spectroscopy ( arpes ) experiment @xcite , sm @xmath0 bands are very flat and distributed broadly from 1 to 4 ev below @xmath6 , which is different from the dft results .
this is due to the strong correlation effect in @xmath0 electrons , which can not be captured in conventional dft calculations .
figure [ black - dmft ] shows the band structure and dos of b - sms ( @xmath9 = 5.9 @xmath10 ) obtained by the charge self - consistent dft + dmft scheme at @xmath72 k. it is seen that the insulating electronic structure is correctly described by the dmft scheme .
the obtained band gap and the valence configuration of sm are 0.46 ev and 2.06 + , respectively .
the band structure in fig .
[ black - dmft ] is very similar to that obtained by the periodic anderson model calculation @xcite , and consistent with arpes data @xcite .
the partial dos in fig .
[ black - dmft]b shows 4@xmath4 multiplets , which are in good agreement with xps data @xcite . since the sm @xmath30-band is above the sm @xmath4-band , b - sms is a topologically trivial insulator .
= 5.9 @xmath10 ) at @xmath72 k. on - site coulomb and exchange energies of @xmath73 ev and @xmath74 ev were adopted .
( b ) partial dos of b - sms .
( c ) xps data for b - sms taken from ref .
, width=8 ]
= 5.6 @xmath10 ) in the gga + soc with 10@xmath59-soc scheme . ( a ) symmetry decomposed band structure with the relativistic double group bases .
sm 5@xmath30 @xmath33 band is seen to be hybridized with @xmath67 doublet bands .
the valence state of sm is obtained to be 2.36 + in this case .
( b ) symmetry projected pdos .
sm @xmath13 dos is projected into the double group bases .
sm @xmath17 @xmath33 dos is magnified 50 times for a clear view . , width=8 ] figure [ golden - dft ] provides electronic structures of g - sms ( @xmath9 = 5.6 @xmath10 ) in the gga + soc with 10@xmath59-soc scheme . the electronic structures in fig . [ golden - dft ]
are quite similar to those of g - sms described by the dmft at low @xmath5 ( see fig .
2(c ) in the main text ) .
but , the number of @xmath4-electron ( @xmath26 ) in the dft + soc with 10@xmath59-soc scheme is 5.64 ( sm@xmath75 ) , while @xmath26 in the dmft scheme is 5.27 ( sm@xmath20 ) .
figure [ t - dep ] presents the temperature ( @xmath5)-dependent physical properties of g - sms ( @xmath9 = 5.6 @xmath10 ) described by the dmft .
figure [ t - dep](a ) and ( b ) show the renormalization factor @xmath76 and effective mass @xmath77 versus @xmath5 . upon
cooling , @xmath76 of @xmath13 ( @xmath24 ) increases ( decreases ) , resulting in decreasing ( increasing ) @xmath77 of @xmath13 ( @xmath24 ) .
relaxation times of both @xmath13 and @xmath24 diverge upon cooling , as shown in fig .
[ t - dep](c ) , suggesting the crossover to the landau fermi liquid behavior at low @xmath5 .
figure [ t - dep](d ) provides the @xmath5-dependent behavior of the hybridization function at @xmath12 ( @xmath78 ) , which is closely related to the kondo interaction strength .
hence the largest @xmath78 value near @xmath55 k suggests that kondo screening behavior becomes strongly enhanced at this temperature range . on the other hand , the total dos at @xmath12 in fig .
[ t - dep](e ) increases first and then decreases below @xmath79 k. note that the total dos at @xmath12 is saturated at low-@xmath5 , indicating the full lattice coherence below @xmath41 k. fig .
[ t - dep](f ) shows the resistivity versus @xmath5 obtained by the drude model . the resistivity drop for @xmath24
is clearly shown below @xmath41 k , which is also captured in the dmft result ( see fig .
3(c ) in the main text ) .
it implies that the crossover from the kondo liquid to the landau fermi liquid occurs below @xmath41 k. 99 p. blaha , k. schwarz , g. k. h. madsen , d. kavasnicka , and j. luitz , wien2k ( karlheinz schwarz , technische universitat wien , austria , 2001 ) .
r. pappalardo , j. chem . phys . * 34 * , 1380 ( 1961 ) .
kang , j .- w .
kim , k. kim , j .- s .
kang , j. d. denlinger , b. i. min , arxiv:1312.5898 ( 2013 ) . c. m. varma , rev .
phys . * 48 * , 219 ( 1976 ) .
t. ito , a. chainani , h. kumigashira , t. takahashi , and n. k. sato , phys .
b * 65 * , 155202 ( 2002 ) . c. lehner , m. richter , and h. eschrig , phys . rev .
b * 58 * , 6807 ( 1998 ) .
m. campagna , e. bucher , g. k. wertheim , and l. d. longinotti , phys .
rev . lett . * 33 * , 165 ( 1974 ) . | we have investigated temperature - dependent behaviors of electronic structure and resistivity in a mixed - valent golden phase of sms , based on the dynamical mean - field theory band structure calculations . upon cooling ,
the coherent sm @xmath0 bands are formed to produce the hybridization - induced pseudogap near the fermi level , and accordingly the topology of fermi surface is changed to exhibit a lifshitz - like transition .
the surface states emerging in the bulk gap region are found to be not topologically protected states but just typical rashba spin - polarized states , indicating that sms is not a topological kondo semimetal . from the analysis of anomalous resistivity behavior in sms , we have identified universal energy scales , which characterize the kondo / mixed - valent semimetallic systems .
kondo and mixed - valent physics in strongly - correlated @xmath0 electron systems have been subject of longstanding controversy .
it includes various interesting phenomena such as @xmath1-wave superconductivity and non - fermi - liquid behavior in heavy - fermion systems @xcite , and recently proposed topological kondo insulator behavior in a typical mixed - valent insulator smb@xmath2 @xcite .
the topological kondo insulator of our present interest has attracted a great deal of recent attention . however , the realization of the topological properties in smb@xmath2 is still under debate .
the immediate question was addressed whether a similar mixed - valent system , sms , also has the topological properties or not .
indeed , the mixed - valent golden phase of sms ( g - sms ) was reported to be a topological kondo semimetal @xcite .
sms has been studied for last four decades @xcite , but there remain several issues still unresolved .
sms crystallizes in a face - centered cubic ( fcc ) structure of rock - salt ( nacl - type ) type . at the ambient pressure
, sms has a so - called black phase , which is a semiconductor with indirect and direct band gaps of 90 mev and 0.4 ev , respectively @xcite . in the black phase of sms ( b - sms ) , the valence state of sm is divalent ( 2 + ) , and so the system is nonmagnetic . under high pressure above 6.5 kbar
, sms undergoes a first - order isostructural phase transition from b - sms to g - sms , in which sm ions are mixed - valent with the average valency of 2.6 + @xmath3 2.8 + @xcite .
this isostructural transition is accompanied by the volume collapse as much as 15% @xcite , as shown in fig .
[ struct](b ) .
the energy gap decreases monotonically with increasing the pressure .
it is controversial whether g - sms has a real gap or a pseudo - gap @xcite .
the resistivity behavior of g - sms is quite anomalous in the sense that the overall behavior is kondo - lattice like but it exhibits a couple of abnormal hump structures @xcite .
applying the pressure further , g - sms has a magnetic instability at about 19.5 kbar with the antiferromagnetic order @xcite . above 19.5 kbar
, the resistivity shows a metallic behavior and the valence state of sm increases toward 3 + @xcite .
there have been several reports on the density - functional theory ( dft)-based band structure study of sms @xcite .
but even the ground - state insulating nature of b - sms is not properly described by the dft - based schemes , and so the advanced methods like the dynamical mean - field theory ( dmft ) should be employed to investigate the topological properties in g - sms that has the strongly - correlated 4@xmath4 electrons .
here we have investigated electronic structures of sms , based on the dmft scheme .
first , we have shown that the electronic properties of b - sms are described properly only by the dmft scheme .
then we have examined the @xmath5-dependent electronic structure evolution in g - sms . upon cooling ,
the @xmath0 states form the coherent @xmath0 bands with a pseudo - gap feature near the fermi level ( @xmath6 ) , and accordingly the topology of the fermi surface ( fs ) is changed , which is reflected well in the anomalous resistivity behavior in g - sms .
we have demonstrated that the surface states realized in g - sms are not the topological states , but are just the typical rashba states .
we have employed the all - electron flapw band method implemented in wien2k @xcite .
we have checked that both the dft and the dft+@xmath7 ( on - site coulomb @xmath7 ) schemes can not describe the ground state insulating electronic structure of sms properly ( see the supplement @xcite ) .
therefore , we have employed the combined dft and dmft ( dft+dmft ) approach implemented in wien2k , which has successfully reproduced many aspects of the strongly - correlated electron systems @xcite .
we used projectors in the large window of 10 ev , and the on - site coulomb and exchange energies of @xmath7 = 6.1 ev , @xmath8 = 0.8355 ev were adopted to fit in x - ray photoemission spectroscopy ( xps ) data ( see fig .
s2 for the dmft band structure of b - sms in the supplement ) @xcite . to solve the impurity problem ,
the non - crossing approximation ( nca ) is used @xcite . to verify the justification of the nca scheme , we have also used the continuous time quantum monte carlo scheme for b - sms and checked that two schemes give the same result . the dmft band structure and dos of g - sms ( @xmath9 = 5.6 @xmath10 ) at @xmath11 k are shown in fig .
[ golden - dmft](a ) .
notable is the flat sm @xmath0 bands near @xmath6 , which yield sharp kondo resonance - like peaks in the dos ( see fig .
[ nf - t](a ) ) .
the @xmath0-band in the vicinity of @xmath12 is mainly of @xmath13 character ( @xmath14 ) , while @xmath15 bands are located at about @xmath160.17 ev from @xmath12 .
these coherent sm @xmath0 bands hybridize strongly with sm @xmath17 band to produce the hybridization gap near @xmath6 ( see fig .
[ golden - dmft](c ) ) . however , the @xmath14 ( @xmath18 ) band at @xmath19 is above @xmath6 , and so the band structure exhibits the metallic nature having the mixed - valent state of sm@xmath20 .
the @xmath18 band is dispersive with the band width of about 0.03 ev , as in smb@xmath2 @xcite .
the dmft band structure near @xmath12 is analogous to the band structure obtained by the dft+soc ( soc : spin - orbit coupling ) , as shown in the supplement @xcite , but the band width of the former is nearly ten times smaller than that of the latter , which is also similar to the case in smb@xmath2 @xcite . even though g - sms has metallic nature , the band inversion occurs at @xmath21 , and so it is tempting to anticipate the topologically protected surface states in sms @xcite . to check the topological property ,
we have examined the surface band structure in fig . [
golden - dmft](b ) , which was obtained by the model slab calculation based on the dft bulk band structure @xcite .
the surface states , however , have a tiny gap instead of a dirac cone and the momentum - dependent splitting of spin states , which reveals that they are not topologically protected states but just rashba spin - polarized surface states .
this result is a contrast to that of li _ et .
al . _
@xcite , who argued that g - sms is to be a topological semimetal .
the charge gap protection is essential to have the topological nature . in this respect ,
la - doped sms can be a better candidate for a topological kondo insulator , because sm@xmath22la@xmath23s under pressure was reported to be an excitonic insulator with an energy gap of 1 mev @xcite .
-dependent behaviors in g - sms ( @xmath9 = 5.6 @xmath10 ) .
( a ) @xmath5-dependent dos variation in the dmft scheme . upon cooling , the @xmath13 dos behaves like a kondo resonance with the pseudo - gap like dip feature near @xmath12 .
the @xmath24 dos peak near @xmath25 ev is also shown .
( b ) the number of occupied @xmath4-electrons ( @xmath26 ) as a function of the logarithmic @xmath5 scale . upon cooling
, @xmath26 increases monotonically from trivalent state to the mixed - valent state .
the effective valence - transition ( vt ) occurs at @xmath27 k , as indicated by the peak in @xmath28 .
( c ) resistivity vs. @xmath5 obtained by the dmft scheme .
the experimental data are taken from ref .
@xcite , which are shifted by 9.3 k toward higher @xmath5 . upon cooling
, sms undergoes several crossover transitions from valence - fluctuation liquid ( vfl ) , kondo - liquid ( kl ) , pseudogap ( pg ) , to landau fermi - liquid ( lfl ) .
, width=8 ] figure [ golden - dmft](c ) shows the @xmath5-dependent band structures of g - sms ( @xmath9 = 5.6 @xmath10 ) . at @xmath29
k , only the sm @xmath30-band is seen to cut @xmath12 . upon cooling ,
sm @xmath0 spectra start to emerge near @xmath12 at @xmath27 k , and @xmath14 bands are observed to be formed below @xmath31 k. at @xmath32 k , the @xmath14 hole bands near @xmath19 ( @xmath18 quartet ) forms the coherent band , and the hybridization gap between sm @xmath17 ( @xmath33 ) and @xmath14 ( @xmath34 ) bands begins to appear at the crossing point near @xmath19 .
the coherence temperature @xmath35 is usually defined in metallic kondo lattices as an onset @xmath5 , at which the hybridization gap appears @xcite . note that @xmath35 is equivalent to @xmath36 that is introduced in the two - fluid model for the kondo lattice @xcite .
hence @xmath32 k is considered to be @xmath35 of g - sms . in the kondo lattice systems ,
@xmath35 can often be identified from the peak position of the resistivity , as will be discussed in fig .
[ nf - t](c ) . with decreasing @xmath5
further , the separation between the upper @xmath14 flat band ( @xmath37 doublet ) along @xmath38 and the lower @xmath14 flat band ( @xmath37/@xmath39 quartet ) becomes enhanced , and eventually the former is shifted up above @xmath12 at @xmath40 k. below @xmath41 k , it is seen that the @xmath14 band becomes completely coherent over the whole brillouin zone . in this case , the @xmath18 hole band is still above @xmath12 , and so sms at very low @xmath5 would exhibit metallic nature ( see fig .
[ nf - t](c ) ) .
figure [ golden - dmft](d ) presents the @xmath5-dependent fs evolution in g - sms .
also presented is the dft - fs , for comparison , which has @xmath21-centered electron ellipses and @xmath19-centered hole pockets . at @xmath29
k , the dmft - fs arises solely from @xmath30-band , which produces the @xmath21-centered ellipses . upon cooling ,
the ellipses become reduced more and more , and so almost disappear at @xmath11 k. their spectral weights , however , become enhanced due to the contribution from the hybridized @xmath14 band so as to have maximum intensity at @xmath42 k. on the other hand , @xmath19-centered hole pockets originating from @xmath14 ( @xmath18 ) band begin to appear at @xmath42 k , and are clearly manifested below @xmath43 k. therefore , the topology of the fs is to be changed twice , at @xmath42 k and @xmath41 k. the former is due to the emergence of the coherent @xmath0 band formation at @xmath19 , while the latter is due to the separation of the upper @xmath14 flat band ( @xmath34 doublet ) from @xmath6 .
this behavior is reminiscent of the lifshitz transition that is a typical continuous quantum phase transition characterized by the topological change of the fs @xcite .
the feature in sms is quite interesting because a lifshitz - like transition occurs with the variation of @xmath5 .
such a lifshitz - like transition in g - sms could be explored by arpes and de hass - van alphen experiments .
the change of the fs topology is well reflected in the @xmath5-dependent resistivity behavior in fig .
[ nf - t](c ) . upon cooling ,
the @xmath14 dos peak at @xmath44 mev becomes sharper and sharper , as shown in fig .
[ nf - t](a ) , resembling the @xmath5-dependent evolution of kondo resonance .
noteworthy is the pseudo - gap feature near @xmath12 , which arises from the @xmath4-@xmath30 band hybridization below @xmath45 k , which is close to @xmath46 k. it is also seen that , upon cooling , the dos at @xmath12 increases first and then decreases below @xmath27 k @xcite .
the @xmath15 dos near @xmath25 ev , which corresponds to the spin - orbit split side - band , behaves similarly to the @xmath14 dos upon cooling . the number of @xmath4-electrons ( @xmath26 ) in g - sms versus @xmath5 is presented in fig . [ nf - t](b ) . upon cooling
, @xmath26 increases monotonically from @xmath47 ( trivalent state ) at high @xmath5 to @xmath48 ( mixed - valence state ) at low @xmath5 .
the increasing trend of @xmath26 upon cooling has also been observed in smb@xmath49 @xcite .
it is worthwhile to observe that @xmath26 curve has an inflection point at @xmath27 k , which indicates that the effective valence - transition ( vt ) occurs at @xmath27 k. figure [ nf - t](c ) shows the resistivity versus @xmath5 of g - sms evaluated in the dmft scheme @xcite .
the overall behavior is that the resistivity increases upon cooling , but decreases below @xmath11 k , as in metallic kondo lattice systems .
however , the detailed @xmath5-dependent behavior is quite anomalous .
the calculated resistivity has a broad maximum at @xmath27 k , and starts to increase again at @xmath31 k to produce hump and dip structure at @xmath50 k and @xmath51 k , respectively . upon further cooling ,
it exhibits another maximum at @xmath11 k. quite similar behavior is indeed observed in the measured resistivity of g - sms @xcite , as shown in fig .
[ nf - t](c ) , even though the feature at @xmath27 k is not so obvious .
the anomalous behavior of the resistivity is closely correlated with the @xmath5-dependent evolution of electronic structure in fig .
[ golden - dmft](c)-(d ) .
namely , the broad maximum at @xmath27 k is considered to arise from the dynamical valence - fluctuation ( vf ) , as manifested by the effective vt in fig .
[ nf - t](b ) and the incoherent sm @xmath0 spectra in fig .
[ golden - dmft](c ) at @xmath27 k. intriguingly , g - sms has a minimum volume at @xmath27 k @xcite , which signifies the close connection of a broad maximum in the resistivity with the mixed - valent nature of sms .
hence we assign @xmath52 k as @xmath53 .
interestingly , smb@xmath2 also exhibits a similar broad maximum feature in the resistivity @xcite and a minimum volume @xcite near @xmath54 k. the resistivity increases below @xmath31 k , at which the coherent @xmath14-band starts to emerge in fig . [
golden - dmft](c ) .
the resistivity hump at @xmath50 k indicates that @xmath35 is around 26 k. in conventional metallic kondo lattices , the resistivity decreases monotonically to zero below @xmath35 .
the behavior near @xmath35 was explained in the two - fluid model by a crossover from the kondo spin - liquid ( ksl ) to kondo fermi - liquid ( kfl ) @xcite .
we use a term of kondo - liquid ( kl ) in fig .
[ nf - t](c ) to comprise both ksl and kfl phases .
thus g - sms exhibits a crossover from mixed - valence to kondo lattice behavior upon cooling @xcite .
the charge - fluctuation that starts to be effective at @xmath52 k becomes almost frozen at @xmath51 k , as shown in fig .
[ nf - t](b ) .
then , the spin - fluctuation becomes dominating near @xmath55 k , so as to activate the kondo screening .
in fact , the effective hybridization obtained in the dmft becomes the largest near @xmath56 k @xcite .
meanwhile , @xmath51 k of the dip structure coincides with @xmath5 , at which the upper @xmath14 flat band ( @xmath34 doublet ) along @xmath38 is detached from @xmath12 .
thus the resistivity up - turn occurs due to the apparent pseudo - gap ( pg ) feature , and so we assign @xmath40 k as @xmath57 .
actually , this kind of hump and dip structure in the resistivity has been observed in several ce compounds of kondo - insulator type , such as cenisn and cerhsb @xcite .
the third resistivity drop at @xmath11 k occurs due to the complete formation of the coherent @xmath14 band over the whole brillouin zone . below @xmath11
k , the imaginary part of self - energy almost vanishes @xcite , whereby the crossover from the kl to the landau fermi - liquid ( lfl ) takes place @xcite .
so we assign @xmath11 k as @xmath58 .
therefore , sms undergoes several crossover transitions upon cooling , from vf liquid ( vfl ) , kl , pg to lfl , as shown in fig .
[ nf - t](c ) .
the energy scales @xmath53 , @xmath35 , @xmath57 , and @xmath58 , which are identified for sms , are considered to be universal to characterize kondo / mixed - valent semimetallic systems . finally , it should be pointed out that the resistivity behavior in sms is quite different from that in smb@xmath49 that is a candidate of topological kondo insulators .
g - sms is not a kondo insulator but close to a kondo semimetal , so that the conventional metallic resistivity behavior is expected to be realized at very low @xmath5 . in conclusion , in g - sms ,
the coherent sm @xmath0 bands are formed upon cooling to produce the hybridization - induced pseudogap feature near @xmath6 , which is accompanied by a lifshitz - like topological transition in the fs .
g - sms is found to be not a topological kondo semimetal in view of that the in - gap surface states are not topological states but are typical spin - polarized rashba states . from the analysis of @xmath5-dependent resistivity of g - sms ,
we have identified characteristic multiple energy scales , which are expected to govern the physics of kondo / mixed - valent semimetallic systems universally .
helpful discussions with junwon kim , j. h. shim , b. h. kim , ki - seok kim , j. -s .
kang , j. d. denlinger , j. w. allen , and k. sun are greatly appreciated .
this work was supported by the nrf ( no .
2009 - 0079947 and no .
2011 - 0025237 ) , postech bk21plus physics division and bsri grant , and the kisti supercomputing center ( no .
ksc-2013-c3 - 010 ) .
99 g. r. stewart , rev .
mod .
phys . * 56 * , 755 ( 1984 ) .
p. coleman , c. ppin , q. si and r. ramazashvili j. phys . : condens .
matter * 13 * r723 ( 2001 ) .
p. gegenwart , q. si and f. steglich , nature physics * 4 * , 186 ( 2008 ) .
m. dzero , k. sun , v. galitski , and p. coleman , phys .
rev .
lett . * 104 * , 106408 ( 2010 ) .
t. takimoto , j. phys .
. jpn . * 80 * , 123710 ( 2011 ) .
m. dzero , k. sun , p. coleman , and v. galitski , phys .
rev .
b * 85 * , 045130 ( 2012 ) .
c .-
j .
kang , j .- w .
kim , k. kim , j .- s .
kang , j. d. denlinger , b. i. min , arxiv:1312.5898 ( 2013 ) .
j. kim , k. kim , c .- j .
kang , s. kim , h. c. choi , j .- s .
kang , j. d. denlinger , and b. i. min , phys .
rev .
b * 90 * , 075131 ( 2014 ) . n. xu _
et al .
_ , phys .
rev .
b * 88 * , 121102(r ) ( 2013 ) .
m. neupane _
et al . _ , nat .
commun . * 4 * , 2991 ( 2013 ) .
d. j. kim , j. xia , z. fisk , nat .
mater .
13 , 466 ( 2014 ) .
d. j. kim , s. thomas , t. grant , j. botimer , z. fisk , j. xia , sci .
rep .
* 3 * , 3150 ( 2013 ) .
j. d. denlinger , j. w. allen , j .- s .
kang , k. sun , j .- w .
kim , j. h. shim , b. i. min , d .- j .
kim , z. fisk , arxiv:1312.6637 ( 2013 ) .
c .-
h .
min , p. lutz , s. fiedler , b. y. kang , b. k. cho , h .- d .
kim , h. bentmann , and f. reinert , phys .
rev .
lett . * 112 * , 226402 ( 2014 ) .
z. li , j. li , p. blaha , and n. kioussis , phys .
rev .
b * 89 * , 121117(r ) ( 2014 ) . v. n. antonov , b. n. harmon , and a. n. yaresko , phys . rev .
b * 66 * , 165208 ( 2002 ) .
a. svane , v. kanchana , g. vaitheeswaran , g. santi , w. m. temmerman , z. szotek , p. strange , and l. petit , phys .
rev .
b * 71 * , 045119 ( 2005 ) . c. m. varma , rev . mod .
phys . * 48 * , 219 ( 1976 ) .
p. villars and l. d. calvert , _ pearson s handbook of crystallographic data for intermetallic phases _ ( asm international , materials park , 1991 ) .
m. campagna , e. bucher , g. k. wertheim , and l. d. longinotti , phys .
rev .
lett . * 33 * , 165 ( 1974 ) .
a. jayaraman , v. narayanamuri , e. bucher , and r. g. maines , phys .
rev .
lett . * 25 * , 1430 ( 1970 ) .
p. p. deen , d. braithwaite , n. kernavanois , l. paolasini , s. raymond , a. barla , g. lapertot , and j. p. sanchez , phys .
rev .
b * 71 * , 245118 ( 2005 )
. t. ito , a. chainani , h. kumigashira , t. takahashi , and n. k. sato , phys .
rev .
b * 65 * , 155202 ( 2002 ) . t. mizuno , t. iizuka , s. kimura , k. matsubayashi , k. imura , h. s. suzuki , and n. k. sato , j. phys .
. jpn . * 77 * , 113704 ( 2008 ) . m. b. maple and d. wohleben , phys . rev
. lett .
* 27 * , 511 ( 1971 ) .
j. m. d. coey , s. k. ghatak , m. avignon , and f. holtzberg , phys .
rev .
b * 14 * , 3744 ( 1976 ) . k. matsubayashi , k. imura , h. s. suzuki , g. chen , n. mori , t. nishioka , k. deguchi , and n. k. sato , j. phys .
. jpn . * 76 * , 033602 ( 2007 ) .
p. wachter , _ handbook of the physics and chemistry of rare earths _ , edited by k. a. gschneidner , l. eyring , and s. hfner ( north - holland , amsterdam , 1994 ) , vol .
19 , p. 177 .
j. flouquet , y. haga , p. haen , d. braithwaite , g. knebel , s. raymond , s. kambe , j. magn .
magn . mater . *
272 - 276 * , 27 ( 2004 ) .
r. m. martin and j. w. allen , j. appl . phys .
* 50 * , 7561 ( 1979 ) .
s. m. shapiro , r. j. birgeneau , and e. bucher , phys .
rev .
lett . * 34 * , 470 ( 1975 ) .
a. barla _
et al . _ ,
phys .
rev .
lett . * 92 * , 066401 ( 2004 ) .
k. imura , k. matsubayashi , h. s. suzuki , n. kabeya , k. deguchi , and n. k. sato , j. phys .
. jpn . * 78 * , 104602 ( 2009 ) .
f. lapierre , m. ribault , f. holtzberg , and j. flouquet , solid state commun .
* 40 * , 347 ( 1981 )
. m. konczykowski , j. morillo , and j. p. senateur , solid state commun . *
40 * , 517 ( 1981 ) .
m. konczykowski , f. lapierre , p. haen j. magn .
magn . mater . *
47 - 48 * , 274 ( 1985 ) .
k. imura , k. matsubayashi , h. s. suzuki , k. deguchi , n. k. sato , physica b * 404 * , 3028 ( 2009 ) .
p. blaha , k. schwarz , g. k. h. madsen , d. kavasnicka , and j. luitz , wien2k ( karlheinz schwarz , technische universitat wien , austria , 2001 ) .
see the supplemental materials for the dft , dft+@xmath7 , and dmft results for b - sms and g - sms , and @xmath5-dependent dmft physical parameters .
g. kotliar , s. y. savrasov , k. haule , v. s. oudovenko , o. parcollet , and c. a. marianetti , rev .
mod .
phys . * 78 * , 865 ( 2006 ) .
k. haule , c .- h .
yee , and k. kim , phys .
rev .
b * 81 * , 195107 ( 2010 ) . for the model slab calculation ,
the wannier downfolding scheme was employed to fit the bulk band structure obtained by the dft+10@xmath59-soc ( see the supplement ) .
we have also performed the dft slab calculation and confirmed that results from the model and dft slab calculations are essentially the same .
p. wachter , a. jung , and p. steiner , phys .
rev .
b * 51 * , 5542(r ) ( 1995 ) .
h. c. choi , b. i. min , j. h. shim , k. haule , and g. kotliar , phys .
rev .
lett . *
108 * , 016402 ( 2012 ) .
s. nakatsuji , d. pines , and z. fisk , phys .
rev .
lett . * 92 * , 016401 ( 2004 ) ; y .- f .
yang and d. pines , phys .
rev .
lett . * 100 * , 096404 ( 2008 ) ; y .- f .
yang and d. pines , proc .
natl .
acad .
sci . * 109 * , e3060 ( 2012 ) .
i. m. lifshitz , sov .
phys .
jetp * 11 * , 1130 ( 1960 ) .
m. mizumaki , s. tsutsui , and f. iga , j. phys .
. ser . * 176 * , 012034 ( 2009 ) .
k. iwasa , t. tokuyama , m. kohgi , n. k. sato , n. mori , physica b * 359 - 361 * , 148 ( 2005 ) .
a. kebede _
et al . _ , physica b * 223 - 224 * , 256 ( 1996 ) . v. a. trounov _ et al . _ , j. phys . :
condens .
matter * 5 * , 2479 ( 1993 ) .
p. coleman , handbook of magnetism and advanced magnetic materials , edited by helmut kronmuller and stuart parkin .
vol 1 : fundamentals and theory .
john wiley and sons , 95 - 148 ( 2007 ) .
kumar and n. s. vidhyadhiraja , j. phys . : condens .
matter * 23 * , 485601 ( 2011 ) .
t. takabatake _ et al .
_ , phys .
rev .
b * 41 * , 9607(r ) ( 1990 ) . s. k. malik and d. t. adroja , phys . rev .
b * 43 * , 6277(r ) ( 1991 ) . * + topological properties and the dynamical crossover from mixed - valence to kondo - lattice behavior in golden phase of sms * |
the order complex of a geometric lattice is one of many simplicial complexes associated to matroids . for a geometric lattice @xmath2
the order complex of @xmath9 is the simplicial complex whose simplices consist of all chains in @xmath10 the number of flats in each rank , also known as the whitney numbers of the second kind , can be viewed as special cases of the flag @xmath11-vector of @xmath12 the euler characteristic of @xmath0 is the mbius invariant of @xmath13 @xcite .
surveys of these topics are @xcite , @xcite and @xcite .
other enumerative invariants of @xmath0 have not received as much attention .
the explicit relationship between the flag @xmath4-vector of @xmath0 and the * cd*-index of oriented matroids and zonotopes discovered in @xcite suggests that it may be time to study the @xmath4-vector and flag @xmath4-vector of @xmath12 we begin with a review of the basic notions associated to geometric lattices , graded posets , @xmath4-vectors and flag @xmath4-vectors .
this is followed by an examination of the geometric lattices which minimize and maximize the flag @xmath4-vector of their order complex . as a consequence
we find the zonotopes with a specified dimension and number of zones which minimize or maximize the * cd*-index .
then we show that @xmath0 has a convex ear decomposition
. an immediate consequence of this decomposition is our main theorem concerning the @xmath4-vectors of order complexes of geometric lattices .
[ main h ] let @xmath13 be a rank @xmath1 geometric lattice .
then , for @xmath14 the @xmath4-vector of @xmath0 satisfies @xmath15 @xmath16 as is frequently the case with theorems of this type , the proof is not combinatorial .
the remaining sections are devoted to understanding the flag @xmath4-vector of @xmath0 with an eye toward providing a combinatorial proof of ( [ going up ] ) .
this will lead us to an examination of the weak bruhat order on the symmetric group .
we will take all posets in this work to be finite .
a poset @xmath17 is _ graded _ if all maximal chains have the same length and we call this length the _ rank _ of @xmath17
. a graded poset has an associated rank function @xmath18 which assigns to each element @xmath19 of @xmath17 a positive integer such that @xmath20 , where @xmath21 is the length of the longest chain of the form @xmath22 .
a _ lattice _ is a poset such that each pair of elements , @xmath23 and @xmath19 , has a least upper bound , or _
join _ , denoted @xmath24 , and a greatest lower bound , or _
meet _ , denoted @xmath25 .
consequently , a lattice has a unique minimal element @xmath26 such that @xmath27 for all @xmath28 , and a unique maximal element @xmath29 with @xmath30 for all @xmath31 .
we call an element of @xmath13 which covers @xmath26 an _ atom _ , and say that @xmath13 is _ atomic _ if every element in @xmath13 can be written as the join of atoms . a _ geometric lattice
_ is a graded atomic lattice whose rank function satisfies the _ semimodular condition _ that for any @xmath23 , @xmath32 , @xmath33 a broad class of geometric lattices arise from the affine dependencies of a finite set of points @xmath34 in euclidean space . in this case
the rank @xmath21 elements of the lattice are subsets of the form @xmath35 where @xmath36 is a @xmath37-dimensional subspace spanned by the elements of @xmath34 .
these subsets are ordered by inclusion .
points are in _ general position _ if every set of @xmath38 points spans a @xmath21-dimensional subspace .
one particularly useful geometric lattice arises from the _ near pencil _ arrangement on @xmath39 points in @xmath40-dimensional space which consists of @xmath41 points on a line with the remaining @xmath42 points in general position .
for ease of reference we shall call this lattice the rank @xmath43 near pencil on @xmath39 atoms ( see figure [ r - labeling ] ) . a _ matroid _
@xmath44 is a set @xmath34 ( for us @xmath34 will always be finite ) with a closure operation satisfying the exchange property ( see ( * ? ? ?
* section 1.4 ) for more details ) . for @xmath45
denote the _ closure of @xmath46 _ by @xmath47 .
simple matroid _ is a matroid such that @xmath48 and @xmath49 for every element @xmath50 .
the closed sets , or _ flats _ , of a matroid , when partially ordered by inclusion form a geometric lattice . in fact a result of birkhoff shows there is a bijection between geometric lattices and simple matroids @xcite .
a set @xmath51 is _ independent _ if @xmath52 for any @xmath53 .
basis _ of a matroid is a maximal independent set and a _ circuit _ is a minimal dependent set .
a _ loop _ in a matroid @xmath54 is an element @xmath55 that is contained in no basis , while a _ coloop _ is an element which is contained in every basis of the matroid .
let @xmath56 be a basis of a matroid @xmath57 . if @xmath58 , then @xmath59 contains a unique circuit , @xmath60 .
the element @xmath61 is in @xmath60 , and we call @xmath60 the _ fundamental circuit of e with respect to b_. by a basis of a geometric lattice we will mean a collection of atoms whose cardinality is the rank of @xmath17 and whose join is @xmath62 . particular bases which will be useful to us are the _ nbc - bases_. in order to define an nbc - basis we first fix an arbitrary linear order @xmath63 on the atoms of the geometric lattice @xmath13 .
a _ broken circuit _ of @xmath64 is a circuit with its least element removed .
the nbc - bases of @xmath64 are the bases of @xmath13 that do not contain a broken circuit .
all such bases must contain the least atom .
indeed , if @xmath56 is a basis of @xmath13 which does not contain the least element , then it will contain the broken circuit formed by removing the least element from the fundamental circuit contained in the union of @xmath56 and the least element .
let @xmath13 be the geometric lattice of the rank 3 matroid on @xmath65 ( with the natural order ) whose bases consist of all triples except @xmath66 and @xmath67 .
then the nbc - bases of @xmath13 are the triples @xmath68 , @xmath69 , @xmath70 , @xmath71 .
notice that @xmath72 is not an nbc - basis because it contains the broken circuit @xmath73 .
let @xmath74 be a simplicial complex ,
i.e. , @xmath74 is a collection of subsets of a vertex set @xmath34 satisfying @xmath75 for any @xmath76 and if @xmath77 and @xmath78 then @xmath79 .
maximal faces of @xmath74 are _ facets _ and @xmath74 is _ pure _ if all its facets have the same dimension . a pure @xmath80-dimensional simplicial complex
is said to be _
shellable _ if there is an ordering of its facets @xmath81 such that @xmath82 is a pure @xmath83-dimensional complex for @xmath84 .
such an ordering is called a _
shelling_. equivalently , a linear ordering @xmath85 on the facets of a complex is a shelling if and only if it satisfies the following criterion ( see , for instance , @xcite ) * property @xmath57 * : for all facets @xmath86 and @xmath87 of @xmath74 such that @xmath88 there is a facet @xmath89 with @xmath90 such that @xmath91 and @xmath92 . given a poset @xmath17 with @xmath93 and @xmath94 , the _ order complex _
@xmath95 of @xmath17 is the simplicial complex whose vertices are the elements of @xmath96 and whose simplices are the chains of @xmath96 .
thus the facets correspond to maximal chains .
we say that a poset @xmath17 is shellable if there exists a shelling of @xmath95 .
a poset @xmath17 admits an _ r - labeling _
if there is a map from the edges of @xmath17 to the positive integers ( or more generally to some partially ordered set ) such that in any interval @xmath97 = \{z\in p : x \leq z \leq y\}$ ] of @xmath17 there is a unique saturated chain with increasing labels ( known as a rising chain ) . for @xmath98 covering @xmath99 in @xmath17 , denote the label on edge @xmath100 by @xmath101
. then a rising chain in @xmath97 $ ] is a maximal chain @xmath102 with @xmath103 .
an _ el - labeling _
@xcite of a poset is an r - labeling in which the unique rising chain in an interval @xmath97 $ ] comes first lexicographically among all of the chains in @xmath97 $ ] .
an el - labeling of a geometric lattice can be obtained by labeling the atoms @xmath104 and labeling the edge @xmath105 with the minimal atom @xmath106 such that @xmath107 ( see figure [ r - labeling ] ) .
we will call this ordering the _ minimal labeling _ of the facets of @xmath0 and for a given facet @xmath86 we denote its minimal label @xmath108 notice with this labeling distinct chains have distinct labeling sequences .
a result of bjorner shows that ordering the maximal chains of an el - labeled poset lexicographically on the chain labels gives a shelling of the associated order complex ( * ? ? ?
* theorem 2.3 ) .
for @xmath17 a rank @xmath43 poset , the number of simplices in @xmath95 of cardinality @xmath21 is denoted @xmath109 , and @xmath110 is known as the _ @xmath11-vector _ of the order complex .
the _ @xmath4-vector _ of @xmath111 is defined as @xmath112 where @xmath113 given a rank @xmath43 poset @xmath17 , and @xmath114= \{1,2,\ldots , r\}$ ] , let @xmath115 be the rank selected subposet of @xmath17 defined by @xmath116 .
the number of maximal chains of @xmath115 is denoted @xmath117 ; that is @xmath117 counts the number of chains in @xmath17 in which the ranks are the elements of @xmath118 .
the collection @xmath119 , @xmath120 $ ] , is known as the _ flag @xmath11-vector _ of the poset .
the flag @xmath11-vector gives a natural refinement of the @xmath11-vector of a poset s associated order complex as @xmath121 .
let @xmath122 denote the rank @xmath43 truncated boolean algebra on @xmath39 atoms .
@xmath122 is isomorphic to the rank @xmath39 boolean algebra @xmath123 for rank @xmath124 , @xmath125 , and rank @xmath43 of @xmath122 consists of the maximal element @xmath62 . every @xmath1-subset of @xmath126 $ ] is a basis of @xmath122 and every @xmath124-subset @xmath127 is independent . we shall see that the flag @xmath11-vector for rank @xmath43 geometric lattices on @xmath39 atoms is maximized by @xmath122 . dowling and wilson @xcite proved that for rank @xmath43 geometric lattices with @xmath39 atoms the singleton flags @xmath128 are minimized by the near pencil lattice . in her ph.d .
thesis @xcite , the first author gave an explicit formula for @xmath129 of the near pencil lattice and proved that this minimizes @xmath129 , @xmath114 $ ] for all geometric lattices of rank @xmath43 with @xmath39 atoms .
in this section we prove the stronger result that @xmath130 is minimized by the near pencil for all @xmath114 $ ] .
define the flag @xmath4-vector @xmath131 , for @xmath114 $ ] , by @xmath132 again , the flag @xmath4-vector refines the @xmath4-vector since @xmath133 .
an extremely useful combinatorial interpretation of the flag @xmath4-vector is that it counts the number of chains with a specified descent set in an r - labeled poset .
we describe this interpretation below . for a maximal chain @xmath134 , the _
descent set _ of @xmath135 is @xmath136 .
the following proposition can be found in ( * ? ? ?
* theorem 2.7 ) and ( * ? ? ?
* theorem 3.13.2 ) .
@xcite@xcite [ combinatorial h_s ] for @xmath17 a graded poset that admits an @xmath137-labeling , @xmath138 is the number of maximal chains of @xmath17 with labels having descent set @xmath118 .
since equation can be inverted to give @xmath139 we see that @xmath117 counts the number of maximal chains in an @xmath137-labeled poset @xmath17 with descent set contained in @xmath118 .
the positions of descents in maximal chains of a rank @xmath43 poset @xmath17 can be encoded using the _ * ab*-index _ @xmath140 of the poset which we describe presently .
assign to each chain in @xmath17 a word in the non - commuting variables `` * a * '' and `` * b * '' by assigning an * a * if consecutive edge labels in the chain increase and * b * if the labels decrease . summing over all the chains in @xmath17
gives the * ab*-index @xmath141 } h_s(p ) u_s\ ] ] where the word @xmath142 is given by @xmath143 if @xmath144 and @xmath145 if @xmath146 .
through the course of this paper we will use the descent set of a chain and its corresponding * ab*-monomial interchangeably where we take * a * to mean an ascent and * b * to indicate a descent in the chain label .
@xmath147 will indicate the set of all permutations with descent set @xmath118 and @xmath148 will refer to the * ab*-monomial with descent set @xmath118 .
next we consider the lattices which minimize and maximize the flag @xmath4-vector .
let @xmath13 be a rank @xmath43 geometric lattice with @xmath39 atoms .
then for all @xmath114 $ ] we have @xmath149 .
[ maxh ] place a linear ordering on the atoms of @xmath13 and @xmath150 we construct an injection from the minimal labelings of @xmath13 to the minimal labelings of @xmath122 which preserves descent sets .
this will prove the proposition since @xmath130 is the number of minimal labelings with descent set @xmath151 let @xmath152 be a minimal label of a facet @xmath86 in @xmath12 the uniqueness of rising chains in any interval implies that @xmath153 is completely determined by the initial segment @xmath154 where @xmath155 is the label preceding the last descent of @xmath108 for instance , if @xmath156 then @xmath157 plus the knowledge that @xmath158 precedes the last descent of the label , determines @xmath108 since every subset of cardinality @xmath124 is a flat of @xmath122 and @xmath159 there is a unique facet @xmath160 of @xmath161 such that @xmath162 and whose final descent is immediately after @xmath163 the function which takes @xmath153 to @xmath164 is the required map . since @xmath139 we immediately have the following result . among rank @xmath43 geometric lattices with @xmath39 atoms , @xmath122 maximizes @xmath117 for all @xmath120 $ ] .
we look again to the near pencil when considering the lattice which minimize the flag @xmath4-vector .
let @xmath114.$ ] fix @xmath165 the number of orderings of @xmath166 such that @xmath43 comes before @xmath124 and has descent set @xmath118 is independent of @xmath167 [ ordering ] suppose @xmath43 is the @xmath21th element of the permutation .
there are @xmath168 ways to choose the elements ( including @xmath124 ) which appear after @xmath43 .
since @xmath43 effectively splits the permutation into two smaller permutations we can consider the elements to the right of @xmath43 as @xmath169 . let @xmath170 be the number of ways to arrange @xmath171 consistent with descent set @xmath118 and let @xmath172 be the number of ways to arrange the elements to the left of @xmath43 consistent with descent set @xmath118 .
then @xmath173 is the number of permutations of @xmath174 $ ] such that @xmath43 appears before @xmath124 .
this number is independent of @xmath124 .
let @xmath13 be a rank @xmath43 geometric lattice with @xmath39 atoms and let @xmath17 be the rank @xmath43 near pencil on @xmath39 atoms .
then @xmath175 [ minh ] let @xmath176 and @xmath177 be the atoms of @xmath17 and @xmath13 respectively and let @xmath111 and @xmath0 be the corresponding order complexes .
order the atoms of @xmath17 so that @xmath178 are the coloops of @xmath179 what are the nbc - bases of @xmath180 any basis must contain the coloops and @xmath181 the least atom in the non - trivial line . on the other hand , any basis of the form
@xmath182 is an nbc - basis of @xmath179 recall that @xmath138 is the number of orderings of the nbc - bases of @xmath17 for which the ordering is a minimal labeling of the corresponding maximal chain of flats in @xmath111 and the descent set is @xmath151 for @xmath183 this is all orderings . for @xmath184 an ordering is a minimal labeling if and only if @xmath185 comes before @xmath186 now order the atoms of @xmath13 so that @xmath187 is a basis of @xmath188 clearly @xmath189 is an nbc - basis of @xmath2 any ordering of @xmath56 is a minimal labeling of the corresponding maximal chain , and the contribution of these orderings to @xmath190 is the same as the contribution of all the orderings of @xmath191 to @xmath192 for each @xmath193 we form a basis @xmath194 of @xmath13 as follows .
let @xmath195 be the fundamental circuit of @xmath196 with respect to @xmath197 let @xmath198 where @xmath199 is the second highest element of @xmath200 for instance , if @xmath201 and the fundamental circuit of @xmath202 is @xmath203 then @xmath204 now one can check that each @xmath194 is an nbc - basis of @xmath13 and that any ordering of @xmath194 where @xmath196 comes before the least element of @xmath195 is a minimal labeling of the corresponding maximal chain . in combination with lemma [ ordering ]
this shows that there are at least as many minimal labelings with descent set @xmath118 for @xmath0 as there are for @xmath205 when the number of atoms of a rank @xmath43 geometric lattice is not specified the standard boolean algebra @xmath206 minimizes the flag @xmath4-vector ( * ? ? ?
* proposition 7.4 ) .
@xcite let @xmath13 be a geometric lattice of rank @xmath43 .
then for all @xmath114 $ ] we have @xmath207 .
hence the * ab*-index @xmath208 is coefficient - wise greater than or equal to the * ab*-index of the boolean algebra @xmath206 .
[ 7.4ber ] oriented matroids are signed versions of standard matroids .
we refer the reader to @xcite for more details .
the elements of the oriented matroid , when partially ordered , form an eulerian poset ( see @xcite ) which is known as the lattice of regions .
a poset is _ eulerian _ if every interval @xmath97 $ ] , where @xmath209 , has the same number of elements of odd rank as even rank . of interest to our work
is the fact that underlying each oriented matroid is a standard matroid along with its associated geometric lattice of flats .
a collection of hyperplanes @xmath210 is _ essential _ if @xmath211 . a special class of oriented matroids , called realizable matroids ,
have an associated essential hyperplane arrangement .
the lattice of regions of the realizable oriented matroid is isomorphic to the face lattice of the corresponding hyperplane arrangement .
every essential hyperplane arrangement has an associated _ zonotope _ , which is the polytope formed by taking the minkowski sum of the normals to the hyperplanes ( see @xcite ) .
it was noted by fine and proved by bayer and klapper @xcite that when @xmath17 is an eulerian poset the * ab*-index of @xmath17 , @xmath140 , can be written in terms of @xmath212 and @xmath213 .
when @xmath140 is expressed in terms of * c * and * d * it is referred to as the * cd*-index .
the face lattice of an @xmath40-dimensional convex polytope is an eulerian poset and the @xmath214-index of the polytope is defined to be the * cd*-index of its corresponding face lattice . when @xmath17 is the lattice of regions of an oriented matroid every occurrence of * d * in the * cd*-index appears as * 2d * and so it is referred to as the * c*-*2d*-index @xcite .
the lattice of flats of the underlying matroid contains all of the information necessary to determine the * c*-*2d*-index of the lattice of regions of an oriented matroid ( * ? ? ?
* theorem 3.4 ) .
this connection is made explicit in @xcite .
the following proposition indicates how to construct the * c*-*2d*-index of a zonotope given the geometric lattice underlying the associated hyperplane arrangement ( * ? ? ?
* corollary 3.2 ) .
@xcite let @xmath13 and @xmath215 be the underlying geometric lattice and zonotope associated to an essential hyperplane arrangement respectively .
then the * c*-*2d*-index of the zonotope @xmath215 is given by @xmath216 where @xmath63 is a linear function which takes * ab * words to * cd * words by replacing each occurrence of * ab * with * 2d * , then replacing the remaining letters with * c * s .
[ 3.1ber ] the hyperplane arrangement @xmath217 for @xmath218 has the boolean algebra @xmath219 as its underlying lattice of flats .
the zonotope corresponding to this arrangement is the @xmath1-dimensional cube .
propositions [ 7.4ber ] and [ 3.1ber ] together imply the following ( * ? ? ?
* corollary 7.6 ) .
@xcite among all zonotopes of dimension @xmath43 , the @xmath1-dimensional cube has the smallest * c*-*2d*-index .
consider the rank @xmath43 near pencil on @xmath39 atoms and the truncated boolean algebra , @xmath122 .
as seen in section 2 the near pencil is associated with the arrangement of @xmath41 points on a line with the remaining @xmath220 points in general position .
similarly @xmath122 can be associated to the arrangement of @xmath39 points in general position in @xmath221 . in both of these point arrangements
we can consider the set of rays from the origin to each point .
taking the hyperplanes normal to these rays gives an essential hyperplane arrangement whose underlying geometric lattice is the near pencil , and @xmath122 respectively .
combining proposition [ 3.1ber ] with theorem [ minh ] and proposition [ maxh ] gives the following analogous result for @xmath1-dimensional zonotopes with @xmath39 zones .
let @xmath222 and @xmath223 denote essential hyperplane arrangements whose underlying geometric lattice is the rank @xmath43 near pencil and truncated boolean algebra on @xmath39 atoms respectively . among all zonotopes of dimension
@xmath43 with @xmath39 zones , the zonotope corresponding to @xmath223 has the largest * c*-*2d*-index , and the zonotope corresponding to @xmath224 has the smallest * c*-*2d*-index .
convex ear decompositions were introduced by chari @xcite .
our convex ear decomposition is motivated by a basis for @xmath225 constructed by bjrner @xcite .
a _ convex ear decomposition _ of a pure @xmath42-dimensional simplicial complex @xmath74 is an ordered sequence @xmath226 of pure @xmath42-dimensional subcomplexes of @xmath74 such that 1 .
@xmath227 is the boundary complex of a simplicial @xmath40-polytope , while for each @xmath228 is an @xmath42-ball which is a proper subcomplex of the boundary of a simplicial @xmath40-polytope .
2 . for @xmath229 3 . @xmath230 .
@xcite [ convex h ] if @xmath74 has a convex ear decomposition , then for @xmath231 the @xmath4-vector of @xmath74 satisfies @xmath232 @xmath16 for the rest of this section @xmath233 is an ordered basis of atoms in @xmath13 with corresponding chain of flats @xmath234 let @xmath233 be a minimal labeling of a facet of @xmath12 then @xmath235 is an nbc - basis of @xmath188 suppose @xmath56 contains a broken circuit . if @xmath236 is the lowest ranked flat which contains a broken circuit , then @xmath237 [ switching lemma ] let @xmath238 be a minimal labeling of a facet of @xmath12 if @xmath239 then @xmath240 is also a minimal labeling of a facet of @xmath12 for two flats @xmath241 in @xmath13 let @xmath242 suppose @xmath240 is not a minimal labeling .
then there exists an atom @xmath61 such that either @xmath243 and @xmath244 or @xmath245 and @xmath246 in the first case , @xmath247 however , this implies that @xmath248 is not the least atom in @xmath249 in the second case , either @xmath250 which implies that @xmath251 is not the minimal atom in @xmath252 or @xmath253 but this last is impossible since @xmath254 and @xmath248 is the least atom in @xmath255 let @xmath56 be a basis of @xmath188 associated to any ordering @xmath233 of @xmath56 is the facet @xmath256 of @xmath12 the _ basis labeling _ of @xmath86 ( with respect to @xmath56 ) is @xmath257 this may or may not be the same as @xmath108 let @xmath258 be the nbc - bases of @xmath13 in lexicographic order . for each @xmath259 let @xmath260 be the union of all the facets of @xmath0 associated to all possible orderings of @xmath261 each @xmath260 is isomorphic to the order complex of the rank @xmath43 boolean algebra and as a simplicial complex is the boundary of the first barycentric subdivision of the @xmath40-simplex
. now define @xmath262 to be the pure subcomplex of @xmath260 whose facets are the facets of @xmath260 whose minimal labeling and basis labeling coincide . except for @xmath263
each @xmath262 is a proper subcomplex of @xmath260 ( * ? ? ?
* lemma 7.6.2 ) . if @xmath264 then @xmath262 is a closed @xmath42-ball .
it is sufficient to show that @xmath262 is nonempty and shellable . to see that @xmath262 is nonempty
, we note that for any basis @xmath56 of @xmath13 the minimal ordering and the basis ordering are the same for the maximal chain corresponding to ordering @xmath56 in reverse if and only if @xmath56 is an nbc - basis .
order the facets of @xmath262 in _ reverse _ lexicographic order with respect to the basis labeling of the corresponding maximal chains .
we show that this ordering satisfies property @xmath265 suppose @xmath160 and @xmath86 are facets of @xmath262 and @xmath266 let @xmath267 be the chain which represents their intersection .
we must find @xmath268 a facet of @xmath269 lexicographically after @xmath86 such that @xmath270 and @xmath271 let @xmath135 be the least index such that the length of @xmath272 $ ] is greater than one . since @xmath153 is lexicographically before @xmath273 there are basis atoms
@xmath274 in @xmath275 such that the chain corresponding to @xmath86 contains as a short saturated chain @xmath276 with @xmath277 and @xmath278 let @xmath279 be the facet corresponding to interchanging @xmath280 and @xmath281 in @xmath108 then @xmath282 and @xmath283 by the switching lemma , @xmath279 is a facet of @xmath284 if @xmath285
then @xmath286 let @xmath287 be a face in @xmath288
. by definition @xmath287 is not a facet .
the boundary of @xmath262 is equal to the boundary of @xmath289 ( topological closure ) .
so it is sufficient to show that @xmath287 is contained in a facet of @xmath289 .
write @xmath290 by assumption @xmath291 a facet of @xmath269 and @xmath292 a facet whose corresponding minimal labeling basis @xmath293 is lexicographically before @xmath261 therefore , there is some pair @xmath294 such that @xmath295 is lexicographically before @xmath296 hence , the unique increasing minimally labeled saturated chain of @xmath297 $ ] is not contained in @xmath261 now let @xmath298 be a facet of @xmath260 obtained as follows . first saturate the interval @xmath297 $ ] by adding in the atoms of @xmath299 in increasing order . then extend this to a saturation of the chain corresponding to @xmath287 in any way which results in a facet of @xmath300 such an @xmath279 contains @xmath287 and must be in @xmath289 since its minimal label and its basis label are not equal .
the two previous propositions show that @xmath0 has a convex ear decomposition .
an immediate consequence is theorem [ main h ] . since @xmath301 a natural question is whether or not the @xmath302-vector of @xmath0 is an @xmath57-vector .
the @xmath302-vector is @xmath303 where @xmath304 a sequence of nonnegative integers is an @xmath57-vector if it is the hilbert function of a quotient of a polynomial ring .
see , for instance , ( * ? ? ?
* theorem 2.2 , pg .
56 ) for an equivalent numerical definition of @xmath57-vector . is the @xmath302-vector of the order complex of a geometric lattice an @xmath57-vector ?
note : after this paper was written the second author discovered a proof that the g - vector of any space with a convex ear decomposition is an m - vector .
let @xmath233 be an nbc - basis of @xmath13 ordered so that @xmath305 then we can identify all the orderings of the basis with @xmath306 the symmetric group on @xmath43 letters .
for @xmath307 we write @xmath308 where @xmath309 the switching lemma tells us that if @xmath310 corresponds to a minimal labeling and @xmath311 is obtained from @xmath312 by interchanging @xmath313 with @xmath314 when @xmath315 then @xmath311 also corresponds to a minimal labeling .
let @xmath316 then @xmath317 if and only if @xmath311 can be obtained from @xmath312 by repeated application of the above switching procedure .
evidently @xmath318 is a partial order .
it is , in fact , the weak bruhat order on @xmath319 an equivalent definition of @xmath318 is the following .
define the _ inversion _ set of @xmath312 to be @xmath320 and @xmath321 then @xmath317 if and only if @xmath322 ) .
see @xcite for more information on the weak order in general coxeter groups .
let @xmath323.$ ] we say @xmath118 _ dominates _
@xmath36 if there exists an injection @xmath324 such that @xmath325 for all permutations @xmath326 if @xmath118 dominates @xmath327 then @xmath328 for all geometric lattices of rank @xmath43 or greater .
the switching lemma and proposition [ combinatorial h_s ] imply that there are at least as many facets which contribute to @xmath130 as @xmath329 at each step in the convex ear decomposition .
how do we find pairs @xmath330 such that @xmath118 dominates @xmath331 first some elementary facts .
[ lengthen ] suppose @xmath118 dominates @xmath332 let @xmath333 be * ab*-monomials , possibly equal to @xmath334
. then 1 .
@xmath335 dominates @xmath336 2 .
@xmath337 dominates @xmath338 3 .
@xmath339 dominates @xmath340 we prove ( 3 ) since the other proofs are virtually identical . by definition , if @xmath341 then @xmath342 and @xmath343 let @xmath344 be an injection which preserves the weak bruhat order .
let @xmath345 be a permutation such that the * ab*-monomial of the @xmath346 is @xmath347 the * ab*-monomial of @xmath348 is @xmath349 and the * ab*-monomial of @xmath350 is @xmath351 identify the ordered set @xmath174 $ ] with the ordered set @xmath352 - \{a_1,a_{r+3}\}$ ] in the canonical way .
define @xmath353 to be the permutation obtained by applying @xmath354 to @xmath355 using this identification .
clearly , @xmath85 is an injection and @xmath356 since @xmath357 and @xmath358 the descent monomial of @xmath353 is @xmath359 [ subset condition ] if @xmath118 dominates @xmath327 then @xmath360 suppose @xmath361 let @xmath312 be a permutation with descent set @xmath36 such that @xmath362 if @xmath363 then @xmath364 must also equal @xmath365 hence @xmath311 also has a descent at @xmath167 another place to look for @xmath366 with @xmath118 dominating @xmath36 is through the symmetries of the flag @xmath4-vector of the boolean algebra . since @xmath328 for all geometric lattices of rank at least @xmath367 we can begin our search by examining @xmath368 this lattice has @xmath369 symmetry . let @xmath370 be the permutation which reverses order , i.e. , @xmath371 we omit the elementary proof of the following : let @xmath36 be the descent set of @xmath372 1
. the descent set of @xmath373 is @xmath374-t.$ ] 2 .
the descent set of @xmath375 is @xmath376 where @xmath377 : r - i+1 \notin t\}.$ ] 3 .
the descent set of @xmath378 is @xmath379,$ ] where @xmath380 = \{i : r - i+1 \in t\}.$ ] combined with proposition [ combinatorial h_s ] , the above proposition shows that in @xmath381-t } = h_{t \circ \beta}.$ ] proposition [ subset condition ] rules out the possibility of @xmath36 being dominated by @xmath382 or @xmath374-t$ ] except when they are equal .
however , if for each @xmath124 at most one of @xmath124 and @xmath383 is in @xmath327 then @xmath384 let @xmath385.$ ] then @xmath386 as the map which sends @xmath387 and @xmath388 shows , @xmath36 is dominated by @xmath389 and hence @xmath390 for all geometric lattices of rank 4 ( or more ) .
[ main conjecture ] if @xmath391 then @xmath389 dominates @xmath332 we have verified this conjecture by computer for @xmath392 for convenience , table [ known cases ] lists all the cases that we have used in section [ i to i+1 ] .
these computations were helped by the observation that if @xmath393 and @xmath389 dominates @xmath327 then @xmath394 dominates @xmath395 to see this , use the fact that if @xmath396 then @xmath397 by combining proposition [ lengthen ] with cases where conjecture [ main conjecture ] is known to hold many pairs @xmath330 with @xmath118 dominating @xmath36 can be constructed .
for instance , @xmath398 dominates @xmath399 and @xmath400 dominates @xmath401 are known cases of conjecture [ main conjecture ] .
hence , @xmath402 dominates @xmath403 if true , conjecture [ main conjecture ] would be surprising .
we are required to find @xmath404 such that @xmath405 yet , @xmath406 for all @xmath407
our goal is to give a combinatorial flag @xmath4-vector proof of ( [ going up ] ) .
one way to do this would be to construct a matching from @xmath408-subsets of @xmath374 $ ] to @xmath124-subsets of @xmath40 such that each @xmath408-subset is matched to an @xmath124-subset which dominates it .
for example , here is such a matching from 2-subsets of @xmath409 $ ] to 3-subsets of @xmath409.$ ] @xmath410 this matching gives a combinatorial flag @xmath4-vector proof that @xmath411 for all geometric lattices whose rank is greater than or equal to @xmath412 .
table [ 3 to 4 ] gives a matching for @xmath413$]-sets of @xmath414 $ ] to @xmath415-sets of @xmath414 $ ] .
as we will see below , this is not possible for all @xmath40 and @xmath416 but it can be done for somewhat smaller @xmath167 the proof is by induction on @xmath419 the case of @xmath420 being trivial .
consider the tree in figure [ tree ] .
each @xmath421-monomial stands for the initial descent pattern of a permutation .
a dot above an ascent at position @xmath21 means that if @xmath36 has the given initial descent pattern , then @xmath36 is dominated by @xmath422 the bottom of each branch shows how the induction hypothesis should be applied . if the monomial ends with an ascent , then the induction hypothesis is applied to everything to the right of the bar . for instance
, the entry @xmath423 says to apply the induction hypothesis to the @xmath424 positions to the right of @xmath425 indeed , this insures @xmath426 supersets which dominate @xmath332 when the monomial ends with @xmath427 the induction hypothesis is applied to the positions to the right of the next ascent .
for instance , the monomial @xmath428 is covered by the branch endpoint @xmath429 let @xmath430 starting from the top of the tree we look for either an interior node which matches @xmath431 exactly or a branch endpoint which matches an initial segment of @xmath432 all of the interior nodes satisfy the theorem and all of the branch endpoints demonstrate how to use the induction hypothesis to prove the theorem for @xmath432 ( 0,190)@xmath334 ( 3,185)(0,-1)15 ( 5,185)(4,-1)25 ( 32,173)@xmath433 ( 0,162)@xmath434 ( 3,157)(0,-1)15 ( 5 , 157)(4,-1)25 ( 32,144)@xmath435 ( -2,134)@xmath421 ( 5,129)(4,-1)25 ( 32,117)@xmath436 ( 3,129)(0,-1)15 ( -5,101)@xmath437 ( 5,96)(4,-1)25 ( 32,82)@xmath438 ( 3,96)(0,-1)15 ( -8,65)@xmath439 ( 3,60)(0,-1)15 ( 5,60)(4,-1)25 ( 32,48)@xmath423 ( -11,37)@xmath440 ( 3,32)(0,-1)15 ( 5,32)(4,-1)25 ( 32,20)@xmath441 ( -13,8)@xmath442 the condition on @xmath124 ensures that @xmath446 hence , by the above lemma , each @xmath408-subset of @xmath374 $ ] has at least @xmath124 supersets which dominate it .
obviously any @xmath124-subset of @xmath374 $ ] has at most @xmath124 subsets of cardinality @xmath447 which it dominates .
the theorem is now an elementary application of hall s marriage theorem .
theorem [ explained i ] is not optimal .
we have already seen that there are suitable matchings for 2-sets to 3-sets in @xmath409 $ ] and 3-sets to 4-sets in @xmath414.$ ] however , it is not always possible to obtain suitable matchings for all @xmath448 let @xmath449.$ ] there are no 5-supersets of @xmath36 in @xmath450 $ ] which dominate @xmath332 this can be seen by directly computing @xmath329 and @xmath130 in @xmath451 where @xmath118 runs over all potential supersets . in each case
@xmath452 instead of insisting on a one - to - one matching we can consider grouping subsets together .
of course , this must be done in moderation .
indeed , @xmath5 is just a reflection of grouping all subsets of the same cardinality together .
let @xmath453 as the following table shows , there is a bijection which respects the weak bruhat order from the permutations in @xmath454 whose descent set is @xmath36 or @xmath118 to those whose descent set is @xmath455 or @xmath456 hence @xmath457 in rank 4 ( or more ) geometric lattices .
* acknowledgment . * as has frequently been the case throughout our careers we benefitted from helpful conversations with louis billera .
the first author was partly supported by an nsf - vigre postdoctoral fellowship .
the second author was partly supported by nsf grant dms-0245623 . | we prove that the order complex of a geometric lattice has a convex ear decomposition . as a consequence , if @xmath0 is the order complex of a rank @xmath1 geometric lattice @xmath2 then for all @xmath3 the @xmath4-vector of @xmath0 satisfies @xmath5 and @xmath6 we also obtain several inequalities for the flag @xmath4-vector of @xmath0 by analyzing the weak bruhat order of the symmetric group . as an application
, we obtain a zonotopal * cd*-analogue of the dowling - wilson characterization of geometric lattices which minimize whitney numbers of the second kind .
in addition , we are able to give a combinatorial flag @xmath4-vector proof of @xmath7 when @xmath8 |
several clusters of galaxies show extended ( @xmath0 mpc size ) synchrotron emission not directly associated with the galaxies but rather diffused into the intracluster medium ( icm ) : these radio sources are called radio halos . in some cooling flow clusters with a central dominant galaxy , the diffuse radio emission is extended on a smaller scale , forming the so called _ mini halos _ ( feretti & giovannini 1996 ) .
+ the radiative life
time of an ensemble of relativistic electrons losing energy by synchrotron emission and inverse compton ( ic ) scattering off the cmb photons is given by @xmath1 $ ] , where @xmath2 is the magnetic field intensity ( in g ) , @xmath3 is the lorentz factor and @xmath4 g . in a cooling flow region ( i.e. for distances @xmath5 , the cooling radius ) the compression of the thermal icm
is expected to produce a significant increase of the strength of the frozen in intracluster magnetic field : @xmath6 for radial compression ( soker & sarazin 1990 ) or @xmath7 for isotropic compression ( tribble 1993 ) .
therefore , in absence of a reacceleration or continuous injection mechanisms , relativistic electrons injected at a given time in these intense fields ( of order of some tens of @xmath8 g , e.g. ge & owen 1993 ) should already have lost most of their energy and the radio emission would not be observable for more than @xmath9 yr .
this short lifetime contrasts with the diffuse radio emission observed in mini
halos , hence it seems plausible that the electrons have been reaccelerated .
the time evolution of the energy of a relativistic electron is determined by the competing processes of losses and reaccelerations ( both related to the magnetic field ) and it is important to study the efficiency of these processes in the cooling flow region as a function of @xmath10 ( isotropic compression of the field is assumed ) : @xmath11 \\
\dot{\gamma}_{+}&=&+ \alpha_+ \gamma \ , = f(x ) \ , \gamma\end{aligned}\ ] ] where @xmath12 is the coefficient of synchrotron and ic losses , @xmath13 the coulomb losses term , @xmath14 the reacceleration coefficient , @xmath15 and @xmath16 is the electron number density at @xmath17 . since the characteristic time of radiative losses is proportional to @xmath18 ,
while that of reacceleration is independent of @xmath3 , the losses dominate the time evolution of the electrons for energies higher than @xmath19 , the break energy ( fig . 1 , left panel ) . in our model ( gitti , brunetti & setti 2002 )
we assume that the relativistic electrons are continuosly reaccelerated by mhd turbulence _ via _ fermi - like processes .
this kind of reacceleration has the correct radial dependence on the parameters in the cooling flow region to naturally balance the radial behaviour of the radiative losses ( fig .
1 , right panel ) . under these assumptions ,
the stationary spectrum of the relativistic electrons is : @xmath20 which is essentially peaked at @xmath19 and where we have parameterized the electron energy density as @xmath21 , @xmath22 being a free parameter .
the other free parameters in the model are @xmath23 and @xmath24 , the leading mhd turbulence scale at @xmath17 .
the first observed and well studied case of radio mini - halo is that in the perseus cluster ( a426 , at @xmath25 ) , which hosts also a massive cooling flow . the diffuse radio emission ( see left panel in fig .
2 ) has a total extension of @xmath26 ( comparable with that of the cooling flow region , @xmath27 kpc ) , a steep spectrum with spectral index @xmath28 ( between 327 mhz and 609 mhz , @xmath29 ) and its morphology is correlated with that of the cooling flow x - ray map ( ettori et al .
1998 ) . on smaller scales ( @xmath30 ) , bhringer et al .
( 1993 ) showed evidence of interaction between the radio lobes of the central radiogalaxy 3c84 and the x ray emitting intracluster gas .
more recent results confirmed this interaction ( fig .
2 , right panel ) and support the interpretation of the holes in the x ray emission as due to buoyant old radio lobes ( fabian et al .
we notice that the spectral index in the lobes ranges from @xmath31 in the centre to @xmath32 in the outer regions of fig .
2 ( right panel ) , i.e. a value similar to the spectrum of the mini halo extended over a scale @xmath33 times larger .
thus it is impossible to find a direct connection between the radio lobes and the mini
halo in terms of simple particle diffusion or buoyancy .
the relativistic electrons of the mini
halo should then be reaccelerated and the morphological connection with the cooling flow suggests a leading role of the cooling flow itself in powering the reacceleration process .
+ therefore , we have applied our model to the perseus cluster , whose radio properties ( brightness profile , integrated spectrum and radial spectral steepening ) can be well reproduced assuming isotropic compression of the field ( gitti et al .
2002 ) . in this case
, the necessary energy rate to reaccelerate the relic population of electrons in a typical time of few @xmath34 yr is @xmath35 erg s@xmath36 , while the energy rate supplied by the cooling flow is of the order of @xmath37 erg s@xmath36 , so only a small fraction of it is required .
the number of relativistic electrons in the cooling flow region is @xmath38 .
this is comparable to the number of the electrons in a typical radiogalaxy and may suggest an important role of the agns ( and possibly of the central agn ) in the injecton of the electron relic population .
in the framework of our model , large scale turbulence in the cluster volume is required to balance the radiative losses of the electrons at @xmath17 ( for perseus we found @xmath39 ) , however at the same time the turbulence can not be too high in order to avoid the disruption of the cooling flow .
this scenario is similar to that of the extended radio halos where , however , higher energy electrons ( with @xmath40 ) are required ( e.g. brunetti et al . 2001 ) : this means that the turbulence efficiency would have to be correspondingly greater hence bringing to the disruption of a possible cooling flow , in agreement with the observed anti - correlation between radio halos and cooling flows .
+ based on these considerations , we suggest that the physical conditions of the icm in a cooling flow cluster with a radio mini halo are intermediate between those which lead to the formation of extended radio halos , hosted by clusters without cooling flows , and those holding in cooling flows clusters without radio halos .
this could explain , at least qualitatively , the rarity of radio mini halos . in this scenario , where the energetics of the turbulence plays the main part in discriminating between the two opposite situations
, perseus may represent a borderline case : indeed there is evidence for a relatively recent merger event ( dupke & bregman 2001 ) which may have left a fossil turbulence , not energetic enough to produce an extended radio halo , but sufficient for reaccelerating the electrons producing the mini halo .
in order to extend the application of our model , we have undertaken the analysis of a2626 ( @xmath41 ) .
this cluster has a radio emission extended on a scale comparable to that of the cooling flow region ( e.g. rizza et al . 2000 ) , and thus it is one of the best mini halo candidates .
+ we present preliminary results for this cluster , obtained from vla archive data at 1.5 ghz and 330 mhz at different resolutions .
the 1.5 ghz map ( fig .
3 , left panel ) shows an unresolved core ( with a flux density @xmath42 mjy , in agreement with roland et al .
1985 ) and a diffuse emission extended by @xmath43 ( @xmath44 mjy ) ; we have estimated a polarization level @xmath45 % . at 330 mhz we do not detect the emission of the core ( for which we extimated an inverted spectral index @xmath46 ) while the flux density of the diffuse emission is @xmath47 jy . the spectral index of the diffuse emission between @xmath48 mhz and @xmath49 ghz is @xmath50 .
these results indicate that the radio source observed in a2626 shows an amorphous structure which is likely to be not related to the central radiogalaxy .
thus it appears as a good candidate to apply our model for electron reacceleration in cooling flows . in fig .
3 ( right panel ) we report a preliminary comparison between the observed flux densities and the total synchrotron spectrum predicted by our reacceleration model . bhringer h. , voges w. , fabian a.c . , edge a.c .
, & neumann d.n .
1993 , , 264 , l25 brunetti , g. , setti , g. , feretti , l. , & giovannini , g. 2001 , , 320 , 365 dupke r.a . , & bregman j.n .
2001 , , 547 , 705 ettori s. , fabian a.c .
, & white d.a .
1998 , , 300 , 837 fabian , a. c. , sanders , j. s. , ettori , s. , taylor , g. b. , allen , s. w. , crawford , c. s. , et al .
2000 , , 318 , 65 fabian , a. c. , celotti a. , blundell k.m .
, kassim n.e . , &
perley r.a .
2002 , , 331 , 369 feretti , l. , & giovannini , g. 1996 , iaus , 175 , 333 ge , j.p . , &
owen , f.n .
1993 , 105 , 3 gitti , m. , brunetti , g. , & setti , g. 2002 , , 386 , 456 rizza e. , loken c. , bliton m. , roettiger k. , & burns j.o .
2000 , 119 , 21 roland , j. , hanish , r.j .
, vron p. , & fomalont , e. 1985 , , 148 , 323 sijbring , d. 1993 , ph.d .
thesis , groningen soker , n. , & sarazin c.l .
1990 , , 348 , 73 tribble , p.c .
1993 , , 263 , 31 | we have developed a model in which the diffuse synchrotron emission from radio mini halos , observed in some cooling flow clusters , is due to a relic population of relativistic electrons reaccelerated by mhd turbulence _ via _ fermi like processes . in this model
the energetics is supplied by the cooling flow itself . here , the model ( successfully applied to the perseus cluster , a426 ) is preliminarily applied to the possible mini - halo candidate a2626 , for which we present vla data .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
the operation of suspension of a topological space @xmath0 is far from being faithful .
whole information about noncommutativity of @xmath1 is lost .
one can read only the abelianization @xmath2 similarly , the product structure in the cohomology @xmath3 is lost .
multiplication in @xmath4 is trivial . recently
a new kind of structures came to attention of topologists .
one considers a subset of _ allowable _ paths @xmath5 in the space of all paths in @xmath6 .
this kind of structure is called a directed structure , or a d - stucture @xcite . any topological space
@xmath6 has many directed structures .
the poorest one contains constant paths only ; this is the discrete structure @xmath7 .
another structure is the richest : all paths are allowed .
this is the total structure @xmath8 .
of course , there are many intermediate structures .
the purpose of our paper is to show that , in some sense , the suspension forgets information about the directed structure . given a topological space @xmath6 with a chosen directed structure , there are many ways of constructing directed structure on @xmath9
. there are two particularly interesting cases to consider .
one is the smash product with the circle @xmath10 with total structure @xmath11 the other one is the smash product with the oriented circle @xmath10 , in which only the paths with non - decreasing angle are allowed .
this suspension is denoted by @xmath12 in both cases , we can consider either the minimal structure , which contains only finite concatenations of directed paths coming from @xmath13 or @xmath14 , or the complete one , which allows infinite concatenations .
obviously , all these directed structures depend on the directed structure in @xmath6 .
we formulate our results in a way that they can be applied to a wider variety of directed structures .
we will not assume explicitly that the directed structure on @xmath9 comes from a directed structure on @xmath6 , only that it satisfies certain conditions .
we will study the spaces of directed loops @xmath15 based at the distinguished point of the suspension .
it is a subspace of the space of all loops @xmath16 .
[ twierdzenieglowne ] suppose that @xmath6 is a connected space .
furthermore , assume that the suspension @xmath9 is equipped with a directed structure that is excisive , finitely generated and translatable ( see def [ fingen ] , [ translatable ] and [ exci ] ) . with the notation as above , the inclusion @xmath17 is a weak homotopy equivalence .
the technical assumptions of theorem [ twierdzenieglowne ] are satisfied for a natural class of directed spaces : the conclusion of the theorem [ twierdzenieglowne ] holds for the suspension of the realization of cubical complexes with minimal and completed directed structures .
the homotopy type of the space @xmath16 does not depend on the directed structure of @xmath6 .
therefore , the homotopy type of @xmath18 do not depend on the directed structure in @xmath9 .
the theorem [ twierdzenieglowne ] is a consequence of the following statements ( theorem [ homomain ] , corollary [ cormain ] and theorem [ spojnosc ] ) : the natural map @xmath19 is a weak homotopy equivalence . here
@xmath20 is the free topological monoid generated by @xmath6 , which is called the james construction . by the classical result of james @xcite ( see also @xcite ) the natural map @xmath21 is also a weak homotopy equivalence . in another paper @xcite , milnor have proved by simplicial methods that @xmath16 is homotopy equivalent to the topological group @xmath22 freely generated by @xmath6 .
these result coincide since @xmath20 is connected , and then the group completion does not change the homotopy type .
if @xmath6 is not connected then one is tempted to prove that ( [ javo ] ) is a homotopy equivalence .
we skip that proof due to point - set topology problems .
we rely only on the homological argument which works fine only for connected spaces .
we note that in general @xmath23 is not a homotopy equivalence . for @xmath24
we have @xmath25 @xmath26 and the inclusion is homotopic to the inclusion of the natural numbers into the integer numbers @xmath27 .
we construct an explicit homotopy inverse of ( [ javo ] ) .
we note that the original proof of james and other proofs which are present in the literature for the classical case do not exhibit an inverse map .
our construction of the inverse map is specific for the directed spaces .
in the course of the proof we meet several technical problems .
most of them are caused by the fact that the suspension @xmath28 might have pretty bad structure in the neighbourhood of the distinguished base point .
for any space @xmath6 , the suspension admits the effect of `` peacock feather eye '' , i.e. there are directed loops which infinitely times pass through the distinguished point .
@xmath29\times x$ ] @xmath9 to avoid this problem we first consider the space of paths which are finite concatenations of paths coming from @xmath29\times x$ ] , i.e. we consider the minimal directed structure and later pass to the closure .
the paper is organized as follows : after introducing the notions in [ directed : spaces ] and [ section : cubical ] we discuss directed structures in the smash product @xmath30 . in [ smash ]
we show under suitable conditions that if @xmath6 and @xmath31 are connected directed spaces , then @xmath32 is connected .
further in [ suspension ] we describe a convenient model of the suspension @xmath9 .
the james construction is recalled in [ james ] .
we modify the classical homological proof of james in the context of directed spaces . that is done in the [ homologicznie ] . in the next section
[ homotopijnie ] , we construct a homotopy inverse of the james map ( [ javo ] ) explicitly .
we note that such a construction was not done in the classical context and might be much harder for non - directed paths . finally , in [ straightening ] we show how to deform in a canonical way a path to another one which corresponds to a configuration of points in @xmath6 , that is to a path which lies in the image of the james map .
let @xmath33 be a pointed hausdorff space , which is locally contractible .
the following definition is due to grandis @xcite _ a d - structure _ on a topological space @xmath6 is a family of paths @xmath34,x),\ ] ] called _ d - paths _ , that satisfies the following conditions : * every constant path a d - path . *
the concatenation of d - paths is a d - path .
* non - decreasing reparametrization of a d - path is a d - path . _ a d - space _ , or _ a directed space _ ,
is a space equipped with a d - structure .
if it is clear what d - structure is considered on a space @xmath6 , it is denoted by @xmath35 .
a map @xmath36 , where @xmath6 , @xmath31 are d - spaces , is _ a d - map _ if it preserves a d - structure , i.e. , if @xmath37 whenever @xmath38 . the d - spaces and d - maps form a category , which is complete and cocomplete @xcite
. for every family @xmath39 there exists the unique minimal d - structure @xmath40 on @xmath6 generated by @xmath41 .
it contains all constant paths and all paths having the form @xmath42 where @xmath43 for @xmath44 and @xmath45\to[0,1]$ ] is a continuous non - decreasing map .
if @xmath6 is a d - space , and @xmath46 , then we denote @xmath47 if @xmath33 is a pointed d - space , we define _ the space of directed loops on @xmath6 _ as @xmath48 a path @xmath49\to x$ ] , for @xmath50 not necessarily equal to @xmath51 , will be called a d - path if its reparametrization @xmath52\ni t\mapsto \alpha(a+t(b - a))\in x$ ] is a d - path . here
we list examples of d - spaces that play an important role in this paper : * any topological space has two extreme d - structures : the total @xmath8 one which contains all paths , and the discrete one @xmath7 which contains only constant paths . * _ a directed euclidean space @xmath53 _ , where d - paths are all paths having non - decreasing coordinates . *
every subspace @xmath54 of a d - space is a d - space , with a d - structure @xmath55 .
particular cases are @xmath56\subseteq \vec{\r}$ ] ( _ the directed interval _ ) and @xmath57 ( _ the directed @xmath58cube _ ) . * for every closed subset @xmath54 of a d - space , the quotient @xmath59 is also a d - space , with a d - structure generated by paths having the form @xmath60 , where @xmath38 and @xmath61 is the projection .
we obtain the minimal quotient d - structure @xmath62 . *
the completed quotient d - structure on @xmath59 , denoted by @xmath63 , that contains all paths @xmath64 for which there exists a ( non - necessarily finite ) cover of @xmath52 $ ] by closed intervals @xmath65 , @xmath66 , such that @xmath67 is a projection of a d - path on @xmath6 . the completed structure is the completion of the minimal d - structure in the sense of @xcite . *
an important special case is _ the directed circle _
@xmath68\cup[1,+\infty)\big).\ ] ] in this case the minimal and the completed d - structure are equal . note that the minimal structure in @xmath59 is not always natural .
for example , take the interval @xmath69 $ ] with the total structure .
then @xmath70 as topological spaces . but
the directed paths @xmath49\to x/\{-1,1\}$ ] of the minimal quotient structure have the property that @xmath71 has only finitely many connected components .
let @xmath6 be a space and let @xmath35 be a d - structure on @xmath6 . _
the completion _ of @xmath35 is a d - structure @xmath72 on @xmath6 such that @xmath73 if and only if there exists a cover @xmath74=[0,1]\ ] ] such that @xmath75}\in \vp(x)$ ] for all @xmath76 .
the completion of the minimal structure on a quotient space will be called _ the completed structure_. clearly , @xmath77 .
semi - cubical complexes @xcite form a family of d - spaces , which is especially important because of their applications in computer science . in this paper
we consider cubical sets : a class which contains semi - cubical sets and is closed with respect to taking suspensions .
_ a cubical set _
@xmath78 is a sequence of disjoint sets @xmath79 , @xmath80 equipped with * face maps @xmath81 , for @xmath82 , @xmath44 , @xmath83 , * degeneracy maps @xmath84 , for @xmath85 , @xmath86 , which satisfy the following cubical relations : * @xmath87 for @xmath88 , * @xmath89 for @xmath88 , * @xmath90 every cubical set has a geometric realization which is a d - space . for @xmath83 and @xmath44 , define d - maps @xmath91 _ the geometric realization _ of a cubical set @xmath78 is the quotient space @xmath92 where @xmath93 is generated by @xmath94 and @xmath95 .
there are ( at least ) two natural d - structures on @xmath96 : * the minimal quotient d - structure @xmath97 , namely , a continues path @xmath64 is a d - path if there exists a sequence of numbers @xmath98 , cubes @xmath99 and d - paths @xmath100\to \vec{i}^n$ ] such that @xmath101 for @xmath102 $ ] .
this is a minimal d - structure such that the inclusions of cubes induce d - maps . * the completed d - structure @xmath103 ,
namely , a path @xmath64 is a d - path if there exists a ( non - necessarily finite ) cover of @xmath52 $ ] by closed intervals @xmath65 , @xmath66 , such that @xmath67 has the form @xmath104 , for @xmath105 and a d - path @xmath106 .
this coincides with the completed d - structure on a quotient space . the geometric realization of @xmath78 with the minimal quotient ( resp .
the completed ) d - structure completed d - structure will be denoted by @xmath107 ( resp .
@xmath108 ) .
we will need the following technical result .
[ cubicalneighborhood ] let @xmath78 be a cubical set , @xmath109 its subset , and @xmath110 .
then there exists a subset @xmath111 such that the interior of @xmath112 contains @xmath113 and the maps @xmath114 induced by the inclusion is a homotopy equivalence , where @xmath115 stands for either @xmath116 or @xmath117 .
the same argument is valid in both cases .
let @xmath118 be a d - map given by the formula @xmath119$,}\\
3(s-\tfrac{1}{2})+\tfrac{1}{2 } & \text{for $ s\in[\tfrac{1}{3},\tfrac{2}{3}]$,}\\ 1 & \text{for $ s\in[\tfrac{2}{3},1]$ , } \end{cases}\ ] ] and let @xmath120 denote its @xmath58th product .
furthermore , let @xmath121 , @xmath122 $ ] , be a homotopy between the identity map on @xmath123 and @xmath124 .
for every @xmath125 , the family @xmath126 , @xmath85 , commutes with the face maps @xmath127 and the degeneracy maps @xmath128 ; as a consequence , the formula @xmath129 defines a homotopy between the identity map on @xmath96 and @xmath130 .
notice that , for every @xmath85 , the interior of @xmath131 contains @xmath132 .
therefore , the interior of @xmath133 contains @xmath113 . finally , we need to show that the inclusion @xmath134 is a homotopy equivalence .
indeed , @xmath135 induces the map @xmath136 which is a homotopy inverse of the inclusion , with suitable homotopies induced by @xmath137 .
here we prove a general fact about directed structures in the smash product @xmath138 of pointed directed spaces .
we will apply it for @xmath139 and @xmath140 .
the minimal directed structure @xmath141 consists of the paths which are concatenations of a finite number of d - paths coming from @xmath142 .
[ fingen ] we say that a directed structure @xmath143 on the smash product of pointed directed spaces @xmath144 is _ finitely generated _ if it contains the minimal structure and , for any pair of points @xmath145 , the space @xmath146 is dense in @xmath147 . in particular , @xmath148 is dense in @xmath32 .
[ p : completeddstructureisfinitelygenerated ] the completed d - structure @xmath149 is finitely generated .
denote @xmath150 , @xmath151 and fix @xmath152 .
let @xmath153 be a decreasing local basis of @xmath154 at the point @xmath155 , i.e. , such that @xmath156 for @xmath157 .
fix @xmath158 .
consider a cover of @xmath52 $ ] with maximal connected subsets of @xmath159 and @xmath160 . by the compactness of @xmath52 $ ]
, there exists a finite number of pairwise disjoint intervals @xmath161 $ ] , @xmath162 such that @xmath163 and @xmath164 whenever @xmath165 .
define a path @xmath166 now @xmath167 is a sequence of paths of @xmath168 that converges to @xmath64 .
we have used here generalized sequences ( or nets ) and our argument works in the general topological setup .
[ spojnosc ] suppose that @xmath6 and @xmath31 are path connected .
we assume that the directed structure in @xmath30 is finitely generated .
then the space of directed loops @xmath32 is connected .
it is enough to show that @xmath148 is connected since the closure of a connected set is connected .
every path in @xmath148 can be lifted to @xmath142 uniquely away from the base point . therefore , every path is a concatenation of a finite number of images of paths @xmath169\to x\times y\ ] ] such that @xmath170 we will deform such paths to constant paths .
let @xmath171 and @xmath172 be the distinguished points in @xmath6 and @xmath31 .
there are four possibilities : 1 .
@xmath173 and @xmath174 , 2 .
@xmath173 and @xmath175 , 3 .
@xmath176 and @xmath177 , 4 .
@xmath176 and @xmath178 .
let @xmath179 be the coordinates of @xmath64 .
the path @xmath64 is homotopic to the concatenation @xmath180 here @xmath181 and @xmath182 denote constant paths of the lengths @xmath183 . in the first case ( @xmath184 , @xmath185 ) such concatenation
is projected to a constant path . in the second case ( @xmath184 , @xmath186 )
, we note that the first component of ( [ rozklad ] ) is projected to a constant path .
since the @xmath31 is path connected , the second component can be deformed to the path of the form @xmath187 .
again , this path is projected to the constant path .
the third and fourth cases are analogous .
let @xmath6 be a space with a distinguished point @xmath171 .
we will describe the suspension @xmath188 as a quotient of @xmath189 .
first , we present the circle @xmath10 as the quotient space @xmath190\cup [ 1,\infty)\big)\ , .\ ] ] then @xmath9 is identified with the quotient of @xmath189 by the subspace @xmath191\cup [ 1,\infty)\big)\times x\cup \vr\times \{x_0\}\,.\ ] ] the point of @xmath9 which is the image of @xmath192 will be denoted by @xmath193 .
the distinguished point of @xmath9 is @xmath194 the coordinates in @xmath9 are defined only away from the distinguished point .
let @xmath195 and @xmath196 be the projections onto the respective coordinates . for a given d - structure @xmath35 on @xmath6 ,
there is a variety of `` natural '' d - structures on @xmath197 which can be defined .
they are : * the directed suspension structure @xmath198 , in which a path is directed if and only if it is a concatenation of a finite number of paths which are the images of directed paths in @xmath199 .
it is a minimal structure such that the quotient map @xmath200 is a d - map . *
the completed directed suspension structure @xmath201 , which allows `` infinite '' concatenations of d - maps . * the completed structure @xmath202 , which is the total d - structure on @xmath9 .
we will consider all these cases simultaneously .
we will even work in a greater generality and assume only that the d - structure satisfies the following definition : [ translatable ] we say that a directed structure on @xmath9 is _ translatable _ if it contains paths @xmath203\ni t\mapsto \l t , x \r\in \sigma x\ ] ] for all @xmath204 , and the maps @xmath205 preserve the directed structure for all @xmath206 , @xmath207 , @xmath208 and @xmath209 .
all d - structures listed above are translatable .
notice that we do not assume that a translatable d - structure is induced , in any way , from a d - structure on @xmath6 . without specifying any
particular directed structure in the suspension , we prove the results in section [ homologicznie ] for @xmath198 , @xmath201 and @xmath210 simultaneously .
we will assume that the directed structure is finitely generated ( see definition [ fingen ] ) ; this is true in the three listed cases thanks to proposition [ p : completeddstructureisfinitelygenerated ] .
we will also need another technical but natural condition , which excludes point - set pathology ( see definition [ exci ] ) .
in section [ homotopijnie ] we assume that all directed paths are non - decreasing along @xmath211 coordinate away from the base point .
this assumption allows to construct a map from the loop space to the james construction .
in this section we will use one - sided moore paths , i.e. , we do not assume that the paths are parametrized by the unit interval , but by an arbitrary interval @xmath212 $ ] for @xmath213 .
the concatenation of such paths is strictly associative ( when it is defined ) .
[ moore - one ] let @xmath31 be a directed space , @xmath214 . by @xmath215
we denote the space of directed moore paths @xmath64 , such that @xmath216 .
formally , @xmath217}\text { is directed},\;\alpha(0)=y_0\,,\;\alpha(s)=\alpha(t^\infty_\alpha)\;\text{for } s > e\,\}\,.\end{gathered}\ ] ] the element of @xmath215 will be regarded as a map @xmath218\to y\,.\ ] ] the space @xmath215 is contractible .
the contracting homotopy is given by the formula @xmath219}$ ] for @xmath220 $ ] .
we will use the following notation : @xmath221 is the evaluation at the end of the path @xmath222 .
if @xmath31 has a total directed structure , i.e. , every path is directed , then the map `` @xmath223 '' is a serre fibration with the fiber @xmath224 .
it is not true in general ; it might happen that the set @xmath225 ( paths from @xmath172 to @xmath226 ) is empty .
fix a space @xmath6 and fix a translatable d - structure on @xmath9 .
denote by @xmath227 ( resp .
@xmath228 ) the upper ( resp
. the lower ) sub - cone of @xmath9 , i.e. , the image of @xmath229\times x$ ] ( resp .
@xmath230 ) in @xmath9 . let @xmath231 for @xmath204 , @xmath232 , @xmath233 let @xmath234\to \sigma x\ ] ] be the path given by @xmath235 note that , according to the definition [ translatable ] , such paths are directed .
in general , we will use the letter @xmath236 to denote various versions of paths , which have @xmath237 fixed .
let us write @xmath238 for the loop @xmath239 and let us consider the map @xmath240
@xmath241 let us recall some details from @xcite .
there a convenient way of constructing a map @xmath244 is presented .
the map @xmath245 does not preserve the base point , since the path @xmath246 has length 2 .
let us modify the space @xmath6 .
let @xmath247=x\sqcup [ 0,1]/{x_0\sim 1}\ ] ] and extend the map @xmath236 @xmath248\end{cases}\ ] ] let @xmath249 be the distinguished point in @xmath250 .
the map @xmath251 preserves the distinguished points ; therefore , it induces a map of monoids : @xmath252 also we have a map @xmath253 induced by the retraction @xmath254 .
this map is a homotopy equivalence .
therefore , we obtain a map defined up to homotopy @xmath255 note that the james construction @xmath256 contains the disjoint union @xmath257 and the restriction of the map @xmath258 to @xmath259 is the concatenation @xmath260 later in
[ homotopijnie ] , we will construct a map @xmath261 such that @xmath262\in j(x)\,.\ ] ] it will follow that @xmath263 is a homotopy inverse of @xmath264 .
in this section , we consider an arbitrary directed structure in @xmath9 which is translatable ( [ translatable ] ) . consider the space of directed loops @xmath265 the homology @xmath266 with coefficients in a field has an algebra structure since @xmath243 is a topological monoid .
all homology groups appearing in this section will have coefficients in a field @xmath267 .
the main result is the following theorem [ homomain ] and its corollaries .
we will use singular homology , which works in a general topological context but we need some general condition allowing to apply mayer - vietoris sequence ( see ( * ? ? ?
* ch 4.6 , p.188 ) for the definition of excisive couple ) .
this technical issue forces a condition which we impose in the directed structure of @xmath9 .
[ exci ] a d - structure on @xmath9 is _ excisive _ if the pair @xmath268 ( defined in [ stozkil ] ) is an excisive couple in @xmath269 in the sense of ( * ? ? ?
* ch 4.6 , p.188 ) .
[ homomain ] let @xmath6 be a space , and let @xmath270 be a directed structure on @xmath9 that is excisive and translatable .
assume that @xmath243 is connected .
let @xmath271 .
then the map @xmath272 induces an isomorphism of algebras @xmath273 where @xmath274 is the free algebra generated by the vector space @xmath275 .
[ cormain ] under assumptions as above , the extension of the inclusion @xmath276 to the james construction @xmath277 is a weak homotopy equivalence . in particular ,
the homotopy type of @xmath243 does not depend on the directed structure on @xmath9 .
it follows that : under the assumptions as above , the inclusion @xmath278 is a homotopy equivalence .
finally , we will show that the assumptions on the d - structure on @xmath197 are satisfied for directed suspensions of cubical complexes .
assume that @xmath6 is a geometric realization of a cubical set @xmath279 , and the directed structure is the directed suspension structure ( minimal or completed ) .
then @xmath280 is both translatable and excisive .
if @xmath6 is connected , then @xmath281 is connected . as before
, the translatability is clear .
let @xmath282 , @xmath283 be @xmath284dimensional cubes with the single non - degenerate vertices @xmath285 and @xmath286 respectively , and let @xmath287 .
there is a d - homeomorphism of triples @xmath288,[-1,0],[0,1])$ ] . as a consequence ,
the geometric realization of @xmath289 is d - homeomorphic to @xmath197 .
let @xmath290 and @xmath291 be the images of @xmath292 and @xmath293 .
there is a d - homeomorphism of triples @xmath294 by [ cubicalneighborhood ] , there are subsets @xmath295 that contain @xmath296 and @xmath297 in their interiors , respectively , and @xmath298 are homotopy equivalences .
the pair @xmath299 is excisive ; this implies that @xmath300 is also excisive .
the last essertion follows from theorem [ spojnosc ] .
fix a d - structure on @xmath9 that is translatable .
let @xmath301 be the map induced defined by @xmath302 this map shrinks @xmath228 to the distinguished point .
the map @xmath303 is d - homotopic ( i.e. , homotopic via d - maps ) to the identity : the homotopy is defined for @xmath122 $ ] : @xmath304 the map @xmath305 shrinks the image of @xmath306\times x$ ] to the distinguished point . the map @xmath303 induces a map of the space of the directed paths . for @xmath307
we have @xmath308 .
therefore [ prop1]the inclusion @xmath309 is a homotopy equivalence .
its inverse is given by @xmath310 similarly we have a map @xmath311 @xmath312 which shrinks @xmath227 to the distinguished point . the homotopy between @xmath313 and the identity is given by the formula @xmath314 the map @xmath315 shrinks the image of @xmath316 to the distinguished point .
we obtain [ prop2]the inclusion @xmath317 is a homotopy equivalence .
the inverse is given by @xmath318 [ prop3]the map @xmath319 given by @xmath320 is a homotopy equivalence .
the inverse is given by @xmath321 the composition @xmath322 is joined with the identity by the homotopy @xmath323 @xmath324 where @xmath305 is given by ( [ phit ] ) .
the opposite composition is equal to @xmath325 the homotopy to the identity is given by @xmath326 @xmath327 proof of theorem [ homomain ] is a modification of the proof in the classical situation @xcite and @xcite .
we consider the decomposition @xmath328 we may apply mayer - vietoris exact sequence since we know , by assumption , that @xmath300 is an excisive couple .
let @xmath329 be the inclusion .
since the space @xmath330 is contractible , the mayer - vietoris exact sequence for the reduced homology gives us an isomorphism @xmath331 due to propositions [ prop1]-[prop3 ] , we obtain an isomorphism @xmath332 the key point is to analyze the isomorphism map in terms of the algebra structure of @xmath266 .
we have a commutative diagram @xmath333 the map @xmath334 @xmath335 is homotopic to the projection on @xmath243 . on the other hand @xmath336 @xmath337 and after a reparametrisation of @xmath236 is equal to @xmath338 this map is homotopic to @xmath339 denote by @xmath112 the algebra @xmath340 .
this is a graded algebra with @xmath341 , by the assumption that @xmath342 is connected .
thus , @xmath343 the reduced homology of @xmath344 can be decomposed as @xmath345 the bottom isomorphism of the diagram [ dia ] is of the form
@xmath346 @xmath347 we subtract one summand @xmath348 from the above mapping and since @xmath349 we obtain an isomorphism @xmath350 @xmath351 this property characterizes the tensor algebra @xmath274 , see ( * ? ? ?
5.3.2 ) .
the spaces @xmath20 and @xmath243 are topological monoids , therefore their fundamental groups are commutative . a map between such spaces is a weak homotopy equivalence if and only if it induces an isomorphism of homology groups .
it is enough to check that the induced maps are isomorphisms for homologies with coefficients in the fields @xmath352 ( for any prime @xmath353 ) and in @xmath354 .
by @xcite the homology of @xmath242 is the free tensor algebra generated by @xmath355 .
the map @xmath356 gives rise to the map of homology : @xmath357 by theorem [ homomain ] the last map is an isomorphism .
we will concentrate on _ the directed suspension _ of a pointed d - space @xmath6 , i.e. , we assume that @xmath280 is either @xmath358 or @xmath359 .
we do not assume that @xmath6 is connected but it is important that , for every directed path @xmath360 , the composition @xmath361 is non - decreasing on any interval on which it is well - defined . to emphasize this , we will write @xmath28 rather than @xmath9 . a path @xmath360 is _ strictly increasing _ if , for every @xmath362 one of the following conditions are satisfied : * there exists @xmath363 such that @xmath364 , * the map @xmath365 is strictly increasing .
let @xmath366 be the space of strictly increasing paths , and let @xmath367 be the space of strictly increasing loops .
the inclusion @xmath368 is a homotopy equivalence . for every @xmath369 , the map @xmath370 @xmath371 where @xmath372 is the normalized parameter of @xmath64 , is a homotopy inverse .
@xmath373 time @xmath374coordinate of @xmath375 for an open neighborhood @xmath376 of @xmath171 , and a sequence @xmath377 of open subsets of @xmath6 such that @xmath378 , define the following subset of @xmath242 @xmath379 let @xmath380 denote the set @xmath242 with the topology generated by the sets @xmath381 .
the identity map @xmath382 is continuous but its inverse is not in general .
nevertheless , it induces a weak homotopy equivalence since it is homeomorphism on each compact subset .
let @xmath383 be the space of directed strictly increasing moore paths on @xmath384 that consist of finitely many segments .
for every path @xmath385 , the inverse image @xmath386 is a finite set . the map @xmath387 @xmath388 is continuous .
let @xmath377 , @xmath389 be open subsets of @xmath6 such that @xmath390 for all @xmath391 .
fix @xmath392 such that @xmath393 .
then there exists a sequence @xmath394 such that @xmath395 and @xmath396 .
furthermore , there exist @xmath397 such that @xmath398 for @xmath399 , @xmath400 for @xmath401t_i , t^b_i\right]$ ] and @xmath402 for @xmath403 $ ] .
let @xmath404 denote the space of paths @xmath64 such that @xmath405 .
the set @xmath406,\{\l s , x \r:\ ; 0<s<1,\ ; x\in u_i \ } ) \end{gathered}\ ] ] is an open neighborhood of @xmath407 contained in @xmath408 .
it is clear from the construction that for @xmath409 we have @xmath410\,,\ ] ] i.e. @xmath411 where @xmath412 is the map induced by the retraction @xmath254 . with the assumption that @xmath342 is connected , it follows from theorem [ homomain ] that @xmath263 is a homotopy equivalence .
below we present a construction of a retraction of @xmath413 to a subspace consisting of the paths which are , up to a reparametrization , in the image of @xmath258 .
if @xmath6 is a discrete d - space , then every directed loop @xmath414 is a non - decreasing reparametrization of a concatenation @xmath415 as a consequence , the map @xmath263 induces a continuous bijection between the space of traces ( cf .
* section 2 ) ) @xmath416 and @xmath380 .
we show a construction which allows to construct a deformation of an arbitrary path to the concatenation of paths of the form @xmath238 .
let @xmath417\to c_-\subset\vsigma x$ ] be a directed increasing path such that @xmath418 in order to construct a deformation , we introduce the map @xmath419
@xmath420 now let us define the homotopy @xmath421 i.e. , @xmath422\\ \l \tfrac{s - a}a - t , p(\alpha(a ) ) \r & \text{for } s\in[a , a(1+t)].\end{cases}\ ] ] the path @xmath423 is of length @xmath424 .
we rescale the parameter linearly in order to obtain a path of length @xmath183 .
the resulting deformation has the property @xmath425 and @xmath426 is a concatenation of a constant path with @xmath427 .
similarly , we deform paths
@xmath417\to c_+\subset\vsigma x$ ] such that @xmath428 by the formula @xmath429 where @xmath430 .
that is @xmath431\\ \psi^+_t(\alpha(s - at))&\text{for } s\in[at , a(1+t)].\\\end{cases}\ ] ] we reparametrize @xmath432 to have a path of length @xmath183 .
finally , we construct a deformation of a directed increasing path satisfying @xmath433 such path has a unique intersection with the set @xmath434 .
let @xmath435 be the parameter for which @xmath436 .
then @xmath64 can be written as the concatenation @xmath437 with @xmath438 of length @xmath439 .
we define a deformation @xmath440 the line coordinate of @xmath441 is piecewise linear ( increasing linearly on @xmath442 $ ] and on @xmath443 and constant elsewhere ) .
the path @xmath444 can be deformed further ( in an uniform way ) to obtain a path @xmath445 we have obtained a canonical deformation @xmath446 @xmath447 we omit the proof that this procedure is continuous with respect to @xmath64 .
the rigorous proof is quite tedious and we leave it to the reader since we do not need it in presence of theorem [ homomain ] and corollary [ cormain ] .
every path @xmath448 is a concatenation of a chain @xmath449 where the paths @xmath450 satisfy ( [ pojedyncza1][pojedyncza2 ] ) and @xmath451 are constant paths .
these constant paths can be deformed to paths of length zero in a canonical way and the paths @xmath450 are deformed using the homotopy defined in the precious subsection .
the resulting retraction @xmath452 leaves unchanged the paths which are the images @xmath453 .
let us summarize the results of the homological and geometric constructions . * for @xmath6 connected , according to corollary [ cormain ] , the map @xmath454 is a weak homotopy equivalence under suitable assumptions on the directed structure , including the minimal or completed structure for cubical complexes .
* we have constructed a map @xmath455 such that @xmath456 is equal to the natural projection . here
@xmath413 is a subset of @xmath342 , which is homotopy equivalent .
it follows that the map @xmath263 is essentially a homotopy inverse of @xmath251 . * in addition , we show how to deform an individual path to a path which is in the image of @xmath251 .
the deformation is done in a canonical way .
john milnor . on the construction @xmath457 . in john
frank adams , editor , _ algebraic topology a student s guide _ , pages 119136 .
cambridge university press , london - new york , 1972 .
london mathematical society lecture note series , no . | we prove that the loop space of the directed suspension of a directed space is homotopy equivalent to the james construction .
in particular , it does not depend on the directed structure of a given directed space .
= 1 |
giant molecular clouds ( gmcs ) are main sites of star formation .
there are approximately three thousand gmcs in our galaxy , with a size of 10 - 80 pc and a mass of 10@xmath11 - 5 @xmath3 10@xmath12 m@xmath13 @xcite .
while only low - mass stars form in dark clouds @xcite , in gmcs both low - mass and high - mass stars form @xcite .
there are many uncertain aspects of star formation in gmcs , such as the core mass function ( cmf ) , its link to the initial mass function ( imf ) , massive star formation , and cluster formation .
hence , studies of star formation in gmcs are imperative to the comprehensive understanding of star formation .
theoretical studies of massive star formation and cluster formation have proposed that external effects on molecular clouds , such as ultraviolet ( uv ) radiation from hii regions , supernova remnants ( snrs ) , molecular outflows , and cloud - cloud collisions are required to trigger the massive - star @xcite and cluster formation @xcite .
furthermore , it is proposed that the shape of the cmf could also be affected by the presence of those external effects @xcite .
hence , it is essential to unveil the global picture of those external effects on molecular gas in gmcs .
it is , however , not yet clear how ubiquitous those external effects and the triggered star formation in gmcs are .
unbiased and extensive mapping observations of gmcs in molecular and dust continuum emission , covering an entire gmc and resolving individual cores with a typical size of @xmath5 0.1 pc , and the comparison with the images at other wavelengths , are required to approach the global picture of star formation in gmcs .
the orion - a gmc , with an extent of @xmath5 @xmath14 ( corresponding to @xmath5 72 pc ) in the southern part of orion constellation , is the nearest ( @xmath15= 400 pc ; @xcite ) gmc and one of the well studied regions in star formation studies @xcite .
many authors have proposed that the orion - a gmc is externally affected , both at small ( @xmath5 0.1 pc ) and large scales ( @xmath5 several @xmath3 10 pc ) , by molecular outflow , ionizing radiation , snrs , and stellar winds generated by the orion ob association , from its morphological structure , velocity structure , and the stellar ages @xcite .
ob stars embedded in the orion - a gmc form the hii region , m 42 , 43 , and ngc 1977 @xcite .
molecular outflows driven from omc-2/fir 3 and fir 6c interact with the surrounding dense cores , which triggers gravitational instabilities in the dense cores and the next generation star formation @xcite .
these facts imply that the orion - a gmc is one of the most suitable gmcs for an unbiased extensive study of triggered star formation . with the advent of multi - beam receivers at millimeter wavelengths , wide - field mapping observations over several square degrees at an angular resolution of several @xmath3 10@xmath7
are now feasible . in this paper
, we present a wide , sensitive 1.1-mm dust - continuum and @xmath0co ( @xmath1=10 ) line imaging of the northern part of the orion a gmc with the aztec camera mounted on the aste telescope and bears mounted on the nobeyama ( nro ) 45-m telescope , respectively .
the aim of this paper is to reveal the overall structure of the orion - a gmc from a cloud ( @xmath16 10 pc ) to a core scale ( @xmath17 0.1 pc ) with the wide coverage and high sensitivity , and specifically , to unveil the global picture of the external triggers of star formation in the gmc from the comparison with the near - infrared ( nir ) and mid - infrared ( mir ) images .
the detailed physical processes of individual triggered star formation will be discussed in the forthcoming papers .
the 1.1 mm dust - continuum emission is optically thin and traces the total mass of molecular gas in cores and envelopes @xcite .
the @xmath0co ( @xmath1=10 ) emission traces low - density gas ( @xmath17 10@xmath18 @xmath19 ) and is usually optically thick and easily thermalized .
hence , the @xmath0co emission traces global structures of gmcs , and the peak intensities of the @xmath0co emission at the emission ridge often reflect kinetic temperature of molecular gas .
this paper is organized as follows : in section 2 , the nro 45-m and aztec / aste observations are described . in section 3 , we present results of the aztec 1.1 mm continuum and the nro @xmath0co ( @xmath1=10 ) mapping of the orion - a gmc . in section 4
, we focus on the possible external triggers of the star formation in the orion - a gmc .
first , we describe the possibility that the entire orion - a gmc is influenced over @xmath5 several @xmath3 10 pc scale by the external effect .
then , we show the region irradiated by the uv radiation from ob stars embedded in the orion - a gmc over a few pc - scale . in section 5 , we summarize this paper .
from october to december 2008 , we carried out 1.1 mm dust - continuum observations toward the @xmath20 @xmath3 @xmath21 region in the northern part of the orion a molecular cloud with the aztec camera @xcite equipped in the aste 10 m telescope @xcite located at pampa la bola ( altitude = 4800 m ) , chile
. the weather conditions during our observing period were moderate and the typical opacity at 225 ghz was in the range of 0.1 - 0.2 .
the aztec camera mounted on the aste telescope is an 144-element bolometric camera tuned to operate in the 1.1 mm atmospheric window , and provides an angular resolution of 28@xmath7 in full width at half maximum ( fwhm)@xcite .
observations were performed in the raster scan mode toward a pair of 100@xmath22 @xmath3 100@xmath22 fields , centered on ( @xmath23,@xmath24)=(@xmath25 ) and ( @xmath26 ) . each field
was observed several times with azimuth and elevation scans .
the separation among scans was adopted to be 117@xmath7 , which is a quarter of the aztec field of view ( fov;@xmath5 @xmath27 ) .
a scanning speed of the telescope was 250@xmath7 sec@xmath6 . totally 30 individual maps in the entire field with an integration time of 26 minutes were taken , and those maps were averaged to produce the final map with the total integration time of 26 minutes @xmath3 30 maps = 13 hours .
the telescope pointing was checked every 2 hours by observing a quasar 0539 - 057 over a 4 @xmath3 4 arcmin@xmath28 region .
the derived pointing offsets were linearly interpolated along the time sequence and the interpolated pointing offset was applied to each target map .
the pointing uncertainty of the aztec map is estimated to be better than 2@xmath7 .
we observed uranus as a flux calibrator twice per night . by observing uranus with each aztec detector element
, we measured the flux conversion factor ( fcf ) from the optical loading value ( in watts ) to the source flux ( in jy beam@xmath6 ) for each detector element . the principal component analysis ( pca , @xcite ) cleaning method
was applied to remove the atmospheric noise .
details of the flux calibration are described by @xcite and @xcite . since the pca method does not have sensitivity to extended sources
, we apply the iterative mapping method ( fruit , @xcite ) to recover the extended components .
the noise level is @xmath5 9 mjy beam@xmath6 in the central region and @xmath5 12 mjy beam@xmath6 in the outer edge , the effective beam size is @xmath5 40@xmath7 , after the fruit imaging .
we describe the performance of fruit in appendix 1 . from december 2007 to may 2008 , we carried out @xmath0co ( @xmath1=10 ; 115.271204 ghz ) line observations toward the @xmath29 @xmath3 @xmath30 region in the northern part of the orion a molecular cloud with the 25-element focal plane receiver bears equipped in the 45 m telescope at nro . at 115 ghz
, the telescope has a fwhm beam size of 15@xmath7 and a main beam efficiency of 32 % .
the beam separation of the bears is @xmath31 on the plane of the sky @xcite . at the back end
, we used 25 sets of 1024 channel auto - correlators ( acs ) which have the bandwidth of 32 mhz and the frequency resolution of 37.8 khz @xcite .
the frequency resolution corresponds to the velocity resolution of @xmath5 0.1 km s@xmath6 at 115 ghz . during the observations , the system noise temperatures were in the range between 250 and 500 k in a double sideband .
the standard chopper wheel method @xcite was used to convert the output signal into the antenna temperature @xmath32 ( k ) , corrected for the atmospheric attenuation .
the telescope pointing was checked every 1.5 hours by observing a sio maser source , ori - kl , and was better than 3@xmath7 during the whole observing period . to correct the gain variation among the 25 beams of the bears
, we used calibration data obtained from the observations toward orion irc 2 using another sis receiver , s100 , with a single sideband ( ssb ) filter and acousto - optical spectrometers ( aoss ) as a back end .
our mapping observations were made with the otf mapping technique @xcite .
we used the emission - free area near ( @xmath5 @xmath33 away ) the observed area as the off position .
we obtained an otf map with a scanning direction both along the ra and dec over the @xmath29 @xmath3 @xmath30 region and combined them into a single map , in order to reduce the scanning effects .
we adopted a convolutional scheme with a spheroidal function @xcite to calculate the intensity at each grid point of the final map - cube data with a spatial grid size of @xmath34 , half of the beam size .
the resultant effective resolution is 21@xmath7 .
the rms noise level of the final map is 0.94 k in @xmath9 ( 1 @xmath35 ) at a velocity resolution of 1.0 km s@xmath6 .
we summarize the parameters for the 1.1 mm continuum and the @xmath0co ( @xmath1=10 ) line observations in table [ obs ] .
figure [ aztec ] shows the distribution of the 1.1 mm dust - continuum emission in the northern part of the orion a molecular cloud .
previous observations of the northern part of the orion a molecular cloud in dust - continuum emissions at 850 @xmath10 m @xcite , 1.2 mm @xcite , and 1.3 mm @xcite have revealed an `` integral - shape '' filamentary feature from the south to the north of ori - kl ( @xmath23 , @xmath24 = @xmath36 ) , where the northern end , region surrounding ori - kl , and the southern region of ori - kl in the integral - shape filament are called omc-2/3 , omc-1 , and omc-4 , respectively .
the global feature seen in our 1.1 mm dust - continuum map is consistent with that found in the previous observations , that is , the integral - shape filament along the north - south direction ( named as `` integral - shape filament '' in figure [ aztec2 ] ) . the central orion - kl region has been masked out , because around the ori - kl region the continuum emission is too bright to be reconstructed as an accurate structure with the aztec observing and data - reduction technique ( see appendix 1 ) .
our mapping area extends @xmath5 @xmath37 along the eastern and the western directions centered on the integral - shape filament , and is wider than that of the previous observations . as a result ,
we have found new dust - continuum features in the eastern and the western areas of the integral - shape filament , as described below . toward the east of the omc-2/3 region ,
our high - sensitivity ( @xmath5 9 mjy ) 1.1 mm continuum observations have revealed the presence of another filamentary structure ( named as `` new filament '' in figure [ aztec2 ] ) .
in addition , a continuum feature extending almost perpendicular to the integral - shape filament centered on mms 9 ( @xmath23 , @xmath24 = @xmath38 ) was detected clearly at a signal to noise ratio of @xmath16 50 @xmath35 , although this feature was not clearly seen in the previous studies ( named as `` east - west extension '' in figure [ aztec2 ] ) . in the most eastern side of this structure , a cavity - like structure with a size of 0.4 pc was found at a central position of @xmath23 , @xmath24 = @xmath39 ( named as `` cavity - shape '' in figure [ aztec2 ] ) .
the distribution of the 1.1 mm dust - continuum emission around the hii region , m 43 , which is excited by a b0.5 star , nu ori ( @xmath23 , @xmath24 = @xmath40 ) @xcite , shows a shell - like structure with a size of @xmath5 0.5 pc ( named as `` shell '' in figure [ aztec2 ] ) . toward the south of m 42 ( omc-4 region )
the 1.1 mm dust - continuum emission shows a `` v - shape '' structure with a size of @xmath5 2 pc , which is also seen in the 850 @xmath10 m continuum emission @xcite , and a cavity - like structure with a central position of ( @xmath23 , @xmath24 = @xmath41)(named as `` cavity '' in figure [ aztec2 ] ) . toward the northwest of m 42 , the 1.1 mm dust - continuum emission forms an `` l - shape '' structure with a size of @xmath5 0.5 pc ( named as `` l - shape '' in figure [ aztec2 ] ) . as well as
the integral - shape filament there appears a branch of the 1.1 mm dust - continuum emission extending toward the southeast of ori - kl .
the branch forms a filamentary structure with a length of @xmath5 3 pc along @xmath23 @xmath5 @xmath42 .
this filament is known as one of photo - dominated regions ( pdrs ) from the cn line observations @xcite and is named as `` dark lane south filament '' ( dlsf , named as `` dlsf '' in figure [ aztec2 ] ) .
furthermore , toward the opposite side of dlsf with respect to the integral - shape filament there is another filamentary structure with a size of @xmath5 3 pc , which is bending at the middle ( named as `` bending structure '' in figure [ aztec2 ] ) . in the southmost region
where an active cluster - forming region of l 1641 n @xcite resides , four filamentary structures with a size of @xmath5 0.5 - 2.0 pc are seen aligned almost in parallel ( named as `` paralleled filament '' in figure [ aztec2 ] ) .
figure [ co_alla ] and [ co_allb ] show the total integrated and the peak intensity maps of the @xmath0co ( @xmath1=10 ) emission in the northern part of the orion a molecular cloud , respectively .
the main feature seen in these maps is the same as that in the 1.1 mm continuum map , namely , integral - shape filament . at the eastern edge of the integral - shape filament
there appears a sharp emission contrast while at the western edge the emission intensity decreases gradually .
hereafter we call the sharp emission contrast at the eastern edge of the integral - shape filament `` co front '' ( see figure [ co_allc ] ) . in the total integrated intensity map the brightest point at the center corresponds to the location of ori - kl . toward the south - east of ori - kl
there appears a bar - like feature with a length of @xmath5 0.6 pc , which is more evident in the peak intensity map .
this feature corresponds to `` orion - bar '' ( see figure [ co_allc ] ) . as well as the @xmath0co components associated with the integral - shape filament
, there appears a branch of the @xmath0co emission extending toward the southeast from ori - kl .
this @xmath0co structure corresponds to dlsf seen in the aztec 1.1 mm dust - continuum emission ( see figure [ co_allc ] as well as figure [ aztec2 ] ) .
similarly to co front , there is a sharp emission contrast at the western side of dlsf .
in addition to these filamentary features , a @xmath0co emission extending to the east from orion - kl and a diffuse co emission at the west of the integral - shape filament are seen .
these @xmath0co emission features are consistent with the features found by the previous @xmath0co ( @xmath1=10 ) observations at an angular resolution of 45@xmath7 and a grid spacing of 45@xmath7 @xcite . toward the emission ridge , the peak intensity map of the @xmath0co ( @xmath1=10 )
emission probably shows the temperature structure , under the condition that the @xmath0co ( @xmath1=10 ) emission is optically thick and thermalized .
the temperature around ori - kl is the highest ( @xmath5 110 k ) , and the temperature around ngc 1977 is relatively high ( @xmath5 64 k ) .
figure [ co_channel ] shows the velocity channel maps of the @xmath0co ( @xmath1=10 ) emission with a velocity width of 1.0 km s@xmath6 . in the high - velocity range (
@xmath43 = 2.5 - 3.5 @xmath44 15 - 18 km s@xmath6 ) , a point - like @xmath0co emission associated with ori - kl is seen at ( @xmath23 , @xmath24 ) = ( @xmath25 ) . from @xmath43
= 2.5 to 6.5 km s@xmath6 , there is another compact @xmath0co component seen toward the west of ori - kl at ( @xmath23 , @xmath24 ) = ( @xmath45 ) .
this @xmath0co component is associated with the aztec 1.1 mm dusty condensation ( named as `` region b '' , see 3.3 ) . from @xmath43
= 4.5 km s@xmath6 to 14 km s@xmath6 the main integral - shape filament shows a velocity gradient from the south to the north of ori - kl , as already reported by previous observations @xcite . on the other hand , from @xmath43 = 5.5 to 8.5 km s@xmath6 the @xmath0co emission originated from dlsf shows a velocity gradient from the north to the south ( see figure [ co_allc ] ) , which suggests that the velocity gradient of the @xmath0co emission associated with dlsf is opposite compared to the large scale velocity gradient of the orion a molecular cloud .
the central velocity of dlsf ( @xmath5 7 km s@xmath6 ) is also different from that of the integral - shape filament . hence , dlsf is likely to be kinematically distinct from the main cloud component . in the velocity range of 12 to 16 km
s@xmath6 , the @xmath0co emission forms a shell - like structure with a size of @xmath5 2 pc toward the south of ori - kl ( named as `` shell around ori - kl '' in figure [ co_allc ] ) . @xcite and @xcite also found this shell - like structure around ori - kl in the @xmath46co ( @xmath1=10 ) and the @xmath0co ( @xmath1=10 ) lines observations . as well as these emission - ridge components , in the velocity range of 3.5 - 7.5 km s@xmath6
there is extended diffuse co component at the eastern region of the integral - shape filament .
the peak intensity and the velocity width of the diffuse co component are @xmath5 5 k and @xmath5 5 km s@xmath6 , respectively . the origin of this rather featureless and extended @xmath0co emission is likely to be different from that of the ridge components .
@xcite also found the diffuse co component with a velocity width of @xmath5 7 km s@xmath6 and a peak intensity of @xmath5 5 k in the east of the l 1641 n region , which locates in the southern part of the aztec 1.1 mm continuum map ( figure [ aztec ] ) and out of our @xmath0co observing region .
the diffuse co component in the east of the l 1641 n region found by @xcite is probably the same component as that in the east of the omc-2/3 region and in the east of dlsf .
their and our results suggest that the diffuse co component distributes along the east of the integral - shape filament over the extent of @xmath16 14 pc
. we will discuss the origin of the diffuse co component in more detail .
figure [ co_alla ] shows the mean velocity map of the @xmath0co ( @xmath1=10 ) emission .
the large - scale velocity gradient from the south to north can be confirmed in this mean velocity map as already shown in the channel maps ( see figure [ co_channel ] ) .
figure [ co_allb ] shows the velocity dispersion map of the @xmath0co ( @xmath1=10 ) emission . the velocity dispersion map shows the increase of the velocity dispersion toward the east of co front in the omc-2/3 region ( see figure [ co_allc ] ) .
moreover , there is a shell - like structure around ori - kl in the velocity dispersion map ( see figure [ co_allc ] ) .
figure [ multi ] ( a)-(d ) show the aztec 1.1 mm , @xmath0co ( @xmath1=10 ) peak intensity , 2mass , and the midcourse space experiment ( msx ) 8 @xmath10 m maps in the @xmath0co observing region , respectively .
the 8 @xmath10 m emission mostly arises from polycyclic aromatic hydrocarbons ( pahs ) and has been observed at the boundary of the hii regions @xcite .
recent laboratory experiments have shown that pahs are produced on the dust grains irradiated by the fuv radiation @xcite .
hence , presence of the strong 8 @xmath10 m emission suggests presence of the fuv radiation from hii regions .
while the 1.1 mm and the @xmath0co images show the distribution of the molecular gas , the 2mass and the 8@xmath10 m images show three bright nebulae , m 42 , m 43 , and ngc 1977 .
although the millimeter and infrared emission distributions appear quite distinct from each other , detailed comparisons among the 1.1 mm , @xmath0co , and the 8 @xmath10 m images show their spatial correlations in several regions .
figure [ pdr ] ( a ) and ( b ) compare the 1.1 mm continuum distribution to the msx 8 @xmath10 m and the @xmath0co ( @xmath1=10 ) emission distributions in the m 43 and dlsf region . in this region
there are two hii regions , m 42 and m 43 , excited by @xmath47@xmath48 ori c and nu ori , respectively .
there are systematic trends of the 8 @xmath10 m , 1.1 mm , and , the @xmath0co emission distributions as a function of the distance from the hii regions .
toward m 43 , there is a shell - like structure in the @xmath0co , 1.1 mm , and the 8 @xmath10 m emissions , however , the radius of the shell decreases systematically from the @xmath0co , 1.1 mm , and the 8 @xmath10 m emissions .
dlsf is likely to be irradiated by @xmath47@xmath48 ori c and the @xmath0co , 1.1 mm , and the 8 @xmath10 m emissions show systematic emission distributions as a function of the distance from @xmath47@xmath48 ori c ; the @xmath0co emission distribution is located farthest and the 8 @xmath10 m closest .
we have also identified other four regions with the systematic @xmath0co , 1.1 mm , and the 8 @xmath10 m emissions distribution .
these regions are labeled as region a - d , and the locations of region a - d are shown in figure [ co_allc ] .
figure [ aztec_8a ] - [ aztec_8e ] show the same maps as those in figure [ pdr ] but for region a - d , respectively . in region
a where the 1.1 mm continuum peak locates at ( @xmath23 , @xmath24)=(@xmath49 ) , the nearest ( @xmath15= @xmath5 1.4 pc ) hii region , m 42 , is located on the southeast from the continuum peak ( see an arrow in figure [ aztec_8a ] ) .
the 8 @xmath10 m , 1.1 mm , and the @xmath0co emissions show the systematic distributions along the direction to the hii region ; the 8 @xmath10 m emission distributes on the southeastern part of the 1.1 mm dust - continuum condensation , while the @xmath0co emission distributes on the central part of the 1.1 mm dust - continuum condensation . in region b where the 1.1 mm continuum peak locates at ( @xmath23 , @xmath24)=(@xmath50 ) , the nearest ( @xmath15= @xmath5 2.3 pc ) hii region , m 42 , is located on the east from the continuum peak ( see an arrow in figure [ aztec_8b ] ) . the 8 @xmath10 m and
the @xmath0co emissions show the positional offset along the direction to the hii region , and the 8 @xmath10 m emission locates on the side of the hii region .
the dusty condensation in region b is associated with a spitzer 24 @xmath10 m source . in region c where the 1.1 mm continuum peak locates at ( @xmath23 , @xmath24)=(@xmath51 ) , the nearest ( @xmath15= @xmath5 3.2 pc ) hii region , m 42 , is located on the north from the continuum peak ( see an arrow in figure [ aztec_8c ] ) .
the 8 @xmath10 m , @xmath0co , and the 1.1 mm emissions show the systematic distributions along the direction to the hii region ; the 8 @xmath10 m emission distributes on the northern part of the 1.1 mm dust - continuum condensation , while the @xmath0co emission distribution is similar to the 1.1 mm emission distribution . in region d where the 1.1 mm continuum peak locates at ( @xmath23 , @xmath24)=(@xmath52 ) , the nearest ( @xmath15= @xmath5 2.0 pc ) hii region , m 42 , is located on the north from the continuum peak ( see an arrow in figure [ aztec_8e ] ) .
the 8 @xmath10 m , 1.1 mm , and the @xmath0co emissions show the systematic distributions along the direction to the hii region ; the 8 @xmath10 m emission distributes on the northern part of the 1.1 mm dust - continuum condensation , while the @xmath0co emission distributes on the southern part of the 1.1 mm dust - continuum condensation .
one of the first findings in our extensive @xmath0co ( @xmath1=10 ) emission observations of the northern part of the orion a molecular cloud is the presence of the diffuse co component over the entire eastern region of the integral - shape filament .
figure [ diffuse ] highlights the total integrated intensity map of the diffuse co component with respect to the peak intensity map at the omc-2/3 region , and figure [ ra_pv ] shows position - velocity ( p - v ) diagrams whose cut lines are along the ra direction shown in figure [ diffuse ] . in the p - v diagrams ,
the central velocity of the diffuse co component is 5 - 6 km s@xmath6 , while that of the main component is 11 - 12 km s@xmath6 .
these two velocity components appear to be connected at co front , raising the @xmath0co velocity dispersion at co front .
figure [ int ] shows peak intensity profiles whose cut lines are the same as that of the p - v diagrams in figure [ ra_pv ] .
while the peak intensity of the diffuse co component is around 4 - 7 k at @xmath9 , that of the main cloud component exceeds 50 k , and hence there is an abrupt increase of the @xmath0co peak intensity at co front .
in contrast , toward the western outskirts of the main cloud component the @xmath0co peak intensity decreases gradually ( see figure [ co_alla ] - [ co_allc ] ) .
these peculiar features extend over the entire co front as can be seen in figures [ co_alla ] - [ co_channel ] , although toward the southern part of co front the difference of the central velocities becomes smaller .
the reason why the velocity difference between the diffuse co component and the main cloud component becomes less clear is the large scale velocity gradient of the main cloud component ( see 3.2 ) .
there are two possible interpretations on the origin of the diffuse co component ; one is the trail of the main cloud component moving away from the galactic plane and the other is the pre - existing surrounding diffuse gas swept up by neighboring external sources .
@xcite has suggested these two possibilities on the origin of the diffuse co component in the east of the l 1641 n region .
it is , however , difficult to explain the difference of the central velocity between the diffuse co component and the main cloud component and the abrupt increase of the @xmath0co peak intensity by the former interpretation .
we consider that the `` l - shaped '' feature of the diffuse co component in the p - v diagrams and the abrupt increase of the @xmath0co peak intensity imply presence of the interaction between the diffuse co component and the main cloud component @xcite , and that the diffuse co component is being accumulated on to the surface of the main cloud component at co front .
in fact , @xcite , @xcite and @xcite have suggested that ori ob 1a and ori ob 1b , @xmath5 100 pc away from the orion a molecular cloud , affect the orion a molecular cloud and disturb the morphology and kinematics of the cloud , respectively .
these ob stars are the possible sources to sweep up the pre - existing diffuse material toward the direction to the main cloud component .
our observational results favor the proposal by @xcite , since ori ob 1b locates in the east of the orion a molecular cloud and the diffuse
co component also distributes toward the same direction from the orion a molecular cloud .
the collision of the diffuse co component with the main cloud component might trigger the next star formation in the cloud @xcite . in the last section ,
we suggest the possibility that the diffuse co component is swept up by ori ob 1b and interacts with the entire orion a molecular cloud . in this section
, we focus on more local effects in a few pc - scale , namely , the hii regions embedded in the orion a molecular cloud . as well as the external sources away from the cloud , newly born ob stars embedded in the cloud can also influence on the structure , chemistry , thermal balance , and the star formation in the cloud @xcite .
there are several ob stars ( such as @xmath47@xmath48 ori c and nu ori ) and hii regions in our mapping area of the orion a molecular cloud and these sources are likely to affect star formation in the orion a molecular cloud .
as shown in 3.3 , we found that there are systematic differences of the distribution among the @xmath0co , 1.1 mm , and the 8 @xmath10 m emissions and that the 8 @xmath10 m , 1.1 mm , and the @xmath0co emissions are distributed as a function of the distance from the ob stars .
the most clear example is in dlsf irradiated by @xmath47@xmath48 ori c , and the intensity profiles as a function of the distance from @xmath47@xmath48 ori c are shown in figure [ int_profile ] .
the presence of the strong 8 @xmath10 m emission suggests the fuv radiation from the hii regions . on the other hand ,
@xmath0co molecules are dissociated by the fuv radiation , since the energy of fuv ( 6 ev - 13.6 ev ) is high enough to dissociate @xmath0co molecules with an dissociative energy of 11.4 ev .
hence , the @xmath0co emission does not distribute at the near side of the exciting star .
the 1.1 mm emission traces both the cold and warm dusts .
hence , the systematic distributions of the 8 @xmath10 m , 1.1 mm , and the @xmath0co emissions can be explained by the pdr model @xcite .
in fact , toward dlsf and m 43 , where the systematic distributions of the @xmath0co , 1.1 mm , and the 8 @xmath10 m emissions are seen , the observations of the cn line , one of the excellent tracers for pdr , suggest the presence of pdr .
region a - d , relatively isolated dusty condensations , also exhibit the systematic distributions of the @xmath0co , 1.1 mm , and the 8 @xmath10 m emissions along the direction to the exciting sources .
these isolated condensations are likely to be pre - existing molecular cloud cores .
when an expanding hii region interacts with a pre - existing molecular cloud core , the strong uv radiation will compress the core .
this compression could induce the gravitational collapse of the core @xcite .
our results suggest that the dusty condensations of region a - d are affected by the uv radiation from the ob stars .
moreover , the dusty condensation of region b is associated with the spitzer 24 @xmath10 m sources , suggesting the presence of the star formation .
we have carried out extensive mapping observations of the northern part of the orion a molecular cloud in the 1.1 mm dust - continuum emission with the aztec camera equipped in the aste telescope and in the @xmath0co ( @xmath1=10 ) emission with the bears receiver equipped in the nro 45 m telescope .
the main results of our wide - field ( @xmath53 ) and high - sensitivity ( @xmath5 9 mjy beam@xmath6 ) observations are summarized as follows ; 1 . we have found new features of the 1.1 mm dust - continuum emission in addition to the previously known integral - shape filament . in the omc-2/3 region , a new dusty filamentary structure is detected toward the east of the integral - shape filament .
a dusty shell - like structure around the hii region , m 43 , and a filamentary structure associated with dlsf are detected . in the southmost region where an active cluster - forming region of l
1641 n is located , four filamentary structures with a size of @xmath5 0.5 - 2.0 pc are seen aligned almost in parallel . 2 . in the @xmath0co ( @xmath1=10 )
line observations , we found diffuse co component over the entire eastern region of the integral - shape filament .
the peak intensity profiles from the diffuse co component to the integrals - shape filament show an abrupt increase at the border line ( co front ) and the p - v diagrams show the sudden velocity change from 5 - 6 km s@xmath6 to 11- 12 km s@xmath6 .
we suggest that these peculiar features of the diffuse co component show that the diffuse material swept up by orion ob 1b , @xmath5100 pc away from the orion a molecular cloud , is being accumulated onto the surface of the integral - shape filament .
comparisons among the @xmath0co , 1.1 mm , and the 8 @xmath10 m emission distribution in m 43 , dlsf , and region a - d have revealed that the 8 @xmath10 m , 1.1 mm , and the @xmath0co emissions are located sequentially as a function of the distance from the neighboring ob stars .
these results suggest the presence of the pdrs around m 43 , dlsf , and region a - d , excited by the nu ori , @xmath47@xmath48 ori c , and @xmath47@xmath48 ori c , respectively .
in particular , the dusty condensation of region b is associated with a spitzer 24 @xmath10 m source , suggesting the presence of the star formation .
region a - d are possible site of star formation triggered by the pdrs .
we thank the referee , charles lada , for his constructive comments that polish the manuscript significantly .
we are grateful to the staffs at the nobeyama radio observatory ( nro ) for both operating the 45 m and helping us with the data reduction , and to nro is a branch of the national astronomical observatory , national institutes of natural sciences , japan .
we also acknowledge the aste staffs for both operating aste and helping us with the data reduction .
observations with aste were ( in part ) carried out remotely from japan by using ntt s gemnet2 and its partner r&e ( research and education ) networks , which are based on accessnova collaboration of university of chile , ntt laboratories , and national astronomical observatory of japan .
this research has made use of the nasa/ ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
this work was supported by grant - in - aid for scientific research a 18204017 .
y. shimajiri is financially supported by a research fellowship from the jsps for young scientists .
in this appendix , we present our imaging simulation with artifact gaussian sources as input models , in order to test the performance of the aztec data reduction . the aztec data set is reduced by using the available aztec data reduction pipeline written in idl and developed by the university of massachusetts according to guidelines in @xcite .
this pipeline includes the despiking , atmosphere removal , and optimal filtering .
the despiking which is removal of the cosmic ray is , however , not applied to orion data reduction to avoid the removal of source emission , since the emission in the orion region is too strong .
principal components analysis ( pca ) cleaning method is applied for removal of the atmospheric noise in this pipeline .
finally , the 30 individual maps were coadded .
this data reduction pipeline has , however , been optimized for the identification of point sources .
the pca cleaning removes the astronomical signals in the case that the emission has an extended structure , since the signals which are a good correlation among each element are regarded as the atmospheric components and are removed .
hence , the flux and size of the extended sources are underestimated . in order to recover the extended features in the orion map which is subtracted by the pca cleaning
, we apply an iterative mapping method to the orion data - set .
hereafter , we call this method , fruit .
this method is based on an iterative mapping method applied to the bolocam data reduction pipeline @xcite .
the fruit data reduction pipeline is also written in idl and developed by the university of massachusetts .
the fruit algorithm is described in @xcite . in order to investigate the performance of fruit
, we have performed the simulation that the twenty five gaussian sources with a peak flux of 1 jy and a fwhm size of 30 , 60 , 90 , 120 , 150 , and 180@xmath7 are inserted to orion data - set and are applied the data reduction pipeline .
figure [ recover ] shows the fraction of the over - estimated ( positive ) or missed ( negative ) flux , @xmath54 as a function of the fwhm size of the input model gaussian . we have applied the aperture photometry to estimate the total flux .
error bars show a standard deviation ( 1 @xmath35 ) of the output total flux among twenty five sources . in the case
that the fwhm size of the input model gaussian is under 150@xmath7 , the output total flux is underestimated by less than 20% .
the total - flux reproductivity of the source with larger fwhm is lower than that with smaller fwhm .
we insert the twenty five sources with a peak flux of 1 , 20 , 40 , 60 , and 80 jy and a fwhm size of 60@xmath7 to the orion data - set and applies the same data reduction pipeline .
figure [ recover_strong ] shows the fraction of the over - estimated ( positive ) or missed ( negative ) flux as a function of the peak flux of the input model gaussian .
the total flux of sources with a peak flux under 20 jy is underestimated by less than 10 % .
the reproductivity of the total flux is , however , low for the source with a peak flux over 20 jy . in the case of orion
, the total flux of sources in our map should be recovered by more than 80 % , since the peak flux and the fwhm size of most sources are smaller than 20 jy and 150@xmath7 .
moreover , the restored image of the 1.1 mm dust continuum emission is consistent with that of the scuba 850 @xmath10 m dust continuum emission as shown in figure [ aztec_scuba2 ] .
lcc telecope / receiver & aste / aztec & nro 45 m / bears + line / wavelength & 1.1 mm & @xmath0co ( @xmath1=10;115.271 ghz ) + observation date & oct .
2008 - dec .
2008 & nov .
2007 - may 2008 + observing mode & raster & otf + observation time & 15 hr & 40 hr + mapping size & @xmath20 @xmath3 @xmath21 & @xmath29 @xmath3 @xmath30 + effective beam size & 40@xmath7 & 21@xmath7 + velocity width & & 1.0 km s@xmath6 + typical r.m.s . & 9 mjy beam@xmath6 & 0.9 k + | we present new , wide and deep images in the 1.1 mm continuum and the @xmath0co ( @xmath1=10 ) emission toward the northern part of the orion a giant molecular cloud ( orion - a gmc ) .
the 1.1 mm data were taken with the aztec camera mounted on the atacama submillimeter telescope experiment ( aste ) 10 m telescope in chile , and the @xmath0co ( @xmath1=10 ) data were with the 25 beam receiver ( bears ) on the nro 45 m telescope in the on - the - fly ( otf ) mode .
the present aztec observations are the widest @xmath2 @xmath3 @xmath4 , corresponding to 12 pc @xmath3 17 pc ) and the highest - sensitivity ( @xmath59 mjy beam@xmath6 ) 1.1 mm dust - continuum imaging of the orion - a gmc with an effective spatial resolution of @xmath5 40@xmath7 .
the @xmath0co ( @xmath1=10 ) image was taken over the northern @xmath8 ( corresponding 9 pc @xmath3 9 pc ) area with a sensitivity of 0.93 k in @xmath9 , a velocity resolution of 1.0 km s@xmath6 , and an effective spatial resolution of 21@xmath7 . with these data ,
together with the msx 8 @xmath10 m , spitzer 24 @xmath10 m and the 2mass data , we have investigated the detailed structure and kinematics of molecular gas associated with the orion - a gmc and have found evidence for interactions between molecular clouds and the external forces that may trigger star formation .
two types of possible triggers were revealed ; 1 ) collision of the diffuse gas on the cloud surface , particularly at the eastern side of the omc-2/3 region , and 2 ) irradiation of uv on the pre - existing filaments and dense molecular cloud cores . our wide - field and high - sensitivity imaging
have provided the first comprehensive view of the potential sites of triggered star formation in the orion - a gmc . |
extensive theoretical and experimental work has been devoted to the study of amorphous alloys of rare earths and transition metals , for instance tbfe@xmath7 and yfe@xmath7 .
they are modeled @xcite by a heisenberg hamiltonian with random uniaxial single - site anisotropy defined on a simple cubic lattice , or , in short , by the random - anisotropy model ( ram ) @xmath8 where @xmath9 is a three - component spin variable , @xmath2 is a unit vector describing the local ( spatially uncorrelated ) random anisotropy , and @xmath10 the anisotropy strength . in amorphous alloys
the distribution of @xmath2 is isotropic , since , in the absence of crystalline order , there is no preferred direction .
random anisotropy is a relevant perturbation of the pure heisenberg model , the crossover exponent being @xcite @xmath11 .
therefore , random - anisotropy systems show a behavior that is different from that observed in pure heisenberg systems .
even though the critical behavior of the ram has been investigated at length in the last thirty years ( see @xcite for a review ) we do not have yet a satisfactory picture of its critical behaviour . the imry - ma argument @xcite forbids the presence of a low - temperature phase with nonvanishing magnetization for @xmath12 .
however , this does not exclude the appearance of a glass transition with a low - temperature phase characterized by quasi - long - range order ( qlro ) , i.e. , a phase in which correlation functions decay algebraically @xcite .
functional renormalization - group calculations @xcite indicate that qlro may set in for small values of @xmath10 , in agreement with a landau - ginzburg calculation @xcite of the equation of state for @xmath13 . in the opposite limit @xmath14
the ram hamiltonian reduces to that of an ising spin glass with a correlated bond distribution . indeed , for @xmath15
one can write @xmath16 , where @xmath17 is an ising spin , and obtain the hamiltonian @xmath18 we call this model strong random - anisotropy model ( sram ) ( we set @xmath19 without loss of generality ) . model ( [ raminfd ] ) differs from the usual ising spin glass model ( isgm ) in the bond distribution : in hamiltonian ( [ raminfd ] ) the random variables @xmath20 on different lattice links are correlated .
for instance , one has @xmath21 , where the product is over the links belonging to a given plaquette and the average is taken with respect to the distribution of the vectors @xmath2 .
an interesting hypothesis , originally put forward in @xcite , is that the sram transition is in the same universality class as that of the isgm .
this conjecture looks very plausible , since the sram is nothing but an ising model with local disorder and frustration . in some sense
we can think of the sram and of the isgm as two different versions of the same model : in the sram disorder is associated with lattice sites , while in the isgm disorder is associated with lattice bonds .
they are analogous to the site - diluted and bond - diluted ising model @xcite , whose hamiltonian is given by ( [ raminfd ] ) with @xmath22 ( site dilution ) and @xmath23 ( bond dilution ) , @xmath24 being a random variable such that @xmath25 ( @xmath26 ) with probability @xmath27 ( resp .
@xmath28 ) . note that the site - diluted model can also be interpreted as a bond - diluted model with a correlated bond distribution , exactly as is the case for the sram .
nonetheless , there is little doubt though a satisfactory numerical check is still missing that the two models belong to the same universality class .
note that the sram is less frustrated than the standard isgm , since in the sram @xmath21 .
this fact does not rule out the conjecture since it is known that maximal frustration is not necessary to obtain glassy behaviour .
for instance , the random - bond ising model with @xmath29 with probability @xmath27 and @xmath30 with probability @xmath28 has a glassy low - temperature phase for @xcite @xmath31 .
the identity of the sram and isgm universality classes was confirmed in two dimensions by a renormalization - group calculation using the large - cell method @xcite , though in this case the critical point is at @xmath32 . in three dimensions instead
, the critical behaviour of the sram has been controversial for a long time . while , for small @xmath10 , numerical simulations @xcite provided some evidence of the existence of a finite - temperature transition ( though qlro was never observed ) , in the sram even the existence of a finite - temperature transition was in doubt @xcite .
in @xcite we study again the sram and find good evidence for the existence of a finite - temperature transition .
the corresponding critical behaviour turns out to be compatible with the conjecture of @xcite .
note that at the transition only the overlap variables , which are the usual order parameters at a spin - glass transition , become critical .
magnetic quantities do not show any critical behaviour , though we expect nonanalyticities induced by the critical modes .
thus , on both sides of the transition the system is paramagnetic .
it is unknown whether this paramagnetic phase survives up to @xmath32 .
note that our results predict an isgm transition also in a generalized sram in which the vectors @xmath33 are @xmath34 dimensional , for @xmath35 .
indeed , as @xmath34 increases , the bond correlation decreases and , for @xmath36 , one reobtains the isgm although with a nonstandard bond distribution . for @xmath37 instead a _
ferromagnetic _ phase transition would be possible since the system is less frustrated than the case we have considered .
note that such a transition would only be observed in correlations of @xmath38 , where the ising variables @xmath39 should be chosen such as to have the couplings @xmath40 ferromagnetic on a maximal set of lattice links .
this ferromagnetic transition would not violate the imry - ma argument , since order in the @xmath41 variables does not imply order in the continuous variables @xmath9 .
most likely , as for @xmath42 , the low - temperature phase would still be paramagnetic even if some ising variables magnetize .
a question that remains open is the behaviour of the ram for finite anisotropy @xmath10 .
if there is indeed a low - temperature phase with qlro for small @xmath10 as predicted in @xcite , then there should be a critical value @xmath43 such that isgm behaviour is observed only for @xmath44 .
nothing is known about @xmath43 and we can not even exclude that @xmath45 , so that isgm behaviour is observed only for model ( [ raminfd ] ) .
in ref . @xcite we study the critical behaviour of the sram by means of monte carlo ( mc ) simulations . since the model is essentially a spin glass we focus on the so - called overlap variables @xmath46 , where @xmath41 and @xmath47 are associated with two different replicas of the model with the same bond variables . for
the sram one can also consider the standard magnetic variables @xmath48 .
we find , in agreement with the results of @xcite , that these quantities are not critical , i.e. on the low - temperature side of the transition the system is still paramagnetic .
we study the behaviour of the sram in the high - temperature phase .
this reduces the algorithmic problems the mc algorithm becomes very slow as temperature is reduced and allows us to consider lattices of size @xmath3 up to @xmath49 . for the mc dynamics , we combine the metropolis algorithm with the random - exchange ( parallel - tempering ) method @xcite . in order to verify whether the system has a critical behaviour , we looked for the occurence of finite - size scaling ( fss ) . for this purpose
, we considered the ratios @xmath50 with @xmath51 , @xmath52 ( both quantities are associated with the two - point function of the overlap variables ) , fixing @xmath53 . if fss holds , as @xmath54 increases all points should eventually fall on a universal curve depending on @xmath55 . our results , reported in figures [ chi ] , [ xi ] , show that this happens : the data corresponding to the pairs @xmath56 , @xmath57 are only slightly different from those with @xmath58 , @xmath59 , the difference being much less than that observed by comparing data with @xmath60 and @xmath61 .
therefore , the data strongly suggest that fss holds , ( though , for @xmath62 , scaling corrections are significant ) and therefore , that the system becomes eventually critical . at the critical point @xmath63
. looking at fig .
[ xi ] we see that all mc data satisfy @xmath64 , which indicates that they all lie in the high - temperature phase .
this allows us to set a lower bound on the position of the critical point , @xmath65 . in order to determine the critical properties of the model , i.e. critical point and critical exponents
, we used the iterative method which was introduced in @xcite and generalized in @xcite to include scaling corrections .
it allowed us to obtain infinite - volume estimates of @xmath66 and @xmath52 up to @xmath67 ( @xmath68 is the infinite - volume second - moment correlation length ) in the high - temperature phase .
the starting point is the fss relation @xmath69 valid for any long - distance quantity . here
@xmath70 is an arbitrary number ( in our calculations we fixed @xmath71 ) , @xmath72 and @xmath73 are universal scaling functions , and @xmath74 is a to - be - determined correction - to - scaling exponent . by iterating ( [ fss - corr ] )
it is possible to extrapolate the finite - volume estimates @xmath75 , obtaining the infinite - volume value @xmath76 . for @xmath71.,scaledwidth=90.0% ] for @xmath71.,scaledwidth=90.0% ] our results show that @xmath77 increases quite rapidly in the range we have studied , @xmath78 , confirming that the system eventually becomes critical . fitting the infinite - volume results , we are able to determine the critical point and critical exponents associated with the critical behaviour of the overlap observables . we obtain : @xmath79
the critical exponents are defined by @xmath80 and @xmath81 .
the suffix @xmath82 is introduced to remind that they refer to the overlap variables and not to the magnetic ones . in the analysis it is crucial to include corrections to scaling in the fss ( corrections behave as @xmath83 ) and in fits of infinite - volume quantities ( corrections behave as @xmath84 as @xmath85 ) .
the exponent @xmath74 was determined by studying the critical behaviour of two universal ratios that involve the four - point and the two - point correlation function of the overlap variables .
we obtained @xmath86 .
estimates ( [ results ] ) are close to those obtained for the isgm and thus support the conjecture that the sram transition is in the same universality class as that of the isgm .
@xcite quotes @xmath87 , @xmath88 , while @xcite reports @xmath89 and @xmath90 .
other estimates are reported in table 1 of @xcite . with the quoted error bars
there is a small discrepancy between our estimate of @xmath91 and those of @xcite .
this difference should not be taken too seriously , since there are similar discrepancies among the estimates obtained by different groups for the bimodal isgm , see table 1 of @xcite . for several lattice sizes.,scaledwidth=90.0% ] for several lattice sizes.,scaledwidth=90.0% ] in mc simulations the critical point is often determined by considering the crossing point @xmath92 of the binder cumulant @xmath93 , where @xmath94 , or of the ratio @xmath95 : indeed , @xmath96 as @xmath97 .
our data for @xmath98 and @xmath95 , which are reported in figures [ biq ] and [ xil ] , are compatible with @xmath99 . indeed ,
the estimates of both @xmath98 and @xmath95 at fixed @xmath54 get closer as @xmath100 increases towards 1.08 , although a crossing point is not clearly observed for those values of @xmath54 ( @xmath101 ) that extend up to @xmath102 .
this is probably due to scaling corrections that are particularly large in the sram .
we can perform a more quantitative check by using the results of @xcite for the critical - point values @xmath103 and @xmath104 .
they quote : @xmath105 ( bimodal distribution ) and @xmath106 ( gaussian distribution ) ; @xmath107 ( bimodal distribution ) and @xmath108 ( gaussian distribution ) .
these results are compatible with ours for @xmath98 and @xmath95 close to @xmath109 .
for @xmath110 we have @xmath111 ( @xmath112 ) , @xmath113 ( @xmath56 ) , and @xmath114 ( @xmath112 ) , @xmath115 ( @xmath56 ) .
these results are very close to the isgm estimates and show the correct trend .
they are therefore consistent with the existence of a critical point at @xmath116 and with the conjecture that the isgm and the sram belong to the same universality class .
999 harris r , plischke m and zuckermann m j , 1973 , 160 .
calabrese p , pelissetto a and vicari e , 2004 e * 70 * , 036104 [ cond - mat/0311576 ] .
dudka m , folk r and holovatch yu , 2005 _ j. magn .
mater . _ * 294 * , 305 [ cond - mat/0406692 ] .
imry y and ma s - k , 1975 , 1399 .
pelcovits r a , pytte e and rudnick j , 1978 , 476 ; _ erratum _ 1982 * 48 * , 1297 .
feldman d e , 2000 b * 61 * , 382 .
feldman d e , 2001 _ int .
_ b * 15 * , 2945 [ cond - mat/0201243 ] .
aharony a and pytte e , 1980 , 1583 ; 1983 b * 27 * , 5872 .
chen j h and lubensky t c , 1977 b * 16 * , 2106 .
pelissetto a and vicari e , 2002 _ phys . rep . _ * 368 * , 549 [ cond - mat/0012164 ] .
folk r , holovatch yu and yavorskii t , 2003 _ physics uspekhi _ 46 , 169 [ cond - mat/0106468 ] .
ozeki y and ito n , 1998 , 5451 .
bray a j and moore m a , 1985 _ j. phys .
_ c : _ solid state _ * 18 * , l139 .
jayaprakash c and kirkpatrick s , 1980 b * 21 * , 4072 .
chakrabarti a , 1987 b * 36 * , 5747 .
fisch r , 1990 b * 42 * , 540 .
fisch r , 1998 b * 58 * , 5684 [ cond - mat/9801033 ] .
itakura m , 2003 b * 68 * , 100405(r ) [ cond - mat/0303552 ] .
parisen toldin f , pelissetto a , and vicari e , 2006 _ j. stat .
_ p06002 [ cond - mat/0604124 ] .
geyer c j , 1991 in _
computer science and statistics : proc .
of the 23rd symposium on the interface _ , edited by e. m. keramidas ( fairfax station : interface foundation ) , p. 156 .
hukushima k and nemoto k , 1996 _ j. phys .
japan _ 65 , 1604 [ cond - mat/9512035 ] .
caracciolo s , edwards r g , ferreira s j , pelissetto a and sokal a d , 1995 , 2969 [ hep - lat/9409004 ] .
calabrese p , martn - mayor v , pelissetto a and vicari e , 2003 e * 68 * , 036136 [ cond - mat/0306272 ] .
jrg t , 2006 b * 73 * , 224431 [ cond - mat/0602215 ] .
katzgraber h , krner m and young a p , 2006 b * 73 * , 224432 [ cond - mat/0602212 ] . | we investigate the nature of the critical behaviour of the random - anisotropy heisenberg model ( ram ) , which describes a magnetic system with random uniaxial single - site anisotropy , such as some amorphous alloys of rare earths and transition metals .
in particular , we consider the strong - anisotropy limit ( sram ) , in which the hamiltonian can be rewritten as the one of an ising spin - glass model with correlated bond disorder : @xmath0 , where @xmath1 and @xmath2 is a random three - component unit vector .
we performed monte carlo simulations of the sram on simple cubic @xmath3 lattices , up to @xmath4 , measuring correlation functions of the replica - replica overlap , which is the order parameter at a glass transition .
the corresponding results show critical behaviour and finite - size scaling . they provide evidence of a finite - temperature continuous transition with critical exponents
@xmath5 and @xmath6 .
these results are close to the corresponding estimates that have been obtained in the usual ising spin - glass model with uncorrelated bond disorder , suggesting that the two models belong to the same universality class .
this is consistent with arguments that suggest that the disorder correlations present in the sram are irrelevant . |
the galactic bulge holds important information towards understanding the formation and evolution of the milky way ( mw ) .
however , our edge - on perspective in the disk plane and the large and variable dust extinction especially in optical bands make it a non - trivial task to study the mw bulge . despite these difficulties ,
our knowledge of the galactic bulge has advanced greatly in the past two decades .
numerous studies indicate that the mw bulge contains a bar structure .
for example , the non - circular motions of gas in the inner galaxy was explained with bar perturbation ( de vaucouleurs 1964 ; fux 1999 ; englmaier & gerhard 1999 ; bissantz & gerhard 2002 ) .
likewise , the boxy shape of the mw bulge revealed by _ cobe / dirbe _ photometry ( dwek et al .
1995 ) was characterized as the perspective effect of a tilted bar ( blitz & spergel 1991 ; binney et al .
more evidence for the barred morphology emerges from star counts using distance indicators like red clump ( rc ) stars ( stanek et al . 1994 ; rattenbury et al .
2007 ; cao et al . 2013 ) .
kinematic information of the mw bulge stellar populations provides additional insights into its origin . the bulge radial velocity assay ( brava ) revealed the fast cylindrical rotation of the galactic bulge extending up to @xmath2 , which is inconsistent with the classical scenario that the bulge is built in mergers and dominated by random motions ( howard et al .
2008 , 2009 ; kunder et al .
shen et al .
( 2010 ) demonstrated that a pure - disk galaxy with a spontaneously formed bar structure can match the brava observation of the radial velocity and velocity dispersion towards the bulge region well .
they also found that the mass of a classical bulge , if in existence , may be relatively small .
therefore , the galactic bulge is likely built through bar - induced internal dynamical processes , with minor contribution from a merger - built classical bulge .
recently , some studies suggested that the galactic bulge shows the signature of a vertical x - shape . towards the bulge region ,
the apparent magnitudes of rcs in certain fields show double - peaked distributions , with increasing peak separations towards higher latitudes ( mcwilliam & zoccali 2010 ; nataf et al .
2010 ; saito et al . 2011 ; ness et al . 2012 ; nataf et al .
2015 ) . since rcs
are well - calibrated standard candles with a roughly constant absolute magnitude , such a double - peaked feature seems to indicate the existence of a vertical x - shaped feature in the galactic bulge .
more detailed studies revealed that certain latitudinal slices of the galactic bulge have two density lumps aligned with the bar , which indeed results in the observed double - peaked apparent magnitude distributions of rcs ( saito et al .
based on the rcs in the vista variables in the via lactea ( vvv ) survey , the symmetrized three - dimensional bulge model of wegg & gerhard ( 2013 ) also clearly illustrates the x - shape in its side - on view and surface density slices .
further made - to - measure modelling reveals an off - centered x - shape in the galactic bulge , and the peanut / x - shaped component accounts for over @xmath3 of the bulge stellar mass ( portail et al .
the x - shaped bulge morphology is frequently seen in edge - on disk galaxies ( e.g. whitemore & bell 1988 ; ltticke et al .
2000 ; bureau et al . 2006 ) , and is presumably similar to the structure in our own galaxy . using a simple and self - consistent @xmath0-body simulation ,
li & shen ( 2012 ) showed that such an x - shape can arise naturally during the evolution of the galactic bar .
numerical simulations demonstrated that a bar forms via the bar instability , and then thickens vertically due to the buckling instability ( combes & sanders 1981 ; raha et al . 1991 ) .
the vertical motion in a peanut / x - shape bulge is enhanced after the buckling episode .
some studies proposed that x1 orbits with 2:1 vertical lindblad resonance , namely _ banana orbits _ for their side - on appearance , can contribute to the peanut / x - shaped feature in edge - on galaxies ( pfenniger & friedli 1991 ; patsis et al .
2002 ; athanassoula 2005 ) .
great efforts have been made to understand the roles of banana orbits ( and their associated orbits ) in forming the peanut / x - shaped bulges ( e.g. patsis & katsanikas 2014a , b ) .
however , recent work by portail et al .
( 2015b ) classified orbital families in the peanut / x - shaped bulges , and suggested the brezel - like orbit as the main contributor to the x - shape .
since all of these orbital families can potentially support the peanut / x - shaped feature , additional kinematic information is required to distinguish their contributions in the galactic x - shape . if the galactic x - shape is dominated by certain orbital families , its stellar kinematics is expected to reflect the underlying orbital motion .
thus , the stellar kinematics is essential to better understand the galactic x - shape .
currently , only a handful of observational studies have explored the kinematics inside the x - shape .
vsquez et al .
( 2013 ) ( hereafter v13 ) analyzed the stellar motion in the x - shape using rcs in @xmath4 .
with the three - dimensional velocity available for @xmath5 stars , they found weak anti - correlation between the longitudinal proper motion and the radial velocity in both bright and faint rcs . meanwhile , no significant correlation was seen between the latitudinal proper motion and the radial velocity in their sample .
they interpreted the correlation with stars on elongated bar orbits , possibly the streaming motion along the arms of the x - shape .
poleski et al .
( 2013 ) ( hereafter p13 ) investigated the proper motion of rcs in fields at @xmath6 . between the probability - distinguished closer and further arms of the x - shape ,
the mean proper motion difference is asymmetric in both @xmath7 and @xmath8 directions , which is linear for @xmath9 , but roughly constant for @xmath10 .
they attributed the linear side to the streaming motion within the x - shape . from a numerical perspective ,
gardner et al . ( 2014 ) ( hereafter g14 ) first explored the kinematic properties of x - shaped bulges .
they compared the kinematics at the closer and further sides of the bar / bulge , and suggested that x - shaped bugles show coherent signature of a minimum along @xmath1 in the mean radial velocity difference between the closer and further sides .
although efforts have been made in both observations and simulations , the kinematics of the x - shape is still poorly understood . in this paper
, we exploit the power of simulations to present a closer look at the x - shape in our own galaxy . using an @xmath0-body model of the milky way - like barred galaxy ,
we explore the kinematic features of the galactic x - shape , and compare our model to the observations and other simulation works .
the numerical model here can not represent the actual galactic bulge in all aspects , but its self - consistency and simplicity still reveal the general properties of the x - shaped bulge in such exploration and comparison .
our results lead to a better understanding of the existing observations , serve as an independent check of other numerical works , and make predictions for future observations . in section 2 , we briefly describe the @xmath0-body model used in this paper . results and comparisons with existing observations
are shown in section 3 . in section 4 , we discuss the nature of the x - shape by analyzing the correlations between proper motions and radial velocities , and test two potential applications to derive galactic parameters based on the galactic x - shape .
our results are summarized in section 5 .
in this work , we use the same @xmath0-body model as in shen et al .
( 2010 ) and li & shen ( 2012 ) .
initially it contains a cold ( toomre s @xmath11 ) exponential stellar disk of @xmath12 particles rotating in a rigid dark matter halo potential .
the bar forms rapidly within the first gigayear , and becomes thickened vertically in the inner region via the buckling instability .
after @xmath13 , the structure of this model reaches a quasi - steady state , with a bar pattern speed of @xmath14 ( shen 2014 ; molloy et al .
the snapshot at @xmath15 was analyzed in shen et al .
( 2010 ) and li & shen ( 2012 ) .
assuming a distance of the sun to the galactic center ( gc ) ( @xmath16 ) and a @xmath17 angle of sun gc line to the bar major axis , this model yields excellent matches with the mean radial velocity and velocity dispersion measurements of brava .
the x - shape in this model was highlighted after subtracting a smooth component from the edge - on projection ( li & shen 2012 ) .
the distance distributions in the model towards the bulge fields can qualitatively match the apparent magnitude distributions of rcs in mcwilliam & zoccali ( 2010 ) .
thus , despite its simplicity , this model is valuable in understanding the kinematics of the galactic x - shape . for consistency
, we adopt the same solar configuration as in shen et al .
( 2010 ) and li & shen ( 2012 ) .
following g14 , we also split the particles in a field into two samples to illustrate the kinematic difference between the two sides of the bar / x - shape .
particles within @xmath18 from the sun are denoted as `` near '' , and those beyond @xmath18 are denoted as `` far '' .
we calculated the heliocentric radial velocity ( or line - of - sight velocity @xmath19 ) and proper motion ( @xmath20 , @xmath21 ) of particles inside the bar / bulge region ( galactocentric radii less than @xmath22 ) .
along the minor axis of the galactic bulge , some specific fields are carefully studied in order to allow for detailed comparisons with observations . across the bulge region , particles are binned in the @xmath23 plane to map the spatial trend of the mean velocity and velocity dispersion . in observations ,
@xmath24 from multi - epoch measurements needs a factor @xmath25 to represent its angular speed in the longitudinal direction , as required by the metric of spherical coordinates . in this paper
we also use @xmath26 as the notation of the metric - corrected longitudinal proper motion .
from the solar perspective , the galactic x - shape features double - peaked magnitude / distance distributions in fields passing through it ( li & shen 2012 ) . in this section , we study the distance distributions towards the x - shape in this model in greater detail , and investigate the properties of the double - peaked feature across the bulge region .
we use kernel density estimator to test if the distance distribution is double - peaked ( silverman 1981 ; hall & york 2001 ) . in each field
, the distance distribution is constructed from the sample particles with a gaussian kernel .
the multimodality of the distance distribution is illustrated with the critical kernel sizes for @xmath27 modes ( @xmath28 ) , and its significance levels are evaluated using bootstraping .
a relatively large and statistically significant @xmath28 implies the existence of more than @xmath27 modes . for each field , we search for the critical kernel size for a single - peaked or double - peaked distribution ( @xmath29 and @xmath30 , respectively ) , and test their significance levels ( @xmath31 , @xmath32 ) via standard bootstrapping .
the details of the method is described in appendix a. to illustrate the properties of the double - peaked feature , we locate the two peaks in the distance distribution using the mean value of @xmath29 and @xmath30 as an optimal kernel size , and find the separation of the two peaks in the distance distribution and the ratio of peak amplitudes ( far to near ) .
it is worth pointing out that the algorithm here is relatively conservative , while the visual identification ( with _ a priori _ knowledge about the double peaks ) is more flexible in distinguishing modes .
+ figure [ fig : double_peak_map ] shows the properties of the double - peaked distance distributions across the bulge region . inside the regions where the double - peaked feature is robustly detected , we find some prominent trends .
the separation of the two peaks in the distance distribution increases with @xmath33 due to the x - shape geometry ( upper left panel ) , which is especially clear in the vertical slice of @xmath1 ( upper right panel ) .
while at the same latitude , the separation is roughly constant .
meanwhile , the ratio of peak amplitude ( far to near ) decreases with @xmath7 .
since the bar is tilted , its near and far sides dominate at @xmath34 and @xmath35 respectively , as illustrated in the latitudinal slice of @xmath36 ( lower right panel ) .
as @xmath7 decreases , the peak amplitude decreases at the near side but increases at the far side , which further increases the ratio of peak amplitude ( far to near ) .
in addition , at both sides , the identified peaks are closer to the observer as @xmath7 increases ( lower right panel ) . in other regions where
the double - peaked feature is not robustly detected ( low @xmath31 ) , the separation and peak ratio are spiky since the secondary peaks are likely to be spurs in the under - smoothed distribution there .
the spatial distributions of the overdensity shown in the right column of figure [ fig : double_peak_map ] are qualitatively consistent with the recent ogle - iii rc observations in figure 9 of nataf et al .
( 2015 ) .
the increasing peak separation with @xmath8 in this model is qualitatively consistent with previous studies ( e.g. mcwilliam & zoccali 2010 ; saito et al .
2011 ; li & shen 2012 ; nataf et al .
the peak separation measured in @xmath37 ( @xmath38 above the disk plane at the gc ) is about @xmath39 , which is similar to the result in figure 8 of mcwilliam & zoccali ( 2010 ) . the varying peak amplitude due to the tilted bar / x - shape agrees with the results in mcwilliam & zoccali ( 2010 ) and saito et al .
( 2011 ) , and naturally leads to the decreasing ratio of peak amplitudes ( far to near ) with @xmath7 ( lower left panel of figure [ fig : double_peak_map ] ) .
some minor disagreements exist on the latitudinal range where the double - peaked feature is detectable .
the two split rcs in mcwilliam & zoccali ( 2010 ) merge around @xmath6 , and saito et al .
( 2011 ) found a similar but slightly lower limit of @xmath40 .
the double - peaked density profile in wegg & gerhard ( 2013 ) emerges at @xmath41 above the galactic plane , corresponding to @xmath42 at @xmath43 . in our model
, the double - peaked distance distribution can be detected at @xmath44 ( @xmath45 above the disk , at @xmath43 ) , which is similar to the result obtained by wegg & gerhard ( 2013 ) .
saito et al . ( 2011 ) obtained a longitudinal range of @xmath46 for the double - peaked feature , while the range is latitude - dependent in our model . towards higher latitudes ,
the longitudinal range for the double - peaked feature in our model becomes much wider than @xmath46 in saito et al .
these minor disagreements could be due to the different peak - finding methods adopted . to further illustrate the statistical significance of the double - peaked feature in the distance distribution ,
the critical kernel sizes ( @xmath29 , @xmath30 ) and their significance levels ( @xmath31 , @xmath32 ) are illustrated in figure [ fig : silverman_test ] .
closer to the disk plane , most regions towards the galactic bulge feature low @xmath31 , @xmath32 and relatively small @xmath29 , @xmath30 , suggesting that the distance distributions are single - peaked there . in two triangular regions
away from the disk plane ( @xmath47 ) , we obtain clearly larger @xmath29 with high significance level , while both @xmath30 and @xmath32 are modest there .
the robustly large @xmath29 but insignificant @xmath30 here indicates that in these regions , the distance distribution is strongly double - peaked due to the presence of the vertical x - shape .
furthermore , the critical kernel size @xmath29 increases with @xmath33 , which also agrees with the larger peak separation at higher latitudes .
since larger kernel sizes are required to over - smooth the gap between the two modes towards higher latitudes , the increasing @xmath29 with @xmath33 also implies the vertical x - shape . in practice
, the identification of the x - shape can be affected by various uncertainties in the determination of the distance , such as sample contamination , photometric errors , non - uniform extinctions and variations in the intrinsic luminosity of rcs . moreover
, the location of the peak depends on the peak - finding method adopted . on the other hand
, this model may not exactly illustrate the x - shaped feature in the mw bulge . despite the possible disagreements between the model and observations , our model is still valuable in studying the x - shaped mw bulge for the broad similarities with the observed x - shape geometry .
modern spectroscopic surveys have acquired precise measurements of stellar radial velocities , which can provide important clues to understanding the dynamics of galactic structures .
previous radial velocity observations have revealed the cylindrical rotation of the galactic bulge , indicating the secular nature of its formation history ( howard et al .
2009 ; shen et al .
2010 ; williams et al .
v13 studied the radial velocity of rcs inside the x - shape towards @xmath48 .
they observed an excess of approaching stars at @xmath49 in the bright rcs ( near side ) , and a similar excess of receding stars at @xmath50 in the faint rcs ( far side ) .
in contrast , inside the plaut s window ( @xmath51 ) , the radial velocity barely differs between the bright and faint rcs ( de propris et al .
the numerical study of g14 related the x - shape morphology to a minimum in the mean difference @xmath19 between the two sides .
therefore , although there have been a handful of kinematic studies of the bulge stars focused in exploring the x - shape , the kinematic imprints of the x - shape are still not well studied .
first we study the radial velocity in fields along the minor axis of the bulge . in calculating the radial velocity
, we set the circular velocity of the sun to be @xmath52 , without considering the solar motion with respect to the local standard of rest ( lsr ) .
since the model is symmetric with respect to the galactic plane , two fields with opposite latitudes are combined in the analysis to achieve better number statistics .
figure [ fig : v_los ] shows the radial velocity distributions in @xmath1 , @xmath53 and @xmath54 . for both sides of the bar / x - shape , the radial velocity spans a wider range at latitudes closer to the galactic plane , which is seen in the decreasing velocity dispersion with @xmath33 . in the same field , we find systematically smaller velocity dispersion at the far side , indicating a decreasing velocity dispersion away from the galactic disk . in all these fields , the near side has more approaching particles ( negative @xmath19 ) , while the far side has more receding particles ( positive @xmath19 ) accordingly .
the excess of approaching / receding particles is evident in the median velocities of the two sides .
the excesses of @xmath19 at the two sides may be subject to the orbital motion inside the bar structure .
figure [ fig : bar_orbit ] sketches a naive bar - supporting orbit in this configuration . in an axisymmetric , unbarred disk , @xmath19 at @xmath55
is expected to be symmetric with respect to @xmath56 at both sides .
if it has coherent non - circular motions , i.e. the alignment of elongated orbits have a preferred direction , asymmetric @xmath19 distributions with respect to @xmath56 should be detected at both sides . in this case , an excess of approaching particles is expected at the near side , and an excess of receding particles should be observed at the far side , as expected from figure [ fig : bar_orbit ] .
although the reported excesses at @xmath57 in v13 are not clear in this model alone , there are clear excesses at @xmath58 , which , however , become less noticeable in the combined field @xmath59 .
perhaps we need better number statistics to confirm this , both in observations and models .
] , there are indeed more approaching or receding particles at the two sides , which agrees well with figure [ fig : bar_orbit ] .
nidever et al . (
2012 ) also used bar orbits to explain the cold high - velocity peaks seen in the commissioning observations of the apache point observatory galactic evolution experiment ( apogee ) .
however , as li et al .
( 2014 ) demonstrated with numerical simulations , bars are not expected to display such a high velocity peak ; instead , the high velocity stars are located around the tangential point between the bar orbits and the sight lines . + , these orbits contribute an excess of approaching stars at the near side , and an excess of receding stars at the far side .
meanwhile , @xmath20 at the near and far sides will peak at the positive and negative sides of @xmath60 respectively , where @xmath60 is the proper motion of the galactic center due to the motion of the sun .
its tilted alignment may also lead to the position shift in both the extrema of @xmath20 , and the zero - velocity line of @xmath61 with respect to @xmath1 . ]
then we study the mean radial velocity ( @xmath61 ) , and its dispersion ( @xmath62 ) across the whole bulge region .
as illustrated in figure [ fig : vlos_map ] , @xmath61 at both sides shows clear cylindrical rotation , and the longitudinal gradient of @xmath61 is strong at @xmath44 , which coincides with the latitude where the x - shape is detectable in this model .
the zero - velocity line of @xmath61 shifts towards @xmath34 at the near side , and towards @xmath35 at the far side .
the shift of zero - velocity lines agrees with the coherent alignment of elongated orbits .
as sketched in figure [ fig : bar_orbit ] , the radial velocity is zero only close to the two tips of such orbits .
if the bar structure is populated by such orbits , the zero - velocity lines at the two sides of the bar will appear at two opposite sides of @xmath1 . also , the decreasing radial velocity dispersion away from the gc indicates a kinematically hot inner bulge .
we find that , the difference of @xmath61 between the near and far sides ( near - far ) is generally negative , while the difference of @xmath62 is mostly positive . since on the @xmath23 plane , the particles at the near side are closer to the disk plane than those at the far side , the positive @xmath63 and negative @xmath64 may reflect the kinematic gradient in the vertical direction .
near @xmath65 , we also notice the minimum of @xmath64 along @xmath1 . considering both the absolute value and the relative depth with respect to the gc , the minimum in our model is quite similar to g14 .
the agreement between our model and g14 indicates that the minimum in the mean difference of radial velocity may be a coherent feature of x - shaped bulges , which may originate from the excess of approaching / receding particles at the near and far sides .
zoccali et al .
( 2014 ) studied the kinematics of the galactic bulge with the _ giraffe _ inner bulge survey ( gibs ) . without distinguishing the near and far sides
, they mapped the line - of - sight velocity , and velocity dispersion of the bulge region using the spectra of @xmath66 rcs observed in 24 fields .
they found an asymmetric @xmath61 across the galactic longitude , and a vertically elongated peak of @xmath62 at the gc .
compared to the results in zoccali et al .
( 2014 ) , @xmath61 of our model is nearly symmetric with @xmath7 ( left panel of figure [ fig : v_los_all ] ) , even in the presence of a tilted bar .
we also find a vertically elongated peak of @xmath62 ( right panel of figure [ fig : v_los_all ] ) .
although the velocity dispersion map in zoccali et al . ( 2014 ) was interpolated from coarsely sampled fields , it still shares great similarity with our model . therefore , the vertically elongated dispersion map in zoccali et al .
( 2014 ) may be physically meaningful ; it may originate from the velocity distribution of bar supporting orbits .
we will study this feature in greater detail in our future work .
the proper motion reflects the angular velocity of stars in the heliocentric frame .
it is useful to study the dynamics of the mw ( e.g. kozowski et al .
2006 ; rattenbury et al .
2007 ; soto et al . 2012 ) .
previous usage of the proper motion was limited by the data quality and the lack of accurate distance estimation .
new space missions like _ gaia _ will obtain precise proper motion and distance measurements for an unprecedented number of stars , and the study of the milky way dynamics will benefit significantly from the upcoming boom of high - quality data .
currently , the potential of the proper motion is still to be fully explored , and less is known about the imprints of the fine structures ( like the x - shape ) on the observed proper motions . in this section , we will study the proper motion towards the bulge region in our model .
+ first , we study the proper motion distribution along the minor axis of the bulge .
figure [ fig : proper_motion ] shows the proper motion in four fields ( @xmath1 , @xmath67 and @xmath68 ) . to ensure robust statistics , as in the previous section , the fields with opposite latitudes are co - added to carry out the analysis . the latitudinal proper motion for fields at @xmath69
is reversed according to the symmetry of the model . for @xmath20 , the near and far sides peak at the positive and negative sides of @xmath70 respectively , where @xmath71 is the proper motion of the gc due to the motion of the sun ( assuming @xmath43 , @xmath72 ) .
the displaced distributions of @xmath20 at the two sides lead to a significant difference in their mean values . for @xmath21 ,
the two sides have similar mean values ( close to zero ) .
we find a larger dispersion of @xmath20 and @xmath21 at the near side , and the dispersions at both sides decrease with @xmath33 .
the difference of @xmath73 between the near and far sides mainly reflects the rotation of the model bulge . in these fields
, @xmath74 is comparable to the dispersion of @xmath73 at both sides , which indicates the strong rotation of the model bulge .
the role of bar - supporting orbits in this case still needs further investigation . in our @xmath75 field
we measured a relatively large @xmath74 of @xmath76 , which is almost three times the result presented in v13 .
such a large difference deserves further exploration and explanation .
since the proper motion scales inversely with distance , particles at the near side will show a larger proper motion dispersion than the far side , assuming similar linear velocity distributions at the two sides .
this agrees well with the larger proper motion dispersion at the near side in this model .
here we test if the larger dispersion at the near side can be explained solely with the shorter distance .
we identify the density peaks in fields along the minor axis of the bulge using the method described in section 2.1 , and pick the @xmath77 particles closest to the identified peak position at each side .
then , we calculate the proper motion dispersion of these peak particles , and find the ratios of dispersions between the two sides ( near to far ) .
if the larger proper motion dispersion is mainly due to the shorter distance , the ratio of proper motion dispersion ( near to far ) should roughly match the ratio of peak distance ( far to near ) .
the results are summarized in table 1 .
the ratio of @xmath78 is marginally consistent with , but slightly larger than , the ratio of peak distance , and both of them increase with latitude .
however , @xmath79 poorly matches the ratio of peak distance , and no clear trend with the latitude is observed .
therefore , the larger proper motion dispersion measured at the near side can not be solely explained by the shorter distance .
similar to the systematically smaller line - of - sight velocity dispersion at the far side , the slightly larger @xmath80 compared to @xmath81 may be due to the decreasing linear velocity dispersion away from the galactic plane .
[ tab : peak ] .distance ratio and proper motion dispersion ratios between the near and far sides [ cols="<,<,<,<",options="header " , ] * notes * : _ separation _ indicates the distance between two peaks identified with kernel density estimator , @xmath74 stands for the difference of mean proper motion , and @xmath82 is the derived angular velocity . _
peak _ is the @xmath77 particles nearest to the identified peaks .
we assume a @xmath83 error in distance , and @xmath84 in @xmath85 . the proper motion of the gc is @xmath86 ( assuming @xmath87 at @xmath43 ) . ,
top panel ) and the corresponding angular velocity ( @xmath88 , bottom panel ) of the x - shaped peaks in the vertical @xmath1 plane as a function of gactocentric radius ( @xmath89 ) . in each panel ,
the blue points represent the peak particles , while the black solid line shows the azimuthally averaged values , regardless of the particle height @xmath90 . for the blue points , we also labelled the corresponding latitude @xmath33 where the particles are selected
. since @xmath91 decreases with increasing the height , at a given radius , the azimuthally ( and vertically ) averaged @xmath91 is smaller than the rotation velocity at lower @xmath92 but larger than that at higher @xmath92 . due to the x - shape geometry ,
the galactocentric radius and the height of the peak particles ( the blue dots ) increase with @xmath33 , resulting in larger @xmath91 at smaller radius and smaller @xmath91 at larger distance compared to the average value ( the black line ) . ] to better understand this behavior , we compare the mean azimuthal velocity of peak particles with the azimuthally averaged rotational velocity of all the particles at the same galactocentric radius , regardless of their height @xmath90 . we study a series of fields within @xmath93 ( with an interval of @xmath94 ) along @xmath1 . for each field ,
the average rotation velocity ( @xmath91 ) and the corresponding angular velocity ( @xmath95 ) is different from the conventional angular frequency derived from the circular - speed curve . ] are computed based on peak particles .
an average radius ( @xmath96 ) for each pair is obtained by averaging the galactocentric radius of the two peaks .
of course , this radius increases for higher galactic latitudes .
therefore , @xmath82 from equation 3 at different galactic latitudes can be naturally translated into a function of radius .
figure 14 shows @xmath91 and @xmath97 as a function of the galactocentric radius @xmath96 .
the solid line represents the azimuthally averaged @xmath91 .
apparently , @xmath91 and @xmath97 of peak particles ( blue dots ) agree very well with the azimuthally averaged values of all the particles ( black solid line ) .
therefore , @xmath82 in equation 3 actually measures the azimuthal streaming motions ( @xmath91 ) , rather than the bar pattern speed .
since the x - shape is symmetric in the @xmath55 plane , it can be used to locate the position of the gc ( mcwilliam & zoccali 2010 ; g14 ) .
the gc can be located by fitting the peak positions in the vertical @xmath1 plane with the x - shape geometry .
g14 compared two ways to locate the gc position , namely fitting the four arms of the x - shape , versus fitting the x - shape with two crossed lines .
they found that method ii yields an accuracy better than @xmath83 in most cases , even if the peak distance has an uncertainty of @xmath3 .
however , the sample limitations ( e.g. latitude coverage , number of fields , etc . ) may affect the results .
we try to locate the gc in our model with these two methods , and further beyond g14 , we test their accuracies under different latitude coverages and total number of fields .
we select @xmath0 pairs of fields along the minor axis of the bulge ( @xmath1 ) , with @xmath0 fields above and @xmath0 fields below the galactic plane respectively .
fields are randomly selected in the latitude range where the x - shape can be clearly identified in this model ( @xmath98 ) . in each field , the two peaks are identified with the same procedure described in section 2.1 .
we then fit the arms of the x - shape with four straight lines , and find their intersection points ( method i ) . in the ideal case , two of their four intersection points are right above / below the gc ( i.e. on the vertical axis of the galaxy ) , and the other two points are exactly on the galactic plane . thus the gc position can be located from the galactic vertical axis , and the galactic plane .
meanwhile , we try to locate the gc by directly fitting the x - shape with two crossed lines ( method ii ) .
the two methods are illustrated in the left panel of figure [ fig : gcfit ] .
we run a series of tests with different @xmath0 to demonstrate the effect of limited field number .
to illustrate the impact of limited latitude coverage , in another series of tests , we also change the latitude range to @xmath99 ( `` lower '' ) or @xmath100 ( `` higher '' ) , and compare the results with the case of full latitude coverage ( @xmath101 ) .
we find that , fitting the x - shape with two crossed lines ( method ii , right panel of figure [ fig : gcfit ] ) yields better results compared to fitting the four individual arms ( method i , middle of figure [ fig : gcfit ] ) , which agrees with the conclusion in g14 .
meanwhile , increasing the number of fields leads to smaller uncertainties for both methods .
we also find that , the accuracy of fitting depends on the latitude range of the fields used .
the position of the gc can be better constrained with the fields at lower latitudes , while the fields at higher latitudes give the worst results . in principle , fields at higher latitudes put stronger constraints on the gc position . however , since the x - shape is well - defined at lower latitudes ( @xmath102 ) due to better number statistics ( also seen in the upper right panel of figure 1 ) , the fitting results from these regions are more accurate compared to higher latitudes ( @xmath103 ) . fitting the x - shaped overdensity is a novel and accurate way to determine the position of the galactic center .
however , it still has some drawbacks .
first it relies on the identification of peaks in the apparent magnitude distributions of the sample , which could have large uncertainties in real observations due to the presence of scatters in the intrinsic luminosity , source incompleteness and sample contamination , etc . in the conversion from magnitude to distance ,
reddening and extinction could also contribute to the uncertainties .
even if the uncertainties due to photometric distance determination can be greatly reduced with geometric distance measurements like _ gaia _ does , there remains uncertainties in method ii .
ideally , if the peaks at the near side have similar heights as the far side , fitting two diagonal lines should trace the exact position of gc .
however , due to the projection effect , the peaks identified at the near side are closer to the disk plane than the far side .
consequently , the intersection of the two best - fit crossed lines tends to be closer to the sun , leading to a slight systematic underestimation of the gc distance .
since the projection effect is stronger at higher latitude , fields at lower latitudes are expected to yield better results than those at higher latitudes , which agrees with the right panel of figure [ fig : gcfit ] . in reality , the number of observational fields are usually not symmetric above and below the galactic plane , which may also affect the best - fit crossed lines to locate the gc .
we explore the kinematic properties of the x - shaped milky way bulge with an @xmath0-body model that well represents the observed morphology and dynamics of the galactic bulge . from the solar perspective
, the two density peaks on the sightlines due to the x - shape are identified using a gaussian kernel density estimator with adaptive kernel size . in this model ,
the distance distributions appear to be double - peaked above @xmath44 along @xmath1 , and the increasing separation between two peaks with @xmath33 agrees with previous observations . as a result of a tilted bar , the height ratio of the two peaks ( far to near ) , as well as their mean distance to the sun , features a longitudinal gradient at higher latitudes .
we study the line - of - sight velocity and proper motion distributions in fields ( @xmath67 and @xmath68 ) along @xmath1 , with the opposite fields properly scaled and added to increase number statistics . to illustrate the possible imprints of the bar / x - shape ,
the particles are divided into two groups ( near and far ) according to their distances to the sun compared to @xmath18 . in these fields ,
the line - of - sight velocity distributions significantly differ between the near and far sides ; the near side shows an excess of approaching particles , while a corresponding excess of receding particles exists in the far side .
due to the rotation of the bar / bulge , the longitudinal proper motions at the near and far sides peak at the positive and negative sides of @xmath70 , respectively .
the measured mean differences of @xmath20 between the two sides are comparable to the proper motion dispersion there , indicating the fast cylindrical rotation of the bar / bulge .
the latitudinal proper motions at the two sides barely differ in their mean values .
we note that the differences of @xmath19 and @xmath20 between the two sides can be qualitatively explained with the coherent alignment of bar - supporting orbits .
we test if the larger proper motion dispersion measured at the near side ( compared to the far side ) is due to the closer distance . in our model ,
only the dispersion ratio of @xmath20 between the two sides ( near to far ) is marginally consistent with the distance ratio of the two peaks ( far to near ) .
the results here indicate that the systematically larger dispersion at the near side is not only due to the closer distance , but also contributed by the differences in their intrinsic velocity distributions .
we also map the mean line - of - sight velocity , mean proper motion and their corresponding dispersions across the bulge region , both with and without distinguishing the near and far sides of the galactic bar / bulge .
the mean @xmath19 at the two sides features strong cylindrical rotation , and the position of the zero - velocity line with respect to @xmath1 agrees with the coherent alignment or bar - supporting orbits .
the mean difference of @xmath19 between the near and far sides shows a shallow minimum along @xmath1 , which is suggested as the signature of an x - shaped bulge ( g14 ) . without distinguishing the near and far sides
, @xmath61 is symmetric with respect to @xmath1 , which barely shows the effect of the tilted bar as reported in zoccali et al .
for the mean @xmath20 at the two sides , due to the coherent alignment of bar - supporting orbits , the extrema shift to the two opposite sides of @xmath1 respectively . at lower latitudes ,
the displacements of the extrema are larger than that at higher latitudes
. this may be due to the buckled inner region of the bar dominating at higher latitudes , while the long and thin component of the bar dominates at lower latitudes .
inside @xmath104 , @xmath105 , the mean differences of @xmath20 and @xmath21 between the two sides are roughly symmetric with respect to @xmath1 .
the breaks of @xmath74 and @xmath106 reported in p13 are absent in our model . for their dispersions , @xmath62 , @xmath78 and @xmath107 at both near and far sides
reveal the kinematically hot central region .
the dispersion ratio is saddle - shaped at both sides , with a diagonal pattern of the cross - correlation factor , indicating that the proper motion anisotropy is tilted towards the gc . without distinguishing the near and far sides of the bar / x - shape , the dispersion of @xmath19 shows a vertically elongated peak similar to zoccali et al .
meanwhile , the proper motion dispersion is larger in the longitudinal direction , and the dispersion ratio ( @xmath108 ) increases away from @xmath1 . we also observe a diagonal pattern of the cross - correlation factor .
compared to the proper motion measurements in rattenbury et al .
( 2007 ) , this model shows slightly smaller dispersion but stronger anisotropy .
we also study the orbital motion inside the x - shape using the correlation of velocity components . in @xmath109 , for @xmath21 vs. @xmath19 and @xmath21 vs. @xmath20 , both the significance of the correlations ( pearson s r value ) and the slope from linear regressions reveal slightly stronger velocity anisotropy near the x - shape . except for @xmath21 vs. @xmath20 at the near side , the correlation is weakest for the peak sample .
we note that at both sides , the direction of the anisotropy roughly coincides with the case of downward - tilt banana orbits ( for @xmath110 ) .
however , the underlying orbital family can not be clearly distinguished here . with this model ,
we extend the azimuthal rotation velocity measured in p13 with the x - shape kinematics to various galactic latitudes .
we find that the angular velocity decreases at higher galactic latitudes , inconsistent with being the bar pattern speed .
the angular velocity derived from the proper motion difference between the two sides of the x - shape actually measures the azimuthal streaming motions , similar to the mean rotation profile @xmath111 . the x - shape is thought to be able to constrain the position of the galactic center due to its symmetry .
we test two methods to do this proposed in gardner et al .
( 2014 ) with our model , namely fitting the four arms of the x - shape to find the center and diagonally joining the four arms by fitting the x - shape with two crossed lines .
we demonstrate that a global - fitting yields better results than fitting the four arms individually . also , the fields at lower latitudes provide stronger constraints on the gc position due to better number statistics there .
we thank the anonymous referee for constructive comments that help to improve the paper , especially section 3.5 . the research presented here
is partially supported by the 973 program of china under grant no .
2014cb845700 , by the national natural science foundation of china under grant nos.11333003 , 11322326 , 11403072 , and by the strategic priority research program `` the emergence of cosmological structures '' ( no .
xdb09000000 ) of the chinese academy of sciences .
zyl is grateful for the support from shanghai sailing program ( no .
14yf1407700 ) . this work made use of the facilities of the center for high performance computing at shanghai astronomical observatory .
hospitality at apctp during the 7th korean astrophysics workshop is kindly acknowledged .
athanassoula , e. 2005 , celestial mechanics and dynamical astronomy , 91 , 9 binney , j. , gerhard , o. , spergel , d. , 1997 , , 288 , 365 bissantz , n. , gerhard , o. , 2002 , , 330 , 591 blitz , l. , spergel , d. n. , 1991 , , 379 , 631 bureau , m. , aronica , g. , athanassoula , e. , et al .
2006 , , 370 , 753 cao , l. , mao , s. , nataf , d. , 2013 , , 434 , 595 combes , f. , sanders , r. h. , 1981 , , 96 , 164 de propris , r. , rich , r. m. , kunder , a. , et al .
, 2011 , , 732 , 36 de vaucouleurs , g. 1964 , proceedings of the international astronomical union symposium no .
20 , 195 dwek , e. , arendt , r. g. , hauser , m. g. , et al .
1995 , , 445 , 716 efron , b. , 1979 , ann .
, 7 , 1 englmaier , p. , gerhard , o. , 1999 , , 304 , 512 fux , r. , 1999 , , 345 , 787 gardner , e. , debattista , v. p. , robin , a. c. , et al .
2014 , , 438 , 3275 hall , p. , york , m. , 2001 , statistica sinica , 11 , 515 howard , c. d. , rich , r. m. , reitzel , d. b. , et al .
2008 , , 688 , 1060 howard , c. d. , rich , r. m. , clarkson , w. , et al .
2009 , , 702 , l153 kunder , a. , koch , a. , rich , r. m. , et al .
2012 , , 143 , 57 kozowski , s. , woniak , p. r. , mao , s. , et al .
2006 , , 370 , 435 li , z .- y . & shen , j. , 2012 , , 757 , l7 li , z .- y . , shen , j. , rich , r. m. , kunder , a. & mao , s. , 2014 , , 785 , 17 ltticke , r. , dettmar , r .- j . & pohlen , m. , 2000 , , 145 , 405 mcwilliam , a. , zoccali , m. , 2010 , , 724 , 1491 molloy , m. , smith , m. c. , shen , j. , evans , n. w. , 2015 , , in press , ( arxiv:1412.4689 ) nataf , d. m. , udalski , a. , gould , a , et al .
, 2010 , , 721 , 28 nataf , d. m. , udalski , a. , skowron , j. , et al . , 2015 , mnras , 447 , 1535 ness , m. , freeman , k. , athanassoula , e. , et al .
2012 , , 756 , 22 nidever , d. l. , zasowski , g. , majewski , s. r. , et al .
2012 , , 755 , l25 patsis , p. a. , skokos , ch . & athanassoula , e , 2002 , , 337 , 578 patsis , p. a. , katsanikas , m. 2014 , , 445 , 3525 patsis , p. a. , katsanikas , m. 2014 , , 445 , 3546 pfenniger , d. , friedli , d. 1991 , , 252 , 75 poleski , r. , udalski , a. , gould , a. , et al . 2013 , , 776 , 76 portail , m. , wegg , c. , gerhard , o. & martinez - valpuesta , i , 2015a , , 448 , 713 portail , m. , wegg , c. , & gerhard , o. 2015b , , in press , ( arxiv:1503:07203 ) raha , n. , sellwood , j. a. , james , r. a. , & kahn , f. d. 1991 , , 352 , 411 rattenbury , n. j. , mao , s. , sumi , t. , smith , m. c. 2007 , , 378 , 1064 saito , r. k. , zoccali , m. , mcwilliam , a. , et al .
2011 , , 142 , 76 shen , j. , rich , r. m. , kormendy , j. , et al .
2010 , , 720 , l72 shen , j. , 2014 , in proceedings of iau symposium 298 , 201 silverman , b. w. , 1981 , j. r. statist .
b , 43 , 97 soto , m. , kuijken , k. , rich , r. m. 2012 , , 540 , a48 stanek , k. z. , mateo , m. , udalski , a. , et al .
1994 , , 429 , 73 vsquez , s. , zoccali , m. , hill , et al .
2013 , , 555 , a91 whitmore , bradley c. , bell , m. 1988 , , 324 , 741 williams , m. j. , zamojski , m. a. , bureau , m. , et al . 2011 , , 414 , 2163 wegg , c. , gerhard , o. , 2013 , , 435 , 1874 zoccali , m. , gonzalez , o. a. , vasquez , s. , et al .
2014 , , 562 , 66
in this section , we present a brief description of the silverman s test used in section 3.1 .
the silverman s test ( silverman 1981 ) is based on the kernel density estimation of the sample . for a univariate sample @xmath112 , @xmath113 , @xmath114 drawn from an unknown probability distribution function ( pdf ) @xmath115 , its kernel density estimator ( kde ) is @xmath116 where @xmath117 is the random variable , @xmath118 is the kernel function which is assumed to be a gaussian distribution function , and @xmath119 is the kernel size .
here @xmath120 serves as an approximation of the unknown original pdf @xmath115 .
it resembles a continuous version of the histogram , and @xmath119 corresponds to the `` bin size '' .
the kernel size @xmath119 controls the amount of smoothing ; larger @xmath119 leads to fewer peaks in the kde with the gaussian kernel . to test the null hypothesis that @xmath115 has @xmath27 or fewer modes ( @xmath121 ) against the alternative one that @xmath115 has more than @xmath27 modes ( @xmath122 ) , we use the critical kernel size of @xmath27 modes @xmath123 large
@xmath28 rejects the null hypothesis , indicating that there are more than @xmath27 peaks . in other words
, the @xmath27-critical kernel size is the smallest kernel size that makes a kde of @xmath27 peaks , for a particular sample .
if the sample has more than @xmath27 modes , a large @xmath28 is required to over - smooth the kde , and reduce the peak number to @xmath27 . on the contrary , if the sample itself has less than @xmath27 modes , only an under - smoothed kde can result in more than @xmath27 peaks , which requires a small @xmath27-critical kernel size . for a particular @xmath27-mode hypothesis ,
now we address the statistical significance of the @xmath27-critical kernel size from the data ( @xmath124 ) . in principle , one can draw a series of samples with equal size @xmath125 from @xmath115 , find their @xmath27-critical kernel sizes ( @xmath126 ) , and compare the critical kernel size from the data to those from resampling the true pdf @xmath115 using the @xmath127 level test : @xmath128 where @xmath129 represents the conditional probability , and @xmath127 is the level of significance . since the peak number decreases with increasing the kernel size , we can address the significance level of @xmath124 by accounting for the peak number of @xmath130 from resampling : @xmath131 @xmath132 however , the true density @xmath115 is always unknown . in order to evaluate the significance of @xmath124 , silverman ( 1981 ) generated the samples using the method in efron ( 1979 ) : @xmath133 where @xmath134 are sampled uniformly from the original data allowing replacement , @xmath135 is the variance of the sample , @xmath124 is the @xmath27-critical kernel size from the data , and @xmath136 is a series of random numbers from a standard normal distribution . here
@xmath137 actually samples a representative pdf @xmath138 , which is at most @xmath27-modal , and has the same variance as the sample .
therefore , the significance level of @xmath124 can be conservatively estimated with such a bootstraping procedure .
besides its conservative nature , the silverman s test is also not asymptotically accurate .
one way to improve the accuracy of silverman s multimodality test is presented in hall & york ( 2011 ) , by using a bootstraping distribution function @xmath139 as suggested in hall & york ( 2011 ) , there exists a unique absolute constant @xmath140 such that the bootstrap test is asymptotically correct : @xmath141 @xmath142 and the constant @xmath140 can be calculated either analytically , or using monte carlo simulations . in addressing the significance level of the double - peaked feature ( @xmath143 ) , we use the tabular value given in hall & york ( 2011 ) . | we explore the kinematics ( both the radial velocity and the proper motion ) of the vertical x - shaped feature in the milky way with an @xmath0-body bar / bulge model . from the solar perspective
, the distance distribution of particles is double - peaked in fields passing through the x - shape .
the separation and amplitude ratio between the two peaks qualitatively match the observed trends towards the galactic bulge .
we confirm clear signatures of cylindrical rotation in the pattern of mean radial velocity across the bar / bulge region .
we also find possible imprints of coherent orbital motion inside the bar structure in the radial velocity distribution along @xmath1 , where the near and far sides of the bar / bulge show excesses of approaching and receding particles . the coherent orbital motion is also reflected in the slight displacement of the zero - velocity - line in the mean radial velocity , and the displacement of the maximum / minimum in the mean longitudinal proper motion across the bulge region .
we find some degree of anisotropy in the stellar velocity within the x - shape , but the underlying orbital family of the x - shape can not be clearly distinguished .
two potential applications of the x - shape in previous literature are tested , i.e. , bulge rotation and galactic center measurements .
we find that the proper motion difference between the two sides of the x - shape can be used to estimate the mean azimuthal streaming motion of the bulge , but not the pattern speed of the bar . we also demonstrate that the galactic center can be located with the x - shape , but the accuracy depends on the fitting scheme , the number of fields , and their latitudinal coverage . |
surface electrons ( se ) levitating over liquid helium has been used as a paradigmatic quasi - two dimensional electron system ( q2des ) .
it constitutes the counterpart of the electron system on semiconductor heterostructures in the sense that the electrons on helium obey the classical statistics because the low densities achievable in experiments ( @xmath0 @xmath1 ) .
one point difference between them comes from the scattering mechanisms . in the former system
, electrons are scattered by surface excitations , the ripplons , and by vapor atoms of the helium surface . in the latter one , scattering by impurities , phonons , and interface roughness are the processes that limit the electron mobility . besides the remarkable phenomena discovered in this low - dimensional electron system , for instance , wigner crystallization@xcite and the existence of edgemagnetoplasmons@xcite se have been used as a probe to investigate the elementary surface excitations of cryogenic liquids and solids .
very recently there is an intensive search for experimental realization of se as qubits leading to a quantum computer.@xcite for a overview of the field see ref . .
for electrons trapped on the bulk helium surface it is well known that the se mobility is dominated by the electron - helium gas atom scattering for @xmath2 k whereas the se scattering by quantized capillary waves ( ripplons ) is responsible for mobility at lower temperatures where the helium vapor density becomes negligible .
the experimental data for se mobility over bulk helium are in reasonable agreement with theoretical calculations.@xcite the situation changes when se are floating over a helium film which in turn is deposited over a solid substrate . in such a condition it is possible to increase significantly the accessible range of electron densitiesmistgunz97 and the electron correlations may become important . furthermore , the electron energy spectrum , the ripplon dispersion relation and the electron - ripplon interaction are modified significantly due to film effects and the van der waals forces from the substrate now play a decisive role in the transport properties . furthermore , besides the usual scattering mechanisms pointed out above , we must consider the se scattering by interface defects at the helium film - substrate boundary , which , as it have been shown in experiments is responsible by the unusual behavior of the se transport properties on a helium film.platzman86,jiangdahm87,jiangstandahm88,hucarmidahm92,dahm97 the present work intends to provide a detailed theoretical study of the mobility of se localized over a helium film deposited on substrates , as solid neon , glass and poly(methyl - methacrylate ) [ pmma ] , which are the materials that have been used in the experiments .
we employ the boltzmann transport equation approach ( bea ) and the force balance equation method ( fbem ) in which coulomb effects on the mobility are taken into account through the dynamical structural factor of the q2des .
we calculate the se mobility as a function of temperature and the film thickness in the range @xmath3
@xmath4 @xmath5 . from the comparison with experimental results we check the role of all scattering processes involved and
verify the reliability of the interface roughness model used in our treatment .
we find that the se mobility is strongly determined by the interface roughness substrate and the results are in quite good agreement with the experiments .
electron states on a liquid helium film are confined in the direction normal to the surface ( @xmath6 ) due to the potential well created by a infinitely high barrier which prevents the electrons penetrate inside the liquid phase and by and attractive potential due to the electron interaction with the polarizable substrates and an applied holding electric field @xmath7 along the @xmath6 direction .
the electron potential energy over flat helium film located at @xmath8 is well known and given by@xcite@xmath9where @xmath10 and @xmath11 with @xmath12 and @xmath13 being the dielectric constants of helium and solid substrate , respectively , and @xmath14 the electron charge . for electrons moving freely along the helium surface with the wave function and the energy spectrum given respectively by @xmath15 and @xmath16 , where @xmath17 and @xmath18 are the wave vector and position vector in the plane @xmath19 of the liquid interface and @xmath20 is surface area occupied by electrons .
unfortunately an analytical solution of schrdinger equation for @xmath21 and subbands energies @xmath22 can not be found for @xmath23 given by eq .
( [ 1 ] ) . in this work ,
we use the variational method by choosing the trial wave function @xmath24 .
the parameter @xmath25 depends strongly on @xmath26 and is determined from the subband energy minimization as was done in refs . and .
one has found that the energy gap between the ground and first excited subbands increases considerably by decreasing @xmath26 .
for example @xmath27 k for @xmath28 cm for the neon substrate and is much larger for a substrate with higher @xmath13 .
@xcite then one can disregard the possibility of electron transition to excited subbands which is proportional to @xmath29 $ ] and take only the ground subband into account in the calculation of the scattering matrix elements .
the electron interaction with helium vapor atoms by a contact type hamiltonian is given by @xmath30where @xmath31 and @xmath32 are three - dimensional positions of electrons and atoms , respectively and @xmath33 is the scattering length.@xcite the electron - ripplon interaction was derived in ref . and the electron interaction with surface roughness was obtained in the case of se over solid hydrogen.sokrinostud95 very recently two of us have constructed an hamiltonian where both electron - ripplon and electron - solid interface interactions are treated on the same footing.@xcite the resulting total interaction potential for any substrate was obtained from the solution of the poisson s equation through the perturbation of the boundary conditions from the flat positions at @xmath34 and @xmath35 due to very small polarizability of liquid helium , the result of our approach for the electron - ripplon interaction coincides with that obtained in ref . .
bearing this in mind , we write the hamiltonian for the electron - ripplon and electron - interface couplings in a unique form as @xmath36with@xcite @xmath37 + \frac{\lambda _ { 1}}{\left ( z+d\right ) ^{2}}+ee_{\bot } \]]and@xcite @xmath38 } { ( z+d)}\right . \\ & & \left .
+ \left ( \frac{\varepsilon _ { s}-\varepsilon _ { he}}{\varepsilon _ { s}+\varepsilon _ { he}}\right ) \frac{q^{2}k_{2}[q(z+d)]}{2}\right\ } , \end{aligned}\ ] ] where @xmath39 is the modified bessel function . here
we used the fourier transform @xmath40 for @xmath41 and @xmath42 and , quantizing the oscillations of free helium surface , one obtains@xmath43with @xmath44 and @xmath45 being the annihilation and creation operators for ripplons with wave number @xmath46 and satisfying the dispersion law @xmath47where @xmath48 is the surface tension coefficient and @xmath49 is the density of helium . in eq .
( [ 6 ] ) , @xmath50 with @xmath51 being the gravity acceleration and @xmath52 the van der waals constant .
even though one can not find from first principles the interface roughness displacement @xmath53 , we will use a reasonable model to calculate the interface scattering contribution to se transport properties .
in order to calculate se mobility along the helium surface in the presence of a driving electric field @xmath54 we apply the well - known kinetic formalism where the electron scattering plays a crucial role in transport properties and does contribute to collision integrals in the boltzmann equation as follows @xmath55where @xmath56 is the electron distribution function and @xmath57 , @xmath58 , and @xmath59 are collision integrals of electron with helium atoms , ripplons and roughness interface between the film and the substrate , respectively . by considering quasielastic scattering processes the solution of eq .
( 7 ) is @xmath60 where @xmath61 is equilibrium distribution given by the boltzmann function for the nondegenerate q2des , @xmath62 is the angle between @xmath17 and @xmath54 , and @xmath63where the collision frequency @xmath64 and @xmath65 .
\label{8}\]]the probability amplitude per unit - time @xmath66 for electron transition from states @xmath17 to @xmath67 is given by fermi s golden rule and the subscript @xmath68 corresponds to each scattering mechanism whose interaction potential is given by eqs .
( [ 3 ] ) and ( [ 4 ] ) .
the se mobility is then given by@xmath69where @xmath70 , with the thermal wave vector @xmath71 the frequency of electron - atom collisions has been calculated in ref .
saitoh77 and the result is @xmath72note that @xmath73 does not depend on @xmath74 being a function of volume concentration of helium atoms @xmath75 where @xmath76 is the vaporization energy and @xmath77 is the @xmath78he mass . as @xmath79 is an increasingly exponential function of @xmath80 @xmath73 becomes negligible in comparison with @xmath81 and @xmath82 for @xmath83 k. the calculation of @xmath81 and @xmath84 is more complicated .
using eqs .
( [ 4 ] ) and ( [ 8 ] ) one can obtain in a straightforward way that both @xmath85 and @xmath86 can be written as @xmath87where @xmath88 means an average over ensemble . considering the electron - ripplon scattering one obtains , from eq .
( [ 5 ] ) , @xmath89 where the ripplon number @xmath90 is given by the bose - einstein function , which leads to @xmath91 for long wavelength ripplons with @xmath92 .
equation ( [ 12 ] ) results into @xmath93where @xmath94 and@xmath95 .\end{aligned}\]]here @xmath96 , \label{16}\end{aligned}\]]with @xmath97 \theta ( 1-y ) \\ & & -\left [ \frac{1}{\sqrt{\left ( y^{2}-1\right ) ^{3}}}\arcsin \left ( \frac{% \sqrt{y^{2}-1}}{y}\right ) \right ] \theta ( y-1),\end{aligned}\ ] ] in the above expressions @xmath98@xmath99 @xmath100 , @xmath101
, @xmath102 is the exponential - integral function , and @xmath103 is the step function . for the scattering by interface roughness , one obtains @xmath104where@xmath105 } { x+2\gamma d}\exp ( -x)dx . \label{18}\]]in eq .
( [ 18 ] ) and thereafter @xmath39 is the modified @xmath68-order bessel function .
now we must to choose an interface roughness model to derive the amplitudes @xmath106 appearing in eq .
( [ 12 ] ) . following prange and nee @xcite
, we use the gaussian correlated model , in which the interface roughness is described by two characteristic sizes , the height @xmath107 and the lateral length @xmath108 these parameters define the autocorrelation function for interface roughness which can be written as @xmath109which leads to @xmath110assuming this model , the collision frequency , given by eq .
( [ 17 ] ) , can be calculated and the result is@xmath111where @xmath112 and@xmath113 the gaussian correlated model was employed due to the successful description of the interface roughness for similar systems , for its simplicity and , since there are only two fitting parameters @xmath114 and @xmath115 monarkha and co - workers@xcite have studied electron correlation effects in the transport as well quantum magnetotransport of se on helium by using the force balance transport equation method ( fbem ) . in this approach the frictional force experienced by the center of mass of the q2des due to electron - scatterer interactions is evaluated through the calculation of the momentum rate absorbed by the scatterers . for the system of @xmath116 electrons the kinetic frictional force @xmath117
is given by@xmath118where @xmath119\right\rangle .
\label{27}\end{aligned}\]]here @xmath120 and @xmath121 are momentum and energy of scatterer system in the state @xmath122 , @xmath123 the energy of the electron system in the state @xmath124 , @xmath125 is the hamiltonian of the interaction between the electron and the particular scatterer .
the angle brackets means a thermodynamic average in the laboratory coordinate system . in ref .
is was shown that @xmath126 can be written in terms of the dynamic form factor of the q2des @xmath127 in the laboratory frame .
@xmath128 is connected with the form factor @xmath129 in the center - of - mass of the electron system , moving at a velocity @xmath130 relative to the laboratory frame , as @xmath131 . here
it is important to point out that electron correlations are considered in the system through the many - particle dynamic structure factor which is related to the density - density response function @xmath133 through the fluctuation - dissipation theorem @xmath134im@xmath135 $ ] , where @xmath136 is the electron density.@xcite in the limit of low velocities @xmath130 satisfying the condition @xmath137 , we may define an effective collision frequency as a proportionality factor between the momentum loss per unit time of the electron system and the average electon velocity as @xmath138the effective collision frequency is the sum of frequencies for a particular interaction hamitonian @xmath139 .
the electron mobility is given then by @xmath140 substituting the interaction hamiltonians , given by eqs .
( [ 3 ] ) and ( 4 ) , into eq .
( [ 27 ] ) one obtains @xmath141 @xmath142 \nonumber \\ & & \times |\langle \chi _ { 1}|v_{rq}(z)|\chi _ { 1}\rangle the calculation of @xmath143 is done in a similar way to that of @xmath144 and the result is straightforward@xmath145 is essential to the evaluation of @xmath146 .
unfortunately @xmath147 can be calculated analytically only in the case of the noninteracting electron system and there are no reliable approximate expressions for @xmath148 in the whole frequency range in contrast to the static structure factor @xmath149 @xmath150 which can be evaluated by appropriate approximation methods . in our calculations , we replace @xmath148 in eqs .
( [ 29]-[31 ] ) by the noninteracting dynamical structure factor@xcite@xmath151and consider the limit @xmath152 .
now we present the results of numerical calculations for se mobility over a helium film from eqs .
( [ 9 ] ) and ( [ meff ] ) .
we have used the parameters @xmath153 and @xmath114 of the electron - interface scattering to have the best fit to the mobility experimental data .
first we must say that the mobility curves in the bea and fbem exhibit the same overall behavior independently of the adjustable parameters for the electron - interface interaction .
results for the mobility dependence on the film thickness in the bea are presented in fig . 1 for a glass substrate @xmath154 for @xmath155 k and the fitting parameters @xmath156 and @xmath157 .
for a sake of comparison , we plot the mobility data taken from ref . .
one can see the decisive role of electron - interface scattering and the best fit is attainable for @xmath158 , where , as expected , the influence of roughness is more pronounced .
we also show separately the contributions to the mobility coming from electron - ripplon and electron - atom scattering processes .
we observe that neither electron - gas nor electron - ripplon scattering mechanisms can explain the experimental data for wide range of @xmath26 . by fitting the experimental data with mobility calculations in the fbem
, we obtain the best results for @xmath159 and @xmath160 . in fig .
2 we depict , in a similar way to fig .
1 , mobility curves as a function of the film thickness for solid neon @xmath161 and @xmath162 k. the best fit of experimental data @xcite is achieved for @xmath163 and @xmath164 in bea and @xmath165 and @xmath166 in fbem .
we see now that @xmath153 and @xmath114 are of the same order of magnitude of @xmath26 which makes unjustifiable the approximation @xmath167 which supports the perturbative approach to obtain the potential @xmath168 of eq .
( [ 4 ] ) , and , hence , the applicability of born approximation for the description of electron - interface scattering .
we would guess that the neon surface , being much rougher than the glass one , it has more irregularities leading to significant fluctuations of @xmath169 moreover , we observe a worse agreement between the theoretical curve and experimental data at large @xmath26 , where the electron - ripplon and electron - gas mechanisms would dominated the scattering .
the temperature dependence of se mobility is depicted in fig .
3 where the same values for @xmath153 and @xmath114 obtained by fitting @xmath170 are used .
as it can be seen we get rather good agreement with the experimental data of ref .
jiangstandahm88 for @xmath171 .
however the agreement becomes less satisfactory for smaller @xmath26 . by considering only the electron - ripplon and electron - atom scattering we found a difference of about more than one order of magnitude between the calculation results and the experimental ones .
unfortunately , to our knowledge , there is no experimental data on the temperature dependence of se mobility in the case of a film over solid neon .
figure 4 shows the influence of the dielectric constant of the substrate on the se mobility calculated within the bea .
we used here the same values of @xmath107 and @xmath114 as in fig . 1 for both glass and pmma @xmath172 substrates .
we see that the mobility varies inversely with respect to @xmath173 for whole of temperatures considered .
this is a direct consequence from @xmath174 appearing in the collision frequencies which in turn are in the denominator of the mobility formulas [ eqs .
( [ 9 ] ) and ( [ meff ] ) ] . in order to make clearer the role of different scattering mechanisms we present in fig .
5 the se mobility , calculated within the fbem , as a function of temperature for a film over glass and @xmath171 .
the calculated mobility curves are shown when the distinct scattering processes are considered separately .
as it can be seen , the inclusion of roughness interface scattering is essential to explain the experimental data .
the nonmonotonic dependence of se mobility must be attributed to the crucial role of electron - interface scattering since it is well known that the mobility limited only by electron - ripplon and electron - gas scattering is a monotonous decreasing function of temperature .
and glass substrate .
the solid line is the contribution from all scattering mechanisms whereas the dashed line is the contribution of gas scattering and the dotted line is the result for only the interface roughness scattering process . ] finally , we compare the mobility results obtained in the two different methods used here by depicting the curves of @xmath170 and @xmath175 in figs . 6 and 7
respectively for the case of a glass substrate .
we plot also with the experimental data .
smaller values of @xmath153 obtained in fbem as compared with those in bea fulfill more satisfactorily the condition latexmath:[$% between experimental data and theoretical curves calculated in fbem for the same values of @xmath153 and @xmath114 .
we point out that the numerical results within the fbem are the same as those obtained in the complete control approximation ( cca).@xcite in the cca , the mobility is calculated with the bea , but in the regime where the electron - electron collision frequency is much higher than the collision frequency due to other mechanisms .
the electrons are supposed to redistribute their momenta , due to collisions between electrons , in such a way that the momentum of the total system does not change but electrons acquire the same drift velocity .
the coincidence of results indicates that in order to establish the specific role of electron - electron interactions we must go beyond the noninteracting dynamical structure factor for calculating the mobility in the fbem as given by eq .
( [ 32 ] ) .
our conclusions then are valid only in the low or intermediate electron density regime .
in this work we have studied the dependence of the mobility of surface electrons over a liquid helium film on the temperature and on the film thickness .
the transport properties are determined from the solution of the boltzmann equation in the relaxation time approximation and in the force balance equation .
we employed the gaussian correlated model to describe the interface roughness between the film and the substrate .
the parameters are adjustable through the evaluation of both temperature and film thickness dependencies of the mobility .
the values obtained for the strength and range of the rough interface are of the order of few atomic layers and @xmath3 respectively which are reliable and in the mesoscopic regime .
we showed that the se mobility is limited by the roughness scattering because electron - ripplon and electron - gas atom scattering can not explain the experimental data .
however , we can not rule out the presence of other mechanisms as se localization by potential wells caused by the underlying solid substrate.@xcite
we are indebted to professor a. j. dahm , whom forthwith provides us values of specific constants .
this work was supported by cnpq and fapesp , brazil . | the mobility of surface electrons localized over helium films underlying solid substrates has been evaluated by solving the boltzmann equation in the time relaxation approximation and the force balance equation in which an effective mobility is obtained in terms of the dynamical structure factor of the nondegenerate electron liquid .
the essential processes of electron scattering by gas atoms , ripplons , and film - solid interface roughness are taken into account . the electron mobility dependence on the film thickness and temperature is determined and compared with experimental data available .
we find that the interface - roughness scattering is the dominant process for explaining the experimental results .
we estimate the extended defect sizes of the underlying substrate within the gaussian correlated model for interface roughness . |
let @xmath6 be a commutative noetherian ring with unit indeed , _ all _ rings encountered this paper will have these properties and let @xmath7 be an ideal . for @xmath8
, define @xmath9 as the largest @xmath10 such that @xmath11 , or @xmath12 if there is no largest @xmath10 ( e.g. if @xmath13 ) . then the *
* associated graded ring is defined as @xmath14 one of its virtues is that it has a map not only to , but from , @xmath15 .
( whereas @xmath6 does nt naturally have a map from @xmath15 . )
moreover , the map @xmath16 is an inclusion ( as the @xmath17 summand ) . following samuel ( our reference is rees book @xcite ) , define @xmath18 , the * * homogenization of the filtration @xmath19 .
samuel proved that this limit exists .
nagata and rees showed that it is rational - valued with bounded denominator .
rees gave a formula for @xmath20 , using rees algebras , which we recall in section [ sec : ogr ] .
nagata proved that @xmath21 is bounded ( * ? ? ?
* theorem 4.21 ) , which implies that @xmath22 is again noetherian ( * ? ? ?
* lemma 2.46 ) .
note that this grading is by @xmath23 , not ( usually ) by @xmath24 .
[ ex : xsquare ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath25 $ ] , @xmath26 .
then @xmath27 , and @xmath28 / { \langle}\bar x^2 { \rangle}$ ] , where the subscripts indicate the degrees .
whereas @xmath29 , and @xmath30 $ ] , or @xmath31 \big/ { \langle}\bar x^2 - y { \rangle}$ ] written for comparison .
we urge the reader to check the details of this example , as examples [ ex : betamap ] and [ ex : fatpointintersection ] build upon this one . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the following intuition seems to be useful . in @xmath32 in the example above , @xmath33 is `` rushed '' into the degree @xmath34 piece ( rather than waiting until degree @xmath35 where it `` belongs '' ) , and by degree @xmath34 standards its square ( which has @xmath36 ) vanishes .
it is this premature appearance of @xmath33 that leads to its nilpotency in @xmath32 .
[ prop : flatdegens ] assume that the ideal satisfies @xmath37 .
* there is a flat degeneration of @xmath6 to @xmath32 .
* @xmath38 has no nilpotents . *
if @xmath6 has no nilpotents , there is a flat degeneration of @xmath6 to @xmath38 .
* there are natural maps @xmath39 .
* there are natural maps @xmath40 .
we start with the second claim .
let @xmath41 be nilpotent , so @xmath42
. then @xmath43 .
hence @xmath44 , a contradiction .
the intersection @xmath45 is the ideal @xmath46 , which under the assumption @xmath47 is just the nilpotents .
then the first and third claims use the rees algebra to provide the flat family @xcite .
the fourth claim is obvious .
for the fifth , we need to compute @xmath48 .
then + @xmath49 much of this paper is concerned with the natural map @xmath50 , which we take up in the next section .
young algebraic geometers are strictly indoctrinated to regard killing nilpotents as a bad habit ; information is being thrown away . to allay their suspicions , we emphasize that @xmath38 is _ not _ just @xmath32 mod its nilpotents ( though as we shall see , it contains that as a subring ) .
if @xmath6 and @xmath7 are graded so that one can speak of hilbert functions , then @xmath51 all have the same hilbert function , whereas @xmath32 modulo its nilpotents will have a smaller hilbert function ( unless it has no nilpotents ) . in this sense , the information usually recorded in nilpotents is just showing up in a different way .
since the denominators in @xmath20 are bounded , one may be tempted to clear them by rescaling the grading .
this seems to carry no benefit , and only serves to make the map @xmath52 no longer graded .
given a subscheme @xmath53 of a scheme @xmath54 , hence an ideal sheaf @xmath7 inside the structure sheaf @xmath6 , we can define the * * normal cone @xmath55 to @xmath53 and the * * balanced normal cone @xmath56 , using @xmath57 and @xmath58 respectively .
the term `` balanced '' is chosen to evoke the idea that the grading is carefully weighted to avoid creating nilpotents .
[ lem : betamaps ] let @xmath60 be an ideal in a ring @xmath6 . then there is a natural map @xmath61 . moreover , if @xmath62 takes @xmath63 for @xmath64 an ideal in @xmath65 , then there is a natural commuting square @xmath66 where the associated gradeds are the obvious ones .
the main fact used is that @xmath67 .
the statements then follow more or less directly from the definitions .
since @xmath38 has no nilpotents , the kernel of @xmath61 is plainly at least the nilpotents .
[ prop : betaalg ] the kernel of @xmath68 is exactly the nilpotent elements in @xmath32 . if @xmath32 has no nilpotents , then @xmath69 is an isomorphism ( and otherwise not ) .
we need the calculation @xmath70 let @xmath71 denote the image of @xmath72 in @xmath73 . if @xmath74 , then @xmath75 , so @xmath76 .
hence @xmath77 .
so the kernel is exactly the nilpotents .
if @xmath32 has no nilpotents , then there does not exist @xmath78 with @xmath42 .
so @xmath79 is not more than @xmath80 ; indeed @xmath81 for all @xmath82 .
hence @xmath83 , and @xmath84 .
thus @xmath85 naturally .
since @xmath38 has no nilpotents , @xmath32 can only be isomorphic to it if it too has no nilpotents .
we now switch over to the geometric point of view , in which we map from the balanced normal cone to the ordinary one .
the above proposition tells us that the map @xmath59 factors as @xmath86 , where @xmath87 denotes the reduction of @xmath55 .
so @xmath87 serves as an intermediary when trying to compare the spaces @xmath55 and @xmath56 .
this motivates our looking at chow groups , since @xmath88 .
[ thm : betageom ] let @xmath53 be a closed subscheme of @xmath54 , where @xmath54 is reduced .
the induced map @xmath89 is proper , with finite fibers .
assume now that @xmath54 is quasiprojective .
then the two maps @xmath90 , @xmath91 induce the same chow class in @xmath92 .
( we expect that the hypothesis on @xmath54 is largely unnecessary . )
note that these two chow classes are induced on @xmath87 in very different ways , as we go over in section [ sec : proof ] .
the inclusion @xmath91 defines a class by taking the sum of the top - dimensional components weighted by the lengths of the local rings on the target . whereas the surjection @xmath90 defines a class by taking the sum of the top - dimensional components weighted by the degree of the map over those components .
[ ex : betamap ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath54 be the line with coordinate @xmath33 , and @xmath53 the doubled origin ( defined by @xmath93 ) . then @xmath55 is the doubled line , whereas @xmath56 is just the ordinary line ; see example [ ex : xsquare ] for these calculations .
the map @xmath86 is the squaring map from the line to the ( reduction of the doubled ) line . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it can happen that @xmath56 has more components than @xmath55 , not because a component collapses ( since we know there are finite fibers ) , but because multiple components of @xmath56 cover the same component of @xmath55 .
when this happens , we get a refinement of the multiplicities in the fundamental class of @xmath55 ; the multiplicity of a component @xmath94 is the sum over those components @xmath95 whose image is @xmath96 , of the degree of the map @xmath97 .
[ ex : refined ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath98 / { \langle}a^2-b^2{\rangle}$ ] , so @xmath99 is the union of two lines .
let @xmath100 , so @xmath53 is a double point at the origin .
@xmath55 is a trivial line bundle over @xmath53 , @xmath101 , @xmath102 , and @xmath103 .
the map @xmath59 maps the two lines onto the reduction of @xmath55 . in this way ,
the fundamental class of @xmath55 is a sum of the two ( equal ) chow classes induced by the lines in @xmath56 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ to prove theorem [ thm : betageom ] , we need a number of basic results about balanced normal cones , which will come in section [ sec : ogr ] .
the proof itself will come in section [ sec : proof ] .
it is a bit involved , which seems to be inherent in the fact that the two classes on @xmath87 are induced in different ways .
a simpler proof will appear in @xcite , where we show that the @xmath32-modules @xmath32 , @xmath38 are @xmath104-equivalent .
this implies theorem [ thm : betageom ] , at least in characteristic @xmath34 and in the rational chow group .
we recall the * * basic construction from @xcite .
let @xmath105 be an inclusion ( soon , a regular embedding ) , and @xmath106 a morphism .
let @xmath53 be the pullback , so we have a square @xmath107 now replace each of the big schemes ( @xmath1 and @xmath54 ) by the normal cones to the subschemes .
this allows us to reverse the horizontal arrows , replacing inclusions by epimorphisms .
@xmath108 this is no longer a pullback diagram ; we only have a map from @xmath55 to the actual pullback @xmath109 .
hence @xmath55 defines a chow class on @xmath109 .
( while it is not hard to check that the map @xmath110 is an inclusion , this property does nt seem to play any role in the construction . ) for purposes of intersection theory , it turns out to be useful to require that @xmath111 be a regular embedding , i.e. that @xmath2 be a vector bundle .
this is because fulton and macpherson s goal is to define a chow class down on @xmath53 ( not up on @xmath109 ) , which they call the `` refined intersection product '' of @xmath0 and @xmath54 .
( it can be thought of as a cap product , where the regular embedding @xmath112 plays the role of the cobordism class and the map @xmath113 that of the bordism class . )
this is done using a thom - gysin isomorphism @xmath114 , which holds if @xmath109 is a vector bundle of some dimension @xmath115 .
this is guaranteed if @xmath2 is a vector bundle , motivating that condition .
this completes the basic construction .
how do things change in what we will call the * * balanced basic construction , where we instead use balanced normal cones ? first , we will require that @xmath54 and @xmath1 are reduced , in order that their degenerations to balanced normal cones be flat degenerations .
( these are not particularly stringent assumptions from the point of view of intersection theory , where the most important case is @xmath0 smooth and @xmath116 .
however one should note that if @xmath0 is regularly embedded in @xmath1 , but not reduced , it does not follow that @xmath3 is a vector bundle over @xmath117 . ) as before , by passing to the cones we can reverse the horizontal arrows .
however , these reversed maps are no longer epimorphisms they only hit the reductions ( thanks to the last part of proposition [ prop : flatdegens ] ) .
@xmath118 if we assume that @xmath0 is smooth and @xmath119 is a regular embedding , then @xmath2 is reduced ( as it is a vector bundle over something reduced ) .
hence @xmath120 , and the pullback to @xmath53 is again @xmath109 .
however , even in this case , @xmath53 and @xmath55 are typically not reduced .
so @xmath56 , which _ is _ reduced , is something new .
it too maps ( though usually _ not _ injectively ) to the pullback bundle @xmath109 , and this map factors as @xmath121 .
[ thm : sameclass ] assume that @xmath122 are reduced , with @xmath112 regularly embedded and @xmath113 a morphism .
fulton and macpherson s refined intersection product @xmath123 , usually calculated with @xmath55 , can be calculated equally well with @xmath56 . since the map @xmath124 factors through @xmath125 , theorem [ thm : betageom ] implies that @xmath56 and @xmath55 induce the same chow class on @xmath109 .
[ ex : fatpointintersection ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath126 be affine lines with coordinates @xmath127 , let @xmath0 be the origin in @xmath1 , and let @xmath128 be the squaring map @xmath129 . then @xmath53 is the doubled origin in @xmath54 , defined by @xmath130 . in ordinary intersection theory ,
the normal cones @xmath131 and the pullback @xmath109 are all trivial line bundles , over @xmath132 respectively .
the map @xmath110 is an isomorphism , inducing the fundamental class on @xmath53 , which is the twice the class of the reduced point @xmath133 . in the balanced basic construction ,
the balanced normal cone @xmath56 is the trivial line bundle over @xmath133 , and the map @xmath124 is the squaring map , rather than an isomorphism .
we calculate this on the algebra side , where the diagram above is @xmath134/{\langle}v^2{\rangle } & { \mathop{\twoheadrightarrow } } & { { \mathbb f } } & { \operatorname*{\hookrightarrow } } & { { \mathbb f}}[v_{(1/2 ) } ] \\
\uparrow & & \uparrow & & \uparrow \\ { { \mathbb f } } & = & { { \mathbb f } } & { \operatorname*{\hookrightarrow } } & { { \mathbb f}}[y_{(1 ) } ] .
\end{array}\ ] ] here the parenthesized subscripts indicate the degree in these graded rings . in the graded map on the right , @xmath135 .
the pushout @xmath136 of that right square is obviously @xmath137 $ ] , so this squaring map is the one @xmath138 claimed above , inducing twice the fundamental chow class of @xmath139 .
the gysin map then takes that to twice the fundamental class of the reduced point @xmath133 , as predicted by theorem [ thm : sameclass ] .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ because the space @xmath56 can have more components than @xmath55 , as in example [ ex : refined ] , we can refine fulton and macpherson s `` refined intersection products '' further as a sum over the components of @xmath56 . [ ex : cubic ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the further refinement in example [ ex : refined ] only reflected the fact that @xmath54 itself was reducible .
this example , revisited in section [ sec : limits ] , shows the refinement can be nontrivial even when @xmath54 is irreducible .
let @xmath140 $ ] , @xmath0 the @xmath141-axis , and @xmath54 the nodal cubic curve @xmath142 .
their intersection @xmath143 is a double point at the origin ( the node of the cubic ) and a reduced point at @xmath144 .
the map from @xmath55 to the pullback @xmath145 of the ( trivial ) normal bundle @xmath2 is an isomorphism , inducing the fundamental chow class on @xmath145 and thereby on @xmath53 . in this case
, the chow ring calculation ( on the projective plane ) gives @xmath146 times the class of a point .
the refined intersection product just calculated splits this as @xmath147 , from the double point and single point . to compute the balanced normal cone @xmath56 , present @xmath148 using @xmath149/{\langle}c - a(a+1 ) , b^2 - ac , b^2(a+1 ) - c^2{\rangle}$ ] . then @xmath150 , so @xmath151 is plainly at least @xmath152 . as we will be able to compute later ( using proposition [ prop : compute ] ) , in fact @xmath153 , @xmath154 , and @xmath155/{\langle}a(a+1 ) , ac , b^2(a+1 ) -
c^2{\rangle}= { { \mathbb f}}[a , b , c]\big/\big({\langle}a+1,c{\rangle}\cap { \langle}a , b - c{\rangle}\cap { \langle}a , b+c{\rangle}\big).\ ] ] so @xmath156 is an isolated line union a pair of intersecting lines , and the sum of these components further refines the intersection calculation as @xmath157 . unlike @xmath6 and @xmath32
, this ring @xmath38 is not generated over @xmath158 by two elements .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it would be interesting to find the branch locus of the map @xmath90 in genuine intersection theory examples , and see what that and more refined degeneracy loci mean for enumerative questions .
it would also be interesting to see how the monodromy group of the branched cover relates to the `` galois group '' of the enumerative problem @xcite .
we now outline the rest of the paper . in section [ sec : ogr ] we describe rees formula for @xmath20 and give the basic results about @xmath38 .
when @xmath32 is reduced , then @xmath159 ; we present a number of examples to show some possible reasons that @xmath160 . in section [ sec :
limits ] we study the intersection of a variety in affine space with a hyperplane , and geometrically describe the normal cone ( and under certain conditions , the balanced normal cone ) as flat limits . in section [ sec : icogr ]
we introduce the ring @xmath161 with which to further study @xmath38 in the case that @xmath7 is principal , and we compute several examples .
finally , in section [ sec : proof ] we prove theorem [ thm : betageom ] .
it is a pleasure to thank valery alexeev , tom graber , mark gross , joseph gubeladze , craig huneke , bernd sturmfels , and ravi vakil .
i give special thanks to bernard teissier for sharing with me his unpublished manuscript @xcite .
many examples in this paper were worked out with the help of the computer algebra system macaulay 2 @xcite .
in this section @xmath6 is a ring without nilpotents . ( and commutative , noetherian , and with unit , as per our standing assumptions . )
assume to begin with that @xmath6 is an integrally closed domain , and @xmath7 is a principal ideal @xmath162 . let @xmath163 be the components of @xmath7 s vanishing set , and @xmath164 the corresponding valuations .
then @xmath165 , or put another way , @xmath166 the same bound follows for @xmath20 .
rees theorem , in this special case , says that @xmath20 is actually _ given _ by this formula .
[ ex : tvgr ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath167 be a lattice polytope , and @xmath6 the homogeneous coordinate ring of the projective toric variety @xmath168 , which has a basis given by lattice points in dilations of @xmath167 .
let @xmath169 be the degree @xmath152 element corresponding to some lattice point @xmath170 .
then the valuations @xmath171 in the formula for @xmath20 correspond to the facets of @xmath167 not containing @xmath172 .
if @xmath173 is any lattice point in @xmath167 and @xmath72 the corresponding ring element , then @xmath174 is the distance of @xmath173 to the @xmath111 facet , measured in lattice units .
let @xmath175 denote the continuous piecewise - linear function measuring the distance of @xmath19 to a far wall of @xmath167 along the straight line connecting @xmath172 and @xmath19 ; it takes the value @xmath152 at @xmath172 , @xmath34 on all facets @xmath96 not containing @xmath172 , and varies linearly on the cone from @xmath172 to @xmath96 . then if @xmath176 is a basis element corresponding to a lattice point @xmath170 , we have @xmath177 .
( more generally , if @xmath176 is a basis element corresponding to a lattice point @xmath173 in the @xmath178-fold dilation @xmath179 of @xmath167 , we have @xmath180 . )
an example is in figure [ fig : toricdegen ] .
the ring @xmath38 is the homogeneous coordinate ring of a _ union _ of projective toric varieties , whose components are ( weighted ) cones on the facets of @xmath167 not containing @xmath172 .
this reducibility arises from the fact that in the associated graded , the product of two basis elements can be zero , which happens if and only if when projecting away from @xmath172 the corresponding points in @xmath167 do not project to a common facet .
more general reduced torus - equivariant degenerations of toric varieties were studied in @xcite . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ several phenomena must be dealt with to generalize rees formula to the case of a general ring @xmath6 and nonprincipal @xmath7 .
first , if @xmath7 is not principal , we work with the blowup algebra .
one interesting aspect of this is that if @xmath181 or @xmath182 are singular along @xmath182 , then the exceptional locus in the blowup may have more components than @xmath182 does itself , and we need them for rees formula .
second , if @xmath6 is not integrally closed , the valuations @xmath171 may not be simply associated to divisors .
for example , let @xmath183/{\langle}xy{\rangle}$ ] , and @xmath184 .
then @xmath7 vanishes at the origin , to order @xmath152 if one approaches along the @xmath33-axis and order @xmath185 if one approaches along the @xmath186-axis .
hence `` the order of vanishing at the origin '' is not well - defined .
( for @xmath100 , the natural condition is that @xmath6 be integrally closed inside @xmath187 $ ] .
this will show up in a different guise in section [ sec : icogr ] . ) finally , if @xmath181 is reducible , @xmath7 may vanish altogether on some components .
if @xmath72 does nt vanish on those , then plainly @xmath188 for any @xmath72 , so @xmath189 .
if contrariwise @xmath72 does vanish on them , we can remove those components by passing to @xmath190 and compute @xmath20 in that ring .
in the rest of this section we assume @xmath100 .
[ lem : fingen ] let @xmath191 be a graded noetherian ring . then @xmath6 is finitely generated as an algebra over @xmath192 .
let @xmath193 be homogeneous generators of the augmentation ideal @xmath194 .
let @xmath72 be a homogeneous element of positive degree .
then @xmath195 for some homogeneous @xmath196 with @xmath197 .
by induction , these @xmath198 are polynomials in the @xmath199 with coefficients from @xmath192 .
hence @xmath72 is such a polynomial too . by the direct sum assumption ,
every element of @xmath6 is a sum of such @xmath72s and an element of @xmath192 .
[ prop : multbyb ] let @xmath100 be a principal ideal in a ring @xmath6 .
* the multiplication map @xmath200 is always onto . *
the multiplication map @xmath200 is @xmath201 for all large @xmath111 . *
the multiplication map @xmath202 is always @xmath152:@xmath152 for @xmath203 .
if @xmath169 is not a zero divisor in @xmath6 , then @xmath204 is also @xmath152:@xmath152 .
* the multiplication map @xmath202 is onto for all large @xmath111 .
note that the first two claims are only interesting for @xmath205 , whereas the second two are interesting for @xmath206 .
the first claim is essentially tautological . for the second ,
consider the ascending chain @xmath207 of annihilator ideals in @xmath32 .
let @xmath208 be the stage at which it stabilizes . then for @xmath209
, @xmath210 implies @xmath211 for any @xmath212 . with this and the first claim
, we see that if @xmath213 , @xmath212 , then @xmath214 .
the third and fourth do nt depend on @xmath6 as much as @xmath6 mod its nilpotents , so we assume now that @xmath6 has none .
for the third , let @xmath215 .
if @xmath169 vanishes on a component of @xmath181 on which @xmath216 does nt , then ( 1 ) @xmath169 is a zero divisor and ( 2 ) @xmath217 .
so now we assume that @xmath216 vanishes on each component of @xmath181 on which @xmath169 vanishes , and we can pass to @xmath218 .
now , @xmath219 .
then @xmath220 so @xmath221 implies @xmath222 .
contrapositively , the only way for @xmath223 to be annihilated by @xmath224 is for @xmath169 to be a zero divisor _ and _ @xmath217 .
this gives the third claim .
for the last claim , since @xmath38 is noetherian , let @xmath225 generate @xmath38 as an algebra over @xmath226 ( using lemma [ lem : fingen ] ) .
then any monomial in the @xmath227 of high degree must involve some @xmath228 to a high power .
since each @xmath229 , having @xmath228 to a high power means that a factor of @xmath230 can be extracted .
this establishes the fourth claim .
the following gives a characterization of @xmath20 that is useful for verifying examples , and in section [ sec : limits ] will also be of use in interpreting balanced normal cones geometrically .
it uses the concept of * * homogeneous filtrations @xmath172 , meaning @xmath231 @xmath232 .
[ prop : compute ] the filtration @xmath20 is the unique minimum homogeneous filtration @xmath172 with @xmath233 . in other words , let @xmath172 be a homogeneous filtration on @xmath6 such that @xmath234 . then @xmath235 @xmath236 .
if @xmath237/i$ ] , and @xmath238 are lower bounds on @xmath239 , then let @xmath172 be the ( possibly inhomogeneous ) filtration induced on @xmath6 from the filtration @xmath240 on the polynomial ring .
if the associated graded to @xmath172 has no nilpotents , then @xmath241 . by the existence of the limit @xmath20 , given @xmath242 , for all large @xmath10 we have @xmath243 .
hence @xmath244 for some @xmath245 .
then @xmath246 hence @xmath247 .
for the second part , saying that the associated graded to @xmath172 has no nilpotents is the same as saying that @xmath172 is homogeneous .
plainly @xmath233 .
so by the first part , @xmath248 . since @xmath172 is the smallest filtration with @xmath249 , and @xmath250 by assumption , we have @xmath251 .
so @xmath241 .
note that not every homogeneous filtration on a polynomial ring mod an ideal is of the form in the second part of the proposition for example , the @xmath252-adic filtration on @xmath253 $ ] .
we will only be able to apply the second part of proposition [ prop : compute ] when the generating set has been chosen felicitously . in some of the examples to come , we will present @xmath6 as a polynomial ring modulo an ideal .
we ll determine some lower bounds @xmath254 on the @xmath20s of the variables , including @xmath255 , and consider the induced ( a priori inhomogeneous ) filtration @xmath172 . to compute the associated graded to @xmath172 , we check that the generating set of the ideal is a grbner basis with respect to some term order respecting this weighting of the variables , and replace each relation by its lowest - weight component . to be sure we re satisfying proposition [ prop :
compute ] , it remains to check that the associated graded has no nilpotents .
when all goes well and that turns out to be true , we learn three things : @xmath256 , our lower bounds @xmath254 were correct , and each filtered piece of @xmath6 intersected with the linear span of the variables is spanned by a subset thereof . here
are some of the nonobvious possible behaviors of @xmath20 and @xmath58 .
one way of thinking about the limit @xmath257 is to take the limit through a subsequence @xmath258 , which is easily seen to be increasing : @xmath259 many people s first guess , upon learning the definition of @xmath20 , is that the limit @xmath260 is achieved for some finite @xmath10 .
this turns out to be true if @xmath6 is integrally closed .
let @xmath6 be an integrally closed domain , and @xmath100 .
then there exists @xmath261 such that @xmath262 .
let @xmath109 be the least common multiple of the valuations @xmath263 , so @xmath264 is @xmath24-valued .
then for any @xmath72 , the rational function @xmath265 satisfies the valuative criterion for integrality .
( we asked that @xmath6 be a domain so that @xmath169 is not a zero divisor . ) since @xmath6 is integrally closed , @xmath266 for some @xmath267 .
hence @xmath268 , but we already knew the opposite inequality . [ ex : nons2 ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this is a variant of example [ ex : refined ] , with the same geometry .
let @xmath98 / { \langle}a^2 - ab{\rangle}$ ] , and @xmath100
. then @xmath269 , and in fact @xmath270 .
taking the limit , @xmath271 .
but for no @xmath10 is @xmath272 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we ve already shown that @xmath273 , if and only if @xmath32 has no nilpotents , if and only if @xmath274 . one obvious reason for @xmath38 to be different from @xmath32 is if @xmath275 for some valuation @xmath164 in rees formula , and @xmath38 to have support in other than integer degrees .
geometrically , this corresponds to the divisor @xmath276 not being generically reduced .
( it is still possible for @xmath38 to be integer - graded , as example [ ex : refined ] shows . )
this raises the question : if the divisor @xmath276 is generically reduced , does that force @xmath159 ? to construct a counterexample , it will suffice to make the divisor generically reduced but not reduced , hence not satisfying serre s criterion @xmath277 .
so the ambient @xmath181 should nt satisfy serre s criterion @xmath278 , the canonical example being the union of two planes in @xmath279-space .
let @xmath280/{\langle}d(b - d ) , dc , e(b - d ) , ec { \rangle}$ ] , the union of the @xmath281 plane and @xmath282 plane .
then the @xmath169-divisor is @xmath283/{\langle}b , d^2 , dc , ed , ec { \rangle}$ ] , supported on the @xmath284 line union the @xmath285 line , with an extra point embedded at the origin . since @xmath286 for all @xmath287 ,
we see @xmath288 .
so @xmath289 , and the lower bounds @xmath254 we can guess for the @xmath20 of the variables are @xmath290 , @xmath291 .
( note that @xmath292 . )
the relations are homogeneous with respect to this weighting , hence the associated graded @xmath38 turns out to be isomorphic to @xmath6 : @xmath293/{\langle}ec , d(b - d),dc , e(b - d){\rangle}.\ ] ] since this has no nilpotents , we can use proposition [ prop : compute ] to know that we have correctly calculated @xmath20 .
( side note : the fact that @xmath103 does nt mean that @xmath38 is boring rather , it has served as a means of discovering a grading with which to better understand @xmath6 itself . ) whereas @xmath294/{\langle}ec , d^2,dc , ed{\rangle}$ ] , whose quotient by @xmath295 is @xmath296/{\langle}ec{\rangle}$ ] .
geometrically , the map @xmath297 corresponds ( in reverse ) to a pair of planes meeting at a point , mapping onto a pair of planes meeting along a line , mapping into a thickening of that scheme along the line .
( this hints at a strengthening of theorem [ thm : betageom ] which is implied by our result to appear in @xcite .
let @xmath298 be the open locus over which @xmath69 is a local isomorphism .
then it seems the map @xmath299 takes the fundamental chow class to the fundamental chow class . )
it is curious that this can only happen if @xmath169 is a zero divisor , as we now prove .
let the ring @xmath6 contain the element @xmath169 , and assume that @xmath169 is not a zero divisor , and @xmath300 has no nilpotents .
then @xmath32 has no nilpotents , so @xmath301 and @xmath159 .
assume @xmath215 is nonzero , and @xmath302 , so @xmath216 has image @xmath303 .
assume also that @xmath304 , so @xmath305
. then we can write @xmath306 and @xmath307 , where @xmath101 .
so @xmath308 , hence @xmath309 .
since @xmath169 is not a zero divisor , @xmath310 , so @xmath311 . hence @xmath312 is a nilpotent element of @xmath313 , contradiction . to find an example in which @xmath32 has nilpotents , but only after degree @xmath34
, we therefore need to allow @xmath169 to be a zero divisor .
the proof above suggests , we might equally well have used @xmath314 , in which case @xmath315 .
we preferred @xmath316 to emphasize that the advantages of @xmath38 over @xmath32 are not merely due to the rational grading .
] @xmath317 , which is almost good enough , we just need to take its radical ( using @xcite ) : @xmath318 let @xmath319/i$ ] .
then @xmath6 has no nilpotents , and neither does @xmath320/{\langle}b , c{\rangle}$ ] . but
@xmath321 , @xmath322 , so @xmath216 gives a nilpotent element of @xmath323 .
in fact @xmath316 , and @xmath324 \big/ { \langle}ab , ac , c^2-b^4 d{\rangle}\ ] ] where the parenthesized subscripts indicate the degrees .
this , too , can be checked with proposition [ prop : compute ] .
a more standard grbner basis calculation tells us @xmath325 \big/ { \langle}c , a^2 b{\rangle}.\ ] ] geometrically , the map @xmath50 corresponds ( in reverse ) to a union of a plane and a surface along a line , mapping to a union of a plane and a double plane along a line , where the map is generically @xmath201 on the first component and @xmath326 on the second .
let @xmath327 $ ] be a polynomial ring in @xmath10 variables , and @xmath7 a radical ideal .
let @xmath328 , and let @xmath329 be the @xmath276 hyperplane .
let @xmath330 , and @xmath143 .
we interpret the `` basic construction '' in this case in terms of a transparent geometric limit , and under the hypotheses of proposition [ prop : compute ] , do the same for the balanced version .
we include this description only for illustration , and in this section do not give full proofs ( though they are quite straightforward from the theory of grbner degenerations ) .
the basic construction , in this case , goes from @xmath331 hence @xmath55 maps into the pullback @xmath145 , inducing a chow class on @xmath145 and thereby on the intersection @xmath53 .
there is a geometric picture of the passage to the normal cone @xmath332 .
let the circle @xmath333 act on @xmath334 by @xmath335 then @xmath55 can be computed as the flat limit @xmath336 , stretching @xmath54 away from @xmath337 .
two things can happen to any particular component @xmath338 under this limit .
if @xmath339 , then @xmath340 for all @xmath341 including @xmath342 . the map @xmath343 restricts to a map @xmath344 , inducing the zero chow class .
it is more interesting when @xmath345 .
then @xmath346 , and the map @xmath343 restricts to an isomorphism @xmath347 , inducing the fundamental class .
the thom - gysin isomorphism then takes that to the fundamental class of @xmath348 inside @xmath53 . in all , the intersection class on @xmath53 is given by the fundamental classes of the thickenings of the @xmath349-dimensional components of @xmath53 , leaving out those components that were components of @xmath54 .
in the balanced basic construction , @xmath350 assume now we are in the case of proposition [ prop : compute ] , where @xmath351 for @xmath352 , and @xmath20 is induced from the filtration @xmath353 .
fix a number @xmath261 such that each @xmath354 . in this case
@xmath56 can also be computed as a limit .
let @xmath333 act on @xmath334 by @xmath355 then it is not hard to show that @xmath356 .
[ ex : parabola ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath54 be the parabola @xmath357 in the @xmath358-plane , so @xmath53 is a double point at the origin . then @xmath359 is the skinny parabola @xmath360 , whose limit as @xmath361 is a double line .
the map @xmath343 is an isomorphism . in the balanced construction , @xmath362 , and we need @xmath109 even .
so @xmath359 is the parabola @xmath363 , which is to say , @xmath364 .
the map @xmath365 is a double cover of the reduction of @xmath145 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ex : cubicagain ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ recall the nodal cubic @xmath148 , @xmath98\big/{\langle}b^2 - a^2(a+1){\rangle}$ ] from example [ ex : cubic ] .
the limit picture of the usual normal cone stretches this nodal cubic vertically , resulting in a line at @xmath366 and a double line at @xmath367 . in this case
@xmath38 was not generated by two variables ; we needed to introduce @xmath368 .
geometrically , @xmath54 is stretched into the third dimension . in terms of the @xmath369-picture
, the points in @xmath370 are left alone , the points elsewhere in @xmath371 are pushed behind the page , and the points in @xmath372 are pulled out of the page . the local picture of an @xmath373 through the origin is rotated a bit about the @xmath169 axis , leaving the @xmath358-plane .
the limit picture of the balanced normal cone stretches not only the vertical dimension , but the new third dimension ( since @xmath154 ) . in the limit , one has a vertical line through the point @xmath374 , and the local picture of an @xmath375 has been stretched to a union of two lines lying in the @xmath367 plane .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it would be interesting to study the relation of balanced normal cones and dynamical intersection theory ( which in the context of this section , defines the intersection as the flat limit of @xmath376 as @xmath377 ) .
very preliminary investigation suggests that where usual dynamical intersection theory studies how solutions collide as @xmath377 , the balanced version keeps track also of how _ fast _ they collide .
we saw in proposition [ prop : multbyb ] that the multiplication operator @xmath224 on @xmath38 is always @xmath201 above degree @xmath34 , so that @xmath379 is an injection .
this proposition also told us that on @xmath380 , multiplying by @xmath169 is @xmath201 and in high degrees , onto .
that suggests that we fill in the holes in small degrees , which we do now .
the map @xmath38 to the fraction ring @xmath381 $ ] has kernel @xmath382 .
define @xmath383$.}\ ] ] [ lem : icperiodic ] let @xmath384 be a @xmath385-graded ring with a homogeneous element @xmath169 such that @xmath386 is @xmath201 for all @xmath10 and onto for large @xmath10 .
let @xmath387 be homogeneous .
then @xmath388 is integral over @xmath6 , and even over @xmath389 $ ] .
pick @xmath390 such that @xmath391 .
then @xmath392 , and to show @xmath72 is integral it is enough to show @xmath393 is integral .
in this way we can reduce to the case @xmath394 , which we assume hereafter .
fix @xmath395 such that @xmath396 is onto for all @xmath397 .
then @xmath398 is a finite module over @xmath192 ( proof : take a homogeneous generating set for the ideal @xmath399 ; the elements in @xmath398 generate @xmath398 as an @xmath192-module ) . since we can use @xmath224 to identify all these @xmath400 for @xmath401
, we will denote this module by @xmath402 . by multiplying by @xmath403 , any homogeneous element
@xmath267 has an image @xmath404 in @xmath402 .
now consider the sequence @xmath405 in @xmath402 .
they generate an @xmath192-submodule of @xmath402 , but only finitely many are needed to generate .
hence for some @xmath406 , we can write @xmath407 with each @xmath408 . lifting back to @xmath6 , this becomes @xmath409 .
so @xmath72 satisfies a monic polynomial with @xmath389$]-coefficients .
[ thm : attaching ] let @xmath6 be a ring and @xmath169 an element , inducing @xmath410 as above .
then the natural map @xmath411 is a graded inclusion , and an isomorphism in high degrees .
the multiplication map @xmath412 is an isomorphism for all rational @xmath413 .
( the common reflex is to conclude from this that @xmath414 $ ] .
but @xmath161 is _ rationally _ graded , not integrally , so this result is merely specifying a periodicity in the grading . )
if @xmath415 is a graded ring with @xmath416 an algebraically closed field , and @xmath417 is homogeneous for this grading , then @xmath161 splits naturally as a finite direct sum of doubly graded rings @xmath418 with each @xmath419 .
the map given is the composite @xmath420 of two graded inclusions ( taking @xmath421 to be degree @xmath34 ) , and hence is one also . using proposition [ prop : multbyb ] ,
choose @xmath261 such that @xmath422 is an isomorphism for all @xmath397 .
then if @xmath423 for @xmath397 , we know @xmath424 , which we can identify with @xmath425 since @xmath426 . then by the assumption on @xmath109 , @xmath427 for some @xmath115 , so @xmath428 .
this shows that the inclusion @xmath429 is onto for @xmath397 . to see that @xmath224 is an isomorphism for all rational @xmath413
, we apply lemma [ lem : icperiodic ] to @xmath38 . since @xmath161 stands between @xmath380 and its full normalization , it is finite over @xmath380 . in particular
@xmath430 is a finite - dimensional @xmath158-algebra .
since @xmath430 has no nilpotents and @xmath158 is algebraically closed , we find @xmath431 . in more detail , @xmath430 has a unique @xmath158-basis @xmath432 up to reordering , with @xmath433 , @xmath434 for @xmath435 .
again since @xmath430 is finite - dimensional without nilpotents , all of @xmath430 must be in @xmath436 .
let @xmath437 as an algebra with unit @xmath438
. then @xmath439 as claimed .
it was to obtain a theorem like this that first led the author to the study of balanced normal cones , to study the hilbert function of @xmath415 in terms of @xmath421 and @xmath161 . in a future publication
@xcite we will use theorem [ thm : attaching ] inductively to study standard bases of homogeneous coordinate rings . for the rest of this section , we make the assumptions of the latter part of the theorem , namely that @xmath415 is a graded ring with @xmath416 an algebraically closed field , and @xmath417 is a homogeneous element .
let @xmath440 and @xmath441 the divisor @xmath276 .
let @xmath442 .
write @xmath443 for @xmath444 .
then by the last part of the theorem above , @xmath443 is a disjoint union of weighted cones @xmath445 .
we can now interpret some of these ring maps geometrically : @xmath446 & \qquad \longleftrightarrow \qquad & { \overline c}_x y\setminus x & { \operatorname*{\hookrightarrow } } & { \overline c}_x y \\ { { \overline{\mathrm{gr}}\,}}r & { \operatorname*{\hookrightarrow } } & r/\sqrt{i } \oplus { { \overline{\mathrm{gr}}\,}}r/{\mathrm{ann}}(b ) & \qquad \longleftrightarrow \qquad & x_{red } \cup \overline{{\overline c}_x y\setminus x } & { \mathop{\twoheadrightarrow } } & { \overline c}_x y \\
{ { \overline{\mathrm{gr}}\,}}r & { \operatorname*{\hookrightarrow } } & r/\sqrt{i } \oplus \bigoplus_a a & \qquad \longleftrightarrow \qquad & x_{red } \cup \bigcup_a { { \rm proj\,}}a & { \mathop{\twoheadrightarrow } } & { \overline c}_x y \end{array}\ ] ] [ ex : vee ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ recall @xmath98 / { \langle}a^2 - ab{\rangle}$ ] from example [ ex : nons2 ] , with @xmath447 .
now @xmath448 is integral , since @xmath449 , and @xmath450/{\langle}f(f-1 ) { \rangle}{{\,\cong\,}}{{\mathbb f}}[b ] \oplus { { \mathbb f}}[b']$ ] .
geometrically , the divisor @xmath276 is a double point at the intersection of the two lines @xmath181 .
the normal cone @xmath451 is the trivial line bundle over the double point .
the balanced normal cone @xmath452 is just the two intersecting lines . whereas @xmath453 pulls apart the two lines ; it is the full normalization . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ex : oval ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath454/{\langle}c(c^2-bd ) { \rangle},$ ] so @xmath455 is the union of a line and a conic in the plane . using proposition [ prop :
compute ] , we find @xmath103 , with @xmath456 . then @xmath457 $ ] is integral , because @xmath458 .
in fact @xmath459/{\langle}c(f - d ) , bf - c^2 , f(f - d){\rangle},\ ] ] so @xmath460 is the disjoint union of the point @xmath461 and the ( doubly fat ) point @xmath462 . whereas @xmath463 is only one ( triply fat ) point .
note that @xmath444 is not the full normalization of @xmath0 , which would pull the two components apart at _ both _ ends .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ex : twoplanesagain ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ recall the ring @xmath280/{\langle}d(b - d ) , e(b - d ) , dc , ec { \rangle}$ ] from subsection [ ssec : twoplanes ] , the union of the @xmath281 plane and @xmath282 plane .
we found that the associated graded @xmath38 turns out to be isomorphic to @xmath6 , with @xmath464 .
the first few graded pieces of @xmath465 are @xmath466/{\langle}ce{\rangle}\\ { { \overline{\mathrm{gr}}\,}}r_1 & = & b ( { { \overline{\mathrm{gr}}\,}}r_0 ) \oplus { { \mathbb f}}d \\ { { \overline{\mathrm{gr}}\,}}r_2 & = & b^2 ( { { \overline{\mathrm{gr}}\,}}r_0 ) \oplus { { \mathbb f}}b d \\ { { \overline{\mathrm{gr}}\,}}r_3 & = & b^3 ( { { \overline{\mathrm{gr}}\,}}r_0 ) \oplus
{ { \mathbb f}}b^2 d \\ & \vdots & \end{aligned}\ ] ] so @xmath467 is @xmath201 , and is an isomorphism in all higher degrees .
this suggests we look at the element @xmath468 .
it is indeed integral , satisfying @xmath469 .
we know that @xmath161 should be the @xmath169-cone over @xmath470 , so should have no relations involving @xmath169 ; each will end up replaced by relations in degree @xmath34 .
in fact @xmath471/{\langle}f(f-1 ) , fc , e(f-1 ) , ec { \rangle},\ ] ] geometrically the cone over the disjoint union of the two lines @xmath472 , @xmath473 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ex : reduceddivisoragain ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ from subsection [ ssec : reduceddivisor ] , recall the ring @xmath324/ { \langle}ab , ac , c^2-b^4 d{\rangle}\ ] ] where the parenthesized subscripts indicate the degrees .
the first few graded pieces are @xmath474 \\
{ { \overline{\mathrm{gr}}\,}}r_1 & = & b ( { { \overline{\mathrm{gr}}\,}}r_0 ) \\ { { \overline{\mathrm{gr}}\,}}r_2 & = & b^2 ( { { \overline{\mathrm{gr}}\,}}r_0 ) \oplus c ( { { \overline{\mathrm{gr}}\,}}r_0 ) \\ { { \overline{\mathrm{gr}}\,}}r_3 & = & b^3 ( { { \overline{\mathrm{gr}}\,}}r_0 ) \oplus bc ( { { \overline{\mathrm{gr}}\,}}r_0 ) \\ { { \overline{\mathrm{gr}}\,}}r_4 & = & b^4 ( { { \overline{\mathrm{gr}}\,}}r_0 ) \oplus b^2 c ( { { \overline{\mathrm{gr}}\,}}r_0 ) \\ & \vdots & \end{aligned}\ ] ] in this example
@xmath169 is a zero divisor , and @xmath475 is not @xmath201 .
all later maps are @xmath201 , but only become onto at and after @xmath476 .
the relation @xmath477 says that @xmath478 is integral , as @xmath479 .
in fact @xmath480 / { \langle}a , e^2-d{\rangle}\ ] ] where we ve lost the component that lived in @xmath276 .
note that this is an example where the inclusion @xmath481 is not an isomorphism in _ all _ positive degrees @xmath10 only @xmath482 .
( that s because @xmath316 , not because @xmath169 is a zero divisor . ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ex : cubicsurface ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ this is an irreducible example with @xmath483 .
consider @xmath484/{\langle}b^2 f + bcd + c^3 { \rangle}$ ] , the homogeneous coordinate ring of a cubic surface .
it has a @xmath485 of singularities , along @xmath486 .
it is easy to see that @xmath487 , and the lower bounds @xmath488 , @xmath489 suggest the degeneration @xmath490/{\langle}bcd + c^3 { \rangle}$ ] . applying proposition [ prop : compute ] , we see that we have correctly computed @xmath20 , and @xmath38 .
the first few graded pieces of @xmath38 are @xmath491 \\ { { \overline{\mathrm{gr}}\,}}r_{\frac{1}{2 } } & = & { { \mathbb f}}[d , f]c\\ { { \overline{\mathrm{gr}}\,}}r_1 & = & { { \mathbb f}}[d , f]b \oplus { { \mathbb f}}[d , f]c^2\\ { { \overline{\mathrm{gr}}\,}}r_{1\frac{1}{2 } } & = & { { \mathbb f}}[d , f]bc\\ { { \overline{\mathrm{gr}}\,}}r_2 & = & { { \mathbb f}}[d , f]b^2 \oplus { { \mathbb f}}[d , f]bc^2\\ & \vdots & \end{aligned}\ ] ] multiplication by @xmath169 should give a `` @xmath152-fold periodicity '' on @xmath161 , suggesting we let @xmath492 fill in the hole seen in degree @xmath34 .
then @xmath493 , so @xmath186 is indeed integral over @xmath6 .
in fact @xmath494 / { \langle}y^2+dy , by - c^2{\rangle}= \left({{\mathbb f}}[d , f , y ] / { \langle}y(y+d){\rangle}\right ) \ , [ b , c]/ { \langle}by - c^2{\rangle}.\ ] ] consider the map @xmath495 after @xmath169 is killed : @xmath496/{\langle}c^3 { \rangle}\to { { \mathbb f}}[c , d , f , y ] / { \langle}y^2+dy , c^2{\rangle}\ ] ] taking @xmath497 , this is a map from a bouquet of two @xmath485s ( one with multiplicity @xmath185 ) onto a single @xmath485 ( with multiplicity @xmath146 ) , @xmath201 at the north pole intersection but otherwise @xmath326 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it will be useful later to know that the integral closure can be taken before or after @xmath58 .
[ prop : iccommutes ] let @xmath6 be a ring and @xmath498 .
let @xmath499 denote the integral closure of @xmath218 in @xmath187 $ ]
. then @xmath500 .
first , define a filtration on @xmath187 $ ] again called @xmath20 by @xmath501 . by rees formula for @xmath20 , this formula is well - defined .
( if @xmath169 is not a zero divisor , the more obvious formula @xmath502 works as well . )
plainly it restricts to @xmath20 on @xmath218 , justifying the reuse of the name . with it , we can define @xmath503)$ ] , easily seen to be isomorphic to @xmath504 $ ] .
now the @xmath58 of @xmath505 \qquad\hbox{gives}\qquad { { \overline{\mathrm{gr}}\,}}(r/{\mathrm{ann}}(b ) ) { \operatorname*{\hookrightarrow}}{{\overline{\mathrm{gr}}\,}}{{\widetilde r}}{\operatorname*{\hookrightarrow}}{{\overline{\mathrm{gr}}\,}}(r[b^{-1 } ] ) = ( { { \overline{\mathrm{gr}}\,}}r)[b^{-1}].\ ] ] our goal is to show that @xmath506 is the integral closure of @xmath507 in @xmath503)$ ] .
we first show @xmath506 is integral over @xmath38 .
let @xmath8 lie over @xmath508 .
then if @xmath509 satisfies a monic polynomial @xmath510 $ ] , its image @xmath511 satisfies @xmath512 $ ] . hence we have maps @xmath513 since the composite is an isomorphism in large degrees , so is the second inclusion .
now it is enough to know that the map @xmath514 is onto for all @xmath413 .
let @xmath515 be the image of @xmath516 , so @xmath517 .
then by rees formula and the valuative criterion for integrality , @xmath518 is integral over @xmath218 , so @xmath519 , giving a preimage of @xmath520 in @xmath521 .
let @xmath0 be a topological space , and @xmath522 a closed subset .
then we can think of @xmath0 as built from @xmath522 with @xmath523 attached .
a standard homotopical operation at this point is to study @xmath524 by collapsing @xmath525 to a point , or at least something contractible .
that is slightly too brutal for us .
first thicken @xmath522 to an open neighborhood @xmath526 such that @xmath526 retracts to @xmath522 and @xmath525 retracts to @xmath527 .
then separate @xmath525 into connected components @xmath53 , with each @xmath528 a closed retract of @xmath53 .
now let @xmath529 be @xmath0 with each @xmath530 collapsed to _ its own _ point .
in good cases , up to homotopy this means we replace @xmath531 with the cone on @xmath532 . [ ex : ovaltop ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath0 be a circle and @xmath522 a point on it
. then @xmath525 is connected , so there is only one connected component @xmath53 , and @xmath53 is already contractible .
but in the ( trivial ) passage from @xmath0 to @xmath529 we do nt replace @xmath53 by the cone on the point @xmath533 ; that would flatten the circle @xmath0 to an interval .
rather , we replace @xmath53 with the cone on @xmath534 , which is _ two _ points . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in bad cases like this example , we are still replacing @xmath53 with a cone
it s just not the cone on @xmath532 , but instead a sort of link of that inside @xmath531 . in the above example
, the link was two points . in the algebraic geometry , the passage from @xmath535 parallels the flat degeneration @xmath536 .
the decomposition of the open set @xmath537 into connected components @xmath538 corresponds to the decomposition of the fraction ring @xmath381 $ ] as a direct sum .
`` being a cone '' is replaced by having a periodic grading .
the most subtle point in the above topological picture is the fact that we do nt replace each @xmath531 with a cone on @xmath532 , but on something that _ maps _ to @xmath532 . in the algebraic geometry , this reflects the fact that the inclusion @xmath539 may not be an isomorphism . in this way , perhaps one should think of the map @xmath539 as sort of an attaching map when building a complex . in example [
ex : cubicsurface ] above , the attaching map is the one from the bouquet of two @xmath485s to a single @xmath485 .
since we have to deal seriously with chow classes in this section , we list some simple properties we will need of them : 1 .
if @xmath540 is proper , there is an induced map @xmath541 of their chow groups , and these maps are functorial .
any scheme @xmath53 has a `` fundamental class '' @xmath542 \in a_{\dim w}(w)$ ] .
consequently , a proper map @xmath543 induces a class @xmath544 : = \pi_*([w ] ) \in a_{\dim w}(v)$ ] .
( for some authors , the fundamental class of a nonequidimensional scheme is inhomogeneous , and ours is merely the component in top degree . )
[ chow : decomposition ] if @xmath545 where each @xmath546 , then @xmath542 = \sum_{i:\ \dim w_i = \dim w } [ w_i { \operatorname*{\hookrightarrow}}w]$ ] .
4 . the inclusion @xmath547 of the reduction of @xmath53 induces a map @xmath548 which is an _
isomorphism_. consequently , we can pull back @xmath542 $ ] to a class on @xmath133 . 5 . [ chow : flatfamily ] if @xmath549 is a flat family of schemes each proper over @xmath54 , then @xmath550 $ ] is constant in @xmath341 .
[ chow : normalization ] if @xmath551 is the normalization of a reduced scheme @xmath54 , then @xmath552 = [ v]$ ]
[ chow : blowup ] if @xmath553 is the blowup of @xmath54 along a subscheme @xmath53 containing no component of @xmath54 , then @xmath552 = [ v]$ ] . 8 .
[ chow : factor ] let a composite @xmath554 of proper maps take the fundamental class of @xmath53 to that of @xmath555 , i.e. @xmath556 = [ z]$ ]
. then @xmath557 = [ v]$ ] and @xmath558 = [ z]$ ] .
one way these interrelate is the following .
consider the flat family of schemes over @xmath54 in which @xmath54 degenerates to @xmath55 .
if @xmath559 , then the map @xmath560 is proper , hence ( by property [ chow : flatfamily ] ) induces the fundamental class on @xmath54 .
but @xmath560 factors as @xmath561 , giving us another way to see the class on @xmath562 induced from the thickening @xmath54 .
we ll justify these reductions in propositions [ prop : nonzerodivisor ] , [ prop : blowup ] , and [ prop : normalalongb ] .
then we ll develop the tools to address what for us is the fundamental case , that @xmath53 is defined in @xmath54 by the vanishing of a nonzero divisor @xmath169 , and @xmath54 is normal along @xmath53 .
we ll call a map @xmath563 * * volumetric if it takes the fundamental class to the fundamental class .
( there does not seem to be a standard term for this , and we do nt seriously propose this as the right name for this concept beyond its frequent use in this section . )
theorem [ thm : betageom ] is thus the statement that @xmath125 is volumetric .
[ prop : nonzerodivisor ] let @xmath54 be a reduced scheme and @xmath53 a closed subscheme .
let @xmath564 and @xmath565 .
then theorem [ thm : betageom ] holds for the pair @xmath566 if it holds for the pair @xmath567 .
it is easy to check that the @xmath7-adic filtration on @xmath6 is restricted from the @xmath7-adic filtration on @xmath571 along the obvious map ( which is an inclusion since @xmath6 has no nilpotents ) .
consequently , the map @xmath572 is an inclusion , and an isomorphism in positive degrees .
geometrically , the map @xmath573 is onto , finite - to - one , and @xmath201 away from @xmath574 .
consequently it is volumetric ( which uses @xmath568 ) .
[ prop : blowup ] let @xmath54 be a reduced scheme with @xmath575 a closed subscheme containing no components .
let @xmath576 denote the blowup of @xmath54 along @xmath53 and @xmath577 denote the exceptional divisor .
then theorem [ thm : betageom ] holds for the pair @xmath566 if it holds for the pair @xmath578 .
consider the commuting diagram @xmath579 here @xmath580 denotes the blowup along @xmath53 ; the isomorphism @xmath581 then follows directly from the definitions .
the vertical map on the right is volumetric by property [ chow : blowup ] .
once we prove that the vertical map on the left is volumetric , we re done . to prove that @xmath582 is volumetric , we show that it sits intermediate to a blowup @xmath583 , and apply properties [ chow : blowup ] and [ chow : factor ] .
if @xmath148 and @xmath53 is defined by @xmath7 , let @xmath64 be the ideal in @xmath38 given by @xmath584 , a sort of @xmath58-analogue of the augmentation ideal .
then @xmath585 let @xmath65 denote the graded ring @xmath586 , and @xmath587 the ideal generated by the @xmath7 in the @xmath34th graded piece .
so @xmath588 , @xmath589 .
our goal is to show that @xmath590 , the coordinate ring of @xmath591 , includes naturally into the coordinate ring of @xmath592 . in degree
@xmath34 they are both @xmath38 , and using @xmath593 we can see @xmath594 , completing the proof .
[ prop : normalalongb ] let @xmath148 be a reduced scheme and @xmath498 a nonzero divisor , with @xmath53 the subscheme defined by @xmath276 .
let @xmath499 denote the integral closure of @xmath6 in @xmath187 $ ] , and @xmath595 , and @xmath596 the subscheme defined by @xmath276 .
then theorem [ thm : betageom ] holds for the pair @xmath566 if it holds for the pair @xmath597 .
since @xmath169 is not a zero divisor , the map @xmath598 is volumetric .
( this uses properties [ chow : normalization ] and [ chow : factor ] , since @xmath599 is a partial normalization of @xmath56 . ) by proposition [ prop : iccommutes ] , @xmath600 .
thus we have a commutative diagram @xmath601 it remains to argue that @xmath602 is volumetric .
since @xmath169 is not a zero divisor , @xmath53 and @xmath596 contain no components of the normal cones .
hence we can test the lengths of all components by passing to the projectivization @xmath603 .
but since @xmath596 and @xmath53 are cartier , this is just the map @xmath604 which is volumetric by properties [ chow : normalization ] and [ chow : factor ] .
consider the induced morphism @xmath607 .
the inclusion @xmath608 induces a class by pulling back @xmath542 $ ] along @xmath609 . the projection @xmath610 induces a class by mapping @xmath542 $ ] forward using @xmath611 . by functoriality , these two maps @xmath612 are the same .
[ lem : goodgens ] let @xmath6 be a finitely generated commutative algebra over a field @xmath158 .
let @xmath613 be a list of elements whose images generate the quotient @xmath614 as an @xmath158-algebra .
then there exists a list @xmath615 of nilpotents in @xmath6 such that the concatenation @xmath616 generate @xmath6 as an @xmath158-algebra .
let @xmath7 be the nilpotent radical of @xmath6 , and consider the associated graded @xmath32 .
this is again noetherian , so its augmentation ideal is finitely generated , and by homogeneous elements ; pick generators @xmath617 the images of some elements @xmath615 from @xmath6 .
the monomials in these generators have a maximum possible degree , so @xmath618 for @xmath82 large enough . knowing @xmath619 generate @xmath32 , we now claim that @xmath616 generate @xmath6 .
let @xmath72 be an element to generate .
then its image @xmath620 can be written as @xmath621 for some @xmath158-polynomial @xmath172 , so @xmath622 , where @xmath19 as before measures the depth in the @xmath7-adic filtration .
so to generate @xmath72 , it is enough to generate @xmath623 , which is deeper in the @xmath7-adic filtration .
since @xmath618 for @xmath82 large enough , this algorithm terminates .
the next lemma is sort of a dual to property [ chow : flatfamily ] : in it the base ring is a quotient of each element of the family , rather than a subring .
it seems unlikely that the quasiprojectivity asked of @xmath53 is necessary , and we hope that someone more fluent with chow classes can sidestep it .
[ lem : thickenings ] let @xmath624 be a quasiprojective flat family over a field @xmath158 , whose reduction @xmath625 is a trivial family @xmath626 .
then each thickening @xmath627 of @xmath53 induces the same chow class on @xmath53 .
these lengths do nt change if we replace @xmath54 by the quasiaffine cone over it , so we can assume @xmath54 quasiaffine .
nor do they change if we algebraically close the base field @xmath158 .
doing so will allow us to test the lengths at general - enough points , rather than the generic points . by imposing @xmath631 general affine - linear conditions on the variables @xmath632 , we can @xmath53 down to a set @xmath633 of reduced points , with at least one general point in each top - dimensional component @xmath96 of @xmath53 .
( we use @xmath158 infinite to guarantee the existence of general - enough linear conditions . ) these same conditions cut @xmath54 down to a new flat family @xmath634 . for each top - dimensional component @xmath96 of @xmath53
, pick a point @xmath635 in the smooth locus of @xmath96 .
shrink @xmath634 to the subscheme supported on @xmath636 . as an @xmath637$]-module ,
the ring of functions on @xmath634 is finitely generated , and torsion - free , hence free .
therefore @xmath638 , the coefficient of @xmath639 $ ] in @xmath640 $ ] , is constant in @xmath341 .
the first claims of the theorem that @xmath59 is proper with finite fibers are implied by the claim that @xmath38 is a finite module over the image @xmath641}$ ] of @xmath32 in @xmath38 . to see that this module is finitely generated ,
let @xmath642 be homogeneous generators of @xmath38 over @xmath226 , which we can take to all be in positive degree ( using lemma [ lem : fingen ] ) . by proposition [ prop : multbyb ]
, after some degree @xmath109 the multiplication map @xmath422 is onto . only finitely
many monomials in the @xmath643 have degree @xmath644 , and these monomials generate @xmath38 as an @xmath641}$]-module .
next , we use propositions [ prop : nonzerodivisor ] , [ prop : blowup ] , and [ prop : normalalongb ] to reduce to the case that @xmath100 is a principal ideal , @xmath169 is not a zero divisor , and @xmath6 is integrally closed in @xmath187 $ ] .
three effects of the integrality are that now consider the following diagrams : @xmath649 \\
\iota \uparrow & & \uparrow & & \downarrow \\ { r/\sqrt{i}\,[b ] } & = & { r/\sqrt{i}\,[b ] } & = & { r/\sqrt{i}\,[b]}\end{array } \qquad\qquad \begin{array}{cccc } { { \overline{\mathrm{gr}}\,}}(r / i ) & \dashleftarrow & r / i \\ \downarrow & & \downarrow \\
r/\sqrt{i } & = & r/\sqrt{i } \end{array}\ ] ] we first define these rings , and the vertical maps .
the dashed arrows mean `` has a flat degeneration to '' , and point from general fiber to special fiber .
we ve already defined @xmath38 , and noted the inclusion @xmath650 as the degree @xmath34 part .
extend this map to @xmath651}\to { { \overline{\mathrm{gr}}\,}}r$ ] by taking @xmath169 to the evident element of @xmath652 .
( in fact , this map is an inclusion , with image the integer - graded part of @xmath38 . )
now filter @xmath38 by the @xmath653-adic filtration , and take the associated graded ; call this @xmath654 .
this operation is trivial on the image of @xmath641}$ ] , so there is still a natural map @xmath641}\to { \mathrm{gr}\,}{{\overline{\mathrm{gr}}\,}}r$ ] .
we must understand this ring @xmath654 .
each homogeneous element of @xmath38 is of the form @xmath655 with @xmath656 .
then the product in @xmath654 is @xmath657 if @xmath658 , and @xmath34 otherwise .
put another way , @xmath659 $ ] .
coming from the other end , the @xmath20-filtration on @xmath6 induces one on @xmath15 , whose associated graded we call @xmath660 . since @xmath646 , this ring can be identified with @xmath661 . with this
, we can further identify the rightmost vertical map in the left diagram with the leftmost in the right diagram , up to adjoining @xmath169 .
@xmath662 & { { \,\cong\ , } } & { { \overline{\mathrm{gr}}\,}}(r / i)[b ] & \dashleftarrow & r / i[b]\quad \\ \iota \uparrow & & \uparrow & & \downarrow & & \downarrow & & \downarrow \pi \\ { r/\sqrt{i}\,[b ] } & = & { r/\sqrt{i}\,[b ] } & = & { r/\sqrt{i}\,[b ] } & = & r/\sqrt{i}[b ] & = & r/\sqrt{i}[b ] \end{array}\ ] ] each of these vertical maps induces a chow class on @xmath641}$ ] , the reduction of @xmath32 . applying property [ chow : flatfamily ] , lemma [ lem : backnforth ] , and lemma [ lem : thickenings ] to squares 1 , 2 , and 4
, we see that the four classes are the same .
in particular @xmath663 ) = ( \pi_*)^{-1}([{{\rm spec}\,}{\mathrm{gr}\,}r ] ) \in a({{\rm spec}\,}{r/\sqrt{i}\,[b]})$ ] .
w. fulton , r. macpherson , intersecting cycles on an algebraic variety , real and complex singularities ( proc .
ninth nordic summer school / navf sympos .
oslo , 1976 ) , pp .
179197 , sijthoff and noordhoff , alphen aan den rijn , 1977 . | let @xmath0 be a subscheme of a reduced scheme @xmath1
. then @xmath1 has a flat _ degeneration to the normal cone @xmath2 of @xmath0 _ , and this degeneration plays a key step in fulton and macpherson s `` basic construction '' in intersection theory .
the intersection product has a canonical refinement as a sum over the components of @xmath2 , for @xmath0 and @xmath1 depending on the given intersection problem .
the cone @xmath2 is usually not reduced , which leads to the appearance of multiplicities in intersection formulae .
we describe a variant of this degeneration , due essentially to samuel , rees , and nagata , in which @xmath1 flatly degenerates to the `` balanced '' normal cone @xmath3 .
this space is reduced , and has a natural map onto the reduction @xmath4 of @xmath2 .
the multiplicity of a component now appears as the degree of this map .
hence intersection theory can be studied using only reduced schemes . moreover , since the map @xmath5 may wrap multiple components of @xmath3 around one component of @xmath2 , writing the intersection product as a sum over the components of @xmath3 gives a further canonical refinement . in the case that @xmath0 is a cartier divisor in a projective scheme @xmath1 , we describe the balanced normal cone in homotopy - theoretic terms , and prove a useful upper bound on the hilbert function of @xmath3 . |
we present some new construction of a generalized gaussian process related to coxeter groups of type b. our construction generalizes boejko and speicher s @xmath6-gaussian process @xcite on the @xmath6-deformed fock space @xmath9 where @xmath10 denotes the vacuum vector and @xmath11 the complexification of some real separable hilbert space @xmath12 on which the creation and its adjoint i.e. annihilation operators satisfies the @xmath6-commutation relation : @xmath13 where @xmath14 .
the inner product on @xmath15 , called the @xmath6-deformed inner product , is the sesquilinear extension of @xmath16 where @xmath17 is the set of all the permutations of @xmath18 and @xmath19 is the number of inversions of @xmath20 .
the study of @xmath6-gaussian distributions has been an active field of research during the last decade .
a noncommutative analog of a brownian motion ( or gaussian process , more generally ) is the family of operators , @xmath21 .
when equipped with the vacuum expectation state @xmath22 the @xmath6-gaussian algebra yields a rich non - commutative probability space . for @xmath23 ( corresponding to the bose statistics ) the @xmath6-deformed operator
@xmath24 is a natural deformation of the standard gaussian random variable i.e. its spectral measure relative to the vacuum state satisfies @xmath25 when @xmath26 .
moreover @xmath27 are commutative in the classical sense .
the case @xmath28 corresponds to the fermi statistics .
it should be stressed that , for @xmath29 , the @xmath6-modification of the ( anti ) symmetrization operator is a strictly positive operator .
therefore , unlike the classical bose and fermi cases , there are no commutation relations between the creation operators . for @xmath30 ,
the @xmath6-fock space recovers the full fock space of voiculescu s free probability @xcite .
for @xmath30 the @xmath6-gaussian random variables are distributed according to the semi - circle law @xmath31 when @xmath26 .
the study of the noncommutative brownian motion @xmath32 was initiated in @xcite . for further generalizations of a noncommutative brownian motion , see @xcite .
in particular , this setting gives rise to @xmath6-deformed versions of the stochastic integrals @xcite , with recent extensions to fourth moment convergence theorem @xcite .
it is worthwhile to mention the work of bryc @xcite , where the laha - lukacs property for @xmath6-gaussian processes was shown .
bryc proved that classical processes corresponding to operators which satisfy the @xmath6-commutation relations have linear regressions and quadratic conditional variances .
after @xcite , a series of papers @xcite appeared , which studied discrete generalizations of the @xmath6-commutation relations .
blitvi @xcite introduced a second - parameter refinement of the @xmath6-fock space , formulated as a @xmath33-fock space @xmath34 .
it is constructed via a direct generalization of boejko and speicher s framework @xcite , yielding the @xmath6-fock space when @xmath35 .
the corresponding creation and annihilation operators now satisfy the commutation relation @xmath36 where @xmath4 is the number operator with respect to the grading on @xmath34 .
these are the defining relations of the chakrabarti - jagannathan deformed quantum oscillator algebra , see @xcite and references therein for more details .
the moments of the deformed gaussian process @xmath37 are encoded by the joint statistics of crossings and nestings in pair partitions . in particular
, it is shown that the distribution of a single gaussian operator is orthogonalized by the @xmath38-hermite polynomials .
another generalization of the ccr and car was proposed in 2013 by bryc and ejsmont @xcite .
they define a pair of non - commutative processes on a perturbed fock space with three parameters .
both processes have the same univariate distributions , and satisfy a weak form of the polynomial martingale property .
the processes give two non - equivalent fock - space realizations of the same classical markov process : the two - parameter bi - poisson processes introduced in @xcite , and constructed in @xcite .
the goal of this paper is to introduce an @xmath0-gaussian process on an @xmath0-fock space .
our strategy is to replace the coxeter group of type a i.e. the permutation group appearing in the sum by the coxeter group of type b. the coxeter group of type b can be written as @xmath39 and hence contains the permutation group as a subgroup .
the parameter @xmath40 corresponds to the @xmath41 part and the parameter @xmath6 corresponds to @xmath17 part .
thus our @xmath0-fock space gives the @xmath6-fock space when @xmath7 , and turns out to give the @xmath42-free probability space @xcite when @xmath8 and @xmath43 ( where @xmath44 .
it would be worth mentioning that free probabilistic considerations of type b first appeared in a paper by biane , goodman and nica in @xcite .
recently , connections between type b and infinitesimal free probability were put into evidence in @xcite .
the commutation relation satisfied by the creation and annihilation operators on the @xmath0-fock space reads @xmath45 where @xmath4 is the number operator .
the orthogonal polynomials arising in the present framework are called @xmath6-meixner - pollaczek polynomials satisfying the recurrence relation @xmath46_q(1 + \alpha q^{n-1})p_{n-1}^{(\alpha , q)}(t ) , \qquad n=0,1,2,\dots\ ] ] where @xmath47 . when @xmath7 then we get @xmath6-hermite orthogonal polynomials and when @xmath8 then we get the orthogonal polynomials associated to a symmetric free meixner distribution .
the moments of the gaussian process are given by @xmath48 where @xmath49 is the set of pair partitions , @xmath50 denotes the number of the crossings of @xmath51 and @xmath52 is the number of blocks of @xmath51 which covers @xmath53 .
this formula recovers the @xmath6 case in @xcite when @xmath7 .
we also give an alternative expression of in terms of pair partitions of type b. the plan of the paper is following : first we present definitions and remarks on the @xmath0-fock space and creation , annihilation operators .
next , natural properties of creation and annihilation operators , including norm estimates and the commutation relation , are presented in the section 2.2 .
the generalized gaussian process of type b , @xmath54 , is studied in the section 3 .
the main theorem is placed in the section 3 .
we show the recurrence relation , give a direct wick formula for @xmath55 and then give the mixed moment formula .
let @xmath56 be the set of bijections @xmath57 of the @xmath58 points @xmath59 such that @xmath60 . equipped with the composition operation as a product
, @xmath56 becomes a group and is called a _ coxeter group of type b _ or a _
hyperoctahedral group_. the coxeter group @xmath56 is generated by @xmath61 , @xmath62 .
these generators satisfy the generalized braid relations @xmath63 , @xmath64 , @xmath65 if @xmath66 .
note that @xmath67 generates the symmetric group @xmath17 .
we express @xmath68 in an irreducible form @xmath69 i.e. , in a form with minimal length , and in this case let @xmath70 these definitions do not depend on the way we express @xmath57 in an irreducible form , and hence @xmath71 and @xmath72 are well defined .
let @xmath73 be a separable real hilbert space and let @xmath74 be its complexification with inner product @xmath75 linear on the right component and anti - linear on the left .
when considering elements in @xmath73 , it holds true that @xmath76 . in order to define an action of @xmath77
, we assume that there exists a self - adjoint involution @xmath78 for @xmath79 .
for example we may take the identity as an involution , or if @xmath74 is spanned by an orthonormal basis @xmath80 ( or @xmath81 ) , then we may define the involution @xmath82 given a self - adjoint involution on @xmath74 we define an action of @xmath56 on @xmath83 by @xmath84 let @xmath85 be the ( algebraic ) full fock space over @xmath11 @xmath86 with convention that @xmath87 is a one - dimensional space normed space along a unit vector @xmath10 . note that elements of @xmath85 are finite linear combinations of the elements from @xmath88 and we do not take the completion .
we equip @xmath85 with the inner product @xmath89 we will deform the inner product on @xmath85 . for @xmath90 $ ]
we define the type b symmetrization operator on @xmath83 , @xmath91 note that with convention @xmath92 , we have @xmath93 and note also that @xmath94 is the @xmath6-symmetrization operator @xcite .
moreover let @xmath95 be the type b symmetrization operator acting on the algebraic full fock space . from boejko and speicher (
* theorem 2.1 ) , the operator @xmath96 and hence @xmath97 is positive .
if @xmath98 then @xmath96 is a strictly positive operator meaning that it is positive and @xmath99 .
we deform the inner product by using the type b symmetrization operator : @xmath100 which is a semi - inner product from the positivity of @xmath97 for @xmath101 $ ] .
we restrict the parameters to the case @xmath102 so that the deformed semi - inner product is an inner product .
for @xmath79 the free right creation and free annihilation operators @xmath103 on @xmath85 are defined by @xmath104 it then holds that @xmath105 and @xmath106 is linear , but @xmath107 is anti - linear . in the literature
the left creation and annihilation operators are commonly used , but we consider the right operators . the right creation and
annihilation operators have the advantage that the embedding @xmath108 looks more natural ( see section [ sec3 ] ) and the commutation relation @xmath109 holds true on @xmath83 for any @xmath79 .
if we consider the right action of coxeter groups then the left creation and annihilation operators are more natural .
note that recently the combination of right and left creation and annihilation operators is of interest in a different context of free probability @xcite . for @xmath110 ,
the algebraic full fock space @xmath85 equipped with the inner product @xmath111 is called the _ fock space of type b _ or the _
let @xmath112 and @xmath113 be its adjoint with respect to the inner product @xmath111 .
the operators @xmath114 and @xmath113 are called _ creation and annihilation operators of type b _ or _
@xmath0-creation and annihilation operators _ , respectively .
it is easy to see that @xmath115 is linear and @xmath116 is anti - linear .
since @xmath117 is the @xmath6-symmetrization , our @xmath0-fock space is the @xmath6-fock space when @xmath7 .
thus our type b setting actually generalizes the type a setting .
there is a natural embedding @xmath118 .
corresponding to the quotient @xmath119 , we decompose the operator @xmath120 .
[ prop1 ] we have the decomposition @xmath121 where @xmath122 it is known that there exist unique right coset representatives for @xmath119 with minimal lengths ( * ? ? ?
* section 1.10 , proposition ) .
stumbo @xcite showed that these coset representatives are given by @xmath123 where @xmath124 is the substring composed by the first @xmath125 elements of @xmath126 .
hence any element @xmath68 has the representation @xmath127 for some unique @xmath128 and @xmath129 and moreover , this representation preserves the irreducibility ( see ( * ? ? ?
* section 1.10 , proposition ) ) and hence @xmath130 for @xmath131 .
since @xmath132 has the representation @xmath133 we have the identity .
the operator @xmath132 plays a central role in this paper .
firstly we can compute the annihilation operator in terms of @xmath132 .
[ prop2 ] for @xmath134 , we have @xmath135 let @xmath136 .
then @xmath137 recall that @xmath138 for @xmath139 and so we get @xmath140 [ thm1 ] let @xmath4 be the number operator , i.e. @xmath141 for @xmath142 .
then @xmath143 where @xmath144 note that @xmath145 is the free right annihilation operator , @xmath146 is the free left annihilation operator and @xmath147 . from propositions [ prop1],[prop2 ] we have @xmath148 where @xmath149 after some computations , we get @xmath150 and @xmath151 which is equal to @xmath152 from the self - adjointness of the map @xmath153 . [ commutation ] for @xmath154
we have the commutation relation @xmath155 this commutation relation is quite similar to @xmath156 which appeared in @xcite . from theorem [ thm1 ]
, it holds that @xmath157 and the conclusion follows .
we will study the norm of the creation operators of type b. let @xmath158_q$ ] be the @xmath6-number @xmath159_q:= 1+q+\cdots+q^{n-1},\qquad n \geq1\ ] ] and let @xmath158_q!$ ] be the @xmath6-factorial @xmath159_q ! : = [ 1]_q \cdots [ n]_q,\qquad n \geq1.\ ] ] let @xmath160 be the @xmath6-pochhammer symbol @xmath161 [ lem4 ] for @xmath162 , we get @xmath163_q !
( -\alpha\langle x,\bar{x}\rangle/\|x\|^2 ; q)_n \|x\|^{2 n } , \qquad n \geq 1 , \label{eq0001}\\ & \|r_{\alpha , q}^{(n)}\|_{0,0 } \leq ( 1+|\alpha| |q|^{n-1})[n]_q,\qquad n \geq 1 .
\end{aligned}\ ] ] \(1 ) let @xmath164 for @xmath165 with convention @xmath166 .
then @xmath167 and @xmath168_q \langle x^{\otimes n } , ( p^{(n-1)}_{\alpha , q}\otimes i ) x^{\otimes n}\rangle_{0,0 } + \alpha q^{n-1}[n]_q \langle x^{\otimes n } ,
( p^{(n-1)}_{\alpha , q}\otimes i)(x^{\otimes ( n-1 ) } \otimes \bar{x})\rangle_{0,0}\\ & = [ n]_q ( \|x\|^2 + \alpha\langle x,\bar{x}\rangle\ , q^{n-1 } ) c_{n-1 } , \qquad n \geq 1 , \end{aligned}\ ] ] and so the conclusion follows by induction .
\(2 ) since @xmath169 for all @xmath170 , the conclusion follows easily by definition of @xmath132 .
suppose that @xmath171 . 1
. [ item1 ] if @xmath172 and @xmath173 , then @xmath174 2 .
[ item2 ] if @xmath172 and @xmath175 , then @xmath176 3 .
[ item3 ] if @xmath177 , then @xmath178 4 .
[ item4 ] if @xmath179 , then @xmath180 5 .
[ item5 ] otherwise , @xmath181 when @xmath182 then reminds us a _ lie ball _
* example 3.10 ) . from proposition [ commutation ] , for @xmath183 and @xmath165 , we get @xmath184 if @xmath185 , then @xmath186 .
the equality is achieved when @xmath187 , and hence holds .
upper bound .
if @xmath188 , then from one may obtain the inequality @xmath189 .
, , lower bound .
it follows from that @xmath190_q!(\|x\|^2 + \alpha \langle x,\bar{x}\rangle\ , q^{n-1 } ) \|x^{\otimes ( n-1)}\|_{\alpha , q}^2 . \end{split}\ ] ] by letting @xmath191 , the lower bound follows .
, upper bound .
the proof follows the line of ( * ? ? ?
* lemma 4 ) .
we have @xmath192 by taking the square root of operators using lemma [ lem4 ] one gets @xmath193_q
( p_{\alpha , q}^{(n-1)}\otimes i ) \\ & \leq \frac{1+|\alpha|}{1-q}p_{\alpha , q}^{(n-1)}\otimes i \end{split}\ ] ] on @xmath83 regarding the inner product @xmath194 .
therefore we have for @xmath195 that @xmath196 and hence the inequality @xmath197 holds . upper bound .
we try to refine the estimate by carefully looking at @xmath198_q $ ] as a function of @xmath199 .
let @xmath200 where @xmath201 and @xmath202 .
then @xmath203 , @xmath204 , and hence for @xmath205 one obtains @xmath206 and hence @xmath207_q = f_n(q)/(1-q ) \leq 1/(1-q)$ ] . following the argument
, we have the upper bound @xmath208 the lower bound was already obtained and so [ item3 ] follows . lower bound .
we will carefully study the coefficient @xmath158_q!(\|x\|^2 + \alpha \langle x,\bar{x}\rangle\ , q^{n-1 } ) = f_n(q ) \|x\|^2/(1-q)$ ] in , now with @xmath209 .
then @xmath210 and hence @xmath211 is decreasing . moreover
@xmath212 and @xmath213 .
hence @xmath211 has a unique zero @xmath214 in @xmath215 .
in particular @xmath216 as @xmath217 . therefore
, for fixed @xmath218 , the zero @xmath214 is larger than @xmath6 for large @xmath219 and hence @xmath220 is strictly positive .
this implies that @xmath221
the operator @xmath222 on @xmath85 is called the _ gaussian operator of type b _ or _
@xmath0-gaussian operator_. the family @xmath223 is called the gaussian process of type b or @xmath0-gaussian process . for a probability measure @xmath224 with finite moments of all orders ,
let us orthogonalize the sequence @xmath225 in the hilbert space @xmath226 , following the gram - schmidt method .
this procedure yields orthogonal polynomials @xmath227 with @xmath228 .
multiplying by constants , we take @xmath229 to be monic , i.e. , the coefficient of @xmath230 is 1 .
it is known that they satisfy a recurrence relation @xmath231 with the convention that @xmath232 .
the coefficients @xmath233 and @xmath234 are called _ jacobi parameters _ and they satisfy @xmath235 and @xmath236 . it is known that @xmath237 moreover , the measure @xmath224 has a finite support of cardinality @xmath4 if and only if @xmath238 and @xmath239 for @xmath240 .
the continued fraction representation of the cauchy transform can be expressed in terms of the jacobi parameters : @xmath241 this representation is useful to calculate the cauchy transform when jacobi parameters are given .
more details are found in @xcite .
for @xmath242 let @xmath243 be the orthogonal polynomials with the recursion relation @xmath244_q(1 + \alpha q^{n-1})p_{n-1}^{(\alpha , q)}(t ) , \qquad n=0,1,2,\dots\ ] ] where @xmath245 .
these polynomials are called _ @xmath6-meixner - pollaczek polynomials_. the orthogonalizing probability measure @xmath246 is known in ( * ? ? ?
* ( 14.9.4 ) ) , supported on @xmath247 and absolutely continuous with respect to the lebesgue measure with density @xmath248 where @xmath249 for @xmath250 the recursion coincides with ( * ? ? ?
* ( 14.9.4 ) ) by considering the dilation @xmath251 with @xmath252 .
hence @xmath253 are orthogonal polynomials associated to @xmath246 for @xmath250 ( note that ( 14.9.2 ) contains an error ; the integral should be performed over the interval @xmath254 $ ] , not over @xmath255 $ ] ) .
the case @xmath256 ( or @xmath257 ) is not covered in ( * ? ? ? * ( 14.9.2 ) ) , but we can resort to the analytic continuation .
let @xmath258 be the moments of @xmath246 and let @xmath259 .
we know that @xmath260 the moments @xmath261 are polynomials of the jacobi parameters @xmath262 ( see accardi - boejko s formula ( * ? ? ?
* corollary 5.1 ) ) and so the moments extend to entire analytic functions of @xmath263 . on the other hand , @xmath264 is analytic in @xmath265 and does not have a zero in an open neighborhood @xmath266 of @xmath267 \times ( -1,1)$ ] , and then the function @xmath268 and hence ( by dominated convergence ) the integral @xmath269 has analytic continuation to the same region @xmath270 . by analytic continuation , formula holds for @xmath271 , and hence the measure @xmath246 is a probability measure giving the moments @xmath258 for the case @xmath256 too .
let @xmath276_q(1 + \alpha \langle x , \bar{x}\rangle\ , q^{n-1})$ ] . from lemma [ lem4 ]
we get @xmath277_q!(-\alpha\langle x,\bar{x}\rangle;q)_n=\gamma_0\gamma_1\cdots \gamma_{n-1}$ ] and hence from it follows that @xmath278 therefore the map @xmath279 defined by @xmath280 is an isometry .
note that @xmath281_q x^{\otimes ( n-1 ) } + \alpha q^{n-1 } [ n]_q \langle x,\bar{x}\rangle\,x^{\otimes ( n-1)}\\ & = x^{\otimes ( n+1)}+[n]_q ( 1 + \alpha\langle x,\bar{x}\rangle q^{n-1})x^{\otimes ( n-1 ) } , \end{aligned}\ ] ] where proposition [ prop2 ] was used on the second line .
hence inductively we can compute @xmath282 and show that @xmath283 . since @xmath284 is an isometry we get @xmath285 for even integers @xmath219 . for odd integers
@xmath219 we can show that @xmath286 . since @xmath275 is compactly supported , probability measures giving the moment sequence @xmath287 are unique and hence @xmath288 . 1 .
the measure @xmath289 is the normal law @xmath290 .
the orthogonal polynomials @xmath291 are hermite polynomials .
the measure @xmath292 is the standard wigner s semicircle law @xmath293 .
the orthogonal polynomials @xmath294 are chebyshev polynomials of the second kind .
the measure @xmath295 is the @xmath6-gaussian law and the orthogonal polynomials @xmath296 are called @xmath6-hermite polynomials .
the measure @xmath297 is the bernoulli law @xmath298 .
the measure @xmath299 is a symmetric free meixner law @xcite .
the measure @xmath300 is the delta measure @xmath301 and so is trivial , but we can find a nontrivial scaling . if we take @xmath302 for @xmath303 then becomes @xmath304_q [ n+2 \gamma-1]_q q_{n-1}^{(\gamma , q)}(t),\ ] ] where @xmath305 in the limit @xmath306 the polynomial @xmath307 tends to a polynomial which we denote by @xmath308 and we get the recursion @xmath309 which gives meixner - pollaczek polynomials corresponding to the symmetric meixner distribution @xmath310 see @xcite for further information on meixner - pollaczek polynomials .
free infinite divisibility of probability measures has been studied by several authors .
of particular interests were probability measures arising from generalized brownian motions .
the moments of typical generalized brownian motions have some combinatorial interpretations in terms of set partitions @xcite .
our probability measure @xmath246 has a combinatorial interpretation too ( see theorem [ thm2 ] ) and it extends several known measures .
the @xmath6-gaussian distribution @xmath295 is freely infinitely divisible if and only if @xmath311 $ ] ; see anshelevich et al .
note that the case @xmath312 corresponds to the classical gaussian distribution and its free infinite divisibility was first proved by belinschi et al .
the free meixner distribution @xmath299 is freely infinitely divisible if and only if @xmath313 $ ] ; see saitoh and yoshida @xcite .
the measure @xmath314 with suitable scaling converges to the classical meixner distribution as @xmath315 and the limit distribution is freely infinitely divisible for @xmath316 ( the case @xmath317 is still open ) ; see @xcite .
moreover , numerical calculations of free cumulants ( thanks to franz lehner ) support the following conjecture .
a pair @xmath321 is called a _ set partition of @xmath158 $ ] of type b _
if @xmath51 is a set partition of @xmath158 $ ] and @xmath322 is a coloring of the blocks of @xmath51 .
we denote by @xmath323 the set of all set partitions of @xmath158 $ ] of type b. we denote by @xmath324 the set of set partitions of @xmath158 $ ] of type b such that given @xmath328 , let @xmath329 be the set of partitions @xmath330 such that when @xmath51 is written as @xmath331,\ ] ] then @xmath332 and @xmath333 for all @xmath334 and @xmath335 for all @xmath336 .
we also let @xmath337 .
let @xmath338 be the subset of @xmath324 defined by @xmath339 .
let @xmath340 .
we introduce some partition statistics .
let @xmath341 be the set of pair blocks of a set partition @xmath51 and @xmath342 be the set of singletons of a set partition @xmath51 .
let @xmath343 be the set of negative blocks ( i.e. blocks colored by @xmath344 ) of a set partition @xmath321 of type b. for a set partition @xmath51 let @xmath50 be the number of crossings of @xmath51 , i.e. @xmath345 for two blocks @xmath346 of a set partition , we say that @xmath347 _ covers _
@xmath319 if there are @xmath348 such that @xmath349 for any @xmath350 . for @xmath351 , let @xmath352 be the number of pairs of a singleton and a covering block : @xmath353 let @xmath354 be the number of pairs of a negative block and a covering block : @xmath355 let @xmath356 be the number of pairs of a negative block and a _ singleton to the left _ : @xmath357 for a block @xmath319 of a set partition @xmath51 , let @xmath358 be the number of blocks of @xmath51 which cover @xmath319 and let @xmath359 be the number of singletons which are placed to the left of @xmath319 .
more precisely @xmath360 should be denoted as @xmath361 , but we use the simpler notations
. a set partition @xmath51 is _ noncrossing _ if @xmath362 . the set of noncrossing partitions of @xmath158
$ ] is denoted by @xmath363 and the set of noncrossing pair partitions is denoted by @xmath364 when @xmath219 is even .
a block of a noncrossing partition is _ inner _ if it is covered by another block .
a block of a noncrossing partition is _
outer _ if it is not inner . for @xmath365 ,
let @xmath366 be the number of inner blocks of @xmath51 and let @xmath367 be the number of outer blocks .
note that @xmath368 equals the number of blocks of @xmath51 .
[ lem101 ] for any @xmath369 and any @xmath328 , we have @xmath370 where @xmath371 denotes the tensor product @xmath372 when @xmath319 is written as @xmath373 , @xmath374 if @xmath375 , then we understand that @xmath376 moreover , we may write the formula in terms of set partitions : @xmath377 if @xmath378 \mid { \varepsilon}(i)=1\}>\#\{i\in[j ] \mid { \varepsilon}(i)=\ast\}$ ] for some @xmath379 $ ] , then we have @xmath380 .
this case is also covered by if we understand the sum over the empty set is 0 since @xmath381 in this case . the proof is given by induction . when @xmath382 , @xmath383 and @xmath384 and hence the formula is true .
suppose that the formula is true for @xmath385 . then for any @xmath386
, we get @xmath387}}(k ) } \alpha^{\nb(\pi , f)}q^{\cr(\pi)+ \ins(\pi)+ 2 \innb(\pi , f ) + 2\slnb(\pi , f ) } \\ & \qquad \times \prod_{\substack{\{i , j\ } \in \pair(\pi)\\ f(\{i , j\})=1 } } \langle x_i , x_j\rangle \prod_{\substack{\{i , j\ } \in \pair(\pi)\\ f(\{i , j\})=-1 } } \langle x_i , \overline{x_j}\rangle \ , \b^{{\varepsilon}(k+1)}(x_{k+1})\left(\bigotimes_{i\in \sing(\pi ) } x_i\right ) .
\end{aligned}\ ] ] the operator @xmath388 equals @xmath389 if @xmath390 and @xmath391 if @xmath392 from theorem [ thm1 ] .
we will show that the action of @xmath388 corresponds to the inductive pictorial description of set partitions of type b. we fix @xmath393}}(k)$ ] and suppose that @xmath51 has singletons @xmath394 ( we understand that @xmath395 ( @xmath396 , respectively ) when there is no singleton to the left ( right , respectively ) of @xmath170 ) , pair blocks @xmath397 with color @xmath325 to the right of @xmath170 , pair blocks @xmath398 with color @xmath344 to the right of @xmath170 and pair blocks @xmath399 which cover @xmath170 . there may be pair blocks to the left of @xmath170 , but they do not matter .
note that when there is no singleton , the arguments below can be modified easily .
if @xmath390 , then the operator @xmath389 acts on the tensor product , putting @xmath400 on the right .
this operation pictorially corresponds to adding the singleton @xmath401 ( with color 1 ) to @xmath402}}(k)$ ] , to yield the new type b partition @xmath403 .
this map @xmath404 does not change the numbers @xmath405 or @xmath406 , which is compatible with the fact that the action of @xmath389 does not change the coefficient .
note that if @xmath390 , then any @xmath407}}(k+1)$ ] has the singleton @xmath401 .
hence the formula holds when @xmath408 and @xmath390 .
if @xmath409 acts on the tensor product , then new @xmath410 terms appear by using . in the @xmath411 term
the inner product @xmath412 appears with coefficient @xmath413 .
pictorially this corresponds to getting a set partition @xmath414}}(k+1)$ ] by adding @xmath415 to @xmath321 and creating the pair @xmath416 with color @xmath325 .
this pair crosses the blocks @xmath417 and so increases the crossing number by @xmath418 but decreases the number of inner singletons by @xmath418 because originally @xmath170 was the inner singleton of @xmath399 .
now the new inner singletons @xmath419 and new inner negative blocks @xmath398 appear . because @xmath170 is not a singleton in @xmath420 , the number of singletons left to negative blocks decreases by @xmath42 .
altogether we have : @xmath421 , @xmath422 , @xmath423 , @xmath424 and @xmath425 .
so the exponent of @xmath6 increases by @xmath426 .
this factor @xmath413 is exactly the factor appearing in when @xmath409 acts on @xmath427 .
if @xmath428 acts on the tensor product , then new @xmath410 terms appear by using . in the @xmath411 term
the inner product @xmath429 appears with coefficient @xmath430 ( note here that @xmath431 ) .
pictorially this means that the new pair @xmath432 is created with color @xmath344 . similarly to case 2a
, we count the change of numbers and get @xmath421 , @xmath422 , @xmath423 , @xmath433 and @xmath434 . altogether , when moving from @xmath321 to @xmath420 , the exponent of @xmath40 increases by @xmath325 and the exponent of @xmath6 increases by @xmath435 , which coincides with the coefficient appearing in the action of @xmath428 , creating the inner product @xmath436 .
note that as @xmath321 runs over @xmath437 , every set partition @xmath438 appears exactly once in one of case 2a and case 2b .
therefore in case 2 , the pictorial inductive step and the actual action of @xmath439 both create the same terms with the same coefficients , and hence the formula is true when @xmath408 and @xmath392 .
case 1 and case 2 show by induction that the formula holds for all @xmath440 .
we show the formula . for @xmath441 where @xmath442 are pairs and @xmath443 are singletons
, we have @xmath444 \mid n_i=-1\}}q^{2\sum_{i=1}^k ( \cov(v_i ) + \sl(v_i ) ) \delta_{n_i,-1}}\prod_{i\in[k],n_i=1 } \langle x_{a_i},x_{b_{i}}\rangle\prod_{i\in[k ] , n_i=-1 } \langle x_{a_i},\overline{x_{b_i}}\rangle\right ) \\ & = \sum_{f}\alpha^{\nb(\pi , f)}q^{2\innb(\pi , f)+2\slnb(\pi , f)}\prod_{\{i , j\}\in\pi , f(\{i , j\})=1 } \langle x_i , x_j\rangle\prod_{\{i , j\}\in\pi , f(\{i , j\})=-1 } \langle x_{i},\overline{x_j}\rangle , \end{aligned}\ ] ] where @xmath445 is defined by @xmath446 and @xmath447 and hence @xmath445 runs over the set of all colorings of @xmath51 .
note here that @xmath448\mid n_i=-1\}$ ] , @xmath449 and @xmath450 . 1 . for @xmath7
, we recover the @xmath6-deformed formula for moments ( * ? ? ?
* proposition 2 ) : @xmath458 2 . for @xmath8 ,
let @xmath459 ( in this case @xmath460 is positive definite for @xmath461 and so we may consider @xmath462 ) and let @xmath463 .
we recover the @xmath42-deformed formula for moments ( * ? ? ?
* theorem 7.3 ) : @xmath464 \(2 ) when @xmath8 , the non zero contributions in corollary [ thm2](2 ) come only when @xmath51 is noncrossing and inner blocks are colored by 1 . given @xmath465 , @xmath466 equals @xmath467 if @xmath319 is outer and 1 otherwise .
hence @xmath468 the conclusion follows by multiplying this formula by @xmath469 and by using the relation @xmath470 . by using corollary [ thm2 ]
, we obtain @xmath476 and by permuting @xmath477 , @xmath478 since @xmath479 , there are two orthogonal unit eigenvectors @xmath480 of the involution @xmath481 , and we take @xmath482 and @xmath483 . note that for some @xmath484 we have @xmath485 . hence @xmath486 therefore the vacuum state is not a trace when @xmath487 .
when @xmath7 , the von neumann algebra becomes the @xmath6-deformed von neumann algebra and the traciality is known in ( * ? ? ?
* theorem 4.4 ) .
* it would be interesting to ask if the von neumann algebra @xmath471 is factorial . in the @xmath7 case
the factoriality was proved by boejko , kmmerer and speicher @xcite and by niady in special cases and then by ricard @xcite in full generality .
moreover nou showed that @xmath488 is not injective if @xmath489 @xcite . if @xmath490 then voiculescu @xcite showed that @xmath491 is isomorphic to the free group factor @xmath492 .
recently guionnet and shlyakhtenko @xcite showed that for each integer @xmath493 there exists @xmath494 such that @xmath488 is isomorphic to the free group factor @xmath492 for @xmath495 .
their method was extended by nelson @xcite to the non tracial case and he showed that hiai s @xmath6-deformed araki - woods algebras @xcite are isomorphic to shlyakhtenko s free araki - woods algebras @xcite for small @xmath496 . * it would be worth to ask whether the von neumanan algebra @xmath471 has the completely bounded approximation property .
* determine all the parameters @xmath0 for which the type b symmetrization operator @xmath96 is positive .
the positivity is known for @xmath101 $ ] and also for @xmath8 , @xmath497 as explained in corollary [ cor13 ] . then determine all the parameters for which @xmath99 for all @xmath134 .
the work was partially supported by the maestro grant dec-2011/02/a/ st1/00119 ( m. boejko ) and opus grant dec-2012/05/b
/ st1/00626 of national centre of science ( m. boejko and w. ejsmont ) .
the work was partially supported by marie curie international incoming fellowships ( project 328112 icncp , t. hasebe ) .
t. hasebe thanks nobuhiro asai for his interests and useful comments .
m. anshelevich , partition - dependent stochastic measures and @xmath6-deformed cumulants , doc .
math . 6 ( 2001 ) , 343384 . m. anshelevich , s.t .
belinschi , m. boejko and f. lehner , free infinite divisibility for q - gaussians , math .
17 ( 2010 ) , 905916 .
d. aurlien , n. salim and i. nourdin , fourth moment theorem and @xmath6-brownian chaos , comm . math .
321 , no . 1 ( 2013 ) , 113134 .
belinschi , m. boejko , f. lehner and r. speicher , the normal distribution is @xmath498-infinitely divisible , adv . math .
226 , no . 4 ( 2011 ) , 36773698 .
belinschi and d. shlyakhtenko , free probability of type b : analytic interpretation and applications , amer .
j. math .
134(1 ) ( 2012 ) , 193234 .
p. biane , free hypercontractivity , comm .
phys . 184 ( 1997 ) ,
p. biane , f. goodman and a. nica , non - crossing cumulants of type b , trans .
soc . 355 ( 2003 ) , 22632303 .
m. boejko and r. speicher , an example of a generalized brownian motion , commun .
phys . 137 ( 1991 ) , 519531 . m. boejko and r. speicher , completely positive maps on coxeter groups , deformed commutation relations , and operator spaces , math .
300 ( 1994 ) , 97120 . m. boejko and r. speicher , interpolations between bosonic and fermionic relations given by generalized brownian motions , math . z. 22 ( 1996 ) , 135159 .
humphreys , reflection groups and coxeter groups , cambridge studies in advances math .
29 , cambridge university press , 1990 .
r. koekoek , p.a .
lesky and r.f .
swarttouw , _ hypergeometric orthogonal polynomials and their q - analogues _ , springer - verlag , berlin , 2010
. i. krolak , wick product for commutation relations connected with yang - baxter operators and new constructions of factors , comm .
phys . 210 ( 2000 ) , 685701 .
h. upmeier , jordan algebras in analysis , operator theory , and quantum mechanics , cbms regional conference series in mathematics , 67 . published for the conference board of the mathematical sciences , washington , dc ; by the american mathematical society , providence , ri , 1987 .
viii+85 pp .
d. voiculescu , symmetries of some reduced free product @xmath499 algebras , operator algebras and their connections with topology and ergodic theory , lect .
notes in math . * 1132 * , springer , berlin ( 1985 ) , 556588 .
d. voiculescu , addition of certain noncommuting random variables , j. funct .
66(3 ) ( 1986 ) , 323346 .
d. voiculescu , free probability for pairs of faces i , commun .
, available online , may 2014 .
d. voiculescu , free probability for pairs of faces ii : 2-variables bi - free partial @xmath500-transform and systems with rank @xmath501 commutation . | in this article we construct a generalized gaussian process coming from coxeter groups of type b. it is given by creation and annihilation operators on an @xmath0-fock space , which satisfy the commutation relation @xmath1 where @xmath2 are elements of a complex hilbert space with a self - adjoint involution @xmath3 and @xmath4 is the number operator with respect to the grading on the @xmath0-fock space .
we give an estimate of the norms of creation operators .
we show that the distribution of the operators @xmath5 with respect to the vacuum expectation becomes a generalized gaussian distribution , in the sense that all mixed moments can be calculated from the second moments with the help of a combinatorial formula related with set partitions .
our generalized gaussian distribution associates the orthogonal polynomials called the @xmath6-meixner - pollaczek polynomials , yielding the @xmath6-hermite polynomials when @xmath7 and free meixner polynomials when @xmath8 . |
consider a compact possibly non - orientable surface properly embedded in a knot exterior in the 3-sphere @xmath0 .
it is called _ essential _ if it is incompressible and boundary - incompressible .
the boundary of an essential surface consists of a parallel family of non - trivial simple closed curves .
thus , on the peripheral torus of the knot , they determine an isotopy class of a non - trivial unoriented simple closed curve , which is called the _ boundary slope _ of the surface . in @xcite , hatcher showed that such boundary slopes are only finitely many .
moreover , by culler and shalen @xcite , it was proved that there always exist at least two such boundary slopes .
therefore one can get a non - empty , finite set of boundary slopes for a knot , which is said to be the _ boundary slope set_. the boundary slope set for a knot @xmath1 in @xmath0 gives a non - empty , finite subset of rational numbers in a standard way .
see @xcite for example . in view of this ,
culler and shalen introduced and studied in @xcite the _ diameter _ of the boundary slope set for @xmath1 , which is defined as the difference between the maximum and the minimum among non - meridional elements
. this will be denoted by @xmath2 in this paper .
[ [ section ] ] actually culler and shalen proved in @xcite that ; @xmath3 holds for a non - trivial knot @xmath1 in @xmath0 if @xmath1 does not have meridional boundary slope .
this can be generalized for an alternating knot @xmath1 by using the result in @xcite as follows : @xmath4 where @xmath5 denotes the minimal crossing number of @xmath1 .
see @xcite . in this paper
, we consider a _ montesinos knot _ , which is obtained by composing a number of rational tangles in line , and give the following : [ thm : diam : lowerbound : bycrossing ] for a montesinos knot @xmath1 , we have the inequality @xmath6 this theorem together with the results in @xcite and @xcite for two - bridge knots gives the following upper and lower bounds .
[ cor : diam : lowerbound : bycrossing ] for a non - trivial montesinos knot k , we have the inequalities @xmath7 in fact , as claimed in @xcite , if @xmath1 is an alternating montesinos knot , we have the equality @xmath8 [ [ section-1 ] ] for an alternating knot @xmath1 , the lower bound of @xmath9 in terms of @xmath5 is achieved by considering the checkerboard surfaces @xmath10 for its reduced alternating diagram .
in fact , these are known to be essential by @xcite . for these surfaces
, the following inequality also holds ; @xmath11 we also generalize this for montesinos knots as follows : [ thm : diam : lowerbound : euler ] let @xmath1 be a montesinos knot
. among its non - meridional boundary slopes , let @xmath12 and @xmath13 be the maximum and the minimum respectively . then there exist two essential surfaces @xmath14 and @xmath15 with boundary slopes @xmath12 and @xmath13 such that @xmath16 where @xmath17 denotes the ratio of the negative of the euler characteristic and the number of sheets for @xmath18 for @xmath19 . here , following @xcite , by the _ number of sheets _ of an essential surface , we mean the minimal number of intersection between the surface and the meridian of the knot .
we remark that our lower bound is optimal in a sense .
see remark [ rem : optimal ] . also it should be compared with an upper bound given in ( * ? ? ?
* theorem 4 ) : we actually showed that @xmath20 with boundary slopes @xmath21 in a montesinos knot exterior .
the minimal geometric intersection number of the curves representing two slopes @xmath22 and @xmath23 is called the _ distance _ of @xmath22 and @xmath23 .
this is usually denoted by @xmath24 .
about the distance of the maximal and minimal boundary slopes for a montesinos knot , we have the following as an immediate corollary to theorem [ thm : diam : lowerbound : euler ] .
[ cor : diam : lowerbound : ondistance : byeulerchara ] let @xmath1 be a non - trivial montesinos knot . among its non - meridional boundary slopes ,
let @xmath12 and @xmath13 be the maximum and the minimum respectively .
then there exist two essential surfaces @xmath14 and @xmath15 with boundary slopes @xmath12 and @xmath13 such that @xmath25 where @xmath26 denote the ratio of the negative of the euler characteristic and the number of boundary components for @xmath18 for @xmath19 . recall that the distance of the slopes @xmath27 and @xmath28 is calculated by @xmath29 .
also note that , for an essential surface @xmath30 with boundary slope @xmath31 , the number of sheets of @xmath30 is equal to the product of the denominator of @xmath31 with the number of boundary components of @xmath30 . with these facts ,
the corollary follows from theorem [ thm : diam : lowerbound : euler ] immediately .
this can be regarded as a generalization of the following result shown by culler and shalen : suppose that @xmath32 is a non - exceptional two - surface knot manifold : that is , @xmath32 is an irreducible , connected , compact , orientable 3 -manifold with single torus boundary such that it has at most two distinct isotopy classes of strict essential surfaces and it is neither seifert fibered nor an exceptional graph manifold .
let @xmath14 and @xmath15 be representatives of the two isotopy classes of connected strict essential surfaces .
let @xmath33 denote the boundary slope of @xmath18 and let @xmath34 denote the number of boundary components of @xmath18 .
then for @xmath35 we have @xmath36 please see @xcite for details .
this paper is organized as follows . in the next section
, we review the algorithm by hatcher and oertel given in @xcite . in section [ sec : twist ] , we give some formulae to calculate the twist and prove theorem [ thm : diam : lowerbound : bycrossing ] . in the last section ,
we introduce the remainder term , give formulae for calculating its value and show theorem [ thm : diam : lowerbound : euler ] .
in this section , we give a brief review of the hatcher - oertel s work given in @xcite , which is the base of our arguments .
we also prepare basic terminologies used in the rest of the paper .
note that the terms marked with `` '' are about notions which do not appear in @xcite and are introduced by the authors in accordance with our argument .
let us start with the definition of montesinos knots .
a _ montesinos knot _ is defined as a knot obtained by putting rational tangles together in a row .
see figure [ fig : montesinosknot ] for example . a montesinos knot obtained from rational tangles @xmath37
will be denoted by @xmath38 . here and in the sequel
, @xmath39 denotes an irreducible fraction or the corresponding rational tangle depending on the situation . in the following ,
we assume that each rational tangle is non - integral , just for normalization .
furthermore , we will always assume that the number of tangles is at least three .
note that the knots with at most two tangles are two - bridge knots . for two - bridge knots ,
theorem [ thm : diam : lowerbound : bycrossing ] holds by @xcite . also theorem [ thm : diam : lowerbound : euler ] holds since non - trivial two - bridge knots are all alternating .
( 109,91 ) ( 0,0 ) we here give a very brief review of the algorithm of hatcher and oertel .
see @xcite or @xcite for detail .
given montesinos knot @xmath40 , divide the 3-sphere @xmath0 into @xmath41 3-balls such that @xmath1 is decomposed into @xmath41 rational tangles @xmath42 . at the same time , an essential surface @xmath30 embedded in the exterior of @xmath1 is divided into surfaces @xmath43 .
each of these surfaces @xmath18 can be isotoped into some standard position , and is represented by an `` edgepath '' @xmath44 in a `` diagram '' @xmath45 , and then , whole @xmath30 is represented by an `` edgepath system '' @xmath46 , which consists of @xmath41 `` edgepaths '' as follows .
[ remark : ambiguity ] an edge " corresponds to a piece of surface called a saddle " .
an edgepath " consists of edges , and so , it corresponds to a surface obtained by combining saddles .
now , in fact , there are two possible choices of making saddles for single edge .
thus , multiple surfaces correspond to an edgepath " , and also to an `` edgepath system '' .
we here remark that , despite this ambiguity , all the surfaces corresponding to the same edgepath system " have the common boundary slope , the common value of @xmath47 , the common twist " , and the common `` remainder term '' defined in a later section .
see @xcite for details .
the graph on the @xmath48-@xmath49 plane defined as follows is called a _ diagram _ and denoted by @xmath45 .
a vertex of @xmath45 is a point @xmath50 denoted by @xmath51 or a point @xmath52 denoted by @xmath53 for an irreducible fraction @xmath27 with @xmath54 , or a point @xmath55 denoted by @xmath56 .
two vertices @xmath51 and @xmath57 are connected by an edge , which is a straight segment , if @xmath58 .
there is another kind of edge called a _ horizontal edge _ , which connects @xmath51 and @xmath53 for @xmath59 .
an important class of non - horizontal edges is the _ vertical edges _ , which connect the vertices @xmath60 and @xmath61 for an arbitrary integer @xmath62
. another important class of the edges is the _
@xmath63-edges _ , which connect @xmath56 and @xmath60 for an integer @xmath62 .
the diagram @xmath45 is illustrated in figures [ fig : diagram ] and [ fig : diagram2 ] .
by @xmath64 , we denote a subgraph of @xmath45 lying in @xmath65 .
a path @xmath66 on @xmath45 is called _ minimal _ if @xmath66 is forbidden from retracing an edge and going along two sides of single triangle in succession .
an _ edgepath _
@xmath66 for a rational tangle associated to the irreducible fraction @xmath27 is defined as a minimal path on @xmath45 such as : ( a ) just one point on a horizontal edge @xmath51@xmath53 , or ( b ) a path starting at @xmath51 proceeding monotonically from right to left or at least vertically .
an edgepath is called a _ constant edgepath _ in the former case , and a _ non - constant edgepath _ otherwise .
an _ edgepath system _
@xmath46 for a montesinos knot @xmath67 is then defined as a tuple @xmath68 of edgepaths where each @xmath44 is an edgepath for the tangle @xmath69 .
we call a non - constant edgepath for a rational tangle @xmath69 whose @xmath48-coordinate of the endpoint is @xmath70 and which has no vertical edges a _ _ basic edgepath__@xmath71for @xmath69 . a _
_ basic edgepath system__@xmath71for a montesinos knot is then defined as a set of basic edgepaths with the starting points associated to the tangles . a basic edgepath @xmath72 for a tangle @xmath69 is extended by combining @xmath72 and the horizontal edge @xmath73@xmath74 .
we call it the _ _ extended basic edgepath__@xmath71@xmath75 of @xmath72 . by an _ _ extended basic edgepath system__@xmath71 , we mean the set of the extended basic edgepaths @xmath76 obtained from a basic edgepath system @xmath77
. an extended basic edgepath @xmath78 is naturally regarded as a continuous piecewise - affine function @xmath79\rightarrow\mathbb{r}$ ] which gives the @xmath49 coordinate of the point @xmath80 on the edgepath for each @xmath48 .
we then regard an extended basic edgepath system @xmath81 as a function defined as the sum @xmath82 of functions @xmath83 .
similarly , a basic edgepath system @xmath84 , or an edgepath system in general , can be regarded as a function , which is a restriction of the function @xmath81 . from an extended basic edgepath system @xmath85 for a montesinos knot @xmath86 , @xmath87 , @xmath88 @xmath89 and an arbitrary @xmath90 with @xmath91 , by _ cutting _ it at @xmath92
, we can make an edgepath system @xmath93 as follows .
for each @xmath94 , if @xmath95 , then we set @xmath44 to be a non - constant edgepath @xmath96 .
if @xmath97 , then we set @xmath44 to be a constant edgepath consisting of the point @xmath98 with coordinates @xmath99 on a horizontal edge @xmath100@xmath101 . in this operation , the edgepath system @xmath46 may include an edge which is a proper subset of an edge of the diagram @xmath45 .
the edge is called a _
partial edge_. in contrast , an edge of the diagram @xmath45 completely included in an edgepath is called a _
complete edge_. in the way developed in @xcite , an edgepath system is associated to a surface embedded in a montesinos knot exterior .
such an edgepath system must satisfy the following condition : for the endpoints of the edgepaths , their @xmath48-coordinates are common and their @xmath49-coordinates @xmath102 are summed up to zero ; @xmath103 we refer these conditions as the _ gluing consistency _ , and call an edgepath system satisfying these conditions a _ candidate edgepath system_. in order to enumerate all boundary slopes for a montesinos knot @xmath1 , we first enumerate all basic edgepath systems for @xmath1 , which is obviously finitely many , and then list candidate edgepath systems up as follows .
( i ) from each basic edgepath system @xmath84 , by making the extended basic edgepath system @xmath81 , solving the equation @xmath104 , getting a solution @xmath105 , and cutting @xmath81 at @xmath92 , we have a candidate edgepath system , which is called a _
type i _ edgepath system .
the endpoints of its edgepaths have the common @xmath48-coordinate @xmath106 .
( ii ) by adding appropriate number of vertical edges to some edgepaths of @xmath84 if necessary , we may have a candidate edgepath system , which is called a _ type ii _
edgepath system .
the endpoints of its edgepaths have the common @xmath48-coordinate @xmath70 .
( iii ) by adding suitable ( possibly partial ) @xmath63-edges to @xmath84 , we have a candidate edgepath system , which is called a _ type iii _
edgepath system .
the endpoints of its edgepaths have the common @xmath48-coordinate @xmath107 .
remark here that our classification is slightly different from that in @xcite , that is , a basic edgepath system satisfying gluing consistency is regarded as type ii in this paper and as type i in @xcite . by corollary 2.4 to proposition 2.10 in @xcite , we can determine whether surfaces corresponding to a candidate edgepath system are essential or not .
we thus obtain a list of edgepath systems corresponding to some essential surface .
the algorithm of hatcher and oertel completes by calculating all the boundary slopes for the surfaces associated to such edgepath systems .
a non - constant edgepath has a roughly right - to - left direction . with this direction , each non-@xmath63 non - horizontal edge in the edgepath
is said to be _ increasing _ or _ decreasing _ according to whether @xmath49-coordinate increases or decreases as a point moves along the edge according to the direction .
for each possibly - partial non-@xmath63-edge @xmath108 in the edgepath , we assign @xmath109 or @xmath110 as the _ sign _ @xmath111 if the edge @xmath108 is increasing or decreasing . at each point @xmath51 with @xmath112 , there exist exactly one increasing leftward edge and exactly one decreasing leftward edge .
hence , for a fixed tangle , there exists the unique basic edgepath consisting of only increasing leftward edges , which is called the _ _ monotonically increasing basic edgepath__@xmath71 .
similarly the unique _ _ monotonically decreasing basic edgepath__@xmath71exists .
note that they are minimal .
then , for a fixed montesinos knot @xmath1 , there exist the unique basic edgepath system consisting of monotonically increasing basic edgepaths and the unique basic edgepath system consisting of monotonically decreasing basic edgepaths , which are called the _ _ monotonically increasing basic edgepath system__@xmath71and the _ _ monotonically decreasing basic edgepath system__@xmath71 .
let @xmath113 and @xmath114 denote them respectively .
when we regard the corresponding extended basic edgepath systems as a function , naturally , @xmath115 holds for any extended basic edgepath system @xmath81 for a fixed @xmath1 and any @xmath65 .
besides , @xmath113 and @xmath114 are convex and concave as a function respectively . for each possibly partial edge in the edgepath , we assign the length @xmath116 , where the length of a complete edge is @xmath117 and the length of a partial edge is less than @xmath117 .
precisely , as calculated in @xcite or @xcite , a partial edge @xmath118 , which is included in a complete edge @xmath51@xmath57 and has @xmath90 as the @xmath48-coordinate of the endpoint , has the length @xmath119 as mentioned in remark [ remark : ambiguity ] , a set of surfaces correspond to a candidate edgepath system . in view of this , a candidate edgepath system is called _ incompressible _ , _ compressible _ or _ indeterminate _ ,
if all the corresponding surfaces are essential , all the corresponding surfaces are inessential , or the set of the corresponding surfaces includes both essential and inessential ones , respectively .
[ remark : pi_1-injectivity ] the aim of the algorithm of hatcher and oertel is to enumerate all the boundary slopes of the orientable essential surfaces .
in fact , they determine @xmath120-injectivity instead of incompressibility . for instance , the difference of these two notions is mentioned in @xcite .
note that @xmath120-injectivity is stronger than incompressibility .
this does not matter in @xcite , since @xmath120-injectivity is equivalent to incompressibility as for orientable surfaces .
though , since we here deal with possibly non - orientable surfaces , we have to be careful about the difference .
nonetheless , in our later argument , the difference does not make any trouble fortunately . here , we recall two notions used in @xcite , and give a lemma about incompressibility of edgepath systems used in later arguments .
two successive edges @xmath121@xmath122@xmath123 on the diagram @xmath45 are said to be _
reversible _ if the two edges lie in two triangles of @xmath45 sharing a common edge .
for example , @xmath56@xmath124@xmath125 is reversible since these edges lie in two triangles sharing the common edge @xmath124@xmath126 .
an edgepath is said to be _ completely reversible _ if all pairs of successive two edges in it are reversible . for an edgepath ,
if the last edge of the edgepath is included in @xmath51@xmath57 , then the _ final @xmath127-value _ of the edgepath is defined to be @xmath128 . in some cases
, we may give the positive or negative sign to @xmath127-values according to whether the last edge is increasing or decreasing respectively . an edgepath system
has a _ cycle of final @xmath127-values _ obtained by collecting the final @xmath127-values of @xmath41 edgepaths in the system . here , we summarize some results about incompressibility in propositions in @xcite used in this paper . [
lem : incompressibility ] for a montesinos knot @xmath40 , where @xmath129 , the following hold . 1 .
[ lem : incompressibility : monotonictypei ] a monotonically decreasing type i edgepath system is incompressible .
[ lem : incompressibility : typeiorii ] suppose that a type i or type ii edgepath system @xmath46 lying in @xmath64 has @xmath130 with @xmath131 as its cycle of final @xmath127-values . then @xmath46 is incompressible unless the last edge of @xmath132 lies in the same triangle of @xmath64 , and has the same ending point , as an edge with @xmath133 .
[ lem : incompressibility : monotonictypeii ] a monotonically decreasing type ii edgepath system is incompressible . + if a basic edgepath system @xmath84 does not satisfy the condition ( * ) : + _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the cycle of final @xmath127-values for the @xmath44 s contains at least one @xmath110 , and each @xmath110 in this cycle is separated from the next @xmath110 by @xmath134 s all but possibly one of which are @xmath135 s _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ + then there exists an incompressible or indeterminate type ii edgepath system obtained from @xmath84 by adding upward vertical edges . here
an edgepath with @xmath135 means that the edgepath has final @xmath127-value @xmath136 and is completely reversible .
[ lem : incompressibility : typeiii ] a type iii edgepath system @xmath46 is compressible if and only if @xmath137 holds and at least @xmath138 edgepaths are completely reversible . on the other hand ,
a type iii edgepath system is incompressible if it is constructed from a basic edgepath system @xmath84 with @xmath139 .
( 1 ) , ( 3)(a ) : by corollary 2.4 and propositions 2.6 , 2.7 and 2.8(a ) in @xcite , if a type i or type ii edgepath system is compressible or indeterminate , then its cycle of final @xmath127-values must include both positive and negative . since all the signed final @xmath127-values of the monotonic edgepath system have the common sign , the edgepath system is incompressible .
+ ( 2 ) : it follows from proposition 2.7(3)(a ) in @xcite .
+ ( 3)(b ) : it follows from proposition 2.9 in @xcite , where the condition ( * ) is stated .
+ ( 4 ) : it follows from proposition 2.5 in @xcite .
this section is devoted to proving theorem [ thm : diam : lowerbound : bycrossing ] .
the key is the estimation of the maximal and minimal twists " defined as follows .
the _ twist _ @xmath140 of an edgepath @xmath66 is defined as the sum of @xmath141 for non-@xmath63-edges @xmath142 included in @xmath66 .
the _ twist _
@xmath143 of an edgepath system @xmath46 is then defined as the sum of @xmath144 for edgepaths @xmath44 included in @xmath93 .
then , as described in @xcite , the boundary slope of an essential surface associated to an edgepath system @xmath46 is calculated by @xmath145 , where @xmath146 denotes the edgepath system @xmath46 to which a seifert surface for @xmath1 is associated . note
that @xmath63-edges and constant edgepaths contribute nothing to the twist . in the following , we call
the maximum among the twists of all the incompressible or indeterminate candidate edgepath systems for a montesinos knot @xmath1 the _ maximal twist _ for @xmath1 .
the _ minimal twist _ for @xmath1 is defined in the same way .
precisely we establish the following estimation about twists .
[ prop : twist : lowerbound ] let @xmath147 and @xmath148 denote the maximal and minimal twists for a montesinos knot @xmath1 , respectively .
let @xmath113 and @xmath114 be the monotonically increasing and decreasing basic edgepaths for @xmath1 with the twists @xmath149 and @xmath150 , respectively . then @xmath147 satisfies @xmath151 also @xmath148 satisfies @xmath152 once the above proposition is established , theorem [ thm : diam : lowerbound : bycrossing ] is proved as follows .
as remarked in subsection [ subsec : ho : montesinos knot ] , theorem [ thm : diam : lowerbound : bycrossing ] holds if the number of tangles @xmath153 in a montesinos knot @xmath1 is at most two .
thus we assume that @xmath154 . in the following we use the same notations as in proposition [ prop : twist : lowerbound ] .
if @xmath155 or @xmath156 holds , then @xmath1 is alternating , and we already have @xmath157 in @xcite . in other cases , as in @xcite , @xmath158 holds . if @xmath159 , we have @xmath160 and @xmath161 .
if @xmath162 , we have @xmath163 and @xmath164 . otherwise , we have @xmath165 and @xmath166 . hence , in all cases , @xmath167 .
note that @xmath159 and @xmath162 can not occur at the same time since the relation @xmath168 holds and we are assuming @xmath169 . in order to prove proposition [ prop : twist : lowerbound ] , we have to estimate the maximal and minimal twists for a montesinos knot @xmath1 . in this subsection , we prepare two lemmas giving formulae for comparing twists of edgepath systems . here
we give a lemma used to compare the twists of type i edgepath systems . in the proof
, we introduce an integration formula to compute twists of edgepath systems , which is of interest independently . in the following , we regard an edgepath system as a continuous and piecewise - affine function @xmath79\rightarrow\mathbb{r}$ ] as explained in [ subsubsec : basic : extendedbasic : edgepathsystem ] .
precisely , for a basic edgepath system or a type i edgepath system @xmath46 , we define a continuous and piecewise - affine function @xmath170\rightarrow\mathbb{r}$ ] as follows .
if @xmath46 is a basic edgepath system @xmath84 , we define @xmath171 as @xmath81 regarded as a function .
if @xmath46 is of type i , assuming that @xmath46 is constructed by cutting a basic edgepath system @xmath84 at @xmath92 , we define @xmath171 so that @xmath172 for @xmath173 and @xmath174 for @xmath175 .
[ lem : twist : typei ] let @xmath114 be the monotonically decreasing basic edgepath system for a montesinos knot @xmath1 . 1 .
[ lem : twist : typei : general ] for any type i edgepath system @xmath46 , its twist @xmath143 satisfies @xmath176 .
[ lem : twist : typei : case2 ] assume that @xmath159 .
* assume further that there exists a solution @xmath92 for the equation @xmath177 .
let @xmath178 be the twist of the type i edgepath system @xmath179 obtained by cutting @xmath114 at @xmath92 .
* * [ lem : twist : typei : case2:1/2 ] if @xmath180 holds , then @xmath181 satisfies @xmath182 .
this twist is maximal among all type i edgepath systems . * * [ lem : twist : typei : case2:2/3 ] if @xmath183 holds , then @xmath181 satisfies @xmath184 .
this twist is maximal among all type i edgepath systems . *
[ lem : twist : typei : case2:other ] assume further that @xmath185 holds .
then , for any type i edgepath system @xmath46 , its twist @xmath143 satisfies @xmath186 .
we first prepare the following claim .
let @xmath187 and @xmath188 be a basic or type i edgepath systems .
assume that @xmath189 and @xmath190 are the corresponding functions @xmath79\rightarrow\mathbb{r}$ ] for the edgepath systems .
if @xmath191 holds for @xmath65 , then , their twists satisfy @xmath192 .
the following subclaim gives an integration formula to compute twists .
let @xmath46 be a basic edgepath system or an edgepath system of type i. then its twist is calculated by the integration of the form @xmath193 we fix a basic edgepath system @xmath84 and the extended basic edgepath system @xmath194@xmath195 , @xmath196 ,
, @xmath197@xmath198 for @xmath84 .
let @xmath199@xmath200 , @xmath201 ,
, @xmath202@xmath198 denote a type i edgepath system obtained from @xmath84 by cutting at @xmath92 in our manner . though @xmath203 may not satisfy the gluing consistency ( [ eq : equationgluingconsistency ] ) , we can calculate the twist @xmath204 and @xmath205 formally .
we regard @xmath206 as a function of @xmath48 from @xmath79 $ ] to @xmath207 .
assume first that @xmath44 ends at a point on a decreasing edge @xmath51@xmath57 , where @xmath208 .
let @xmath209 denote the twist of complete edges of @xmath44 included completely in @xmath210 .
the derivative of the twist of the edgepath @xmath211 at @xmath90 is calculated as @xmath212 with the formula ( [ eq : formula : lengthofpartialedge ] ) of length of a partial edge in subsubsection [ subsubsec : lengthofedgepath ] .
note that @xmath209 is constant at @xmath213 .
besides , @xmath214 hence , @xmath215 this identity also holds if @xmath44 ends at a point on an increasing edge or a horizontal edge . since @xmath205 is @xmath70 for @xmath90 close to @xmath117 , by summing up ( [ eq : dtwistdu ] ) and performing integration
, we have @xmath216 by this subclaim , we have ; @xmath217_{1}^{0 } \\ & & - \int_1^{0 } \frac{d}{du } \left(-\frac{2}{(u-1)^2 } \right ) \cdot ( { \varphi}_a(u)-{\varphi}_b(u ) ) du \\ & \le & 0 . \ ] ] note that @xmath218 holds by the gluing consistency .
besides , @xmath187 and @xmath188 comes from the same montesinos knot @xmath67 .
edges of the edgepaths @xmath219 and @xmath220 near @xmath221 are both the horizontal edge corresponding to @xmath69 .
hence , there exists @xmath222 such that @xmath223 holds for @xmath224 . eventually , the square bracket has the value zero . in the following ,
let @xmath225 denote the function for the monotonically decreasing basic edgepath system @xmath114 .
+ ( 1 ) : let @xmath226\rightarrow \mathbb{r}$ ] be a type i edgepath system @xmath46 as a function . from @xmath114 ,
by replacing the last segment @xmath227@xmath228 with a polygonal line @xmath229@xmath230@xmath228 , we have a function @xmath231\rightarrow \mathbb{r}$ ] , which is pretty close to @xmath114 in @xmath232 $ ] and is pretty close to vertical edges connecting @xmath233 and @xmath124 . since @xmath234 holds for @xmath65 , by the method in the above claim , we have @xmath235 for each sufficiently small @xmath236 .
now , @xmath237 hence , @xmath238 note that @xmath239 is necessary to give a type i edgepath system .
( [ lem : twist : typei : case2:1/2])(a1 ) : let @xmath240 denote the function corresponding to the monotonically decreasing type i edgepath system @xmath179 . then @xmath240 and @xmath225 coincide for @xmath210 , while @xmath241 and @xmath242 hold for @xmath173 . by the above subclaim
, @xmath243 suppose that there exists another type i edgepath system @xmath46 whose twist is greater than @xmath244 and which is obtained by cutting a basic edgepath system @xmath84 at @xmath245 .
then we note that @xmath84 must include a part below the horizontal line @xmath246 by the above claim . on the other hand
, @xmath84 satisfies @xmath247 , @xmath248 and @xmath249 for some @xmath250 and @xmath251 satisfying @xmath252 .
though , this is impossible since @xmath84 is affine in the interval @xmath253 .
( [ lem : twist : typei : case2:2/3])(a2 ) : again , let @xmath240 denote the function corresponding to the monotonically decreasing type i edgepath system @xmath179 .
let @xmath254 be a function defined by @xmath255 for @xmath173 and @xmath256 for @xmath175 .
the graph of @xmath254 includes a segment connecting @xmath257 and @xmath258 .
thus @xmath259 holds for @xmath65 . note that @xmath114 is a concave function
then , by the above claim and subclaim we have @xmath260 also let @xmath261 be a function defined by @xmath262 for @xmath253 , @xmath263 for @xmath264 , and @xmath265 for @xmath175 .
the graph of @xmath261 includes a piecewise affine path connecting @xmath257 , @xmath266 and @xmath258 .
thus @xmath267 holds for @xmath65 .
again , by the above claim and subclaim , we have @xmath268 here , @xmath269 denotes the integration value of ( [ eq : twistbyintegration ] ) for the function @xmath171 .
suppose that @xmath46 is another type i edgepath system whose twist @xmath178 is greater than @xmath244 .
a basic edgepath system @xmath84 including @xmath46 must include a part below the line @xmath246 . since @xmath84 is affine in the intervals @xmath270 and @xmath271 , @xmath84 satisfies @xmath272 , @xmath273 and @xmath274 .
though , this is impossible as follows .
@xmath273 means that only one of the edgepaths satisfies @xmath275 for some integer @xmath62 , and thus , at most one of the basic edgepaths can end with an increasing edge .
thus , @xmath276 or @xmath110 .
( [ lem : twist : typei : case2:other])(b ) assume that @xmath46 is a type i edgepath system with the function @xmath171 .
let @xmath254 be a function defined by @xmath277 for @xmath278 and @xmath279 for @xmath280 .
the graph of @xmath254 includes a segment connecting @xmath229 and @xmath281 .
then , @xmath282 holds for @xmath65 .
note that @xmath283 holds for @xmath278 . by the above claim and subclaim
, we have ; @xmath284 the second lemma is also for calculating twist and is useful for type ii and type iii edgepath systems .
we first introduce a classification of edgepath systems .
+ 1 . for a basic edgepath @xmath285 , we call it : ( a ) a class a basic edgepath if it is a monotonically decreasing basic edgepath @xmath286 , ( b ) a class b basic edgepath if @xmath285 and @xmath286 bound single triangle and single vertical edge in the strip @xmath64 , ( c ) a class c basic edgepath if @xmath285 and @xmath286 bound two triangles and no vertical edges in the strip @xmath64 .
a class b basic edgepath is obtained from the monotonically decreasing basic edgepath by replacing an edge @xmath287@xmath288 by an edge @xmath61@xmath288 for some integer @xmath62 .
2 . for a basic edgepath system @xmath84 , we call it : ( a ) a class a basic edgepath system if it is a monotonically decreasing basic edgepath system , ( b ) a class b / c basic edgepath system if exactly one basic edgepath in the basic edgepath system is class b / c and all the other edgepaths are monotonically decreasing ( class a ) , 3 . for a type ii edgepath system , it is called class a / b / c if it is obtained from a class a / b / c basic edgepath system by extending by vertical edges
. 4 . for a type iii edgepath system , it is called class a / b / c if it is obtained from a class a / b / c basic edgepath system by extending by @xmath63-edges for all edgepaths . then we obtain the following lemma .
[ lem : twist : typeiioriii ] let @xmath114 be the monotonically decreasing edgepath system with twist @xmath289 for a montesinos knot @xmath1 . 1 .
let @xmath290 be a type ii edgepath system with twist @xmath291 . then ,
[ lem : twist : typeiioriii : typeii ] ( a ) @xmath292 if and only if @xmath290 is of class a , ( b ) @xmath293 if and only if @xmath290 is of class b , ( c ) otherwise , @xmath294 .
2 . let @xmath295 be a type iii edgepath system with twist @xmath296 .
then , [ lem : twist : typeiioriii : typeiii ] ( a ) @xmath297 if and only if @xmath295 is monotonically decreasing , ( b ) @xmath298 if and only if @xmath295 is of class b or c , ( c ) otherwise , @xmath299 .
let @xmath300 be the monotonically decreasing basic edgepath system .
we first prepare two combinatorial values defined for edgepath systems .
we consider a basic edgepath system @xmath84@xmath301@xmath302 , @xmath303 , @xmath304 , @xmath305@xmath198 for @xmath1 .
let @xmath306 and @xmath307 denote the number of triangles and the number of vertical edges in the strip @xmath64 bounded by @xmath308 and @xmath309 respectively . set @xmath310 and @xmath311 .
then we have the following claim .
let @xmath312 and @xmath313 denote the corresponding values for @xmath84 .
then the twist @xmath314 is calculated by @xmath315 let @xmath316 be the triangles bounded by @xmath308 s and @xmath309 s .
let @xmath317 be the directed loop bounding the triangle @xmath318 with anti - clockwise direction .
let @xmath319 be the bounded downward vertical edges .
let @xmath320 and @xmath321 denote the sets of directed edgepaths of @xmath114 and @xmath84 with the right - to - left direction .
for a leftward edge @xmath322 and @xmath323 with the opposite direction , we here define the effect of @xmath323 on twist to be the negative of the effect of @xmath322 .
the effects of @xmath322 and @xmath323 on twist cancel out .
then , directed arcs and loops satisfy the following identity .
@xmath324 now , three vertices of a triangle in the strip @xmath64 can be expressed as @xmath325 , @xmath326 and @xmath327 .
note that the @xmath48-coordinate of @xmath328 is greater than those of @xmath329 and @xmath330 , and the @xmath49-coordinate of @xmath328 lies between those of @xmath329 and @xmath330 .
thus , the shapes of triangles are somehow restricted . as the result ,
contribution of @xmath331 is @xmath136 .
besides , contribution of a downward vertical edge @xmath332 is @xmath333 .
thus , we have @xmath334 .
the twist of @xmath84 is calculated by @xmath335 .
( [ lem : twist : typeiioriii : typeii ] ) a type ii edgepath system @xmath290 is obtained from a basic edgepath system @xmath84 by extending by vertical edges .
let @xmath312 and @xmath313 denote the corresponding values for @xmath84 .
now , the effect on the twist of @xmath290 by vertical edges is @xmath336 .
thus , by the claim above , we have ; @xmath337 .
note that if @xmath338 , then @xmath339 holds , and @xmath340 is non - minimal since @xmath341 must hold for some @xmath94 .
thus , if @xmath338 , then possible @xmath342 is @xmath229 or @xmath343 .
these correspond to class a and class b type ii edgepath systems .
( [ lem : twist : typeiioriii : typeiii ] ) the twist of a type iii edgepath system is same as its corresponding basic edgepath system @xmath84 .
let @xmath312 and @xmath313 denote the corresponding values for @xmath84 .
then , by the claim above , the twist @xmath314 is calculated as @xmath344
. here we consider the cases satisfying @xmath345 . by definition
, @xmath307 is @xmath70 or @xmath117 for each @xmath94 . if @xmath346 , then @xmath347 , since some increasing edges must appear in @xmath72 .
hence , @xmath348 holds for each @xmath94 , and so does @xmath349 .
if @xmath340 , then @xmath341 holds for some @xmath94 , and this means that @xmath72 is not minimal . eventually , if @xmath345 , then @xmath342 is @xmath229 , @xmath343 or @xmath350 . moreover , if @xmath342 is @xmath343 or @xmath350 , then @xmath351 is @xmath343 or @xmath350 for the unique @xmath94 respectively , since @xmath341 is impossible .
thus , at most one basic edgepath of a basic edgepath system can be non - monotonically - decreasing .
this implies the assertions ( a ) , ( b ) and ( c ) in ( [ lem : twist : typeiioriii : typeiii ] ) .
this subsection is devoted to proving proposition [ prop : twist : lowerbound ] . in the following proof
, we divide the set of montesinos knots into some classes by the nature of their monotonic basic edgepath systems . for each class
, we find an edgepath system which is incompressible or indeterminate , and then calculate or estimate its twist . since the edgepath system is taken so that its twist gives an appropriate bound of the maximal / minimal twist , it suffices to prove proposition [ prop : twist : lowerbound ] . in fact
, our main targets to study are monotonic or nearly monotonic edgepath systems .
we only prove the assertion about the maximal twist .
it is sufficient since the assertion about the minimal twist must hold by symmetry .
that is , with the fact that @xmath352 , @xmath353 , @xmath88 @xmath354 has its mirror image @xmath355 , @xmath356 , @xmath88 @xmath357 , the assertion about the minimal twist of the knot @xmath1 is immediately obtained from the assertion about the maximal twist of the knot @xmath358 .
let @xmath114 be the monotonically decreasing edgepath system with twist @xmath289 for a montesinos knot @xmath1 . by lemma [ lem : incompressibility ] , depending on the value of @xmath359
, we may be able to obtain an incompressible type ii or type iii edgepath system .
hence , we first divide the montesinos knots by the value of @xmath359 , as follows .
case 1 : : @xmath1 satisfies @xmath360 .
case 2 : : @xmath1 satisfies @xmath361 .
case 3 : : @xmath1 satisfies @xmath362 . remember that @xmath359 is an integer . for a montesinos knot with @xmath360 ,
there exists a monotonically decreasing type ii edgepath system @xmath363 such that an essential surface is associated to @xmath363 and its twist @xmath364 satisfies that @xmath365 . from @xmath114 , by extending one of its edgepaths by some downward vertical edges if necessary
, we have a monotonically decreasing type ii edgepath system @xmath366 in this case .
this is incompressible by lemma [ lem : incompressibility]([lem : incompressibility : monotonictypeii])(a ) .
its twist @xmath367 satisfies @xmath368 by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeii])(a ) . for a montesinos knot with @xmath362 , the monotonically decreasing type iii edgepath system @xmath369 such that an essential surface is associated to @xmath369 and its twist @xmath370 satisfies that @xmath371 .
since @xmath372 in this case , @xmath369 is incompressible by lemma [ lem : incompressibility]([lem : incompressibility : typeiii ] ) .
its twist is @xmath373 by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeiii])(a ) .
as we saw in subsection [ subsec : ho : incompressibility ] , the final @xmath127-values of a basic edgepath system are important in showing the existence of a type ii edgepath system with an incompressible surface by lemma [ lem : incompressibility]([lem : incompressibility : monotonictypeii])(b ) .
let @xmath374 , @xmath375 , @xmath88 @xmath376@xmath198 denote the final @xmath127-values of the monotonically decreasing basic edgepath system .
note that all @xmath134 s are negative naturally .
we here divide case 2 into several cases according to the numbers of basic edgepaths of @xmath377 , @xmath378 or @xmath379 , as follows .
case 2 - 1 : : @xmath1 satisfies @xmath380 .
case 2 - 2 : : @xmath1 satisfies @xmath381 . + case 2 - 2 - 1 ; ; moreover , @xmath1 satisfies @xmath382 .
case 2 - 2 - 2 ; ; moreover , @xmath1 satisfies @xmath383 .
+ case 2 - 2 - 2 - 1 : : moreover , @xmath1 satisfies @xmath384 . or equivalently , @xmath385 .
case 2 - 2 - 2 - 2 : : moreover , @xmath1 satisfies @xmath386 . or equivalently , @xmath387 .
case 2 - 2 - 3 ; ; moreover , @xmath1 satisfies @xmath388 .
case 2 - 3 : : @xmath1 satisfies @xmath389 .
+ case 2 - 3 - 1 ; ; moreover , @xmath114 of @xmath1 satisfies ( * ) in lemma [ lem : incompressibility]([lem : incompressibility : monotonictypeii])(b ) .
case 2 - 3 - 2 ; ; moreover , @xmath114 of @xmath1 does not satisfy ( * ) in lemma [ lem : incompressibility]([lem : incompressibility : monotonictypeii])(b ) . for a montesinos knot in these cases ,
an essential surface is associated to a type ii edgepath system @xmath390 obtained by extending the monotonically decreasing basic edgepath system by one upward vertical edge .
it has the twist @xmath391 .
@xmath114 does not satisfy the condition ( * ) in lemma [ lem : incompressibility]([lem : incompressibility : monotonictypeii])(b ) in all these three cases : for there is no @xmath377 in case 2 - 1 , while there is @xmath392 or more @xmath379 despite @xmath381 in case 2 - 2 - 3 .
thus , by lemma [ lem : incompressibility]([lem : incompressibility : monotonictypeii])(b ) , we have a class a type ii edgepath system @xmath393 to which an incompressible surface is associated . by lemma
[ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeii])(a ) , @xmath390 has twist @xmath394 . for a montesinos knot in these cases , the monotonically decreasing type i edgepath system @xmath179 exists ,
an essential surface is associated to @xmath179 , and @xmath179 has twist @xmath178 satisfying @xmath395 .
we consider a part of the graph of the function @xmath114 in the region @xmath278 .
the slope of the last segment of @xmath114 is @xmath396 .
the equation @xmath397 is expressed by @xmath398 in the interval @xmath399 . in these cases , @xmath31 is @xmath392 or greater since the final @xmath127-values include two @xmath134 s equal to @xmath110 or include one @xmath134 equal to @xmath110 and two @xmath134 s equal to @xmath136 . hence , the equation has a solution @xmath400 $ ] .
thus , the monotonically decreasing type i edgepath system @xmath179 exists , and its endpoints have the common @xmath48-coordinates @xmath401 . by lemma [ lem : incompressibility]([lem : incompressibility : monotonictypei ] )
, @xmath179 is incompressible . as in lemma
[ lem : twist : typei]([lem : twist : typei : case2])(a1 ) , the twist of @xmath179 is @xmath182 . here ,
we prove the claim for the remaining case 2 - 2 - 2 - 1 . in this case , a montesinos knot @xmath1 satisfies @xmath385 , @xmath159 , @xmath381 , @xmath384 and @xmath383 . without loss of generality , we can assume that @xmath402 , @xmath403 and @xmath404 .
furthermore , we can assume that @xmath405 satisfies @xmath406 and others satisfy @xmath407 .
that is , any knot @xmath1 in this case can be isotoped into the normalized one .
we here divide further the montesinos knots in case 2 - 2 - 2 - 1 into three cases .
case a : : @xmath1 satisfies @xmath408 .
case b : : @xmath1 satisfies @xmath409 .
+ case b - a ; ; @xmath1 satisfies @xmath410 or @xmath411 .
case b - b ; ; @xmath1 satisfies @xmath131 . for a montesinos knot in these cases , the monotonically decreasing type i edgepath system @xmath179 exists ,
an essential surface is associated to @xmath179 , and @xmath179 has twist @xmath178 satisfying @xmath412 .
@xmath185 holds since @xmath413 , @xmath414 , and @xmath415 hold . besides
, @xmath416 is obtained from @xmath417 , @xmath418 and @xmath419 in case a , from @xmath420 , @xmath418 and @xmath421 in case b - a . then the solution @xmath183 exists for the equation @xmath177 .
hence , there exists the monotonically decreasing type i edgepath system @xmath179 with @xmath422 . by lemma [ lem : incompressibility]([lem : incompressibility : monotonictypei ] )
, @xmath179 is incompressible . by lemma [ lem :
twist : typei]([lem : twist : typei : case2])(a2 ) , @xmath179 has @xmath412 . for a montesinos knot in this case , from @xmath114 , a minimal class b type ii edgepath system @xmath423 is obtained by replacing a @xmath424@xmath425 edge of @xmath426 by a @xmath124@xmath425 edge .
an essential surface is associated to @xmath423 , and @xmath423 has twist @xmath178 satisfying @xmath427 .
let @xmath428 denote the @xmath94-th edgepath of @xmath423 .
note that the edgepath @xmath429 is still minimal since @xmath430 holds and @xmath429 does not include the edge @xmath425@xmath431 .
@xmath423 is incompressible by lemma [ lem : incompressibility]([lem : incompressibility : typeiorii ] ) since the cycle of final @xmath127-values is @xmath130 with @xmath432 .
@xmath433 has twist @xmath427 , by lemma [ lem : twist : typeiioriii](2)(b ) .
consequently , with these six claims , we have shown proposition [ prop : twist : lowerbound ] .
[ rem : signofmaxmintwists ] the twists of the maximal and minimal boundary slopes are always non - negative and non - positive respectively , as follows . since any basic edgepath has at least length @xmath117 and any basic edgepath system has @xmath169 edgepaths , we have @xmath434 and @xmath435 .
as described in the argument above , for the maximal or minimal boundary slope , both @xmath436 and @xmath437 hold . hence , @xmath438 .
[ rem : bestpossibilityofthm : diam : lowerbound : bycrossing ] though it is not completely confirmed , the lower bound given in theorem [ thm : diam : lowerbound : bycrossing ] seems to be sharp in a sense .
we think about a family of montesinos knot @xmath439 , which correspond to claim 5 . by calculation
, @xmath440 seems to become arbitrary close to @xmath441 as @xmath442 goes to infinity .
however , it seems that no montesinos knots @xmath1 can achieve @xmath443 .
in this section , we will prove theorem [ thm : diam : lowerbound : euler ] .
the keys are the `` remainder term '' of an edgepath system and proposition [ prop : diam : lowerbound : main ] , which is a technical proposition about the remainder term .
the comparison of the twist and the euler characteristic ; strictly speaking , the ratio of the euler characteristic and the number of sheets for an essential surface with the maximal or minimal boundary slope , plays a main role in the proof of theorem [ thm : diam : lowerbound : euler ] .
let @xmath46 be an edgepath system with twist @xmath143 . then
, as we saw in @xcite , the ratio @xmath444 for a surface @xmath30 associated to @xmath46 is roughly the sum of lengths of the edgepaths in @xmath46 .
remark that this ratio is determined independently from the choice of the surfaces corresponding to @xmath46 .
see @xcite for details .
thus we also use the notation @xmath445 for the value corresponding to a surface associated to @xmath46 . on the other hand , by definition
, @xmath143 is roughly calculated as twice of the sum of signed lengths of the edges in @xmath46 .
thus , if @xmath46 is nearly monotonic , @xmath143 is roughly twice of @xmath445 . in view of these
, we define the _ remainder term _
@xmath446 of @xmath46 as @xmath447 .
[ prop : diam : lowerbound : main ] for a montesinos knot @xmath40 with @xmath169 , there exists an essential surface @xmath30 such that its twist @xmath449 is maximal and its remainder term @xmath448 is non - negative .
this assertion also holds for an essential surface with the minimal twist .
for a non - two - bridge montesinos knot @xmath1 , by proposition [ prop : diam : lowerbound : main ] , there exist two essential surfaces for @xmath1 such that their twists @xmath450 and @xmath451 are maximal and minimal and their remainder terms @xmath452 and @xmath453 are both non - negative .
let @xmath12 and @xmath13 be the maximum and the minimum among non - meridional boundary slopes for @xmath1 , respectively .
then , together with the definition of the twist , we have @xmath454 note that @xmath450 and @xmath451 can be confirmed to be non - negative and non - positive as in remark [ rem : signofmaxmintwists ] .
next , assume that a montesinos knot @xmath1 is also a two - bridge knot .
the monotonically decreasing and monotonically increasing type iii edgepaths correspond to essential surfaces by @xcite and give the maximal and minimal slopes by the estimation of the twist of an edgepath in the previous section .
we can easily check that the remainder terms are @xmath392 for both essential surfaces .
hence , @xmath455 we fix an edgepath system @xmath46 with the twist @xmath143 .
we recall one more definition used in ( * ? ? ?
* subsection 5.2 ) .
we collect all non-@xmath63-edges @xmath456 of all non - constant edgepaths @xmath44 in @xmath46 , divide them into two classes according to the sign @xmath457 , and then sum up the lengths of edges for each class . with the total lengths @xmath458 and @xmath459 , let @xmath460 denote @xmath461 , and call it the _ cancel _ of the edgepath system . with the cancel , we have the following . 1 .
[ lem : remainderterm : typei ] assume that @xmath46 is a type i edgepath system .
let @xmath464 and @xmath465 denote the constant edgepaths in @xmath46 and the number of the constant edgepaths , where each constant edgepath @xmath44 is a point on a horizontal edge @xmath100@xmath101 .
then , the remainder term is calculated by @xmath466 2 .
[ lem : remainderterm : typeii ] for a type ii edgepath system @xmath46 , we have @xmath467 .
thus , @xmath468 holds if @xmath469 .
[ lem : remainderterm : typeiii ] for a `` non - augmented '' type iii edgepath system @xmath46 with complete @xmath63-edges , we have @xmath470 .
thus , @xmath468 holds if @xmath471 , that is , the type iii edgepath system is monotonic . for non-@xmath63-edges of an edgepath system , let @xmath472 be the sum of length of the edges , @xmath473 the sum of signed length of the edges .
then , we have @xmath474 , @xmath475 , @xmath476 , @xmath477 and @xmath478 . ( [ lem : remainderterm : typei ] ) it is just a calculation of @xmath479 with the following formula in @xcite . @xmath480 our strategy to prove the proposition is as follows : in the proof of proposition [ prop : twist : lowerbound ] , we divided the set of montesinos knots into some classes , and found for each class an edgepath system whose twist gives a lower bound on the maximal twist .
as in that proof , we will divide the set of montesinos knots , and for a montesinos knot in each class , collect edgepath systems with the twist equal to or greater than that lower bound , and then prove that their reminder terms are all non - negative . 1
. @xmath485 and @xmath486 hold for some essential surface @xmath15 .
2 . @xmath487 holds for some essential surface @xmath15 .
3 . @xmath488 holds .
4 . @xmath485 and @xmath489 hold for some essential surface @xmath15 . by virtue of this lemma , in the following proof of the proposition
, we can ignore edgepath systems satisfying ; the condition ( i ) , for it is sufficient to check if @xmath489 ; the condition ( ii ) , for @xmath490 can not be the maximum ; the conditions ( iii ) and ( iv ) , for even if @xmath14 gives the maximal slope , it gives @xmath468 .
[ lem : specialedgepathsystem2 ] let @xmath1 be a montesinos knot such that the monotonically decreasing basic edgepath system @xmath114 satisfies @xmath159 or @xmath70 .
assume that @xmath14 denotes an essential surface with twist @xmath490 associated to a class b or class c type iii edgepath system for @xmath1 .
then there exists an essential surface @xmath15 with twist @xmath491 associated to the monotonically decreasing type iii edgepath system such that @xmath487 holds . for a montesinos knot satisfying @xmath360
, there exists an essential surface associated to a monotonically decreasing type ii edgepath system such that its twist is maximal and its remainder term is non - negative . as in the proof of proposition
[ prop : twist : lowerbound ] , there exists an incompressible , a monotonically decreasing type ii edgepath system @xmath363 .
its twist @xmath364 is maximal by lemmas [ lem : twist : typei]([lem : twist : typei : general ] ) and [ lem : twist : typeiioriii ] .
its remainder term @xmath492 is @xmath493 by lemma [ lem : remainderterm]([lem : remainderterm : typeii ] ) . for a montesinos knot satisfying @xmath362
, there exists an essential surface associated to the monotonically decreasing type iii edgepath system such that its twist is maximal and its remainder term is non - negative . as in the proof of proposition
[ prop : twist : lowerbound ] , the monotonically decreasing type iii edgepath system @xmath369 is incompressible .
its twist @xmath370 is maximal by lemmas [ lem : twist : typei]([lem : twist : typei : general ] ) and [ lem : twist : typeiioriii ] .
its remainder term @xmath494 is @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeiii ] ) . for a montesinos knot in these cases
, there exists an essential surface associated to a class a type ii edgepath system @xmath390 or the monotonically decreasing type iii edgepath system such that its twist is maximal and its remainder term is non - negative . as in the proof of proposition
[ prop : twist : lowerbound ] , there exists an incompressible , class a type ii edgepath system @xmath390 .
its twist @xmath495 is maximal among those of type i or type ii edgepath systems by lemmas [ lem : twist : typei]([lem : twist : typei : general ] ) and [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeii ] ) .
its remainder term @xmath496 is @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeii ] ) , since its cancel @xmath497 is equal to @xmath117 . only the monotonically decreasing type iii edgepath system @xmath369 can have the twist greater than @xmath495 by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeiii ] ) .
its remainder term @xmath494 is equal to @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeiii ] ) . for a montesinos knot in these cases
, there exists an essential surface associated to a monotonically decreasing type i edgepath system or a class a type ii or iii edgepath system such that its twist is maximal and its remainder term is non - negative . as in the proof of proposition
[ prop : twist : lowerbound ] , there exists an incompressible monotonically decreasing type i edgepath system @xmath179 .
its twist @xmath244 satisfies @xmath498 and is maximal among those of type i edgepath systems by lemma [ lem : twist : typei]([lem : twist : typei : case2])(a1 ) .
the piecewise linear equation ( [ eq : equationgluingconsistency ] ) has the form @xmath500 for @xmath270 where @xmath501 . in this case
, @xmath502 is equal to or greater than @xmath503 .
other @xmath127-values are all @xmath136 .
thus , the mean value of @xmath504 is @xmath505 or greater .
hence , we have @xmath506 and a solution @xmath507 .
note that @xmath179 does not include any constant edgepath since @xmath508 holds as shown in the proof of proposition [ prop : twist : lowerbound ] .
eventually , by lemma [ lem : remainderterm]([lem : remainderterm : typei ] ) , @xmath509 . only class
a type ii edgepath systems can have the twist greater than @xmath244 , which is @xmath510 , among all type ii edgepath systems by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeii ] ) .
its remainder term is equal to @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeii ] ) . only class
a type iii edgepath system can have the twist greater than @xmath244 , which is @xmath289 , among all type iii edgepath systems by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeiii ] ) .
its remainder term is equal to @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeiii ] ) . for a montesinos knot in these cases
, there exists an essential surface associated to a monotonically decreasing type i edgepath system , a class a or class b type ii edgepath system , or the monotonically decreasing type iii edgepath system such that its twist is maximal and its remainder term is non - negative . as in the proof of proposition
[ prop : twist : lowerbound ] , there exists an incompressible monotonically decreasing type i edgepath system @xmath179 .
its twist @xmath244 satisfies @xmath511 and is maximal among those of type i edgepath systems by lemma [ lem : twist : typei]([lem : twist : typei : case2])(a2 ) . recall that in this case @xmath385 holds and the final @xmath127-values are @xmath512 with @xmath404 .
since the solution @xmath90 satisfies @xmath513 , at most one constant edgepath system exists in @xmath179 , and if exists , the constant edgepath is on the edge @xmath425@xmath514 .
by lemma [ lem : remainderterm]([lem : remainderterm : typei ] ) , if @xmath515 , then @xmath516 .
otherwise , @xmath517
. only class a or class b type ii edgepath systems can have the twist greater than @xmath244 among type ii edgepath systems by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeii ] ) .
their remainder terms are @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeii ] ) , since their cancel @xmath497 are equal to @xmath117 .
only the monotonically decreasing , class b , or class c type iii edgepath systems can have the twist greater than @xmath244 among all type iii edgepath systems by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeiii ] ) .
however , in this case , class b or class c type iii edgepath systems can not have the maximal twist by lemma [ lem : specialedgepathsystem2 ] .
the remainder term of the monotonically decreasing type iii edgepath systems is equal to @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeiii ] ) . for a montesinos knot in this case
, there exists an essential surface associated to a class a or class b type ii edgepath system or the monotonically decreasing type iii edgepath system such that its twist is maximal and its remainder term is non - negative . as in the proof of proposition
[ prop : twist : lowerbound ] , there exists an incompressible class b type ii edgepath system @xmath423 .
its twist @xmath518 is equal to @xmath519 .
this is maximal among those of type i edgepath systems by lemma [ lem : twist : typei]([lem : twist : typei : case2])(b ) , for @xmath520 is obtained from @xmath413 , @xmath414 and @xmath521 .
its remainder term is @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeii ] ) , since its cancel @xmath522 is equal to @xmath117 .
only a class a type ii edgepath system can have the twist , which is @xmath510 , greater than @xmath518 among all type ii edgepath systems by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeii ] ) .
its remainder term is equal to @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeii ] ) .
only the monotonically decreasing type iii edgepath system can have the twist , which is @xmath289 , greater than @xmath518 , among all type iii edgepath systems by lemma [ lem : twist : typeiioriii]([lem : twist : typeiioriii : typeiii ] ) .
its remainder term is equal to @xmath70 by lemma [ lem : remainderterm]([lem : remainderterm : typeiii ] ) .
[ rem : optimal ] we can easily confirm that the lower bound ( [ eq : diam : lowerbound : main ] ) of the diameter is best possible .
for example , assume that , a montesinos knot @xmath1 has @xmath493 tangles , and its monotonically increasing and decreasing basic edgepath systems satisfy @xmath523 and @xmath524 .
as in claim 2 in the proof of proposition [ prop : diam : lowerbound : main ] , the monotonically decreasing type iii edgepath system and the monotonically increasing type iii edgepath system give the maximal and the minimal twists , and both have @xmath525 .
thus , the diameter for @xmath1 satisfies the equality in the inequality ( [ eq : diam : lowerbound : main ] ) . though the argument in this subsection is necessary , it is technical and a kind of supplement .
the precise definitions of special edgepaths which we have treated separately in the previous subsection are as follows : * there is an edge @xmath526@xmath56 called the _ augmented edge_. for a type iii edgepath system , in some cases , the augmented edge can be attached to some of edgepaths .
the edgepath system thus obtained is called an _ augmented type iii edgepath system_. * for a basic edgepath system @xmath84 satisfying @xmath276 , we can attach partial @xmath63-edges with the common @xmath48-coordinate of ending points to edgepaths in @xmath84 on making a type iii edgepath system .
the edgepath system thus obtained is called a _ type iii edgepath system with partial @xmath63-edges_. * a type ii edgepath system is called a _ type ii edgepath system with redundant vertical edges _ if it includes both upward vertical edges and downward vertical edges .
if there is an essential surface @xmath14 associated to an augmented type iii edgepath system , as mentioned in @xcite , some essential surface @xmath15 is associated to the corresponding non - augmented type iii edgepath system . thus @xmath14 and @xmath15 satisfy ( i ) .
assume that a surface @xmath14 corresponding to a type iii edgepath system with partial @xmath63-edges is essential .
let @xmath84 be the basic edgepath system corresponding to @xmath14 .
if @xmath84 is monotonically decreasing , then @xmath488 holds .
that is , @xmath14 satisfies ( iii ) .
if @xmath84 is of class b or class c , since @xmath159 or @xmath70 holds , by lemma [ lem : specialedgepathsystem2 ] , an essential surface @xmath15 is associated to the monotonically decreasing type iii edgepath system . then , @xmath14 and @xmath15 satisfy ( ii ) .
if @xmath527 , then @xmath14 may give the maximal twist in claim 5 .
though , then , at the same time , the monotonically decreasing type i edgepath system has the twist equal to @xmath528 . even in this case
, it will be proved that @xmath468 holds for a type i edgepath system with remainder term @xmath462 .
thus @xmath14 satisfies ( iv ) .
if @xmath84 is any other basic edgepath system , since @xmath529 holds , @xmath14 can not give the maximal twist and satisfies ( ii ) .
assume that a type ii edgepath system @xmath530 has redundant vertical edges and that @xmath530 is constructed from some type ii edgepath system @xmath531 without redundant vertical edges by adding upward and downward vertical edges .
assume further that an essential surface @xmath14 is associated to @xmath530 . by combining the latter half of the proof of proposition 2.9 in @xcite and exercise just after the proof
, we can see that if all surfaces corresponding to @xmath531 are not-@xmath120-injective , then all surfaces corresponding to @xmath530 are also not-@xmath120-injective .
since we have a non - minimal part in a deformed edgepath system in the argument , not-@xmath120-injective surfaces obtained are moreover compressible in fact .
eventually , if no essential surface is associated to @xmath531 , then no essential surface is associated to @xmath530 .
now , by assumption , @xmath14 associated to @xmath530 is essential .
hence , there exists an essential surface @xmath15 associated to @xmath531 .
@xmath532 holds since contributions of the redundant vertical edges cancel out each other .
moreover , @xmath533 holds since @xmath15 is simpler by the effect of redundant vertical edges .
let @xmath534 and @xmath535 be a class b or class c type iii edgepath system and the monotonically decreasing type iii edgepath system respectively .
assume that @xmath536 or @xmath70 .
assume that an essential surface @xmath14 is associated to @xmath534 .
the compressibility of a type iii edgepath system is determined by use of completely reversibility of edgepaths .
as in lemma [ lem : incompressibility]([lem : incompressibility : typeiii ] ) , a type iii edgepath system @xmath46 is compressible if and only if @xmath137 holds and at least @xmath138 edgepaths are completely reversible .
now , since @xmath536 or @xmath70 , the first condition holds for both @xmath534 and @xmath535 .
suppose that @xmath537@xmath538 , @xmath539 , @xmath88 @xmath540 @xmath198 is compressible and that @xmath541@xmath542 , @xmath543 , @xmath88 @xmath544 @xmath198 is not compressible .
then , at least @xmath138 edgepaths of @xmath535 and at most @xmath545 edgepaths of @xmath534 are completely reversible .
the difference appears in exactly one pair of edgepaths @xmath546 and @xmath547 .
assume that @xmath546 is completely reversible and @xmath547 is not so . since @xmath546 is a monotonically - decreasing completely - reversible edgepath , the edgepath is limited to an edgepath of the form @xmath56@xmath60@xmath288@xmath304@xmath548@xmath549 for some integer @xmath62 and @xmath550 .
only the other possible minimal type iii edgepath system for the @xmath551-tangle is @xmath552@xmath61@xmath549 .
this edgepath @xmath66 can not be a class c type iii edgepath .
for this edgepath @xmath66 to be a class b type iii edgepath , @xmath553 must be @xmath392 .
though , @xmath56@xmath61@xmath288 is also completely reversible .
thus , if @xmath535 is compressible , then @xmath534 is also compressible .
similarly to the argument for ( c ) , a not-@xmath120-injective surface obtained by proposition 2.5 in @xcite is moreover compressible .
eventually if no essential surface is associated to @xmath535 , then no essential surface is associated to @xmath534 .
now , by assumption , @xmath14 associated to @xmath534 is essential .
hence , there exists an essential surface @xmath15 associated to @xmath535 .
obviously , we have @xmath554 . | in this paper , two lower bounds on the diameters of the boundary slope sets are given for montesinos knots .
one is described in terms of the minimal crossing numbers of the knots , and the other is related to the euler characteristics of essential surfaces with the maximal / minimal boundary slopes . |
since dirac established relativistic quantum mechanics , many scientists such as h. weyl , w. heisenberg , have attempted to associate the elementary particles with the eigenstates of the nonlinear spinor equation@xcite . in 1951 , r. finkelsten solved some rigorous solutions of the nonlinear spinor equation by numerical simulation , and pointed out that the corresponding particles have quantized mass spectra@xcite .
however these researches have not realized their authors goal due to the mathematical difficulties in analyzing nonlinear spinor equation . in recent years , a great effort has been made along this line of research .
the theoretical proof about the existence of solitons was investigated in @xcite .
the symmetries and many conditional exact solutions of the nonlinear spinor equations are collected in @xcite .
the present work is a development of some previous works@xcite . in this paper , we define some functions which reflect the properties of eigen solutions to the nonlinear spinor equations , and compute the typical values , then extract some important information from the data .
the following are some general knowledge for the nonlinear spinors .
denote the minkowski metric by @xmath0 , pauli matrices by @xmath1 define @xmath2 hermitian matrices as follows @xmath3 where @xmath4 , @xmath5 and @xmath6 . in this paper
, we adopt the hermitian matrices ( [ 1.2 ] ) instead of dirac matrices @xmath7 for the convenience of calculation . for dirac
s bispinor @xmath8 , the quadratic forms of @xmath8 are defined by @xmath9 where the superscript ` + ' stands for the transposed conjugation . by the definition ( [ 1.3 ] ) we have @xmath10 etc . , where @xmath11 is the dirac conjugation@xcite .
by transformation law of @xmath8 , one can easily check that @xmath12 is a contra - variant 4-vector , @xmath13 a true scalar and @xmath14 a pseudo - scalar .
one can construct some other covariant quadratic forms , but they are not independent on ( [ 1.3 ] ) for some pauli - fierz identities@xcite , such as @xmath15 . in general ,
the lagrangian of the nonlinear bispinor @xmath8 with a vector potential @xmath16 and scalar @xmath17 is given by@xcite @xmath18 where @xmath16 and @xmath17 include the self and external potential , @xmath19 and @xmath20 are used to stand for the repulsive or attractive self interaction , e.g. @xmath16 stands for repulsive electromagnetic potential if @xmath21 , but stands for attractive interactive potential similar to strong interaction if @xmath22 .
@xmath17 stands for a scalar interactive potential like higgs field , which is repulsive if @xmath23 , and attractive if @xmath24 .
however , in this paper , we take ( [ lag ] ) as one system , and only the internal interactions are considered .
if @xmath25 , the eigen solution might be absent , so we only consider the case @xmath26 is a concave function satisfying @xmath27 the corresponding dynamical equation is given by @xmath28 the hamiltonian form of ( [ 1.04 ] ) is given by @xmath29 .
\label{1.4}\end{aligned}\ ] ] where @xmath30 is the momentum operator . for ( [ 1.4 ] ) , by the current conservation law , we have the normalizing condition @xmath31 let @xmath32 be the angular momentum operator @xmath33 then the eigenfunctions of @xmath34 are given by @xmath35 where the index ` t ' stands for transpose , @xmath36 .
for all the eigenfunctions , @xmath37 is commutative with the nonlinear hamilton operator like the linear case , so under a suitable choice of coordinate system , the solutions of ( [ 1.4 ] ) take the following form@xcite , @xmath38 where @xmath39 are real functions independent of @xmath40 and @xmath41 .
the system ( [ 1.4 ] ) has many symmetries such as the global gauge invariance , the spatial reversal invariance etc .
the above assumptions have removed the uncertainty caused by the symmetry , and the solutions are determined except for a signature .
this procedure is quite important for the numerical solving and stability analysis .
if @xmath42 take the form of polynomials , the solutions are analytic functions of @xmath43 , and then they can be expressed as the taylor series of @xmath43 . in the cases of @xmath44 , which are the only cases for free particles
, we have the formal solution as follows @xmath45 where @xmath46 are real free parameters determined by boundary conditions and the normalizing condition , but @xmath47 are real numbers determined by @xmath48 with @xmath49 . the normalizing condition ( [ 1.6 ] )
the simplest case of ( [ 1.4 ] ) is the following dynamical equation @xmath51.\label{2.1}\end{aligned}\ ] ] different from the linear case , the nonlinear spinor equation generally has continuous spectra if the restriction of ( [ 1.6 ] ) is absent , so the normalizing condition is a quantizing condition for nonlinear spinors , and the nonlinear coupling coefficient @xmath52 is meaningful only if the solution satisfies the normalizing condition ( [ 1.6 ] ) .
the eigen solutions to ( [ 2.1 ] ) with spin @xmath53 can be solved rigorously as follows @xmath54 where @xmath55 corresponds to even parity , and @xmath56 corresponds to odd parity . for the above eigenfunctions
, we have @xmath57 the radial equation of even parity satisfies @xmath58f,\\ \frac d { dr } f=-\frac{1}{\hbar c}[(\mu - m)c^2-wc(g^2-f^2)]g-\frac{2}{r}f .
\end{array}\r .
\label{2.10}\end{aligned}\ ] ] for the odd parity , we have @xmath59f,\\ \frac { d}{dr}f=-\frac{1}{\hbar c}[(\mu+m)c^2+wc(g^2-f^2)]g-\frac{2}{r}f .
\end{array}\r .
\label{2.11}\end{aligned}\ ] ] the initial data of ( [ 2.10 ] ) and ( [ 2.11 ] ) satisfy @xmath60 . for ( [ 2.10 ] ) and ( [ 2.11 ] ) , we have positive mass @xmath61 if and only if @xmath62@xcite . making transformation @xmath63 where @xmath64 is equivalent to the spectrum , @xmath65 takes the unit of length .
( [ 2.10 ] ) and ( [ 2.11 ] ) can be rewritten in a dimensionless form . for ( [ 2.10 ] ) we have @xmath66 where prime stands for @xmath67 . for ( [ 2.11 ] ) we have @xmath68 the normalizing condition ( [ 2.9 ] ) becomes @xmath69 where @xmath70 is a dimensionless constant to be determined .
the computation shows that , for any given @xmath71 , there exists a sequence of initial data @xmath72 , such that ( [ 2.10 ] ) and ( [ 2.11 ] ) have eigen solutions .
the theoretical analysis proves that there are infinite eigen solutions for every @xmath64@xcite . in @xcite
we have shown three families of eigen solutions with even parity and the first family of eigen functions with odd parity . to describe the properties of the eigen solutions , we define the following dimensionless functions , which are continuous functions of spectrum @xmath64 for the same family solutions . 1 . the dimensionless norm @xmath73 @xmath74 for the same family of eigen solution
, @xmath75 is a continuous function of @xmath64 . by ( [ 2.16 ] ) , the normalizing condition is equivalent to the equation @xmath76 .
the dimensionless energy @xmath77 @xmath78 this definition of energy is in the nther s sense .
3 . the mean diameter of an eigen solution @xmath79 @xmath80 where @xmath81 is a universal compton wave length for all solutions .
4 . the total dimensionless inner pressure @xmath82 @xmath83 the physical meanings of @xmath73 , @xmath77 and @xmath79 are evident .
now we examine the meanings of @xmath82 .
for the perfect fluid in relativity , the energy momentum tensor is given by @xmath84 for the static fluid , we have the 4-dimensional speed @xmath85 , and then @xmath86 for the nonlinear spinor ( [ 2.1 ] ) in curved space - time with diagonal metric , we define the corresponding concepts as follows@xcite @xmath87 for static spinor , we have @xmath88 ) .
the curves of the dimensionless functions defined above are shown in fig.[fig1 ] and fig.[fig2 ] . in fig.[fig1 ] , the normalizing condition @xmath89 and @xmath90 are derived from the anomalous magnetic moment(amm ) of an electron according to different definition of mass , as computed in the next section .
a rough computation was once given in @xcite .
, dimensionless energy @xmath91 and mean diameter @xmath79 of a spinor .
only the solutions corresponding to the intersection @xmath92 are meaningful in physics , width=453 ] and inner pressure @xmath82 of a dark spinor , width=453 ] for an electron , we have @xmath93 kg , @xmath94j.s , @xmath95 m/s . by ( [ 2.16 ] ) and @xmath96 , we can estimate the value @xmath97 in this case , the nonlinear spinor equation has only two valid eigen solutions corresponding to @xmath98 and @xmath99 .
the norm function @xmath73 of all other families of eigen solutions have no intersection points with @xmath100 .
the radial functions @xmath101 of solutions with even parity are shown in fig.[fig3 ] , where @xmath102 the unit of the coordinate @xmath43 is the universal compton wave length @xmath103 .
so the images of different solutions are visually comparable in fig.[fig3 ] .
more images see @xcite .
the nonlinear spinor with self electromagnetic interaction was researched by a few authors . in 1966
, m. wakano has approximately analyzed the cases of @xmath104 dominance and @xmath105 dominance when @xmath106 , and reached the following conclusions@xcite . in the case of @xmath105 dominance , the eigen solutions or the solitons do not exist for the first order approximation . in the case of @xmath104 dominance ,
the eigen solutions exist but all with negative energy .
in fact , the negative mass is equivalent to change the sign of @xmath104 , which implies to transform the repulsive potential of @xmath104 into the absorbent one . m. soler and a. f. raada calculated the eigen solutions of ( [ 1.4 ] ) by omitting @xmath105 .
but they neglected the normalizing condition and did not use the true value of @xmath107 .
their explanation for the results seems to be misguided by some inadequate ideas@xcite . besides
, the eigen solutions with born - infeld potential were studied in @xcite .
the detailed non - relativistic approximation of the many - spinors equations was given in @xcite in general , the coordinates @xmath43 and @xmath108 can not be separable for nonlinear spinor with vector potential due to the term @xmath105 . however @xmath109 and @xmath110 can be conveniently expressed by fourier series with respect to @xmath108 as ( [ 1.14 ] ) and ( [ 1.15 ] ) , and the equations of the radial functions can be derived via variation principle , because the eigen solutions are the critical points of the following energy functional @xmath111 so the problem ( [ 1.4 ] ) can be changed into an ordinary differential equation system which can be solved by numerical computation . in this paper , we only consider the eigen solutions with @xmath112-spin and even parity , which is the only valid case for a free electron .
in the dimensionless form , we have the magnitude for the fields @xmath113 where @xmath64 is the dimensionless spectrum .
since the high order terms are caused by the vector potential @xmath114 , for adequately large @xmath64 , we only keep the first order approximation for simplicity .
then we have @xmath115 where @xmath116 and @xmath117 are real functions of @xmath43 with @xmath118 . for large spectrum @xmath119 ,
the relative error of the approximation is less than @xmath120 , so the approximation is accurate enough to reveal the anomalous magnetic moment of a spinor with electromagnetic field . the less the value of @xmath64 , the large the error of approximation . the quadratic forms of @xmath8 are given by @xmath121
correspondingly we have @xmath122 substituting ( [ 3.2 ] ) , ( [ 3.3 ] ) into ( [ 1.7 ] ) we get the energy functional @xmath123+(\mu - m)cg^2 -(\mu+m)cf^2-\frac 1 2w(g^2-f^2)^2 \r .
+ \nn \\ & & \l .
e(g^2+f^2)a_0-\frac 4 3 egf a+\frac 1 2 a_0(\pa_r^2+\frac 2 r\pa_r)a_0- \frac 1 3 a(\pa_r^2+\frac 2 r \pa_r-\frac 2 { r^2})a\r\}+mc^2.~~ \label{3.4}\end{aligned}\ ] ] the approximation is only caused by the vector potential @xmath105 . by variation
, we get a closed system of ordinary differential equations @xmath124f-\frac 2 { 3\hbar}ea g,\\ f'=-\frac{1}{\hbar}[(\mu - m)c+ea_0-w(g^2-f^2)]g+(\frac 2 { 3\hbar}ea-\frac{2}{r})f,\\ a_0''+\frac 2 r a_0'=-e(g^2+f^2),\quad a''+\frac 2 r a'-\frac 2 { r^2 } a=-2egf.\\ \end{array}\r
. \label{3.5}\end{aligned}\ ] ] make transformation @xmath125 where @xmath126 is dimensionless potential , which can not be confused with the pressure defined in ( [ 2.20 ] ) . substituting them into ( [ 3.5 ] ) , we get the dimensionless form @xmath127 in ( [ 3.8 ] ) only @xmath64 is a free parameter , which acts as the spectrum similar to the dark case of @xmath128 .
the normalizing condition is still ( [ 2.16 ] ) .
( [ 3.8 ] ) is independent on the undetermined coefficient @xmath52 , but it becomes a global problem .
the natural boundary conditions are given by @xmath129 the solutions of @xmath130 can be expressed as @xmath131\tau^2d\tau d\rho } , \label{3.10 } \\ q & = & \frac { -4\al}{3\rho^2\int^\infty_0(u^2+v^2)\rho^2d\rho } { \int_0^\rho \rho^2 \int_\rho^\infty u(\tau)v(\tau)d\tau d\rho}. \label{3.11 } \end{aligned}\ ] ] we have @xmath132 for the meaningful solutions .
the solution of ( [ 3.10 ] ) and ( [ 3.11 ] ) can be soundly solved by iterative algorithm .
the total energy of the system in nther s sense is given by @xmath133 substituting ( [ 3.6 ] ) , ( [ 3.7 ] ) into it , we get the dimensionless form @xmath134\rho^2 d\rho}{\int^\infty_0(u^2+v^2)\rho^2 d\rho}. \label{3.13 } \end{aligned}\ ] ] the mass of a particle is a complex classical concept , which depends on the method of measurement and the context of theory . using different definition of mass , we will get different spectrum @xmath64 and constant @xmath70 .
different contribution to the energy has different energy - speed relation , which can be detected by elaborate experiment@xcite . such experiment might be a key to disclose the structure of elementary particles .
we give some more discussions in the next section . in what follows ,
we take @xmath135 and @xmath136 as the classical mass for computation . to get the anomalous magnetic moment
, we introduce an infinitesimal external magnetic field @xmath137 adding @xmath138 to ( [ 3.3 ] ) and substituting it into ( [ 3.4 ] ) , we get the increment of the energy @xmath139 where @xmath140 is the magnetic moment of the spinor .
the dimensionless form is given by @xmath141 where the constant @xmath142 is the bohr magneton , @xmath143 by ( [ 3.16 ] ) , we get the anomalous magnetic moment of a particle @xmath144 the empirical value of the amm of an electron is @xmath145 .
the computational result suggests that ( [ 3.17 ] ) might be the truth of the amm . to compare with the dark spinor
, we also define the dimensionless norm by ( [ 2.17 ] ) .
the normalizing condition ( [ 2.16 ] ) is equivalent to @xmath146 .
the dimensionless functions @xmath147 are all continuous functions of @xmath64 for the same family of solutions .
fig.[fig4 ] shows how to determine the spectrum @xmath64 by the empirical amm .
different definition of mass leads to different value of @xmath64 . ) vs. the spectra @xmath64 , the true value for an electron is @xmath145 or @xmath148,width=453 ] .
the constant @xmath70 is determined by the intersection points a or b , which correspond to the empirical anomalous magnetic moment , width=453 ] in fig.[fig4 ] , the trends of @xmath149 shows that @xmath149 is a decreasing function of @xmath64 , and @xmath150 as @xmath151 . by the empirical data of @xmath152
, we can compute the following undetermined parameters , if taking @xmath153 , we have @xmath154 if taking @xmath155 , we have @xmath156 fig.[fig5 ] shows the realistic values of some parameters such as the total energy @xmath157 , the norm function @xmath73 . the constants @xmath70 or @xmath52
is determined by normalizing condition @xmath146 , and then all other parameters can be computed . by fig.[fig5 ] , we learn that , the value of @xmath64 is larger than that of dark spinor , namely , the electromagnetic interaction increases the rest mass @xmath158 of a spinor .
correspond to spinor fields .
@xmath159 correspond to the dimensionless potentials.,width=453 ] since @xmath160 is quite small , by ( [ * 3.2 ] ) we learn that , if @xmath161 , the electromagnetic field only have a little influence on the eigen solution .
the numerical results also show this conclusion , fig.[fig6 ] shows the comparison of the dimensionless fields when @xmath119 .
form the above results and some other computation and analysis , we find some special but interesting properties of the nonlinear spinor .
these unusual properties might have close relationship with the nature of the elementary particles . 1 . by @xmath162 and ( [ 2.20 ] ) , for @xmath163 we find @xmath164 more calculations show that such relation also hold for other kind nonlinear potential @xmath42 satisfying @xmath165 , namely we always have @xmath166 .
an interesting problem is whether the error is just caused by numerical approximation and @xmath167 is a rigorous relation generally valid for nonlinear spinors ? 2 .
all dimensionless energy @xmath77 have a similar trend @xmath168 . for large enough @xmath64
, we always have @xmath169 .
3 . taking equation system ( [ 3.8 ] ) as a developing system with respect to @xmath170 , we find the initial value of @xmath171 etc . are determined by @xmath172 ,
then the eigen solutions satisfying boundary condition ( [ 3.9 ] ) only depend on @xmath172 .
this is a general feature for all eigen solutions of any nonlinear spinor , which implies the eigen equation of the nonlinear spinor is over - determined .
this fact might be the underlying reason of the pauli principle@xcite .
4 . for the nonlinear spinor equation with a scalar interaction @xmath173 similar to ( [ 3.10 ] ) and
( [ 3.11 ] ) , @xmath17 can be expressed as @xmath174 so the solution to ( [ 4.2 ] ) can be soundly solved by iteration . for the amm @xmath152 defined by ( [ 3.17 ] ) ,
computations show that we always have @xmath175 similar to the above cases with electromagnetic interaction .
this result implies that , it is inadequate to describe the strong interaction by a scalar field .
some rough calculations show that , the amm of a proton might be explained by the following nonlinear spinor with a strongly coupling vector interaction @xmath176 , @xmath177 if @xmath178 , we have the following conclusions : ( i ) . in the dimensionless form ,
the absolute values of @xmath179 are comparable with @xmath180 near the center , so the first order approximation ( [ 3.1 ] ) is invalid .
the first family of the eigen solutions is absent , namely , the fields @xmath181 have intersections with the horizontal axis near the origin , so such spinor has complicated interior structure .
although the solution has @xmath182 , but the solution @xmath8 includes components with orbital angular momentum , which has strong influence on the amm , so the value of @xmath183 is much larger than that caused by the weakly coupled electromagnetic interaction .
the energy functional of the nonlinear spinor system ( [ 1.7 ] ) is indefinite , so the stability of the solutions is special .
there are some works on this problem@xcite .
however , the usual treatment for the positive definite system might be inadequate for the nonlinear spinors .
7 . from the derivations of some previous works
, we find that we could almost reconstruct the physical theories based on the nonlinear spinors , so some further investigations on the nonlinear spinors are worthwhile .
the einstein s mass - energy relation @xmath184 is one of the most fundamental formulae in physics , but it has not been seriously tested by an elaborated experiment , and only some indirect evidences in nuclear reaction suggested that it holds to high precision . from the above calculation , we found the interaction potentials of a particle will yield detectable effects , which lead to different energy - speed relation , which can be used as the fingerprints of the interactive potentials of elementary particles .
so the experiment may shed lights on the nature of the interaction and elementary particles . in what follows
, we give a detailed description for the experiment .
the experiment only involves low energy accelerator of particles and measurement of speed . in this section ,
@xmath185 stand for the speed of a particle , which can not be confused with the fields defined in ( [ 1.12 ] ) or ( [ 2.13 ] ) . in einstein s original paper @xcite , he derived the kinetic energy of a particle @xmath186 , @xmath187 which implies the total energy and the speed of a particle have the following simple relation @xmath188 however , this relation is based on the linear classical mechanics , and it has not been directly tested by elaborated experiment .
there were once some indirect evidences in the nuclear reaction .
the most accurate one is provided by s. rainville _
et al_@xcite , which indicates that the mass - energy relation @xmath189 holds to an error level less than 0.00004% in the process of neutron capture by nuclei of sulfur and silicon resulting in @xmath190-radiation .
as pointed out by e. bakhoum@xcite , although the authors claimed it is a direct test , it is actually a test for the energy conversion @xmath191 at low speed of the particles .
as one of the most fundamental relation , a direct test for the original energy - speed relation ( [ t1.2 ] ) is necessary and significant .
what more important is that , for the nonlinear spinors with interactive potentials such as electromagnetic one @xmath16 , the detailed calculation shows the interaction terms result in the fine structure of the energy - speed relation@xcite , and the fine structure can be used as fingerprints of the interactions .
( [ 3.21 ] ) and ( [ 3.22 ] ) show that , the nonlinear potential yields energy to a magnitude of 1kev , and the electromagnetic interaction to a magnitude of 100ev , which can be easily detected by elaborated experiments . hereafter we take @xmath192 as unit of speed .
the general representation of the energy - speed relation is given by @xmath193 where @xmath194 are all constants of mass dimension , and @xmath195 is the total static mass of the particles , @xmath196 corresponds to interactions such as electromagnetic potential , @xmath197 corresponds to the nonlinear self - interactive potential .
the detailed explanation of the parameters see @xcite . because of the little value of @xmath198 and the function @xmath199 ,
the nonlinear effects can be easily concealed behind @xmath195 .
for example , when an electron get kinetic energy 30mev .
the corresponding speed reaches @xmath200 , but @xmath201 that is to say , ( [ t1.3 ] ) is a stiff equation of the coefficients @xmath194 .
so we have to make some transformation to get meaningful solution .
we propose the following experimental project to measure the coefficients @xmath194 in the energy - speed relation .
the flow chart and experimental scheme are illustrated in fig.[tfig ] .
the particles with unit charge are produced by the particles source , and the ones at given initial speed @xmath202 are selected by a homogeneous magnetic field . by adjusting the radius @xmath43 , we can control the initial speed of the particles @xmath202 .
the series - wound accelerator is constructed by a set of uniform electrodes , which can be charged with high voltage @xmath203 .
when the selected particles pass though one pair electrodes , each particle receives an energy increment @xmath204 , which converts into its kinetic energy .
if @xmath205 pair electrodes are charged , then we get the total kinetic energy increment for each particle @xmath206 ( [ t1.3 ] ) and ( [ t1.4 ] ) establish the connection between the speed @xmath207 and @xmath208 .
the final speed @xmath209 of the particles can be measured by the position @xmath210 of the particle counter or film .
then we can determine the constants @xmath194 by fitting the curve @xmath211 as defined by ( [ t1.4 ] ) .
now we make some simplification of ( [ t1.4 ] ) . at first
, we can solve the static mass @xmath195 at low energy @xmath212 as follows .
assume the voltage @xmath213 , in this nonrelativistic case , we have the approximation of ( [ t1.4 ] ) as follows @xmath214 then we get @xmath215 where @xmath216 is the non - relativistic static mass of the particle in classical sense .
therefore , we only need to determine two little coefficients @xmath198 at high energy . by ( [ t1.3 ] ) , ( [ t1.4 ] ) and ( [ t1.6 ] ) , we have the following relation @xmath217\sqrt{1-u_1 ^ 2}-m_s \r\ } , \qquad\qquad \qquad\qquad\quad~ { \rm ( by~ ( \ref{t1.4})~ and~ ( \ref{t1.6}))}\nn\\ & \dot = & ( nev+\frac 1 2 m_s u_0 ^ 2)\sqrt{1-u_1 ^ 2}-\frac { m_s u_1 ^ 2}{1+\sqrt{1-u_1 ^ 2}}+(2m_1-m_\ga)\sqrt{1-u_1 ^ 2}. \label{t1.7}\end{aligned}\ ] ] denoting @xmath218 and substituting it into ( [ t1.7 ] )
, we get @xmath219m_\ga = ( { 1-\chi } ) u - \chi m_s.\label{t1.8}\end{aligned}\ ] ] ( [ t1.8 ] ) is a linear equation of @xmath198 , which can be easily solved by the method of least squares from a sequence of measured data @xmath220 .
define the @xmath221-dimensional vectors @xmath222 by , @xmath223,\\ \vec f & = & [ ( { 1-\chi_1})u_1 -\chi_1 m_s , ( { 1-\chi_2})u_2 -\chi_2 m_s,\cdots,({1-\chi_n})u_n -\chi_n m_s].\end{aligned}\ ] ] then the solution of the least square is given by @xmath224\cdot \vec f,\quad m_\ga=\frac 1 d [ \vec a^2\vec b-(\vec a\cdot\vec b)\vec a]\cdot \vec f,\quad d = \vec b~^2 \vec a^2 -(\vec a\cdot\vec b)^2,\end{aligned}\ ] ] from the above computation , for an electron we have the typical order of magnitude for the parameters in ( [ t1.8 ] ) , @xmath225 the synchrotron radiation is much less than @xmath196 and @xmath197 , so it can be omitted .
a meaningful test strongly depends on the precision of the measurement data @xmath226 , which should be of relative errors less than @xmath227 .
how to promote the precision of the measurement is the key for the success of a test . some possible solutions and its implications : 1 .
if @xmath228 and @xmath229 , which means the einstein s mass - energy strictly holds , and the particles can not be described by the classical fields .
if @xmath228 and @xmath230 , this kind of particles has not nonlinear self - interaction , and the balance of the particles should be explained by scalar and vector interactions .
3 . if @xmath231 , which means the standard model of particles is incomplete , and some calculations in quantum field theory should be modified . so
no matter what result the experiment provides , the implication is always important and fundamental .
although the standard model of particles has achieved a lot of progresses in explanation of the behavior of micro particles , it is essentially a phenomenology .
so the test of the energy - speed relation might be a shortcut to disclose the secrets of the fundamental particles and their interactions .
we have solved the particle - like eigen solutions to some nonlinear spinor equations , and computed several functions which reflect their characteristics .
the numerical results show that , the nonlinear spinor equations have positive discrete mass spectra and anomalous magnetic moment .
the nonlinear potential and interactions yield different contributions to the total energy of the system , which can be used as fingerprints of these terms .
the magnitude of the components can be easily detected by elaborate experiments .
the weird properties of the nonlinear spinors might be closely related with the elementary particles and their interactions , so some deeper investigations on them are significant .
the author is grateful to prof .
ta - tsien li , prof .
han - ji shang and prof .
guang - jiong ni for their guidance .
thanks to prof .
yue - liang wu and dr .
ci zhuang of the chinese academy of science for encouragement of this program .
99 d. iuanenko , sov . phys . * 13 * , 141 - 149 ( 1938 ) h. weyl , phys . rev .
* 77 * , 699 - 701 ( 1950 ) w. heisenberg , physica * 19 * , 897 - 908 ( 1953 ) k. johnson , phys
78b * , 259 - 262 ( 1978 ) p. mathieu , phys . rev . *
d29 * , 2879 - 2883 ( 1984 ) a. f. ranada , _ classical nonlinear dirac field models of extended particles _ , in:_quantum theory , groups , fields and particles _ , edited by a.o.barut , amsterdam , reidel , 1983 r. finkelsten , et al , phys . rev .
* 83(2 ) * , 326 - 332(1951 ) m. soler , phys . rev .
* d1(10 ) * , 2766 - 2767(1970 ) t. cazenave , l. vazquez , comm .
phys.*105 * , 35 - 47 ( 1986 ) f. merle , j. diff . eq . * 74 * , 50 - 68 ( 1988 ) m. balabane , et al , comm .
phys . * 119 * , 153 - 176 ( 1988 ) m. balabane , et al , comm .
phys . * 133 * , 53 - 74 ( 1990 ) m. j. esteban , e. sere , c. r. acad .
pavis , * t. 319 * , serie i , 1213 - 1218 ( 1994 ) m. j. esteban , e. sere , comm . math
* 171 * , 323 - 350(1995 ) wilhelm fushchych , renat zhdanov , _ symmetries and exact solutions of nonlinear dirac equations _ , kyiv mathematical ukraina publisher , ukraine ( 1997 ) , arxiv : math - ph/0609052 y. q. gu , ph .
d. thesis , inst . of math . , fudan univ . , china,1995 y. q. gu , _ some properties of the spinor soliton _ ,
appl . cliff .
* 8(1 ) * , 17 - 29(1998 ) , http://www.clifford-algebras.org/v8/81/gu81.pdf
y. q. gu , _ spinor soliton with electromanetic field _ ,
appl . cliff .
* 8(2 ) * , 271 - 282(1998 ) , http://www.clifford-algebras.org/v8/82/gu82.pdf c. itzykson , j .- b .
zuber , _ quantum field theory _ ,
mc graw - hill inc.1980 j. keller , adv .
appl . cliff
* 3(2 ) * , 147 - 200(1993 ) y. a. rylov , adv . appl .
* 5(1 ) * , 1 - 40 ( 1995 ) y. q. gu , _ new approach to n - body relativistic quantum mechanics _ ,
a22:2007 - 2020(2007 ) , arxiv : hep - th/0610153 y. q. gu , _ mass - energy relation of the nonlinear spinor _
, hep - th/0701030 y. q. gu , _ commutative relations for the nonlinear dirac equation _ , arxiv:0802.1958 y. q. gu , t. t. li , _ eigen equation of the nonlinear spinor _
, arxiv:0704.0436v2 y. q. gu , _ a cosmological model with dark spinor source _ , int . j. mod .
a22:4667 - 4678(2007 ) , arxiv : gr - qc/0610147 y. q. gu , _ the vierbein formalism and energy - momentum tensor of spinors _ , + arxiv : gr - qc/0612106v2 y. q. gu , _ nonlinear spinors as the candidate of dark matter _ , arxiv:0806.4649v2 m. wakano , prog . theor . phys .
* 35(6 ) * , 1117 - 1141(1996 ) m. soler , phys . rev . *
d 8(10 ) * , 3424 - 3429(1973 ) a. f. baada , m. soler , phys . rev . *
d 8(10 ) * , 3430 - 3433(1973 ) a. f. baada , j. usn , l. vzquez , phys . rev . *
d 22(10 ) * , 2422 - 2424(1980 ) y. q. gu , _ the electromagnetic potential among nonrelativistic electrons _
, adv in appl .
* v9(1 ) * , 55 - 60(1999 ) , http://www.clifford-algebras.org/v9/91/gu91.htm i. l. bogolubsky , phys . lett . * 73a(2)*,87 - 90(1979 ) j. werle , acta phys . po1 . *
b12 * , 601(1981 ) a. alvarez , m. soler , phys .
lett * 50(17 ) * , 1230 - 1233(1983 ) einstein , a. ann .
18 , 639 - 641 ( 1905 ) .
s. rainville , _ et al .
, a direct test of @xmath232 _ , nature , 438 , 22 , 2005 , p.1096 . ezzat g. bakhoum ,
_ why @xmath232 emerges in the process of neutron capture _ , + arxiv:0705.3191 | in this paper , we solve the eigen solutions to some nonlinear spinor equations , and compute several functions reflecting their characteristics .
the numerical results show that , the nonlinear spinor equation has only finite meaningful eigen solutions , which have positive discrete mass spectra and anomalous magnetic moment .
the nonlinear potential and interactions yield different contributions to the total energy , and these components of the energy lead to different energy - speed relation .
the magnitude of these components can be detected by elaborate experiments .
the weird properties of the nonlinear spinors might be closely related with the elementary particles and their interactions , so some deeper investigations on them are significant .
keywords : _ nonlinear spinor field , anomalous magnetic moment , mass - energy relation _ |
when the structure of a few - body system is very stable , it may become a basic constituent of many - body systems .
for an example , the structures of light nuclei can be explained based on the cluster model , where the @xmath0-particle is a building block.@xcite another famous example is the cooper pair in condensed matter.@xcite this pair is responsible for the superconductivity . for bose - einstein condensates of spin-1 and spin-2 atoms ,
basic few - body structures have already been proposed by theorists . for spin-1 condensates ,
the interaction can be written as @xmath1 where @xmath2 is the combined spin of @xmath3 and @xmath4 and has two choices 0 and 2 .
@xmath5 is the projector of the @xmath2-channel , @xmath6 is the strength proportional to the @xmath7-wave scattering length of the @xmath2-channel .
when @xmath8 is positive , the ground state will have total spin @xmath9 ( 1 ) when the particle number @xmath10 is even ( odd).@xcite the ground state wave function @xmath11 is proportional to the pair - state @xmath12^{n/2}$ ] or @xmath13^{(n-1)/2}$ ] , where @xmath14 denotes the spin - state of a spin-1 atom , a pair of them are coupled to zero , @xmath15 is the symmetrizer ( a summation over the @xmath16 permutation terms ) . for higher states ,
say , the one with @xmath17 is proportional to @xmath18^{(n-2)/2}$ ] .
thus , the singlet pair @xmath19 appears as a common building block .
these pairs , together with a few other substructures , constitute the low - lying states . for spin-2 condensates
, the interaction can also be written as eq .
( [ e01_vij ] ) , but with @xmath20 , @xmath21 , and @xmath22 .
a detailed classification of the spin - states based on the seniority and @xmath23 was given in the ref .
@xcite . when @xmath10 is even and @xmath24 , @xmath11 is also dominated by the singlet pairs , and is proportional to the pair - state @xmath25^{n/2}$ ] , where @xmath26 denotes the spin - state of a spin-2 atom .
when @xmath10 is a multiple of 3 , and @xmath27 is negative and smaller than @xmath28 , @xmath11 is nearly proportional to the triplex - state @xmath29^{n/3}$ ] .
obviously , the triplex @xmath30 acts as a building block .
furthermore , when the parameters of interaction fall inside the indicated domain , the low - lying states are also dominated by these building blocks accompanied by a few other substructures .
it is evident that the existence of building blocks originates from the special feature of interaction .
the pairs @xmath31 and @xmath32 may appear as building blocks whenever @xmath6 is sufficiently negative . however , for spin-@xmath33 atoms with @xmath33 even , instead of the pairs , a more favorable substructure is a triplex .
let @xmath34 denote the spin - state of a spin-@xmath33 particle .
for the triplex - state @xmath35 , the symmetrizer @xmath15 is not necessary when @xmath36 because @xmath37 itself is symmetric ( i.e. , @xmath38 as can be verified by recoupling the spins ) .
it implies that every two spins are coupled to @xmath33 .
thereby the binding would be maximized if @xmath39 is sufficiently negative .
intuitively speaking , the three spins in the @xmath36 triplex will be coplanar and form a regular triangle .
this is shown in fig .
[ fig1]b , where the angle between every two spins is 120@xmath40 so that they are coupled to @xmath33 .
therefore , for spin-@xmath33 condensates , the @xmath36 triplex might serve as building blocks when @xmath39 is sufficiently negative ( note that the @xmath36 triplex is prohibited when @xmath33 is odd ) .
the ground state structure of spin-3 condensates has already been studied based on the mft.@xcite in this theory the spin structure is described by the spinors .
they will have the polar phase ( corresponding to the pair - state ) when @xmath41 is sufficiently negative , and have the cyclic phase ( corresponding to the triplex - state ) when @xmath41 becomes positive .
if a magnetic field is applied , a number of phases will emerge . instead of using the spinors ,
it is an attempt of this paper to describe the spin structures based on the basic constituents . from the experience of spin-1 and -2 condensates ,
it is expected that pairs and triplexes might also appear in spin-3 condensates when the parameters of interaction are appropriate .
we are going to search for these basic constituents . due to the difficulty in calculation ,
only small condensates ( @xmath10 is small ) are concerned .
we believe that the knowledge from small systems will help us to understand better the larger systems . due to the prohibition of the @xmath36 triplex
, only the @xmath42 triplexes could emerge in spin-3 systems .
since each @xmath43-triplex has @xmath44 magnetic components , additional complexity will arise as shown below .
let @xmath10 spin-3 atoms be confined by an isotropic and parabolic trap with frequency @xmath45 .
the interaction between particle @xmath3 and @xmath4 is @xmath46 , where @xmath47 is the dipole - dipole interaction .
let the wave function for the relative motion of @xmath3 and @xmath4 be @xmath48 , where @xmath49 is the relative orbital angular momentum , and @xmath49 the @xmath2 are coupled to @xmath50 . then , for @xmath51 , the matrix element @xmath52 is non - zero only if @xmath53 .
it implies that @xmath47 would play its role only if a @xmath54-wave spatial excitation is accompanied .
it is assumed that @xmath45 is so large that @xmath55 . in this case
the effect of @xmath47 is suppressed so that it can be neglected .
note that , in general , @xmath47 would be important in spin evolution where higher partial waves emerge .
however , this is not the case of equilibrium state in strong isotropic trap at very low temperature.@xcite furthermore , we consider a small condensate so that the size of the condensate is smaller than the spin healing length . in this case
the single spatial mode approximation ( sma ) is reasonable and is therefore adopted.@xcite let the common spatial wave function be @xmath56 . under the sma , after an integration over the spatial degrees of freedom , we arrive at a model hamiltonian @xmath57 where @xmath58 , @xmath59 . for diagonalizing @xmath60 , the set of normalized and symmetrized fock - states @xmath61
are used as basis functions , where @xmath62 is the number of particles in the @xmath63-spin - component , @xmath64 and @xmath65 , where @xmath66 is the magnetization . since @xmath67 as a whole form a complete set , once the matrix elements @xmath68 have been calculated , exact eigenenergies and eigenstates of @xmath60 can be obtained via the diagonalization .
obviously , both the total spin @xmath23 and its @xmath69-component @xmath66 are conserved when @xmath47 is neglected .
let the @xmath3-th eigenstate be denoted as @xmath70 with total spin @xmath71 and magnetization @xmath72
. the state can be expanded as @xmath73 . since one can extract a particle from a fock - state
, one can also extract a particle from @xmath70 via the expansion as @xmath74 where @xmath75 is the spin - state of a @xmath76 particle in component @xmath63 ( from @xmath77 to @xmath78 ) . with eq .
( [ e03_psii ] ) , we know that the probability of a particle in @xmath63 is just @xmath79 they fulfill @xmath80 .
@xmath81 is called the 1-body probability , and @xmath82 is just the average population of the @xmath63 component .
if one more particle is further extracted , in a similar way , we have @xmath83 we define @xmath84 they fulfill @xmath85 , @xmath86 , and @xmath87 .
when two particles are observed simultaneously , obviously , the probability of one in @xmath63 and the other one in @xmath88 is @xmath89 ( if @xmath90 ) or @xmath91 ( if @xmath92 ) .
@xmath93 is called the correlative probability of spin - components .
( [ e05_psii ] ) be rewritten as @xmath94 then , the probability of a pair of particles coupled to @xmath2 and @xmath95 is @xmath96 one can prove from symmetry that @xmath97 when @xmath98 , and the @xmath99 members @xmath100 are equal to each other when @xmath101 . furthermore , we define @xmath102 which is the probability that the spins of an arbitrary pair are coupled to @xmath2 . in general , @xmath103 can provide information on the possible existence of two - body substructures as shown below .
as an example , let the normalized pair - state be denoted as @xmath104^{n/2 } , \label{e10_psipolar}\ ] ] where @xmath105^{-1/2}$ ] is the constant for normalization ( refer to appendix i ) .
since @xmath106^{n/2 } & = & \sum_{s , m_s } ( \chi(1)\chi(2))_{s , m_s}b_{s , m_s } \nonumber \\ & & \tilde{p}_{n-2 } ( \chi\chi)_{s , m_s } [ ( \chi\chi)_0]^{(n-4)/2 } , \label{e11_pn}\end{aligned}\ ] ] where @xmath107 by making use of the formulae in appendix i , one can prove that all the @xmath100 of the pair - state are equal to @xmath108 with the only exception @xmath109 .
from these data , @xmath103 of the pair - state , denoted as @xmath110 , are listed in tab .
[ tab1 ] .
.@xmath110 , the probability that a pair of particles in the polar state are coupled to @xmath2 . [ cols="<,<,<,<,<",options="header " , ]
instead of using the mft , a language from the few - body theory is used in this paper .
we have shown theoretically the existence of stable 2-body and 3-body structures as building blocks in small spin-3 condensates .
the ratios @xmath111 and @xmath112 defined in this paper are crucial in the search of these basic constituents . the reason leading to the appearance of these constituents is explained based on the feature of interaction . whereas , in the mft , the physics underlying the appearance of a specific spinor is not easy to clarify . the calculation in this paper concerns only small spin-3 condensates ( @xmath113 ) . for spin-2 condensates ,
it has been proved theoretically that the fact that pairs and triplexes appear as building blocks does not depend on @xmath10 ( in fact , the picture of the triplexes would become even clearer when @xmath114 @xcite ) .
it has also been proved that the existence of the pairs in spin-3 condensates does not depend on @xmath10.@xcite thus the existence of triplexes as building blocks in large spin-3 condensates is very probable , nonetheless it deserves a further study .
the support from the nsfc under the grant 10874249 is appreciated .
for the pair - state and odd - pair - state the following equations relating a @xmath10-body and a @xmath115-body systems are very useful : @xmath116^{n/2 } | \tilde{p}_n[(\chi\chi)_0]^{n/2 } \rangle , \nonumber \\ & = & \frac{1}{7 } n^2 ( n-1 ) ( n+5 ) \mathcal{n}_0^{(n-2 ) } , \\
\label{ea2 } \mathcal{n}_{\mathrm{odd}}^{(n ) } & \equiv & \langle \tilde{p}_n \chi_{\mu } [ ( \chi\chi)_0]^{(n-1)/2 } | \tilde{p}_n \chi_{\mu } [ ( \chi\chi)_0]^{(n-1)/2 } \rangle \nonumber \\ & = & \frac{1}{7 } n ( n-1)^2 ( n+6 ) \mathcal{n}_{\mathrm{odd}}^{(n-2 ) } , \\ \label{ea3 } \mathcal{n}_{s , m_s}^{(n ) } & \equiv & \langle \tilde{p}_n ( \chi\chi)_{s m_s } [ ( \chi\chi)_0]^{(n-2)/2 } | \tilde{p}_n ( \chi\chi)_{s m_s } \nonumber \\ & & [ ( \chi\chi)_0]^{(n-2)/2 } \rangle \nonumber \\ & = & \frac{1}{7 } n ( n-1 ) ( n-2 ) ( n+7 ) \mathcal{n}_{s , m_s}^{(n-2)},\end{aligned}\ ] ] where @xmath15 denotes a summation over the @xmath16 permutation terms , eq .
( [ ea3 ] ) holds only if @xmath117 ( if @xmath20 , then eq . ( [ ea1 ] )
should be used ) .
making use of these equations related matrix elements can be derived via iteration . for examples
, the constant of normalization @xmath118 can be obtained from eq .
( [ ea1 ] ) , @xmath119 from eq .
( [ ea2 ] ) .
one can extract a particle ( say , particle 1 ) from an odd - pair - state as @xmath120^k & = & \sum_{\nu } \chi_{\nu}(1 ) \ { \delta_{\mu,\nu } \tilde{p}_{n-1 } [ ( \chi\chi)_0]^k \nonumber \\ & & -(n-1 ) \frac{(-1)^{\nu}}{\sqrt{7 } } \sum_s c_{3\mu,3,-\nu}^{s,\mu-\nu } \nonumber \\ & & \tilde{p}_{n-1 } ( \chi\chi)_{s,\mu-\nu } [ ( \chi\chi)_0]^{k-1 } \ } , \label{ea4}\end{aligned}\ ] ] where @xmath121 , the clebsch - gordan coefficients have been introduced , and only even @xmath2 are included in the summation .
then , from the definition of the 1-body probability , we have @xmath122 it is mentioned that the expression of @xmath123 depends on whether @xmath2 is zero or nonzero . with this in mind , after a simplification , eq .
( [ ea5 ] ) leads to the expressions given in the text .
one can extract two particles ( say , 1 and 2 ) from a pair - state as @xmath124^{n/2 } & = & n \sum_{\mu\nu } \chi_{\mu}(1 ) \chi_{\nu}(2 ) \ { \delta_{\mu,-\nu } c_{3\mu,3\nu}^{0,0 } \tilde{p}_{n-2 } [ ( \chi\chi)_0]^{(n-2)/2 } \nonumber \\ & & + ( n-2 ) \sum_s ( 2s+1 ) u \left ( \begin{array}{ccc } 3 & 3 & 0 \\ 3 & 3 & 0 \\ s & s & 0 \end{array } \right ) c_{s,\mu+\nu , s,-\mu-\nu}^{0,0 } c_{3\mu,3\nu}^{s,\mu+\nu } \tilde{p}_{n-2 } ( \chi\chi)_{s,-\mu-\nu } [ ( \chi\chi)_0]^{(n-4)/2 } \ } , \label{ea6}\end{aligned}\ ] ] where the clebsch - gordan and @xmath125-coefficients have been introduced , and only even @xmath2 are included in the summation . then , from the definition of the correlative probability , we have @xmath126 . \label{ea7}\end{aligned}\ ] ] the suppression of @xmath47 in strong isotropic trap can also be seen under the frame work of mft . in this theory , the effect of @xmath47 is embodied by the factor @xmath127 , where @xmath128 is the strength and is usually small ( @xmath129 for @xmath130cr ) , and @xmath131 , where @xmath132 is the spatial wave function of the @xmath3-th particle , @xmath133 , and @xmath134 .
@xcite a part of the integrand can be in general expanded as @xmath135 , where @xmath136 and @xmath137 denote the partial waves and they are coupled to @xmath138 . since @xmath139 is close to be isotropic when the trap is strong , the @xmath140 terms can be neglected .
this leads to @xmath141 .
thus the effect of @xmath47 is suppressed . | it was found that stable few - body spin - structures , pairs and triplexes , may exist as basic constituents in small spin-3 condensates , and they will play the role as building blocks when the parameters of interaction are appropriate .
specific method is designed to find out these constituents . |
in recent years , extensive efforts have been made to understand the decay of light di - nuclear systems ( a@xmath460 ) formed through low - energy heavy - ion reactions @xcite . in most of the reactions
studied , the properties of the observed fully energy damped yields have been successfully explained in terms of either a fusion - fission ( ff ) mechanism or a deep - inelastic ( di ) orbiting mechanism behavior .
the strong resonance - like structures observed in elastic and inelastic excitation functions of @xmath5mg+@xmath5 mg and @xmath0si+@xmath0si have indicated the presence of shell stabilized , highly deformed configurations in the @xmath6cr and @xmath7ni compound systems respectively @xcite .
the present work aims to investigate the possible occurence of highly deformed configurations in the @xmath8ca di - nucleus produced in the @xmath0si+@xmath2c reaction through the study of light charged particle ( lcp ) emission .
the experiment was performed at the ires strasbourg vivitron tandem facility using 112.6 mev @xmath0si beams on a @xmath2c(160 @xmath9g/@xmath10 ) target . both the heavy ions and their associated lcp s
were detected using the * icare * charged particle multidetector array @xcite .
the heavy ions were detected in eight telescopes , each consisting of an ionisation chamber ( ic ) followed by a 500 @xmath9 m si detector .
the in - plane coincident lcp s were detected using four triple telescopes ( si 40 @xmath9 m , si 300 @xmath9 m , 2 cm csi(tl ) ) , 16 double telescopes ( si 40 @xmath9 m , 2 cm csi(tl ) ) and two double telescopes ( ic , si 500 @xmath9 m ) located at the most backward angles .
typical inclusive @xmath11 energy spectra are shown in fig.1.a .
the data analysis was performed using cacarizo @xcite , the monte carlo version of the statistical - model code cascade .
the angular momenta distribution , needed as the main input to constrain the calculation , was taken from @xmath0si+@xmath2c complete fusion data @xcite .
the other ingredients such as the nuclear level densities and the barrier transmission coefficients , are usually deduced from the study of the evaporated lcp spectra .
standard statistical - model calculations are not able to reproduce the shape of experimental @xmath11-particle energy spectra satisfactorily @xcite .
several attempts have been made to explain this anomaly either by changing the emission barrier or by using a spin - dependent level density .
in hot rotating nuclei forme density at higher angular momentum should be spin dependent . in cacarizo
, the level density , @xmath12 , for a given angular momentum @xmath13 and energy @xmath14 is given by the well known fermi gas expression : @xmath15^{1/2}),\ ] ] where @xmath16 is the level density parameter , @xmath17 is the pairing correction , @xmath18 and @xmath19 with @xmath20 the rigid body moment of inertia and @xmath21 the deformability parameters . by changing the deformability parameters one can simulate the deformation effects on the level densities . for the @xmath0si + @xmath0si reaction @xcite
, the shape of the inclusive and exclusive @xmath11 energy spectra are well reproduced by using large deformation effects @xcite .
similarly the experimental inclusive @xmath11 energy spectra for @xmath0si + @xmath2c of fig.1.a are better described by using deformation effects ( dotted lines ) than with the standard liquid - drop deformation ( dashed lines ) .
+ the exclusive energy spectra of @xmath11-particle in coincidence with individual s and p er s shown in figs.1.b and 1.c are quite interesting .
the dotted lines are the predictions of cacarizo using non - zero values of the deformability parameters .
the energy spectra associated with s are completely different from those associated with p. the latter are reasonably well reproduced by the cacarizo curves whereas the model could not predict the shape of the spectra obtained in coincidence with s ( fig.1.b ) .
this is due to the fact that an additional component might be significant in this case .
one could suggest the hypothesis of a contribution arising from the decay of unbound @xmath22be produced in a binary reaction @xmath8ca @xmath23 @xmath24s+@xmath22be . in order to determine the sources of both the @xmath11 emission and @xmath22be breakup the invariant cross sections in coincidence with p and s are plotted in fig.2 in the ( v@xmath25,v@xmath26 ) plane .
fig.2.b shows the invariant cross sections in coincidence with p which maxima are centered on the compound nucleus velocity as expected for a fusion - evaporation mechanism .
fig.2.a presents two additional contributions ( in circles ) for angles close to 30@xmath27 and 60@xmath27 arising from the binary decay of unbound @xmath3be .
this conclusion is also consistent with the er kinematical analysis of the s and p exclusive energy spectra ( not shown here ) .
the question of the real nature ( ff or orbiting ) of this decay process remains to be explored .
the @xmath11-particle energy spectra measured in coincidence with s have an additional component which may come from the decay of @xmath3be , which is unbound and produced through the binary decay of @xmath8ca @xmath28s + @xmath3be .
work is in progress to analyse the proton energy spectra as well as the in - plane angular correlations of both the @xmath11-particles and the protons .
sanders , a. szanto de toledo , and c. beck , _ phys .
rep . _ * 311 * , 487 ( 1999 ) .
t. bellot , ph.d . thesis , strasbourg university , report * ires 97 - 34*. g. viesti , b. fornal , d. fabris , k. hagel , j.b .
natowitz , g. nebbia , g. prete , and f. trotti , phys .
c * 38 * , 2640 ( 1988 ) .
harmon , b.a .
harmon , s.t .
thornton , d. shapira , j. gomez del campo , m. beckerman , _ phys .
_ c * 34 * , 552 ( 1986 ) .
vineyard , j.f .
mateja , c. beck , s.e .
atencio , l.c .
dennis , a.d .
frawley , d.j .
henderson , r.v.j .
janssens , k.w .
kemper , d.g .
kovar , c.f .
maguire , s.j .
padalino , f.w .
prosser , g.s.f .
stephans , m.a .
tiede , b.d .
wilkins , and r.a .
zingarelli , _ phys .
_ c * 47 * , 2374 ( 1993 ) . c. bhattacharya , m. rousseau , c. beck , v. rauch , s. belhabib , a. dummer , r.m .
freeman , a. hachem , r. nouicer , d. mahboub , e. martin , s.j .
sanders , o. stezowski , and a. szanto de toledo , _ nucl .
* a * , ( 1999 ) _ in press_. | inclusive as well as exclusive energy spectra of the light charged particles emitted in the @xmath0si ( @xmath1=112.6 mev ) + @xmath2c reaction has been measured using the * icare * multidetector array .
the data have been analysed by statistical - model calculations using a spin - dependent level density parametrization .
the results suggest significant deformation effects at high spin and cluster emission of @xmath3be .
2.0 cm pacs numbers : 25.70 |
it is well known that the fundamental particles that form hadronic matter are the quarks and the gluons , whose interactions are described by the lagrangian of quantum chromo - dynamics ( qcd ) . however , exact qcd - based calculations can not yet be performed to explain the properties of hadrons in terms of their constituents .
one has to resort to phenomenological functions to interpret experimental measurements in order to understand how qcd works at the `` long distances '' at play when partons are confined in nucleons .
typical examples of such functions include form factors and parton distributions .
generalized parton distributions ( gpds ) , which unify and extend the information carried by both form factors and parton distributions , are nowadays the object of an intense effort of research , for a more complete understanding of the structure of the nucleon .
the gpds describe the correlations between the longitudinal momentum and the transverse position of the partons inside the nucleon , they give access to the contribution of the orbital momentum of the quarks to the nucleon spin , and they are sensitive to the correlated @xmath5 components .
the original articles , general reviews on gpds and details on the formalism can be found in refs .
@xcite .
the gpds are universal nucleon - structure functions , which can be accessed experimentally via the exclusive leptoproduction of a photon ( dvcs , deeply virtual compton scattering ) or of a meson from the nucleon at high momentum transfer .
more precisely , the virtuality @xmath6 of the photon exchanged with the nucleon by the initial lepton , defined as @xmath7 where @xmath8 and @xmath9 are the four momenta of , respectively , the incoming ( @xmath10 ) and scattered electron ( @xmath11 ) , must be sufficiently large for the reaction to happen at the quark level .
figure [ fig : dvcs ] illustrates the leading - twist @xcite process for dvcs , also called the `` handbag mechanism '' , on a proton target .
the virtual photon interacts with one of the quarks of the proton , which propagates radiating a real photon .
@xmath12 is the squared four - momentum transfer between the initial ( @xmath13 ) and final proton ( @xmath14 ) . @xmath15 and @xmath16 are the longitudinal momentum fractions of the quark before and after radiating the real photon , respectively .
@xmath17 is defined as @xcite : @xmath18 which at leading twist ( @xmath19 ) becomes @xmath20 where @xmath21 is the bjorken scaling variable @xmath22 @xmath23 is the proton mass , and @xmath24 in the bjorken limit , defined by high @xmath6 , high @xmath25 and fixed @xmath21 , the dvcs process can be factorized into a hard - scattering part ( @xmath26 ) that can be treated perturbatively , and a soft nucleon - structure part , described by the gpds . at leading - order qcd and at leading twist , considering only quark - helicity conserving quantities and the quark sector , the dvcs process is described by four gpds , @xmath27 , which account for the possible combinations of relative orientations of nucleon spin and quark helicity between the initial and final state .
the gpds depend upon the variables @xmath28 , @xmath17 and @xmath29 .
the fourier transform , at @xmath30 , of the @xmath29 dependence of a gpd provides the spatial distribution in the transverse plane for partons having a longitudinal momentum fraction @xmath28 @xcite .
model - independent sum rules link the first moment in @xmath28 of the gpds to the elastic form factors ( ffs ) @xcite : @xmath31 where @xmath32 and @xmath33 are the dirac and pauli ffs , and @xmath34 and @xmath35 are the axial and pseudoscalar ffs .
among the three variables @xmath28 , @xmath17 and @xmath29 , which appear in the dvcs formalism , only two , @xmath17 and @xmath29 , are experimentally accessible , since @xmath28 appears only in the quark loop and is integrated over .
indeed , the dvcs amplitude is proportional to sums of integrals over @xmath28 of the form : @xmath36\ ] ] where @xmath37 represents a generic gpd and the top and bottom signs apply , respectively , to the quark - helicity independent , or unpolarized , gpds ( @xmath38 ) and to the quark - helicity dependent , or polarized , gpds ( @xmath39 ) .
each of these 4 integrals , which are called compton form factors ( cffs ) , can be decomposed into their real and imaginary parts , as @xmath40f(x,\xi , t ) \\
\im{\rm m}{\cal f}(\xi , t)&= & -\pi [ f(\xi,\xi , t)\mp f(-\xi,\xi , t ) ] , \end{aligned}\ ] ] where @xmath41 is cauchy s principal value integral . the information that can be extracted from the experimental data at a given ( @xmath42 ) point depends on the observable involved .
when measuring observables sensitive to the real part of the dvcs amplitude , such as double - spin asymmetries , beam - charge asymmetries or unpolarized cross sections , the real part of the cff , @xmath43 , is accessed . when measuring observables sensitive to the imaginary part of the dvcs amplitude , such as single - spin asymmetries or cross - section differences , the imaginary part of the cff , @xmath44 , can be obtained . however , knowing the cffs does not define the gpds uniquely .
a model input is necessary to deconvolute their @xmath28 dependence .
the dvcs process is accompanied by the bethe - heitler ( bh ) process , in which the final - state photon is radiated by the incoming or scattered electron and not by the nucleon itself .
the bh process , which is not sensitive to the gpds , is experimentally indistinguishable from dvcs and interferes with it .
however , considering that the nucleon form factors are well known at small @xmath29 , the bh process is precisely calculable .
it is clearly a non - trivial task to measure the gpds .
it calls for a long - term experimental program comprising the measurement of different observables .
such a dedicated experimental program , concentrating on a proton target , has started worldwide in the past few years .
jefferson lab ( jlab ) has provided the first measurement , in the valence region , of beam - polarized and unpolarized dvcs cross sections at hall a @xcite , providing a @xmath6-scaling test that supports the validity of the leading - order , leading - twist handbag mechanism starting at values of @xmath6 of 1 - 2 ( gev/@xmath45)@xmath46 . hall b provided pioneering measurements of beam @xcite and target @xcite spin asymmetries with the clas detector @xcite , and afterwards obtained beam - spin asymmetries ( bsa ) over a large kinematic range and with high statistics @xcite .
beam - charge asymmetries , longitudinally and transversely polarized target - spin asymmetries ( tsas ) , as well as double - spin asymmetries ( dsas ) , have also been measured by the hermes collaboration @xcite .
this paper is focused on the extraction of longitudinal tsas , bsas and dsas for proton dvcs over a wide phase space using dedicated data taken at jefferson lab with the clas detector .
an analysis of dvcs observables , including the asymmetries of interest in this work , in terms of fourier moments with respect to the azimuthal angle , was carried out by belitsky _
@xcite , up to a twist-3 approximation .
these asymmetries allow one to extract separate components of the angular dependence of the @xmath47 cross section , that are related to the compton form factors , defined in eq . .
the five - fold cross section for exclusive real photon electroproduction @xcite @xmath48 depends on @xmath21 , @xmath29 , the lepton energy fraction @xmath49 , with @xmath50 , @xmath51 , and , in general , two azimuthal angles @xmath52 and @xmath53 .
the observables of interest here do not depend on @xmath53 , the azimuthal angle of the scattered electron in the laboratory frame .
the angle @xmath52 is formed by the leptonic and hadronic planes , as shown in fig .
[ fig : dvcs_phi ] .
the charge of the electron is denoted with @xmath10 and @xmath54 represents the fine - structure constant . ,
formed by the leptonic and hadronic planes.,width=325 ] the amplitude @xmath55 is the coherent sum of the dvcs and bethe - heitler amplitudes ( @xmath56 , @xmath57 ) : @xmath58 with the interference term defined as @xmath59 the azimuthal angular dependence of each of the three terms in eq .
is given by @xcite : @xmath60,\end{aligned}\ ] ] @xmath61\ } \
, , \end{aligned}\ ] ] @xmath62 \ } \ , , \end{aligned}\ ] ] where @xmath63 and @xmath64 are intermediate lepton propagators ( for more details and definitions , see @xcite ) . among the fourier coefficients
@xmath65 , @xmath66 appearing in the previous expansions , the ones appearing in @xmath67 ( p = bh ) depend on the well - known electromagnetic ffs , while the ones appearing in @xmath68 ( p = dvcs ) and @xmath69 ( p = @xmath69 ) depend on the compton form factors , the latter linearly . in the next sections ,
the sensitivity to the various cffs of the dvcs spin observables presented in this paper will be outlined .
the use of a longitudinally polarized ( lp ) target allows the extraction of the target - spin asymmetry @xmath70 , where the first letter in the subscript refers to the beam polarization ( `` unpolarized '' , in this case ) and the second to the target polarization ( `` longitudinally polarized '' , in this case ) , and which is given at twist-2 by : @xmath71 where the ellipses in the denominator represent smaller , higher - twist terms , and the @xmath72 subscript stands for `` unpolarized '' .
the @xmath73 coefficient @xmath74 , originating from the dvcs / bh interference term , at leading twist is proportional to a linear combination of the imaginary parts of the four cffs ( eq . ) , @xmath75 ~.\end{aligned}\ ] ] due to the relative values of the proton form factors @xmath76 and @xmath77 , and given the typical jlab kinematics , the coefficients in eq .
enhance the contribution to @xmath70 coming from @xmath78 over the ones from other cffs
. however , given that @xmath79 is expected to be about twice as big as @xmath80 , @xmath70 will also be sensitive to @xmath79 . in the kinematical range of the data presented here
, higher - twist effects would appear in eq .
as additional @xmath52-dependent terms , the dominant of which is a @xmath81 term in the numerator @xcite .
the expression at twist - two of the beam - spin asymmetry is @xmath82 where @xmath83.\ ] ] thus , the beam - spin asymmetry for dvcs on a proton target is particularly sensitive to the imaginary part of the cff of the unpolarized gpd @xmath3 . as for the tsa , more
terms must be added if one goes beyond the leading - twist approximation , the larger of which is a @xmath84 term in the numerator . as mentioned in section [ sec_intro ] , the slope in @xmath29 of the gpds
is related via a fourier transform to the transverse position of the struck parton .
therefore , a measurement of @xmath78 and @xmath79 provides , respectively , information on the axial and electric charge distributions in the nucleon as a function of @xmath21 ( see eq . ) . a measurement of both the tsa and bsa at the same kinematics is needed to truly distinguish between the two contributions .
the use of a polarized electron beam along with a longitudinally polarized target allows also the determination of the double - spin asymmetry @xmath85 . unlike @xmath70 ,
the bethe - heitler process alone can generate a double - spin asymmetry . at twist-2 level , this observable takes the form : @xmath86 with @xmath87,\end{aligned}\ ] ] where terms depending on powers of @xmath88 were neglected . in this expression , the interference terms , related to the gpds , are expected to be smaller than the known bh terms @xcite .
moreover , both the constant and the @xmath89-dependent terms contain contributions from both bh and the dvcs / bh interference . nonetheless , given the fast variation of bh depending on the kinematics , it is important to sample @xmath85 over a wide phase space to find possible regions of sensitivity to @xmath90 and @xmath91 , which dominate in eq .
the data were taken in hall b at jefferson lab from february to september 2009 , for a total of 129 days .
a continuous polarized electron beam was delivered by the cebaf accelerator onto a solid ammonia target , polarized along the beam direction .
frozen beads of paramagnetically doped @xmath92nh@xmath93 , kept at temperatures of about 1k and in a 5 t magnetic field , were dynamically polarized by microwave irradiation @xcite .
the target was 1.45 cm long and 1.5 cm in diameter .
the target system included a @xmath94he evaporation refrigerator and a superconducting split - coil magnet .
the magnet , which provided a uniform polarizing field ( @xmath95 ) and at the same time focused the low - energy mller electrons toward the beam line , was inserted in the center of the clas detector @xcite .
clas was a spectrometer , based on a toroidal magnetic field , providing a wide acceptance .
the clas magnetic field was generated by six superconducting coils arranged around the beamline to produce a field pointing primarily in the azimuthal direction .
the particle detection system consisted of drift chambers ( dc ) to determine the trajectories and the momenta of charged particles curved by the magnetic field , gas erenkov counters ( cc ) for electron identification , scintillation counters for measuring time - of - flight ( tof ) and electromagnetic calorimeters ( ec ) to detect showering particles ( electrons and photons ) and neutrons .
the segments were individually instrumented to form six independent magnetic spectrometers with a common target , trigger , and data - acquisition system .
an additional detector , the inner calorimeter , constructed for a previous dvcs - dedicated experiment @xcite to complete the photon acceptance at low polar angles ( from 4@xmath96 to 15@xmath96 ) , was placed at the center of clas .
figure [ setup ] shows the whole setup , including the polarized target , clas and the ic . a totally absorbing faraday cup ( fc ) , downstream of clas , was used to determine the integrated beam charge passing through the target .
the data presented here come from the first two of the three parts in which the experiment was divided , and for which a @xmath1nh@xmath2 target was used .
the main differences between these two parts ( referred to as a and b ) were the target position with respect to the center of clas ( @xmath97 cm for part a and @xmath98 cm for part b ) and the electron - beam energy ( @xmath99 gev for part a and @xmath100 gev for part b ) .
the beam energy values were obtained from elastic - scattering analysis on these data , using the @xmath1nh@xmath2 target .
the beam was rastered over the target in a spiral motion in order to assure a homogeneous depolarization over the whole volume of the target .
the beam polarization was frequently monitored in mller runs , via the measurement of the asymmetry of elastic electron - electron scattering .
the target polarization was continuously monitored by a nuclear magnetic resonance ( nmr ) system
. runs on a carbon target were taken periodically throughout the duration of the experiment for unpolarized - background studies .
the selection of the data to be analyzed was done by monitoring the stability over time of faraday - cup normalized yields per sector .
for the results presented here , runs taken with the @xmath1nh@xmath2 and carbon targets from parts a and b were used .
the two parts were analyzed separately and the final results were found by combining the results from the two parts , as will be described in section [ sec_merge ] .
this paper reports on the extraction of three kinds of asymmetries , the experimental definitions of which are given here .
in all of the formulae below , the first sign in the superscript on the number of normalized dvcs / bh events @xmath101 is the beam helicity ( @xmath102 ) and the second sign is the target polarization ( @xmath29 ) .
@xmath101 is obtained from @xmath103 events ( @xmath104 ) , normalized by the corresponding faraday - cup charge ( @xmath105 ) after subtraction of the @xmath106 background as follows : @xmath107 where @xmath108 is the relative @xmath106 contamination , outlined in section [ sec_pi0_back ] .
the beam - spin asymmetry is calculated as : @xmath109 where @xmath110 is the polarization of the beam and @xmath111 are the values of the polarization of the target for its two polarities .
the target - spin asymmetry is computed as : @xmath112 where @xmath113 @xmath114 is the dilution factor to account for the contribution of the unpolarized background ( section [ sec_dilution ] ) and @xmath115 represents a correction applied to define the target - spin asymmetry with respect to the virtual - photon direction ( section [ transverse_sec ] ) .
the double ( beam - target ) spin asymmetry is obtained as : @xmath116 where @xmath117 and @xmath118 is the analog of @xmath119 for the double - spin asymmetry ( section [ transverse_sec ] ) .
+ in the following , the steps leading to the extraction from the data of all the terms composing these asymmetries will be presented .
the final state was selected requiring the detection of exactly one electron and one proton , and at least one photon .
the electrons were selected among all the negative tracks with momenta above 0.8 gev/@xmath45 , but requiring that their energy deposited in the inner layers of the ec @xcite be greater than 0.06 gev , in order to reject negative pions , and that their hits in the cc and in the sc be matched in time .
fiducial cuts were also applied to eliminate the events at the edges of the ec ( where the energy of a particle can not be correctly reconstructed because a large part of the induced electromagnetic shower is lost ) , and to remove the `` shadow '' of the ic , which limits the clas acceptance for charged particles .
the effect of these cuts on the distribution of the total energy deposited in the ec divided by the momentum is shown as a function of the momentum in fig .
[ etot_p_before_after ] . the comparison of the top and middle plots of fig .
[ cc_nphe ] shows the effect of these same cuts on the number of cc photoelectrons .
most of the events in the single - photoelectron peak , which come from either pions or electronic noise in the pmts , are removed by our particle identification ( pid ) cuts for electrons .
the rest of the background is eliminated by the exclusivity cuts applied afterwards as explained in section [ sec_excl_cuts ] ( fig .
[ cc_nphe ] , bottom ) . , divided by the particle momentum @xmath13 as a function of @xmath13 for all the negative tracks .
top : negative charged particles , before cuts .
bottom : after minimum - momentum , @xmath120 , fiducial and timing cuts.,width=325 ] exclusivity cuts.,width=325 ] the protons were selected from the correlation between their velocity , deduced from the time - of - flight measurement using the sc , and the proton momentum as measured by the dcs ( fig .
[ p_beta ] ) .
fiducial cuts on @xmath121 and @xmath52 were also applied in order to remove the shadow of the ic . as a function of @xmath13 for positively charged particles .
the black lines represent the cut applied to select protons.,width=325 ] for the photons , two different sets of cuts were adopted , depending on whether the photon was detected in the ic or in the ec . for the ic case , a low - energy threshold of 0.25 gev and a cut on @xmath121 versus @xmath122 to remove the background coming from mller electrons ( fig .
[ ic_photon_en ] ) were applied , as well as fiducial cuts on @xmath28 and @xmath123 , to remove the outer and inner edges of the detector , where clusters could not be fully reconstructed . as a function of the reconstructed energy for ic hits , showing the cut on the minimum energy at 0.25 gev , as well as the `` triangular '' cut to remove the low - energy / low-@xmath121 accidental background , applied to select photons.,width=325 ] for the ec case , all neutrals passing a low - energy threshold of 0.25 gev and having @xmath124 ( fig .
[ ph_ec_beta ] ) to select the in - time photons were retained .
fiducial cuts as for the electrons were also adopted to remove the edges of the detector and ic - frame cuts were applied to remove the shadow of the ic over the ec . vs energy for neutrals measured by time - of - flight with ec .
the events for which @xmath124 ( black line ) were retained as photon candidates.,width=325 ] monte - carlo simulations were used in this analysis for the three following tasks : * determination of angular and momentum corrections for electrons and protons to compensate for multiple scattering and energy losses in the target and detector materials , as well as for imperfections in the trajectory reconstruction ; * optimization of the dvcs and @xmath125 exclusivity cuts ( sections [ sec_excl_cuts ] and [ sec_pi0_back ] , respectively ) ; * evaluation of the @xmath125 contamination in the @xmath103 event sample ( section [ sec_pi0_back ] ) .
the two sets of generated events , dvcs / bh and @xmath125 ( section [ sec_dvcs_gen ] and [ sec_pi0_gen ] ) , were fed to the clas geant3-based simulation package ( gsim ) to simulate the response of the clas detector .
the output of gsim was then fed to a post - processing code ( gpp ) that simulates dead or inefficient dc wires and smears the dc and tof resolutions to more closely agree with the experiment .
the output of gpp was finally fed to the clas reconstruction package , and reduced ntuples and root files were obtained from the reconstructed files in the same way as was done in the data processing . a dvcs / bh event - generator code , which produced @xmath103 events according to the formalism of belitsky _ et al . _
@xcite , was used for the monte - carlo simulation .
figure [ data_mc_dvcs_kinvars ] shows a comparison of the distributions of the relevant kinematic variables for the data ( black lines ) and the monte - carlo simulation ( shaded areas ) . here ,
pid and @xmath103 exclusivity cuts , which will be described in section [ sec_excl_cuts ] , were included for both the data and the monte carlo .
the agreement between data and simulation is quite good , especially given the purposes of the simulations in these analyses : they are not used for absolute acceptance corrections , but only to help in the determination of cut widths and for background subtraction .
slight differences between data and mc , especially visible in the high-@xmath29 and central-@xmath52 region , are coming from events in the ec topology .
this is probably due to the fact that the data , unlike the mc , are not only pure @xmath103 events , but are contaminated by exclusive @xmath106 events .
the fact that the @xmath125 contamination is larger in the ec topology , as will be reiterated in section [ sec_excl_cuts ] , can explain the data / mc discrepancies .
, @xmath126 , @xmath21 and @xmath52 are plotted .
the histograms are normalized to each other via the ratio of their integrals.,width=325 ] exclusive @xmath125 events were generated using a code for meson electroproduction that included a parametrization of the @xmath125 differential cross sections that have recently been measured by clas @xcite .
figure [ data_mc_pi0_kinvars ] shows a comparison of the distributions of the relevant kinematic variables for the data ( black ) and the monte - carlo simulation ( shaded areas ) . here ,
pid and @xmath125-exclusivity cuts , which will be described in section [ sec_pi0_back ] , were included for both the data and the monte carlo .
the agreement between data and simulation is quite good
. channel .
comparison of data ( black lines ) and monte - carlo simulation ( shaded areas ) . starting from the top left , @xmath6 , @xmath126 , @xmath21 and @xmath52 are plotted .
the histograms are normalized to each other via the ratio of their integrals.,width=325 ] after selecting events with exactly one electron and one proton , and at least one photon , and having applied the momentum corrections , further cuts need to be applied to ensure the exclusivity of the @xmath103 final state .
two kinds of backgrounds need to be minimized : the nuclear background coming from scattering off the nitrogen of the @xmath1nh@xmath2 target , and the background coming from other channels containing an electron , a proton and at least one photon in the final state .
having measured the four - momenta of the three final - state particles , one can construct several observables , hereafter referred to as `` exclusivity variables '' , on which cuts can be applied to select the @xmath103 channel . here ,
the following quantities were studied : * the squared missing mass of the @xmath127 system , mm@xmath128 ; * the angle @xmath129 between the measured photon and the calculated photon , using the detected electron and proton together with @xmath130 kinematics ; * @xmath131 , the difference between two ways to compute the angle @xmath52 between the leptonic and the hadronic plane .
the two ways concern the definition of the normal vector to the hadronic plane : one is via the cross product of the momentum vectors of the proton and the real photon , and the other one is via the cross product of the momentum vectors of the proton and the virtual photon . for the @xmath103 final state
, @xmath131 should be distributed as a gaussian centered at zero , with width determined by the experimental resolution ; * @xmath132 , the transverse component of the missing momentum of the reaction @xmath133 , given by @xmath134 , in the laboratory frame .
the definition of the exclusivity cuts is quite a delicate step in the dvcs analysis .
it is important that the cuts are determined in a consistent way for the data and for the monte - carlo simulation , because the latter will be used to evaluate the @xmath106 background contamination . in this analysis
it was chosen to define each exclusivity cut by fitting the corresponding variable with a gaussian and cutting at @xmath135 around the fitted mean .
this procedure was done separately for the data and for the simulation .
this way , the same fraction of events was kept , for both data and simulation .
the exclusivity variables to be fitted were plotted after applying _ preliminary _ cuts that included : * kinematic cuts to be above the region of the nucleon resonances and to approach the regime of applicability of the leading - twist gpd formalism : @xmath136 ( gev/@xmath45)@xmath46 , @xmath137 , and @xmath138 gev/@xmath139 ; ( where @xmath140 ) * @xmath141 gev , since the real photons of interest are expected to have high energy ; * @xmath142 cut around the mean of mm@xmath128 to eliminate from the experimental data the background from channels other than @xmath103 or @xmath125 ( visible in the top left plot of figs . [ ic_cuts ] and [ ec_cuts ] , where peaks from @xmath143 and @xmath144/@xmath145 are evident ) . in order to eliminate broadening on the widths of the peaks due to events from electron scattering on the nitrogen , the fits to the exclusivity variables were done on the spectra obtained after subtracting carbon data from the @xmath1nh@xmath2 data .
the two datasets were normalized to each other via the ratio of their faraday - cup counts multiplied by a constant that accounts for different densities of materials for the two target types ( see section [ sec_dilution ] ) .
the method to define the cuts described above was adopted for the topology where the photon was detected in the ic , since the comparison with monte carlo showed that these data are strongly dominated by the dvcs / bh channel .
figure [ ic_cuts ] shows , for the ic topology , the effects of the exclusivity cuts , which appear successful both in extracting quite a clean @xmath103 final state ( shaded areas ) and in minimizing the background originating from the nitrogen part of the target ( black areas ) . a different strategy was found to be necessary for the ec case , which displayed , before cuts , a larger contribution from @xmath125 events .
the peaks in the exclusivity variables for the data in this topology are very broad , when visible , and not necessarily produced by dvcs / bh candidates . in the first plot of fig .
[ ec_cuts ] , for example , the distribution of the squared missing mass of the @xmath127 system is shown for the ec case .
as is clearly visible , the peak of the distribution is not centered at zero but around the squared @xmath106 mass , indicating a significant contamination from the exclusive @xmath106 events that will be subtracted later through the procedure described in section [ sec_pi0_back ] .
thus it was decided , for the ec topology , not to fit the distributions of the exclusivity variables to extract cut means and widths . instead it was chosen to fit only the peaks of the dvcs / bh monte - carlo simulations .
to correct for the discrepancies in resolutions between data and simulation , the widths of the various exclusivity variables obtained from the fits were then multiplied by appropriate scaling factors .
these factors were obtained from the comparisons of data / mc for the @xmath125 channel in the ec - ec topology . the cuts and their effects
are shown in fig .
[ ec_cuts ] .
the dvcs reaction can be described by four independent kinematic variables .
the typical variables used to interpret the results in terms of generalized parton distributions are @xmath6 , @xmath21 , @xmath126 and @xmath52 . in accordance with the choice made in previous dvcs analyses @xcite , the binning of the data in the @xmath6-@xmath21 plane was done making 5 slices in the polar angle @xmath146 of the electron and in @xmath21 .
the limits of the slices are given in table [ tab_binning ] , as well as the bin averages for @xmath6 and @xmath21 , defined as the weighted average over the distribution of events in each bin .
the size of the bins was optimized to have comparable statistics .
the top plot of fig .
[ binning_q2_x_t ] shows the chosen grid in the @xmath6-@xmath21 plane .
ten equally spaced bins in @xmath52 and 4 bins in @xmath126 were adopted . the bin limits and data - averaged bin centers
are summarized in table [ tab_binning_t ] . the bottom plot of fig .
[ binning_q2_x_t ] shows the binning in the @xmath29-@xmath21 plane .
.definition of the bins in @xmath21 and @xmath146 ( @xmath6 ) , and average kinematics for @xmath21 and @xmath6 ( ( gev/@xmath45)@xmath46 ) for each bin . [ cols="^,^,^,^,^",options="header " , ] afanasev _ et al .
_ @xcite have computed the radiative corrections for the dvcs and bh processes for clas kinematics .
it was found that , given the strict kinematic cuts adopted to select the final state , the undetected radiated photon can only have small energies . in this case , therefore , the main contribution to the radiative correction comes from spin - independent soft - photon emission that does not affect the polarization observables - while instead it can affect unpolarized cross sections even up to the 20% level .
the approximation of negligible contribution from the radiative corrections to the bsa , tsa and dsa , compared to the size of the asymmetries , is valid at clas kinematics at the 0.1% level @xcite . given the statistical uncertainties and the larger size of other systematic effects , it was chosen to neglect this contribution .
in the following sections the experimental asymmetries are compared to the predictions of four gpd models : vanderhaeghen - guichon - guidal ( vgg ) @xcite , goloskokov - kroll ( gk ) @xcite , kumericki - mller ( kmm12 ) @xcite , and goldstein - gonzalez - liuti ( gll ) @xcite .
both the vgg and gk models are based on double distributions @xcite to parametrize the @xmath147 dependence of the gpds , and on regge phenomenology for their @xmath29 dependence .
the main differences between these two models are in the parametrization of the high-@xmath29 part of the electromagnetic form factors and in the fact that the parameters of the gk model are tuned using low-@xmath21 dvmp data from hera , which are particularly sensitive to gluon and sea - quark gpds .
therefore , the gk model is also suited for gluonic gpds that are not accounted for in vgg .
however , given the @xmath21 range of the results presented here , the description of the valence - quark gpds is sufficient .
kmm12 is a hybrid model designed for global fitting , in which sea - quark gpds are modeled in a mellin - barnes integral representation ; valence quarks are modeled in terms of these gpds on the cross - over line @xmath148 .
the parameters of the model were fixed using unpolarized - proton dvcs data from clas and hermes , as well as the polarized - proton hermes data .
the kinematic range of applicability of this model is defined by the relation @xmath149 .
finally , the ggl model provides a diquark model based parametrization of gpds that incorporates regge behavior by introducing a spectral function for the spectator diquark s invariant mass distribution .
the parameters of the model were obtained by fitting both deep inelastic scattering ( dis ) structure functions and the recently available flavor - separated nucleon form factors data @xcite .
hereafter , the results for the three asymmetries are presented , discussed and compared to the gpd models described in section [ sec_gpd_models ] . the values of the asymmetries for each 4-dimensional bin , along with their uncertainties , are listed in table [ table_all_data ] and in ref .
@xcite . the harmonic structure in @xmath52 of the asymmetries versus ( @xmath6-@xmath21 ) and @xmath126
was studied by fitting their @xmath52 distributions . for this goal , the fact that the three asymmetries have the same denominator has been exploited .
in fact , by performing a simultaneaous fit of the 3 asymmetries , the common denominator can be constrained to be the same for the three different observables .
thus bsa , tsa and dsa were fitted , respectively , with three functions that shared a common denominator @xmath150 ) . in the highest @xmath126 bin of the third ( @xmath6-@xmath21 ) bin ,
@xmath151 was set to zero due to the limited @xmath52 coverage .
figure [ bsa_phi ] shows the beam - spin asymmetry as a function of @xmath52 for each slice in the @xmath6-@xmath21 space ( rows ) and for each bin in @xmath126 ( columns ) .
each asymmetry is fitted with the function @xmath152 and shows a clear @xmath73-like modulation , with a decreasing amplitude as @xmath126 increases , ranging from @xmath153 down to @xmath154 .
the dependence in the other kinematic variables appears less marked , although a slight decrease in the @xmath126 slope seems to happen at the highest @xmath6 and @xmath21 values .
this is confirmed by fig .
[ bsa_tdep ] , which shows the beam - spin asymmetry at @xmath155 ( i.e. the @xmath156 coefficient in eq . ) as a function of @xmath126 for each @xmath6-@xmath21 bin .
the choice of the fitting function was motivated by the physics ( see eq . ) .
the data are compared to the predictions of the vgg , gk , kmm12 and ggl models .
as expected , the vgg and gk models do nt show strong differences between each other .
with respect to the data , they overestimate the amplitude of the experimental asymmetries especially at low @xmath126 and at low @xmath6-@xmath21 . at the highest @xmath126 values , the vgg model gets closer to the data , while the gk model is systematically higher .
both models expect a steeper @xmath126 slope than the data display
. this can be due to the fact that these models are based on double - distributions , where the @xmath126 dependence is factorized with respect to the @xmath157 dependence .
the data , instead , seem to point to more complex correlations between the three variables .
the ggl model is in good agreement with the data at low @xmath126 especially for the first and third @xmath6-@xmath21 bin , while it diverges away from the data in the high-@xmath21 bins .
the discrepancy observed for larger @xmath21 values is an indication that using only dis and form factor data one can only provide a loose constraint on the @xmath17 dependence of the model .
the best fit to the data is provided by the kmm12 model , which however does not cover our whole set of kinematics .
dependence for each @xmath6-@xmath21 bin of the @xmath156 term of the beam - spin asymmetry .
the curves show the predictions of four gpd models for the bsa at @xmath158 : i ) vgg @xcite ( red dashed ) , ii ) kmm12 @xcite ( blue dotted ) , iii ) gk @xcite ( blue dash - dotted ) , and iv ) ggl @xcite ( orange dashed - three - dotted ) .
the square black points are the results obtained from the present analysis , the triangular green data come from the previous clas experiment with unpolarized proton target @xcite.,width=325 ] for consistency , our beam - spin asymmetries were also compared to those obtained from previous clas data @xcite ( e1-dvcs experiment ) . for this task ,
the results for the @xmath156 coefficient were used , taking the kinematic bins from the e1-dvcs data that were closest to our own .
the comparison is shown in fig .
[ bsa_tdep ] , where the e1-dvcs results are represented by the ( green online ) triangles .
the agreement is good , especially considering the imperfect kinematical overlap . the results for the target - spin asymmetry @xcite
are presented in fig .
[ tsa_phi ] as a function of @xmath52 for each slice in the @xmath6-@xmath21 space ( rows ) and for each bin in @xmath126 ( columns ) . as for the bsa ,
it is fitted with the function @xmath159 and shows the typical @xmath73-like dependence , with amplitudes ranging from 0.1 to 0.3 , but its evolution with @xmath126 is quite different from the bsa , in shape and magnitude .
in fact , the amplitude of the target - spin asymmetry seems rather constant as a function of all kinematic variables , @xmath126 included , apart from the expected systematic drop towards @xmath160 .
figure [ tsa_tdep ] shows the @xmath29-dependence for each bin in @xmath6-@xmath21 of the @xmath73 fitting coefficient @xmath161 ( eq . ) , which appears fairly constant , unlike what was observed for the beam - spin asymmetry . as mentioned above , the variable @xmath29 yields the fourier conjugate of the impact parameter , describing the transverse position of the partons in the reference frame where the proton goes at the speed of light .
therefore , a steep @xmath29 slope is equivalent to a rather flat spatial distribution , and vice versa .
equations and point to the proportionality between , respectively , tsa and @xmath162 and bsa and @xmath163 .
thus , the @xmath29 behavior of the tsa compared to that of the bsa suggests that the axial charge ( linked to @xmath4 ) is more concentrated in the center of the proton than the electric charge ( linked to @xmath3 ) .
this fact was already observed in a paper @xcite devoted to the extraction of the cffs @xmath163 and @xmath162 from the hermes data .
this finding is clearly not predicted by the vgg or gk models , which instead display a similar drop with @xmath29 for the tsa as what was computed for the bsa .
these models approximately reproduce the low-@xmath29 magnitude of the asymmetry in some kinematics ( namely , in @xmath6-@xmath21 bins 1 and 3 ) , with a slightly better fit of the data for vgg .
gk predicts an increase of the tsa with @xmath21 that is not observed in the experiment . by comparing the behavior of these two models for the two single - spin asymmetries ,
it can be observed how both reach good agreement with the data at high @xmath126 for the bsa and a low @xmath126 for the tsa .
these data can therefore provide strong guidance to correct the @xmath29 dependence of the parametrizations of both @xmath3 and @xmath164 .
the ggl model also predicts a drop in @xmath126 not confirmed by the data , and moreover it overestimates the magnitude of the asymmetry by at least a factor of 2 .
the best fit to the data is provided , in the bins where it applies , by the kmm12 model .
dependence , for each @xmath6-@xmath21 bin , of the @xmath161 term of the target - spin asymmetry .
the curves show the predictions of four gpd models for the tsa at @xmath158 : i ) vgg @xcite ( red dashed ) , ii ) kmm12 @xcite ( cyan dotted ) , iii ) gk @xcite ( blue dash - dotted ) , and iv ) ggl @xcite ( orange dashed - three - dotted).,width=325 ] our target - spin asymmetries were compared to those obtained from hermes and to the results of a previous exploratory clas experiment @xcite .
given the different kinematical coverages of the three experiments , it was decided to integrate the data over all values of @xmath6-@xmath21 of the phase space , and obtain a set of tsas only as a function of @xmath52 and @xmath126 , with a finer binning in @xmath126 than for our four - fold asymmetries .
for this comparison , each @xmath52 distribution for a given @xmath126 bin was fitted with the function @xmath165 , which was the one used for both the hermes and the old clas data .
the comparison of the @xmath166 coefficients is shown in fig .
[ tsa_comparison ] .
our data have at least a factor 5 smaller error bars than the previously published data , and extend the @xmath126 range up to 1.7 ( gev/@xmath45)@xmath46 .
dependences of the @xmath73 term of the @xmath103 target - spin asymmetries for the present data , integrated over @xmath6 and @xmath21 ( black circles ) , the previous clas experiment @xcite ( magenta triangles ) , and hermes @xcite ( green squares).,width=325 ] a further test was made including in the fit an additional @xmath167 term for the tsa .
thus the adopted fitting function was @xmath168 the data from hermes @xcite , in fact , show a non - negligible contribution of the @xmath169 moment to the tsa .
it must be pointed out , however , that in the expansion in @xmath73 moments done by the hermes collaboration , the contribution of the denominator could mix with the @xmath170 terms , while in our analysis it is treated with its own parameter . in the belitsky - mller - kirchner formalism
, the parts of the dvcs / bh interference term depending on @xmath169 appear only at twist - three level @xcite .
therefore , observing a sizeable @xmath84 component in the target - spin asymmetry would mean having some sensitivity to twist - three cffs .
a recent paper @xcite pointed to a possible way to access the quarks orbital angular momentum via the measurement of twist - three gpds .
first of all , the stability of @xmath161 was verified by comparing the @xmath73 parameter obtained with and without @xmath169 term in the fitting function .
the @xmath84 term appears to be much smaller , at least in the low-@xmath29 region , than the @xmath73 term , often compatible with zero , with a slight tendency to increase at high @xmath126 in some kinematic bins towards negative values .
the @xmath73 component is always dominant .
however , given the limited statistics and the relatively small number of @xmath52 bins , the uncertaintes on the denominator parameter highly affect the extraction of such a potentially small higher - twist @xmath84 modulation in the numerator , so no reliable @xmath84-extraction can be achieved with the present fitting procedure . the double - spin asymmetry is plotted in fig .
[ dsa_phi ] as a function of @xmath52 for each bin in @xmath126 ( columns ) and for each slice in the @xmath6-@xmath21 space ( rows ) .
it is larger in magnitude than the single - spin asymmetries presented in the previous sections ( around 0.6 ) , seems rather flat as a function of @xmath52 , and presents a slow decrease as a function of @xmath126 .
the data were fitted with the function @xmath171 where the denominator parameter @xmath151 is the same as for the fits to the two single - spin asymmetries .
the two sets of fit parameters of the numerator , @xmath172 and @xmath173 , are shown , as functions of @xmath126 and for each @xmath6-@xmath21 bin in figs .
[ dsa_tdep_const ] and [ dsa_tdep_cosphi ] , respectively .
the constant term dominates the asymmetry , while the @xmath89 term of the numerator is compatible with zero for most kinematics . in figs .
[ dsa_phi ] , [ dsa_tdep_const ] , and [ dsa_tdep_cosphi ] , the three fit parameters of the double - spin asymmetries are compared to the four model predictions for dvcs+bh and to the calculations for bh only ( green dot - dashed curve ) .
it seems that bethe - heitler fully dominates the constant term , and all models - except for ggl , which misses both the magnitude and the @xmath29-dependence of this observable - predict this and correctly reproduce it . the best match for this term
is provided by the vgg and gk models , which show sizeable differences only at the highest @xmath126 values , where the dvcs contribution is expected to start to play a role .
the models suggest a slight contribution from dvcs in the @xmath89 term but the statistical precision of the data does not allow us to draw conclusions on which prediction provides the better fit .
dependence for each @xmath6-@xmath21 bin of the constant term @xmath172 of the double - spin asymmetry .
the pink two - dot - dashed curves are the calculations of the dsa for the bethe - heitler process alone .
the curves show the predictions for the full @xmath103 amplitude of four gpd models : i ) vgg @xcite ( red dashed ) , ii ) kmm12 @xcite ( cyan dotted ) , iii ) gk @xcite ( blue dash - dotted ) , and iv ) ggl @xcite ( orange dashed - three - dotted).,width=325 ] dependence for each @xmath6-@xmath21 bin of the @xmath89 term @xmath173 of the double - spin asymmetry .
the pink two - dot - dashed curves are the calculations of the dsa for the bethe - heitler process alone .
the curves show the predictions for the full @xmath103 amplitude of four gpd models : i ) vgg @xcite ( red dashed ) , ii ) kmm12 @xcite ( cyan dotted ) , iii ) gk @xcite ( blue dash - dotted ) , and iv ) ggl @xcite ( orange dashed - three - dotted).,width=325 ]
in recent years , various groups have developed and applied different procedures to extract compton form factors from dvcs observables .
the approach adopted here @xcite is based on a local - fitting method at each given experimental @xmath174 kinematic point . in this framework , instead of four complex cffs defined as in eq .
[ def_cffs ] , there are eight real cffs defined as @xmath175 @xmath176,\ ] ] where the sign convention is the same as for eq . .
these cffs are the almost - free parameters - their values are allowed to vary within @xmath177 times the values predicted by the vgg model - that are extracted from dvcs observables using the well - established dvcs+bh theoretical amplitude .
the bh amplitude is calculated exactly while the dvcs amplitude is taken at the qcd leading twist .
the expression of these amplitudes can be found , for instance , in @xcite .
the three sets of asymmetries ( bsa , tsa and dsa ) for all kinematic bins were processed using this fitting procedure to extract the compton form factors . in the adopted version of the fitter code , @xmath178 is set to zero , as @xmath179 is assumed to be purely real - it is parametrized in the vgg model by the pion pole @xmath180 .
thus seven out of the eight real and imaginary parts of the cffs are left as free parameters in the fit .
figure [ cff_comp ] shows @xmath181 ( black full squares ) and @xmath182 ( red full circles ) , which are obtained from the fit of the present data , as a function of @xmath126 for each of our 5 @xmath6-@xmath21 bins .
these are the two cffs that appear to be better constrained by the present results . given that the size of the error bars reflects the sensitivity of the combination of observables to each cff , it is evident that , as expected , our asymmetries are mostly sensitive to @xmath183 .
dependence for each @xmath6-@xmath21 bin of @xmath181 ( black squares ) and @xmath184 ( red circles ) .
the full points are obtained by fitting the present data ( tsa , bsa and dsa ) .
the empty points were obtained by fitting the bsa results from @xcite integrated over all values of @xmath6 at @xmath185 , and the tsas from @xcite.,width=325 ] the results for @xmath181 and @xmath182 confirm what had been previously observed in a qualitative way by direct comparison of the @xmath29-dependence of our tsas and bsas in section [ sec_results_tsa ] : the @xmath29-slope of @xmath79 is much steeper than that of @xmath183 , hinting at the fact that the axial charge ( linked to @xmath183 ) might be more `` concentrated '' in the center of the nucleon than the electric charge ( linked to @xmath79 ) . this effect seems stronger at the lowest values of @xmath21 , while both cffs tend to flatten out as @xmath21 increases .
it is also interesting to compare the results obtained for the two equal-@xmath21 bins ( @xmath186 ( gev/@xmath45)@xmath46 and @xmath187 ( gev/@xmath45)@xmath46 ) : within the limits imposed by the size of the error bars and by the @xmath6 lever arm ( only 0.44 ( gev/@xmath45)@xmath46 ) , both sets of cffs are compatible , at the 1-@xmath188 level , which supports the validity of the scaling hypothesis . in fig .
[ cff_comp ] the values of @xmath181 and @xmath182 that were obtained @xcite using the same fitting code with the results from @xcite for the beam - spin asymmetry and from @xcite for the target - spin asymmetry , are also shown . aside from the much larger kinematic coverage for the polarized - target observables made accessible by our data , in the kinematics where the previous extraction had been attempted ,
our data improve the precision of @xmath183 .
for the first time four - dimensional single - beam , single - target , and double ( beam - target ) spin asymmetries for deeply virtual compton scattering were extracted over a large phase space at the same kinematics .
this experiment used the clas detector in conjunction with the ic calorimeter and the hall - b longitudinally polarized @xmath1nh@xmath2 target .
165 4-dimensional bins in @xmath6 , @xmath21 , @xmath126 and @xmath52 , covering a wide kinematic range ( @xmath189 ( gev/@xmath45)@xmath46 , @xmath190 , @xmath191 ( gev/@xmath45)@xmath46 , @xmath192 ) were obtained for the three asymmetries , with systematic uncertainties largely below the statistical uncertainties .
the @xmath52 dependence of the obtained asymmetries was studied .
the dominance of the leading - order handbag mechanism is supported by the prevalence , especially at low @xmath126 , of the @xmath73 term over higher @xmath193 components in both single - spin asymmetries .
the dsa is found to be mostly dominated by the constant term , which contains both bh and dvcs / bh interference , although in our kinematics the bh contribution is the strongest .
these data bring important constraints to gpd parametrizations , especially for @xmath3 and @xmath4 . using one method among the various ones that are currently in development
, these data allow us to extract the imaginary parts of the @xmath194 and @xmath195 compton form factors and to gain insight , via their relative @xmath29 slopes , about the spatial distribution of the electric and axial charges in the proton .
the extraction of the compton form factors will be further improved once the new clas results on dvcs cross sections will become available @xcite .
furthermore , the extensive dvcs - devoted program planned at jefferson lab for the 12-gev era will extend our knowledge of the generalized parton distributions of the proton in terms of both kinematical coverage and statistical precision .
+ + we thank the staff of the accelerator and the physics divisions and of the target group at jefferson lab for making the experiment possible .
special thanks to m. guidal , f. sabati , s. liuti , d. mller and k. kumericki for the theoretical support .
this work was supported in part by the u.s .
department of energy and national science foundation , the french centre national de la recherche scientifique and commissariat lenergie atomique , the french - american cultural exchange ( face ) , the italian istituto nazionale di fisica nucleare , the chilean comisin nacional de investigacin cientfica y tecnolgica ( conicyt ) , the national research foundation of korea , and the uk science and technology facilities council ( stfc ) . the jefferson science associates ( jsa ) operates the thomas jefferson national accelerator facility for the united states department of energy under contract de - ac05 - 06or23177 . 1.68 & 0.194 & 0.11 & 25 & 0.37 @xmath196 0.23@xmath196 0.01 & 0.44 @xmath196 0.26 @xmath196 0.02 & 0.0018 & 0.58 @xmath196 0.33 @xmath196 0.04 & 0.0116 + 1.68 & 0.190 & 0.12 & 60 & 0.34 @xmath196 0.11@xmath196 0.03 & 0.12 @xmath196 0.14 @xmath196 0.02 & 0.0024 & 0.34 @xmath196 0.16 @xmath196 0.13 & 0.0070 + 1.58 & 0.186 & 0.13 & 92 & 0.23 @xmath196 0.06@xmath196 0.01 & 0.18 @xmath196 0.07 @xmath196 0.01 & 0.0144 & 0.58 @xmath196 0.08 @xmath196 0.02 & 0.0072 + 1.54 & 0.178 & 0.13 & 128 & 0.13 @xmath196 0.05@xmath196 0.01 & 0.13 @xmath196 0.06 @xmath196 0.02 & 0.0184 & 0.56 @xmath196 0.07 @xmath196 0.02 & 0.0021 + 1.50 & 0.174 & 0.13 & 161 & 0.05 @xmath196 0.05@xmath196 0.01 & 0.07 @xmath196 0.06 @xmath196 0.01 & 0.0086 & 0.46 @xmath196 0.07 @xmath196 0.02 & -0.0024 + 1.51 & 0.173 & 0.13 & 198 & -0.05 @xmath196 0.05@xmath196 0.00 & -0.08 @xmath196 0.06 @xmath196 0.01 & -0.0086 & 0.51 @xmath196 0.07 @xmath196 0.01 & -0.0021 + 1.52 & 0.177 & 0.13 & 235 & -0.21 @xmath196 0.06@xmath196 0.01 & -0.13 @xmath196 0.07 @xmath196 0.02 & -0.0184 & 0.51 @xmath196 0.08 @xmath196 0.02 & 0.0018 + 1.60 & 0.185 & 0.13 & 266 & -0.28 @xmath196 0.06@xmath196 0.01 & -0.18 @xmath196 0.08 @xmath196 0.01 & -0.0138 & 0.44 @xmath196 0.09 @xmath196 0.02 & 0.0059 + 1.68 & 0.190 & 0.12 & 305 & -0.15 @xmath196 0.10@xmath196 0.02 & -0.01 @xmath196 0.12 @xmath196 0.03 & -0.0017 & 0.46 @xmath196 0.14 @xmath196 0.04 & 0.0082 + 1.61 & 0.190 & 0.21 & 97 & 0.19 @xmath196 0.19@xmath196 0.06 & 0.15 @xmath196 0.21 @xmath196 0.04 & 0.0193 & -0.05 @xmath196 0.26 @xmath196 0.10 & -0.0011 + 1.57 & 0.182 & 0.23 & 128 & 0.17 @xmath196 0.10@xmath196 0.03 & -0.01 @xmath196 0.12 @xmath196 0.03 & 0.0209 & 0.54 @xmath196 0.14 @xmath196 0.05 & 0.0034 + 1.51 & 0.179 & 0.23 & 161 & 0.09 @xmath196 0.08@xmath196 0.02 & 0.13 @xmath196 0.09 @xmath196 0.02 & 0.0111 & 0.64 @xmath196 0.10 @xmath196 0.06 & 0.0044 + 1.54 & 0.178 & 0.23 & 199 & -0.12 @xmath196 0.07@xmath196 0.02 & 0.11 @xmath196 0.09 @xmath196 0.01 & -0.0085 & 0.60 @xmath196 0.10 @xmath196 0.03 & 0.0036 + 1.62 & 0.184 & 0.23 & 233 & -0.18 @xmath196 0.09@xmath196 0.03 & -0.24 @xmath196 0.11 @xmath196 0.01 & -0.0221 & 0.38 @xmath196 0.13 @xmath196 0.11 & 0.0022 + 1.64 & 0.191 & 0.22 & 263 & -0.12 @xmath196 0.17@xmath196 0.05 & -0.41 @xmath196 0.19 @xmath196 0.08 & -0.0201 & 0.75 @xmath196 0.21 @xmath196 0.08 & 0.0055 + 1.63 & 0.188 & 0.49 & 96 & 0.39 @xmath196 0.24@xmath196 0.09 & 0.23 @xmath196 0.27 @xmath196 0.06 & 0.0193 & 0.81 @xmath196 0.32 @xmath196 0.14 & 0.0016 + 1.54 & 0.181 & 0.47 & 127 & 0.18 @xmath196 0.11@xmath196 0.05 & 0.14 @xmath196 0.14 @xmath196 0.04 & 0.0213 & 0.34 @xmath196 0.16 @xmath196 0.04 & 0.0023 + 1.36 & 0.173 & 0.49 & 162 & 0.21 @xmath196 0.07@xmath196 0.02 & 0.19 @xmath196 0.08 @xmath196 0.01 & 0.0117 & 0.50 @xmath196 0.09 @xmath196 0.05 & 0.0073 + 1.42 & 0.174 & 0.46 & 196 & -0.07 @xmath196 0.08@xmath196 0.02 & -0.03 @xmath196 0.10 @xmath196 0.01 & -0.0095 & 0.54 @xmath196 0.11 @xmath196 0.03 & 0.0071 + 1.60 & 0.181 & 0.43 & 234 & -0.22 @xmath196 0.12@xmath196 0.02 & -0.17 @xmath196 0.14 @xmath196 0.04 & -0.0194 & 0.36 @xmath196 0.16 @xmath196 0.09 & 0.0020 + 1.56 & 0.185 & 0.51 & 261 & -0.04 @xmath196
0.21@xmath196 0.12 & -0.29 @xmath196 0.24 @xmath196 0.10 & -0.0218 & -0.02 @xmath196 0.29 @xmath196 0.12 & -0.0048 + 1.62 & 0.184 & 1.35 & 20 & 0.13 @xmath196 0.16@xmath196 0.07 & 0.23 @xmath196 0.19 @xmath196 0.02 & 0.0016 & 0.36 @xmath196 0.23 @xmath196 0.07 & -0.0051 + 1.56 & 0.178 & 1.21 & 56 & 0.01 @xmath196 0.14@xmath196 0.05 & -0.05 @xmath196 0.17 @xmath196 0.07 & 0.0009 & 0.50 @xmath196 0.20 @xmath196 0.06 & -0.0017 + 1.46 & 0.178 & 1.02 & 92 & 0.12 @xmath196 0.15@xmath196 0.14 & 0.21 @xmath196 0.20 @xmath196 0.10 & 0.0029 & 0.08 @xmath196 0.21 @xmath196 0.18 & -0.0005 + 1.48 & 0.176 & 1.01 & 128 & 0.20 @xmath196 0.19@xmath196 0.05 & 0.36 @xmath196 0.24 @xmath196 0.04 & 0.0021 & 0.44 @xmath196 0.27 @xmath196 0.06 & 0.0062 + 1.36 & 0.170 & 1.00 & 164 & -0.09 @xmath196 0.08@xmath196 0.01 & 0.41 @xmath196 0.09 @xmath196 0.06 & 0.0034 & 0.12 @xmath196 0.11 @xmath196 0.05 & 0.0067 + 1.35 & 0.171 & 0.97 & 192 & 0.06 @xmath196 0.14@xmath196 0.11 & 0.07 @xmath196 0.18 @xmath196 0.07 & 0.0011 & 0.41 @xmath196 0.19 @xmath196 0.06 & 0.0095 + 1.40 & 0.171 & 1.07 & 239 & -0.22 @xmath196 0.17@xmath196 0.07 & 0.09 @xmath196 0.22 @xmath196 0.03 & 0.0017 & 0.24 @xmath196 0.23 @xmath196 0.06 & 0.0048 + 1.41 & 0.178 & 1.03 & 266 & 0.04 @xmath196 0.10@xmath196 0.05 & -0.27 @xmath196 0.12 @xmath196 0.10 & -0.0034 & 0.30 @xmath196 0.14 @xmath196 0.08 & 0.0015 + 1.58 & 0.179 & 1.24 & 306 & 0.01 @xmath196 0.14@xmath196 0.09 & 0.04 @xmath196 0.17 @xmath196 0.10 & -0.0009 & 0.51 @xmath196 0.20 @xmath196 0.07 & -0.0016 + 1.62 & 0.183 & 1.35 & 340 & -0.11 @xmath196 0.18@xmath196 0.12 & -0.20 @xmath196 0.20 @xmath196 0.10 & -0.0013 & 0.36 @xmath196 0.24 @xmath196 0.17 & -0.0051 + 2.03 & 0.255 & 0.13 & 19 & 0.03 @xmath196 0.03@xmath196 0.01 & -0.03 @xmath196 0.03 @xmath196 0.01 & -0.0023 & 0.60 @xmath196 0.04 @xmath196 0.03 & 0.0328 + 1.99 & 0.250 & 0.14 & 52 & 0.17 @xmath196 0.03@xmath196 0.01 & 0.05 @xmath196 0.04 @xmath196 0.01 & 0.0027 & 0.60 @xmath196 0.04 @xmath196 0.03 & 0.0297 + 1.93 & 0.239 & 0.14 & 89 & 0.20 @xmath196 0.03@xmath196 0.01 & 0.05 @xmath196 0.04 @xmath196 0.01 & 0.0165 & 0.57 @xmath196 0.05 @xmath196 0.02 & 0.0179 + 1.91 & 0.235 & 0.14 & 124 & 0.16 @xmath196 0.05@xmath196 0.01 & 0.12 @xmath196 0.06 @xmath196 0.02 & 0.0247 & 0.63 @xmath196 0.07 @xmath196 0.02 & 0.0025 + 1.84 & 0.228 & 0.14 & 160 & 0.03 @xmath196 0.06@xmath196 0.01 & 0.20 @xmath196 0.07 @xmath196 0.01 & 0.0139 & 0.65 @xmath196 0.08 @xmath196 0.03 & -0.0105 + 1.89 & 0.231 & 0.14 & 199 & 0.02 @xmath196 0.06@xmath196 0.01 & -0.01 @xmath196 0.07 @xmath196 0.02 & -0.0116 & 0.59 @xmath196 0.08 @xmath196 0.02 & -0.0111 + 1.93 & 0.235 & 0.14 & 238 & -0.21 @xmath196 0.05@xmath196 0.02 & -0.13 @xmath196 0.06 @xmath196 0.01 & -0.0246 & 0.50 @xmath196 0.07 @xmath196 0.02 & 0.0009 + 1.94 & 0.239 & 0.14 & 270 & -0.24 @xmath196 0.04@xmath196 0.01 & -0.16 @xmath196 0.04 @xmath196 0.01 & -0.0176 & 0.63 @xmath196 0.05 @xmath196 0.02 & 0.0186 + 1.97 & 0.248 & 0.13 & 308 & -0.16 @xmath196 0.03@xmath196 0.01 & -0.16 @xmath196 0.03 @xmath196 0.01 & -0.0041 & 0.56 @xmath196 0.04 @xmath196 0.03 & 0.0289 + 2.03 & 0.255 & 0.13 & 340 & -0.07 @xmath196 0.03@xmath196 0.00 & -0.09 @xmath196 0.04 @xmath196 0.01 & -0.0007 & 0.55 @xmath196 0.04 @xmath196 0.03 & 0.0320 + 2.23 & 0.276 & 0.22 & 21 & 0.05 @xmath196 0.05@xmath196 0.01 & -0.00 @xmath196 0.06 @xmath196 0.01 & -0.0037 & 0.58 @xmath196 0.06 @xmath196 0.02 & 0.0172 + 2.12 & 0.268 & 0.23 & 53 & 0.22 @xmath196 0.04@xmath196 0.01 & 0.14 @xmath196 0.05 @xmath196 0.02 & 0.0044 & 0.56 @xmath196 0.06 @xmath196 0.03 & 0.0169 + 2.01 & 0.251 & 0.23 & 91 & 0.27 @xmath196 0.04@xmath196 0.01 & 0.16 @xmath196 0.05 @xmath196 0.01 & 0.0246 & 0.49 @xmath196 0.06 @xmath196 0.01 & 0.0105 + 1.91 & 0.248 & 0.23 & 124 & 0.06 @xmath196 0.06@xmath196 0.01 & 0.06 @xmath196 0.07 @xmath196 0.01 & 0.0347 & 0.66 @xmath196 0.07 @xmath196 0.03 & 0.0060 + 1.79 & 0.246 & 0.24 & 161 & 0.03 @xmath196 0.06@xmath196 0.01 & 0.07 @xmath196 0.07 @xmath196 0.01 & 0.0185 & 0.55 @xmath196 0.08 @xmath196 0.02 & -0.0017 + 1.80 & 0.245 & 0.24 & 199 & 0.02 @xmath196 0.06@xmath196 0.01 & -0.03 @xmath196 0.08 @xmath196 0.02 & -0.0174 & 0.70 @xmath196 0.08 @xmath196 0.03 & 0.0004 + 1.97 & 0.242 & 0.23 & 235 & -0.11 @xmath196 0.06@xmath196 0.01 & -0.18 @xmath196 0.07 @xmath196 0.01 & -0.0322 & 0.62 @xmath196 0.08 @xmath196 0.04 & 0.0041 + 2.01 & 0.248 & 0.23 & 270 & -0.31 @xmath196 0.04@xmath196 0.01 & -0.21 @xmath196 0.05 @xmath196 0.01 & -0.0247 & 0.53 @xmath196 0.06 @xmath196 0.02 & 0.0109 + 2.14 & 0.268 & 0.23 & 307 & -0.28 @xmath196 0.04@xmath196 0.02 & -0.15 @xmath196 0.05 @xmath196 0.02 & -0.0044 & 0.57 @xmath196 0.06 @xmath196 0.03 & 0.0167 + 2.21 & 0.277 & 0.22 & 340 & -0.05 @xmath196 0.04@xmath196 0.01 & 0.00 @xmath196 0.05 @xmath196 0.01 & 0.0039 & 0.51 @xmath196 0.06 @xmath196 0.02 & 0.0167 + 2.34 & 0.286 & 0.40 & 26 & 0.22 @xmath196 0.16@xmath196 0.01 & -0.17 @xmath196 0.18 @xmath196 0.01 & -0.0069 & 0.66 @xmath196 0.22 @xmath196 0.05 & 0.0039 + 2.20 & 0.276 & 0.45 & 56 & 0.21 @xmath196 0.06@xmath196 0.02 & 0.15 @xmath196 0.08 @xmath196 0.02 & 0.0075 & 0.43 @xmath196 0.09 @xmath196 0.04 & 0.0000 + 2.04 & 0.254 & 0.44 & 90 & 0.14 @xmath196 0.06@xmath196 0.01 & 0.29 @xmath196 0.07 @xmath196 0.04 & 0.0310 & 0.55 @xmath196 0.08 @xmath196 0.03 & 0.0022 + 1.82 & 0.255 & 0.45 & 127 & 0.08 @xmath196 0.06@xmath196 0.04 & 0.14 @xmath196 0.08 @xmath196 0.02 & 0.0422 & 0.51 @xmath196 0.09 @xmath196 0.02 & 0.0047 + 1.73 & 0.252 & 0.48 & 163 & 0.06 @xmath196 0.04@xmath196 0.01 & 0.12 @xmath196 0.05 @xmath196 0.01 & 0.0209 & 0.45 @xmath196 0.06 @xmath196 0.02 & 0.0056 + 1.74 & 0.254 & 0.47 & 197 & 0.02 @xmath196 0.05@xmath196 0.01 & -0.13 @xmath196 0.06 @xmath196 0.03 & -0.0211 & 0.48 @xmath196 0.07 @xmath196 0.02 & 0.0060 + 1.88 & 0.251 & 0.44 & 233 & -0.13 @xmath196 0.07@xmath196 0.02 & -0.12 @xmath196 0.09 @xmath196 0.04 & -0.0380 & 0.44 @xmath196 0.10 @xmath196 0.03 & 0.0029 + 1.99 & 0.253 & 0.46 & 271 & -0.20 @xmath196 0.05@xmath196 0.03 & -0.17 @xmath196 0.06 @xmath196 0.02 & -0.0310 & 0.50 @xmath196 0.07 @xmath196 0.03 & 0.0018 + 2.20 & 0.274 & 0.45 & 303 & -0.30 @xmath196 0.06@xmath196 0.03 & -0.09 @xmath196 0.07 @xmath196 0.01 & -0.0066 & 0.41 @xmath196 0.09 @xmath196 0.04 & -0.0003 + 2.36 & 0.287 & 0.40 & 334 & -0.21 @xmath196 0.12@xmath196 0.03 & -0.04 @xmath196 0.14 @xmath196 0.02 & -0.0042 & 0.34 @xmath196 0.16 @xmath196 0.02 & -0.0007 + 2.03 & 0.261 & 1.41 & 17 & -0.04 @xmath196 0.05@xmath196 0.02 & -0.08 @xmath196 0.05 @xmath196 0.01 & -0.0026 & 0.48 @xmath196 0.06 @xmath196 0.02 & -0.0031 + 1.98 & 0.258 & 1.12 & 54 & 0.10 @xmath196 0.06@xmath196 0.02 & 0.14 @xmath196 0.07 @xmath196 0.02 & 0.0060 & 0.34 @xmath196 0.08 @xmath196 0.03 & -0.0038 + 1.78 & 0.256 & 1.18 & 87 & 0.05 @xmath196 0.07@xmath196 0.03 & 0.20 @xmath196 0.08 @xmath196 0.04 & 0.0152 & 0.33 @xmath196 0.09 @xmath196 0.03 & 0.0005 + 1.82 & 0.261 & 1.13 & 128 & 0.14 @xmath196 0.09@xmath196 0.03 & 0.46 @xmath196 0.12 @xmath196 0.06 & 0.0188 & 0.26 @xmath196 0.13 @xmath196 0.02 & 0.0027 + 1.79 & 0.252 & 1.03 & 164 & 0.03 @xmath196 0.05@xmath196 0.01 & 0.16
@xmath196 0.06 @xmath196 0.02 & 0.0066 & 0.40 @xmath196 0.06 @xmath196 0.01 & 0.0062 + 1.78 & 0.254 & 1.03 & 195 & -0.03 @xmath196 0.07@xmath196 0.02 & -0.09 @xmath196 0.08 @xmath196 0.03 & -0.0056 & 0.34 @xmath196 0.09 @xmath196 0.03 & 0.0055 + 1.84 & 0.257 & 1.23 & 238 & -0.05 @xmath196 0.08@xmath196 0.02 & -0.30 @xmath196 0.11 @xmath196 0.04 & -0.0145 & 0.34 @xmath196 0.12 @xmath196 0.05 & 0.0039 + 1.75 & 0.257 & 1.14 & 270 & -0.02 @xmath196 0.05@xmath196 0.02 & -0.26 @xmath196 0.07 @xmath196 0.04 & -0.0169 & 0.21 @xmath196 0.07 @xmath196 0.02 & -0.0018 + 2.01 & 0.260 & 1.12 & 307 & -0.07 @xmath196 0.07@xmath196 0.01 & -0.20 @xmath196 0.08 @xmath196 0.02 & -0.0068 & 0.56 @xmath196 0.09 @xmath196 0.04 & 0.0005 + 2.03 & 0.261 & 1.40 & 343 & 0.00 @xmath196 0.04@xmath196 0.01 & 0.02 @xmath196 0.05 @xmath196 0.02 & 0.0018 & 0.36 @xmath196 0.06 @xmath196 0.02 & 0.0051 + 2.45 & 0.265 & 0.13 & 20 & -0.02 @xmath196 0.04@xmath196 0.01 & 0.02 @xmath196 0.05 @xmath196 0.01 & -0.0002 & 0.76 @xmath196 0.06 @xmath196 0.02 & 0.0271 + 2.36 & 0.253 & 0.14 & 53 & 0.07 @xmath196 0.05@xmath196 0.02 & 0.01 @xmath196 0.05 @xmath196 0.02 & 0.0018 & 0.62 @xmath196 0.06 @xmath196 0.04 & 0.0211 + 2.32 & 0.249 & 0.14 & 88 & 0.17 @xmath196 0.05@xmath196 0.01 & -0.01 @xmath196 0.07 @xmath196 0.01 & 0.0071 & 0.76 @xmath196 0.07 @xmath196 0.02 & 0.0125 + 2.30 & 0.246 & 0.14 & 125 & 0.20 @xmath196 0.08@xmath196 0.02 & 0.14 @xmath196 0.10 @xmath196 0.01 & 0.0109 & 0.61 @xmath196 0.11 @xmath196 0.03 & -0.0021 + 2.26 & 0.242 & 0.14 & 162 & 0.20 @xmath196 0.10@xmath196 0.03 & 0.25 @xmath196 0.12 @xmath196 0.05 & 0.0064 & 0.81 @xmath196 0.13 @xmath196 0.04 & -0.0117 + 2.27 & 0.242 & 0.14 & 199 & -0.07 @xmath196 0.09@xmath196 0.04 & 0.07 @xmath196 0.10 @xmath196 0.04 & -0.0044 & 0.81 @xmath196 0.11 @xmath196 0.05 & -0.0116 + 2.28 & 0.243 & 0.14 & 237 & -0.29 @xmath196 0.07@xmath196 0.02 & -0.17 @xmath196 0.09 @xmath196 0.02 & -0.0110 & 0.84 @xmath196 0.09 @xmath196 0.02 & -0.0007 + 2.32 & 0.248 & 0.14 & 271 & -0.22 @xmath196 0.06@xmath196 0.02 & -0.15 @xmath196 0.07 @xmath196 0.01 & -0.0082 & 0.75 @xmath196 0.07 @xmath196 0.02 & 0.0124 + 2.36 & 0.253 & 0.14 & 307 & -0.14 @xmath196 0.04@xmath196 0.01 & -0.10 @xmath196 0.05 @xmath196 0.01 & -0.0023 & 0.56 @xmath196 0.06 @xmath196 0.02 & 0.0206 + 2.45 & 0.265 & 0.14 & 340 & -0.01 @xmath196 0.05@xmath196 0.02 & -0.07 @xmath196 0.06 @xmath196 0.02 & -0.0002 & 0.67 @xmath196 0.07 @xmath196 0.03 & 0.0265 + 2.57 & 0.281 & 0.22 & 24 & 0.03 @xmath196 0.08@xmath196 0.02 & 0.03 @xmath196 0.09 @xmath196 0.03 & -0.0008 & 0.55 @xmath196 0.10 @xmath196 0.02 & 0.0130 + 2.50 & 0.269 & 0.23 & 53 & 0.11 @xmath196 0.05@xmath196 0.02 & 0.09 @xmath196 0.06 @xmath196 0.02 & 0.0031 & 0.68 @xmath196 0.07 @xmath196 0.05 & 0.0122 + 2.40 & 0.258 & 0.23 & 90 & 0.19 @xmath196 0.06@xmath196 0.02 & 0.07 @xmath196 0.08 @xmath196 0.01 & 0.0111 & 0.58 @xmath196 0.09 @xmath196 0.03 & 0.0076 + 2.35 & 0.252 & 0.23 & 126 & 0.16 @xmath196 0.09@xmath196 0.02 & 0.10 @xmath196 0.11 @xmath196 0.02 & 0.0150 & 0.55 @xmath196 0.12 @xmath196 0.03 & 0.0014 + 2.36 & 0.254 & 0.23 & 161 & 0.10 @xmath196 0.10@xmath196 0.03 & -0.18 @xmath196 0.12 @xmath196 0.02 & 0.0057 & 0.39 @xmath196 0.14 @xmath196 0.07 & -0.0045 + 2.37 & 0.254 & 0.23 & 199 & 0.02 @xmath196 0.09@xmath196 0.02 & 0.15 @xmath196 0.11 @xmath196 0.01 & -0.0059 & 0.70 @xmath196 0.11 @xmath196 0.06 & -0.0026 + 2.36 & 0.254 & 0.23 & 236 & -0.17 @xmath196 0.08@xmath196 0.02 & -0.23 @xmath196 0.10 @xmath196 0.02 & -0.0160 & 0.63 @xmath196 0.11 @xmath196 0.03 & 0.0019 + 2.42 & 0.260 & 0.23 & 270 & -0.23 @xmath196 0.06@xmath196 0.03 & -0.14 @xmath196 0.08 @xmath196 0.02 & -0.0118 & 0.61 @xmath196 0.09 @xmath196 0.04 & 0.0079 + 2.48 & 0.266 & 0.23 & 308 & -0.15 @xmath196 0.05@xmath196 0.02 & -0.21 @xmath196 0.06 @xmath196 0.01 & -0.0039 & 0.59 @xmath196 0.07 @xmath196 0.02 & 0.0115 + 2.56 & 0.279 & 0.22 & 336 & -0.01 @xmath196 0.07@xmath196 0.01 & -0.09 @xmath196 0.08 @xmath196 0.02 & 0.0001 & 0.49 @xmath196 0.10 @xmath196 0.03 & 0.0127 + 2.60 & 0.284 & 0.41 & 31 & -0.18 @xmath196 0.22@xmath196 0.09 & -0.01 @xmath196 0.23 @xmath196 0.06 & 0.0000 & 0.01 @xmath196 0.31 @xmath196 0.11 & -0.0045 + 2.55 & 0.275 & 0.44 & 52 & 0.17 @xmath196 0.06@xmath196 0.01 & 0.07 @xmath196 0.07 @xmath196 0.02 & 0.0055 & 0.54 @xmath196 0.08 @xmath196 0.05 & 0.0003 + 2.45 & 0.264 & 0.44 & 91 & 0.20 @xmath196 0.07@xmath196 0.02 & 0.04 @xmath196 0.08 @xmath196 0.04 & 0.0157 & 0.59 @xmath196 0.09 @xmath196 0.03 & 0.0022 + 2.37 & 0.254 & 0.44 & 122 & 0.23 @xmath196 0.10@xmath196 0.03 & 0.15 @xmath196 0.12 @xmath196 0.01 & 0.0188 & 0.69 @xmath196 0.13 @xmath196 0.03 & 0.0043 + 2.33 & 0.250 & 0.41 & 159 & -0.15 @xmath196 0.13@xmath196 0.04 & -0.24 @xmath196 0.16 @xmath196 0.04 & 0.0063 & 0.59 @xmath196 0.18 @xmath196 0.04 & 0.0043 + 2.36 & 0.253 & 0.41 & 199 & -0.02 @xmath196 0.12@xmath196 0.04 & 0.00 @xmath196 0.14 @xmath196 0.03 & -0.0079 & 0.59 @xmath196 0.16 @xmath196 0.04 & 0.0044 + 2.36 & 0.254 & 0.43 & 235 & -0.28 @xmath196 0.10@xmath196 0.03 & -0.05 @xmath196 0.12 @xmath196 0.03 & -0.0184 & 0.71 @xmath196 0.13 @xmath196 0.04 & 0.0045 + 2.43 & 0.261 & 0.45 & 270 & -0.26 @xmath196 0.07@xmath196 0.02 & -0.12 @xmath196 0.09 @xmath196 0.03 & -0.0161 & 0.60 @xmath196 0.10 @xmath196 0.02 & 0.0023 + 2.54 & 0.274 & 0.45 & 307 & -0.24 @xmath196 0.07@xmath196 0.03 & -0.22 @xmath196 0.08 @xmath196 0.03 & -0.0065 & 0.43 @xmath196 0.09 @xmath196 0.05 & -0.0004 + 2.61 & 0.284 & 0.42 & 330 & -0.15 @xmath196 0.19@xmath196 0.07 & -0.03 @xmath196 0.18 @xmath196 0.03 & 0.0004 & 0.59 @xmath196 0.27 @xmath196 0.14 & 0.0000 + 2.43 & 0.261 & 1.22 & 20 & -0.01 @xmath196 0.08@xmath196 0.02 & 0.08 @xmath196 0.09 @xmath196 0.01 & 0.0007 & 0.30 @xmath196 0.11 @xmath196 0.02 & -0.0068 + 2.45 & 0.265 & 1.06 & 52 & 0.14 @xmath196 0.09@xmath196 0.03 & -0.00 @xmath196 0.11 @xmath196 0.03 & 0.0034 & 0.60 @xmath196 0.12 @xmath196 0.02 & -0.0035 + 2.37 & 0.255 & 0.96 & 88 & 0.18 @xmath196 0.20@xmath196 0.12 & 0.21 @xmath196 0.22 @xmath196 0.05 & 0.0085 & 0.65 @xmath196 0.26 @xmath196 0.07 & 0.0004 + 2.43 & 0.261 & 0.96 & 273 & -0.31 @xmath196 0.17@xmath196 0.05 & 0.09 @xmath196 0.20 @xmath196 0.08 & -0.0069 & 0.49 @xmath196 0.23 @xmath196 0.06 & 0.0006 + 2.44 & 0.262 & 1.08 & 308 & -0.16 @xmath196 0.10@xmath196 0.04 & -0.04 @xmath196 0.11 @xmath196 0.03 & -0.0037 & 0.45 @xmath196 0.13 @xmath196 0.05 & 0.0045 + 2.46 & 0.265 & 1.23 & 340 & -0.04 @xmath196
0.07@xmath196 0.02 & -0.09 @xmath196 0.09 @xmath196 0.01 & -0.0007 & 0.44 @xmath196 0.10 @xmath196 0.03 & 0.0059 + 2.55 & 0.319 & 0.16 & 19 & 0.13 @xmath196 0.05@xmath196 0.02 & 0.07 @xmath196 0.07 @xmath196 0.01 & 0.0020 & 0.73 @xmath196 0.07 @xmath196 0.06 & 0.0546 + 2.50 & 0.319 & 0.16 & 50 & 0.25 @xmath196 0.07@xmath196 0.03 & 0.10 @xmath196 0.09 @xmath196 0.01 & 0.0045 & 0.60 @xmath196 0.10 @xmath196 0.06 & 0.0513 + 2.53 & 0.320 & 0.16 & 88 & 0.37 @xmath196 0.14@xmath196 0.08 & 0.16 @xmath196 0.17 @xmath196 0.03 & 0.0066 & 0.74 @xmath196 0.19 @xmath196 0.05 & 0.0464 + 2.55 & 0.317 & 0.16 & 240 & -0.13 @xmath196 0.22@xmath196 0.01 & -0.19 @xmath196 0.26 @xmath196 0.05 & -0.0064 & 0.55 @xmath196 0.29 @xmath196 0.03 & 0.0345 + 2.59 & 0.321 & 0.16 & 273 & -0.12 @xmath196 0.15@xmath196 0.05 & -0.24 @xmath196 0.17 @xmath196 0.07 & -0.0078 & 0.93 @xmath196 0.19 @xmath196 0.09 & 0.0479 + 2.53 & 0.319 & 0.16 & 309 & -0.21 @xmath196 0.07@xmath196 0.02 & -0.11 @xmath196 0.08 @xmath196 0.01 & -0.0045 & 0.77 @xmath196 0.09 @xmath196 0.04 & 0.0528 + 2.55 & 0.318 & 0.16 & 342 & -0.12 @xmath196 0.06@xmath196 0.02 & 0.02 @xmath196 0.07 @xmath196 0.01 & -0.0005 & 0.70 @xmath196 0.08 @xmath196 0.05 & 0.0534 + 2.72 & 0.338 & 0.24 & 19 & 0.08 @xmath196 0.03@xmath196 0.00 & 0.10 @xmath196 0.03 @xmath196 0.01 & 0.0000 & 0.72 @xmath196 0.04 @xmath196 0.03 & 0.0398 + 2.68 & 0.338 & 0.24 & 51 & 0.20 @xmath196 0.03@xmath196 0.01 & 0.11 @xmath196 0.04 @xmath196 0.01 & 0.0050 & 0.69 @xmath196 0.05 @xmath196 0.03 & 0.0387 + 2.67 & 0.337 & 0.24 & 89 & 0.16 @xmath196 0.05@xmath196 0.02 & 0.15 @xmath196 0.07 @xmath196 0.02 & 0.0223 & 0.69 @xmath196 0.07 @xmath196 0.03 & 0.0293 + 2.43 & 0.337 & 0.25 & 122 & 0.13 @xmath196 0.07@xmath196 0.01 & 0.19 @xmath196 0.08 @xmath196 0.01 & 0.0373 & 0.72 @xmath196 0.09 @xmath196 0.04 & 0.0164 + 2.47 & 0.326 & 0.25 & 161 & 0.14 @xmath196 0.12@xmath196 0.01 & 0.25 @xmath196 0.14 @xmath196 0.03 & 0.0214 & 0.59 @xmath196 0.15 @xmath196 0.07 & -0.0042 + 2.55 & 0.322 & 0.25 & 201 & -0.07 @xmath196 0.09@xmath196 0.03 & -0.25 @xmath196 0.11 @xmath196 0.06 & -0.0179 & 0.75 @xmath196 0.12 @xmath196 0.02 & -0.0020 + 2.55 & 0.338 & 0.24 & 238 & -0.20 @xmath196 0.07@xmath196 0.02 & -0.18 @xmath196 0.09 @xmath196 0.01 & -0.0346 & 0.71 @xmath196 0.09 @xmath196 0.03 & 0.0142 + 2.73 & 0.337 & 0.24 & 271 & -0.26 @xmath196 0.06@xmath196 0.03 & -0.28 @xmath196 0.07 @xmath196 0.02 & -0.0233 & 0.68 @xmath196 0.07 @xmath196 0.03 & 0.0278 + 2.69 & 0.336 & 0.24 & 309 & -0.26 @xmath196 0.03@xmath196 0.02 & -0.16 @xmath196 0.04 @xmath196 0.01 & -0.0058 & 0.73 @xmath196 0.05 @xmath196 0.03 & 0.0385 + 2.68 & 0.338 & 0.24 & 341 & -0.04 @xmath196 0.03@xmath196 0.00 & -0.08 @xmath196 0.03 @xmath196 0.01 & 0.0002 & 0.59 @xmath196 0.04 @xmath196 0.03 & 0.0395 + 2.81 & 0.356 & 0.44 & 20 & 0.09 @xmath196 0.03@xmath196 0.01 & 0.04 @xmath196 0.03 @xmath196 0.01 & -0.0055 & 0.59 @xmath196 0.04 @xmath196 0.02 & 0.0123 + 2.74 & 0.346 & 0.46 & 51 & 0.20 @xmath196 0.03@xmath196 0.01 & 0.13 @xmath196 0.03 @xmath196 0.01 & 0.0049 & 0.66 @xmath196 0.04 @xmath196 0.02 & 0.0143 + 2.68 & 0.346 & 0.45 & 90 & 0.15 @xmath196 0.05@xmath196 0.02 & 0.18 @xmath196 0.06 @xmath196 0.02 & 0.0358 & 0.68 @xmath196 0.06 @xmath196 0.02 & 0.0135 + 2.24 & 0.349 & 0.47 & 126 & 0.07 @xmath196 0.05@xmath196 0.01 & 0.19 @xmath196 0.06 @xmath196 0.03 & 0.0636 & 0.53 @xmath196 0.06 @xmath196 0.02 & 0.0097 + 2.25 & 0.339 & 0.49 & 162 & 0.06 @xmath196 0.06@xmath196 0.01 & 0.25 @xmath196 0.07 @xmath196 0.03 & 0.0325 & 0.68 @xmath196 0.08 @xmath196 0.02 & 0.0061 + 2.23 & 0.339 & 0.49 & 202 & -0.13 @xmath196 0.06@xmath196 0.02 & -0.15 @xmath196 0.07 @xmath196 0.02 & -0.0296 & 0.54 @xmath196 0.08 @xmath196 0.04 & 0.0022 + 2.43 & 0.347 & 0.47 & 232 & -0.06 @xmath196 0.06@xmath196 0.01 & -0.21 @xmath196 0.08 @xmath196 0.03 & -0.0567 & 0.38 @xmath196 0.09 @xmath196 0.02 & 0.0034 + 2.72 & 0.344 & 0.47 & 271 & -0.22 @xmath196 0.05@xmath196 0.02 & -0.26 @xmath196 0.06 @xmath196 0.02 & -0.0362 & 0.62 @xmath196 0.06 @xmath196 0.04 & 0.0118 + 2.77 & 0.347 & 0.45 & 309 & -0.21 @xmath196 0.03@xmath196 0.01 & -0.18 @xmath196 0.04 @xmath196 0.01 & -0.0058 & 0.64 @xmath196 0.04 @xmath196 0.02 & 0.0137 + 2.77 & 0.355 & 0.45 & 340 & -0.15 @xmath196 0.02@xmath196 0.01 & -0.07 @xmath196 0.03 @xmath196 0.01 & 0.0049 & 0.59 @xmath196 0.03 @xmath196 0.02 & 0.0129 + 2.59 & 0.352 & 1.30 & 20 & 0.02 @xmath196 0.03@xmath196 0.00 & 0.04 @xmath196 0.04 @xmath196 0.01 & -0.0041 & 0.36 @xmath196 0.04 @xmath196 0.03 & -0.0051 + 2.59 & 0.348 & 1.09 & 52 & 0.13 @xmath196 0.04@xmath196 0.01 & 0.15 @xmath196 0.05 @xmath196 0.02 & 0.0088 & 0.39 @xmath196 0.06 @xmath196 0.02 & -0.0035 + 2.28 & 0.350 & 1.28 & 87 & 0.12 @xmath196 0.06@xmath196 0.02 & 0.09 @xmath196 0.08 @xmath196 0.06 & 0.0326 & 0.36 @xmath196 0.08 @xmath196 0.03 & -0.0008 + 2.27 & 0.353 & 1.09 & 129 & 0.08 @xmath196 0.06@xmath196 0.03 & 0.19 @xmath196 0.08 @xmath196 0.05 & 0.0375 & 0.30 @xmath196 0.08 @xmath196 0.04 & -0.0025 + 2.33 & 0.345 & 1.06 & 161 & -0.00 @xmath196 0.06@xmath196 0.02 & 0.21 @xmath196 0.08 @xmath196 0.03 & 0.0188 & 0.52 @xmath196 0.08 @xmath196 0.02 & 0.0015 + 2.29 & 0.347 & 1.06 & 201 & -0.00 @xmath196 0.06@xmath196 0.02 & -0.19 @xmath196 0.08 @xmath196 0.03 & -0.0184 & 0.52 @xmath196 0.08 @xmath196 0.02 & 0.0020 + 2.33 & 0.352 & 1.23 & 236 & -0.01 @xmath196 0.06@xmath196 0.02 & -0.25 @xmath196 0.09 @xmath196 0.06 & -0.0384 & 0.40 @xmath196 0.09 @xmath196 0.02 & 0.0003 + 2.25 & 0.347 & 1.23 & 269 & -0.04 @xmath196 0.05@xmath196 0.01 & -0.19 @xmath196 0.06 @xmath196 0.04 & -0.0356 & 0.30 @xmath196 0.06 @xmath196 0.02 & -0.0027 + 2.61 & 0.349 & 1.09 & 308 & -0.11 @xmath196 0.05@xmath196 0.01 & -0.14 @xmath196 0.06 @xmath196 0.02 & -0.0086 & 0.48 @xmath196 0.06 @xmath196 0.03 & -0.0016 + 2.57 & 0.352 & 1.28 & 340 & -0.02 @xmath196 0.03@xmath196 0.00 & -0.07 @xmath196 0.03 @xmath196 0.01 & 0.0036 & 0.43 @xmath196 0.04 @xmath196 0.02 & -0.0034 + 3.21 & 0.414 & 0.27 & 19 & -0.24 @xmath196
0.18@xmath196 0.02 & 0.00 @xmath196 0.00 @xmath196 0.00 & 0.0000 & 0.90 @xmath196 0.21 @xmath196 0.02 & 0.0046 + 3.16 & 0.414 & 0.27 & 51 & 0.22 @xmath196 0.21@xmath196 0.09 & -0.04 @xmath196 0.26 @xmath196 0.08 & -0.0007 & 0.30 @xmath196 0.29 @xmath196 0.09 & 0.0070 + 3.36 & 0.443 & 0.51 & 17 & 0.07 @xmath196 0.04@xmath196 0.01 & -0.02 @xmath196 0.04 @xmath196 0.01 & -0.0034 & 0.82 @xmath196 0.05 @xmath196 0.03 & 0.0398 + 3.41 & 0.444 & 0.49 & 52 & 0.21 @xmath196 0.05@xmath196 0.02 & 0.13 @xmath196 0.06 @xmath196 0.02 & 0.0066 & 0.77 @xmath196 0.06 @xmath196 0.04 & 0.0395 + 3.25 & 0.448 & 0.49 & 89 & 0.24 @xmath196 0.07@xmath196 0.03 & 0.27 @xmath196 0.09 @xmath196 0.04 & 0.0035 & 0.77 @xmath196 0.10 @xmath196 0.05 & 0.0397 + 2.84 & 0.430 & 0.53 & 124 & 0.12 @xmath196 0.09@xmath196 0.05 & 0.27 @xmath196 0.11 @xmath196 0.03 & 0.0548 & 0.64 @xmath196 0.12 @xmath196 0.07 & 0.0284 + 2.93 & 0.420 & 0.57 & 155 & -0.08 @xmath196 0.24@xmath196 0.08 & 0.07 @xmath196 0.31 @xmath196 0.08 & 0.0227 & 0.44 @xmath196 0.30 @xmath196 0.08 & 0.0040 + 2.94 & 0.421 & 0.58 & 207 & -0.44 @xmath196 0.23@xmath196 0.06 & -0.25 @xmath196 0.28 @xmath196 0.07 & -0.0273 & 0.50 @xmath196 0.31 @xmath196 0.07 & 0.0082 + 2.95 & 0.435 & 0.50 & 238 & -0.04 @xmath196 0.11@xmath196 0.05 & -0.18 @xmath196 0.13 @xmath196 0.05 & -0.0497 & 0.75 @xmath196 0.14 @xmath196 0.07 & 0.0309 + 3.33 & 0.440 & 0.48 & 271 & -0.08 @xmath196 0.08@xmath196 0.02 & -0.24 @xmath196 0.10 @xmath196 0.01 & -0.0322 & 0.82 @xmath196 0.11 @xmath196 0.04 & 0.0381 + 3.46 & 0.445 & 0.50 & 310 & -0.28 @xmath196 0.05@xmath196 0.02 & -0.23 @xmath196 0.06 @xmath196 0.01 & -0.0879 & 0.68 @xmath196 0.06 @xmath196 0.04 & 0.0364 + 3.35 & 0.443 & 0.52 & 342 & -0.08 @xmath196 0.03@xmath196 0.01 & -0.16 @xmath196 0.04 @xmath196 0.01 & -0.0010 & 0.77 @xmath196 0.04 @xmath196 0.03 & 0.0392 + 3.37 & 0.465 & 1.12 & 18 & 0.06 @xmath196 0.03@xmath196 0.01 & 0.08 @xmath196 0.04 @xmath196 0.01 & -0.0064 & 0.51 @xmath196 0.04 @xmath196 0.02 & 0.0015 + 3.46 & 0.465 & 1.13 & 52 & 0.21 @xmath196 0.05@xmath196 0.02 & 0.16 @xmath196 0.07 @xmath196 0.03 & 0.0095 & 0.60 @xmath196 0.07 @xmath196 0.03 & 0.0071 + 3.18 & 0.459 & 1.27 & 89 & -0.00 @xmath196 0.07@xmath196 0.02 & 0.15 @xmath196 0.10 @xmath196 0.05 & 0.0507 & 0.55 @xmath196 0.10 @xmath196 0.04 & 0.0078 + 3.13 & 0.456 & 1.15 & 126 & 0.11 @xmath196 0.06@xmath196 0.01 & 0.04 @xmath196 0.08 @xmath196 0.05 & 0.0558 & 0.55 @xmath196 0.08 @xmath196 0.02 & 0.0007 + 3.30 & 0.434 & 1.18 & 159 & 0.07 @xmath196 0.11@xmath196 0.09 & 0.12 @xmath196 0.15 @xmath196 0.04 & 0.0241 & 0.43 @xmath196 0.15 @xmath196 0.07 & -0.0116 + 3.11 & 0.430 & 1.09 & 203 & -0.10 @xmath196 0.13@xmath196 0.03 & 0.00 @xmath196 0.16 @xmath196 0.06 & -0.0219 & 0.78 @xmath196 0.16 @xmath196 0.06 & -0.0021 + 3.16 & 0.455 & 1.25 & 236 & -0.06 @xmath196 0.07@xmath196 0.01 & -0.30 @xmath196 0.10 @xmath196 0.06 & -0.0644 & 0.46 @xmath196 0.09 @xmath196 0.03 & -0.0026 + 3.12 & 0.454 & 1.35 & 269 & -0.02 @xmath196 0.06@xmath196 0.02 & -0.21 @xmath196 0.09 @xmath196 0.05 & -0.0541 & 0.56 @xmath196 0.08 @xmath196 0.03 & 0.0083 + 3.49 & 0.467 & 1.10 & 310 & -0.31 @xmath196 0.06@xmath196 0.04 & -0.25 @xmath196 0.07 @xmath196 0.03 & -0.0119 & 0.77 @xmath196 0.08 @xmath196 0.05 & 0.0115 + 3.36 & 0.464 & 1.15 & 342 & -0.07 @xmath196 0.03@xmath196 0.01 & -0.08 @xmath196 0.04 @xmath196 0.01 & 0.0063 & 0.51 @xmath196 0.04 @xmath196 0.02 & 0.0014 + 99 d. mller , d. robaschik , b. geyer , f .-
dittes and j. horejsi , fortschr .
phys . * 42 * , 101 ( 1994 ) .
x. ji , phys .
lett . * 78 * , 610 ( 1997 ) ; phys .
d * 55 * , 7114 ( 1997 ) .
radyushkin , phys .
b * 380 * , 417 ( 1996 ) ; phys .
d * 56 * , 5524 ( 1997 ) .
collins , l. frankfurt and m. strikman , phys .
d * 56 * , 2982 ( 1997 ) .
k. goeke , m. v. polyakov and m. vanderhaeghen , prog .
. phys . * 47 * , 401 ( 2001 ) .
m. diehl , phys .
rept . * 388 * , 41 ( 2003 ) .
a.v . belitsky and a.v .
radyushkin , phys . rept .
* 418 * , 1 ( 2005 ) .
jaffe , `` spin , twist and hadron structure in deep inelastic processes '' lecture notes in physics volume * 496 * , 178 ( 1997 ) .
belitsky , d. mller and a. kirchner , nucl .
b * 629 * , 323 ( 2002 ) .
m. burkhardt , phys .
d * 62 * , 71503 ( 2000 ) . c. muoz camacho _
( hall - a collaboration ) , phys .
lett . * 97 * , 262002 ( 2006 ) .
s. stepanyan _
( clas collaboration ) , phys .
lett . * 87 * , 182002 ( 2001 ) .
( clas collaboration ) , phys .
lett . * 97 * , 072002 ( 2006 ) .
b.a . mecking _
et al . _ ,
a503 * , 513 ( 2003 ) .
f .- x . girod _
( clas collaboration ) , phys .
lett . * 100 * , 162002 ( 2008 ) .
a. airapetian _
( hermes collaboration ) , jhep06 * 0806 * , 066 ( 2008 ) ; a. airapetian _
( hermes collaboration ) , jhep * 06 * , 019 ( 2010 ) .
et al . _ ,
a501 * , 327 ( 2003 ) .
m. amarian _
et al . _ , nucl .
a460 * , 239 ( 2001 ) .
i. bedlinskiy _
( clas collaboration ) , phys .
lett . * 109 * , 112001 ( 2012 ) .
t. w. donnelly and a. s. raskin , annals phys .
* 169 * , 247 ( 1986 ) .
a. puckett _ et al .
lett . * 104 * , 242301 ( 2010 ) .
m. diehl and s. sapeta , eur .
j. c * 41 * , 515 ( 2005 ) .
m. vanderhaeghen , p.a.m .
guichon and m. guidal , phys .
d * 60 * , 094017 ( 1999 ) ; m. guidal , m.v .
polyakov , a.v .
radyushkin , m. vanderhaeghen , phys .
d * 72 * , 054013 ( 2005 ) .
a.v . afanasev _
et al . _ ,
. phys . * 102 * , 220 ( 2006 ) .
p. kroll , h. moutarde and f. sabati , eur .
j. c * 73 * , 2278 ( 2013 ) .
k. kumericki , d. mller and m. murray , arxiv:1301.1230 [ hep - ph ] .
s. liuti , private communication , and g. r. goldstein , j. o. gonzalez - hernandez and s. liuti , phys .
d * 84 * , 034007 ( 2011 ) ; j. o. gonzalez - hernandez , s. liuti , g. r. goldstein and k. kathuria , phys .
c * 88 * , 065206 ( 2013 ) .
radyushkin , phys .
d * 59 * , 014030 ( 1999 ) .
cates , c.w
. de jager , s. riordan and b. wojtsekhowski , phys .
lett . * 106 * , 252003 ( 2011 ) .
http://clasweb.jlab.org/physicsdb/intro.html e. seder _ et al .
_ ( clas collaboration ) , phys . rev
. lett . * 114 * , 032001 ( 2015 ) .
m. guidal , phys .
b * 693 * , 17 ( 2010 ) .
a. courtoy _
et al . _ ,
b * 731 * , 141 ( 2014 ) .
m. guidal , eur .
j. a * 37 * , 319 ( 2008 ) ; m. guidal , eur .
j. a * 40 * , 119 ( 2009 ) , erratum .
m. guidal and h. moutarde , eur .
j. a * 42 * , 71 ( 2009 ) .
m. guidal , h. moutarde , m. vanderhaeghen , rep .
. phys . * 76 * , 066202 ( 2013 ) .
m. guidal , phys .
b * 689 * , 156 ( 2010 ) .
et al . _ , in preparation . | single - beam , single - target , and double - spin asymmetries for hard exclusive photon production on the proton @xmath0 are presented .
the data were taken at jefferson lab using the clas detector and a longitudinally polarized @xmath1nh@xmath2 target .
the three asymmetries were measured in 165 4-dimensional kinematic bins , covering the widest kinematic range ever explored simultaneously for beam and target - polarization observables in the valence quark region .
the kinematic dependences of the obtained asymmetries are discussed and compared to the predictions of models of generalized parton distributions .
the measurement of three dvcs spin observables at the same kinematic points allows a quasi - model - independent extraction of the imaginary parts of the @xmath3 and @xmath4 compton form factors , which give insight into the electric and axial charge distributions of valence quarks in the proton . |
the hankel transform of order @xmath2 of a function @xmath3\rightarrow\mathbb{c}$ ] is defined as ( * ? ? ? * chap . 9 )
@xmath4 where @xmath5 is the bessel function of the first kind with parameter @xmath6 . when @xmath7 , is essentially the two - dimensional fourier transform of a radially symmetric function with support on the unit disc and appears in the fourier
abel cycle ( * ? ? ?
furthermore , the hankel transform of integer order @xmath8 is equivalent to the fourier transform of functions of the form @xmath9 , where @xmath10 . throughout this paper
we take @xmath6 to be an integer .
various semi - discrete and discrete variants of are employed in optics @xcite , electromagnetics @xcite , medical imaging @xcite , and the numerical solution of partial differential equations @xcite . at some stage
these require the computation of the following sums : @xmath11 where @xmath6 is an integer , @xmath12 , and @xmath13 .
special cases of that are considered in this paper are : ( 1 ) the evaluation of schlmilch expansions at @xmath14 @xcite ( @xmath15 or @xmath16 ) ; ( 2 ) the evaluation of fourier bessel expansions of order @xmath0 at @xmath14 ( @xmath7 and @xmath17 , where @xmath18 is the @xmath19th positive root of @xmath20 ) ; and
( 3 ) the discrete hankel transform of order @xmath0 @xcite ( @xmath7 , @xmath17 , and @xmath21 ) . for algorithms that rely on the fast fourier
transform ( fft ) , such as the algorithm described in this paper , tasks ( 1 ) , ( 2 ) , and ( 3 ) are incrementally more difficult .
we consider them in turn in sections [ sec : schlomilch ] , [ sec : fourierbessel ] , and [ sec : discretehankeltransform ] , respectively .
our algorithm is based on carefully replacing @xmath22 by an asymptotic expansion for large arguments that , up to a certain error , expresses @xmath22 as a weighted sum of trigonometric functions .
this allows a significant proportion of the computation involved in to be achieved by the discrete cosine transform of type i ( dct ) and the discrete sine transform of type i ( dst ) .
these are @xmath23 fft - based algorithms for computing matrix - vector products with the matrices @xmath24 and @xmath25 , where @xmath26 asymptotic expansions of special functions are notoriously tricky to use in practice because it is often difficult to determine the precise regime for which they are accurate approximations .
however , for transforms involving orthogonal polynomials , such as the chebyshev
legendre transform @xcite , successful fast algorithms have been derived based on asymptotic expansions . at the point of evaluation , these algorithms are competitive with multipole - like approaches @xcite , but have the advantage of no precomputation .
this allows for more adaptive algorithms such as a convolution algorithm for chebyshev expansions @xcite . in this paper
we will use error bounds to determine the precise regime for which an asymptotic expansion of bessel functions is an accurate approximation and from there derive all the algorithmic parameters to achieve a near - optimal computational cost for any accuracy goal ( see table [ tab : algorithmicparameters ] ) .
for each task ( 1)-(3 ) there are just two user - defined inputs : a vector of coefficients @xmath27 in and a working accuracy @xmath28 .
the values of @xmath29 in are then calculated to an accuracy of @xmath30 .
the resulting algorithm has a complexity of @xmath31 , where @xmath32 for task ( 1 ) , @xmath33 for task ( 2 ) , and @xmath34 for task ( 3 ) . for the majority of this paper we state algorithmic complexities without the dependency on @xmath35 .
previous algorithms based on asymptotic expansions for computing @xcite have either a poor accuracy or a dubious numerical stability @xcite . here
, we find that once an asymptotic expansion of bessel functions is carefully employed the resulting algorithm is numerically stable , can be adapted to any accuracy goal , and requires no precomputation .
moreover , we show how the equally - spaced restriction , which is inherited from the reliance on dcts and dsts , can be alleviated .
we use the following notation . a column vector with entries @xmath36
is denoted by @xmath37 , a row vector is denoted by @xmath38 , a diagonal matrix with diagonal entries @xmath36 is written as @xmath39 , and the @xmath40 matrix with @xmath41 entry @xmath42 is denoted by @xmath43 where @xmath44 .
the paper is organized as follows . in section [ sec : existing ] we briefly discuss existing methods for computing and in section [ sec : besselfunctionproperties ] we investigate the approximation power of three expansions for bessel functions . in section [ sec : schlomilch ] we first describe an @xmath45 algorithm for evaluating schlmilch expansions before deriving a faster @xmath1 algorithm . in sections [ sec : fourierbessel ] and [ sec : discretehankeltransform ] we use the neumann addition formula to extend the algorithm to the evaluation of fourier bessel expansions and the computation of the discrete hankel transform .
we now give a brief description of some existing methods for evaluating that roughly fall into four categories : ( 1 ) direct summation ; ( 2 ) evaluation via an integral representation ; ( 3 ) evaluation using asymptotic expansions , and ; ( 4 ) butterfly schemes .
for other approaches see @xcite .
one of the simplest ideas is to evaluate @xmath46 for @xmath47 and then naively compute the sums in . by using the taylor series expansion in for small @xmath48 , the asymptotic expansion in for large @xmath48 , and miller s algorithm ( * ? ? ?
3.6(vi ) ) ( backward recursion with a three - term recurrence ) otherwise , each evaluation of @xmath49 costs @xmath50 operations @xcite . therefore
, this simple algorithm requires a total of @xmath51 operations and is very efficient for small @xmath52 ( @xmath53 ) .
however , the @xmath51 cost is prohibitive for large @xmath52 .
note that in general there is no evaluation scheme for @xmath49 that is exact in infinite precision arithmetic . at some level a working accuracy of @xmath28 must be introduced .
thus , even in the direct summation approach , the relaxation of the working accuracy that is necessary for deriving a faster algorithm , has already been made .
the sums in are equivalent to the matrix - vector product @xmath54 .
one reason that an @xmath1 algorithm is possible is because the argument , @xmath55 , is a rank @xmath56 matrix .
a fact that we will exploit several times .
when @xmath6 is an integer , @xmath22 can be replaced by its integral representation ( * ? ? ?
* ( 10.9.2 ) ) , @xmath57 where @xmath58 is the imaginary unit and the last equality holds because @xmath59 is even about @xmath60 .
the integrand in is @xmath61-periodic and the integral can be approximated by a @xmath62-point trapezium rule .
substituting this approximation into gives a double sum for @xmath63 that can be evaluated with one fft followed by an interpolation step in @xmath64 operations @xcite .
the major issue with this approach is that the integrand in is highly oscillatory for large @xmath48 and often @xmath65 , resulting in @xmath51 operations .
one can replace @xmath22 in by an asymptotic expansion that involves trigonometric functions . a fast , but perhaps numerically unstable ( see figure [ fig : partition ] ) , algorithm then follows by exploiting dcts and dsts .
existing work has approximated @xmath49 using asymptotic expansions with rational coefficients @xcite , the approximation @xmath66
@xcite , and for half - integer @xmath6 the asymptotic expansion in @xcite .
however , none of these approaches have rigorously determined the regime in which the employed asymptotic expansion is an accurate approximation and instead involve dubiously chosen algorithmic parameters . in this paper
we will use a well - established asymptotic expansion and rigorously determine the regime in which it is an accurate approximation . in 1995 ,
kapur and rokhlin described an algorithm for computing the integral in when @xmath67 @xcite .
the main idea is to write as a product of two integrals as follows ( * ? ? ?
* ( 6 ) ) : @xmath68 the inner integral can be discretized by a ( corrected ) trapnezium rule and computed via a discrete cosine transform , while the outer integral involves a singular kernel and can be computed with the fast multipole method .
as described in @xcite the algorithm can be used to evaluate schlmilch expansions in @xmath23 operations ( see section [ sec : schlomilch ] ) . with modern advances in fast algorithms for nonuniform transforms , it could now be adapted to the computation of the discrete hankel transform though we are not aware of such work .
recently , a new algorithm was developed for a class of fast transforms involving special functions @xcite .
the algorithm compresses the ranks of submatrices of a change of basis matrix by constructing skeleton decompositions in a hierarchical fashion .
this hierarchical decomposition of the matrix means that it can then be applied to a vector in @xmath23 operations .
the compression step of the algorithm is the dominating cost and is observed to require @xmath51 operations ( * ? ? ?
* table .
fortunately , in many applications the compression step can be regarded as a precomputational cost because it is independent of the values of @xmath27 in , though it does depend on @xmath52 .
the methodology behind butterfly schemes is more general than what we present in this paper as it does not require knowledge of asymptotics . here
, we will derive a fast transform that has no precomputational cost and is better suited to applications where @xmath52 is not known in advance
. some of the ideas in the spherical harmonic transform literature @xcite may be useful for reducing the precomputational cost in @xcite , though we are not aware of work in this direction .
in this section we investigate three known expansions of bessel functions ( * ? ? ? * chap .
10 ) : ( 1 ) an explicit asymptotic expansion for large arguments ; ( 2 ) a convergent taylor series expansion for small arguments ; and ( 3 ) the neumann addition formula for perturbed arguments .
mathematically , these three expansions are infinite series , but for practical use they must be truncated . here
, we will focus on deriving criteria to ensure that the errors introduced by such truncations are negligible .
related techniques were successfully used to derive a fast algorithm for the discrete legendre transform @xcite .
the graph of @xmath22 on @xmath69 is often said to look like an oscillating trigonometric function that decays like @xmath70 .
this observation can be made precise by an asymptotic expansion of @xmath22 for large @xmath48 .
let @xmath6 and @xmath71 be integers and @xmath72 .
the first @xmath73 terms of the hankel asymptotic expansion of @xmath22 is given by ( * ? ? ?
* ( 10.17.3 ) ) ( also see @xcite ) : @xmath74 where @xmath75 , @xmath76 is the error term , @xmath77 , and @xmath78 the first term in shows that the leading asymptotic behavior of @xmath22 as @xmath79 is @xmath80 , which is an oscillating trigonometric function that decays like @xmath70 .
the first @xmath73 terms show that , up to an error of @xmath76 , @xmath22 can be written as a weighted sum of @xmath81 and @xmath82 for @xmath83 .
figure [ fig : asydivergence ] ( left ) shows the error term @xmath84 on @xmath85 $ ] for @xmath86 .
as expected , the asymptotic expansion becomes more accurate as @xmath48 increases .
in particular , for sufficiently large @xmath48 the error term is negligible , i.e. , @xmath87 .
it is important to appreciate that is an asymptotic expansion , as opposed to a convergent series , and does not converge pointwise to @xmath22 as @xmath88 . in practice , increasing @xmath89 will eventually be detrimental and lead to severe numerical overflow issues .
figure [ fig : asydivergence ] ( right ) shows the error @xmath90 as @xmath88 and @xmath91 .
thus , the appropriate use of is to select an @xmath71 and ask : `` for what sufficiently large @xmath48 is the asymptotic expansion accurate ? '' .
for example , from figure [ fig : asydivergence ] ( left ) we observe that for @xmath92 we have @xmath93 for any @xmath94 and hence , we may safely replace @xmath95 by a weighted sum of @xmath81 and @xmath82 for @xmath96 when @xmath94 . besselasymptotics ( 50,0 )
@xmath48 ( -1.5,27 ) ( 75,51.5 ) ( 74,41 ) ( 73,32 ) ( 71,25.2 ) ( 68,19 ) ( 44,20 ) ( 35.2,22.2 ) fixedz ( 50,0 ) @xmath89 ( -1.5,27 ) more generally , it is known that the error term @xmath76 is bounded by the size of the first neglected terms @xcite ) .
since @xmath97 and @xmath98 we have @xmath99 this is essentially a sharp bound since for any @xmath100 there is a @xmath72 such that @xmath101 is larger than the magnitude of the first neglected terms . of the form @xmath102 for sufficiently large @xmath103 . ]
let @xmath104 be the smallest real number so that if @xmath105 then @xmath106
. from we can take @xmath104 as the number that solves @xmath107 in general , the equation above has no closed - form solution , but can be solved numerically by a fixed - point iteration .
we start with the initial guess of @xmath108 and iterate as follows : @xmath109 we terminate this after four iterations and take @xmath110 .
table [ tab : smtable ] gives the calculated values of @xmath111 for @xmath112 and @xmath113 .
cccccccccccc ' '' '' @xmath89 ' '' '' & 3 & 4 & 5 & 6 & 7 & 8 & 9 & 10 & 11 & 12 + ' '' '' @xmath114 ' '' '' & 180.5 & 70.5 & 41.5 & 30.0 & 24.3 & 21.1 & 19.1 & 17.8 & 17.0 & 16.5 + @xmath115 & 185.2 & 71.5 & 41.9 & 30.2 & 24.4 & 21.1 & 19.2 & 17.9 & 17.1 & 16.5 + @xmath116 & 200.2 & 74.8 & 43.1 & 30.8 & 24.8 & 21.4 & 19.3 & 18.0 & 17.2 & 16.6 + @xmath117 & 2330.7 & 500.0 & 149.0 & 64.6 & 41.4 & 31.4 & 26.0 & 22.9 & 20.9 & 19.6 + for an integer @xmath6 , the taylor series expansion of @xmath22 about @xmath118 is given by ( * ? ? ?
* ( 10.2.2 ) ) @xmath119 in contrast to , the taylor series expansion converges pointwise to @xmath22 for any @xmath48 .
however , for practical application the infinite series in must still be truncated .
let @xmath120 be an integer and consider the truncated taylor series expansion that is given by @xmath121 where @xmath122 is the error term . as @xmath123 the leading asymptotic behavior of @xmath122 matches the order of the first neglected term and hence , @xmath124 .
in particular , there is a real number @xmath125 so that if @xmath126 then @xmath127 . to calculate the parameter @xmath125 we solve the equation @xmath128 .
remarkably , an explicit closed - form expression for @xmath129 is known @xcite and is given by @xmath130 where @xmath131 is the generalized hypergeometric function that we approximate by @xmath56 since we are considering @xmath48 to be small ( * ? ? ?
* ( 16.2.1 ) ) . solving @xmath128
we find that @xmath132 table [ tab : tttable ] gives the calculated values of @xmath133 for @xmath134 and @xmath113 .
ccccccccccc ' '' '' @xmath135 ' '' '' & 1 & 2 & 3 & 4 & 5 & 6 & 7 & 8 & 9 + ' '' '' @xmath136 ' '' '' & @xmath137 & 0.001 & 0.011 & 0.059 & 0.165 & 0.337 & 0.573 & 0.869 & 1.217 + @xmath138 & @xmath139 & 0.003 & 0.029 & 0.104 & 0.243 & 0.449 & 0.716 & 1.039 & 1.411 + @xmath140 & 0.001 & 0.012 & 0.061 & 0.168 & 0.341 & 0.579 & 0.876 & 1.225 & 1.618 + @xmath141 & 0.484 & 0.743 & 1.058 & 1.420 & 1.823 & 2.262 & 2.733 & 3.230 & 3.750 + the neumann addition formula expresses @xmath142 as an infinite sum of products of bessel functions of the form @xmath143 for @xmath144 .
it is given by ( * ? ? ?
* ( 10.23.2 ) ) @xmath145 here , we wish to truncate and use it as a numerical approximation to @xmath142 .
fortunately , when @xmath146 is small ( so that @xmath147 can be considered as a perturbation of @xmath48 ) the neumann addition formula is a rapidly converging series for @xmath148 .
let @xmath6 be an integer , @xmath72 , and @xmath149 .
then , for @xmath150 we have @xmath151 where @xmath152 is euler s number .
[ lem : neumannbound ] let @xmath6 be an integer , @xmath72 , @xmath149 , and @xmath150 . denote the left - hand side of the inequality in by @xmath153 so that @xmath154 since @xmath155 ( * ? ? ?
* ( 10.14.1 ) ) and @xmath156 ( * ? ? ?
* ( 10.4.1 ) ) , we have @xmath157 by kapteyn s inequality ( * ? ? ?
* ( 10.14.8 ) ) we can bound @xmath158 from above as follows : @xmath159 where the last inequality comes from @xmath160 for @xmath161 .
therefore , by substituting into we find that @xmath162 where we used @xmath163 and @xmath164 .
lemma [ lem : neumannbound ] shows that @xmath165 approximates @xmath166 , up to an error of @xmath35 , provided that @xmath167 .
equivalently , for any @xmath168 table [ tab : neumanntable ] gives the values of @xmath169 for @xmath170 .
ccccccccccc ' '' '' @xmath62 ' '' '' & 3 & 4 & 5 & 6 & 7 & 8 & 9 & 10 + ' '' '' ' '' '' & @xmath171 & @xmath172 & 0.001 & 0.002 & 0.004 & 0.008 & 0.013 & 0.020 + at first it may not be clear why lemma [ lem : neumannbound ] is useful for practical computations since a single evaluation of @xmath173 is replaced by a seemingly trickier sum of products of bessel functions .
however , in sections [ sec : fourierbessel ] and [ sec : discretehankeltransform ] the truncated neumann addition formula will be the approximation that will allow us to evaluate fourier
bessel expansions and compute the discrete hankel transform .
the crucial observation is that the positive roots of @xmath95 can be regarded as a perturbed equally - spaced grid .
let @xmath18 denote the @xmath19th positive root of @xmath95 .
we know from that the leading asymptotic behavior of @xmath95 is @xmath175 for large @xmath48 .
therefore , since @xmath176 for @xmath177 the leading asymptotic behavior of @xmath18 is @xmath178 for large @xmath19 . similarly , the leading asymptotic behavior of @xmath179 is @xmath180 for large @xmath103 and @xmath52 .
the next lemma shows that @xmath181 and the ratios @xmath182 can be regarded as perturbed equally - spaced grids .
let @xmath18 denote the @xmath19th positive root of @xmath95 .
then , @xmath183 and @xmath184 [ lem : besselinequalities ] the bounds on @xmath185 are given in ( * ? ? ?
3 ) . for the bound on @xmath186
we have ( since @xmath187 and @xmath188 ) @xmath189 the result follows from @xmath190 and @xmath191 .
lemma [ lem : besselinequalities ] shows that we can consider @xmath181 as a small perturbation of @xmath192 and that we can consider @xmath182 as a small perturbation of @xmath193 .
this is an important observation for sections [ sec : fourierbessel ] and [ sec : discretehankeltransform ] .
a schlmilch expansion of a function @xmath194\rightarrow \mathbb{c}$ ] takes the form @xcite : @xmath195 here , we take @xmath196 to be an integer and develop a fast @xmath1 algorithm for evaluating schlmilch expansions at @xmath197 , where @xmath198 .
the evaluation of at @xmath197 is equivalent to computing the matrix - vector product @xmath199 , i.e. , @xmath200 where @xmath201 and @xmath15 .
the @xmath41 entry of @xmath202 is @xmath203 , which takes a similar form to the @xmath204 entry of @xmath205 , @xmath206 , and the @xmath41 entry of @xmath207 , @xmath208 , see .
we will carefully use the asymptotic expansion in to approximate @xmath202 by a weighted sum of @xmath205 and @xmath207 . to derive a fast matrix - vector product for @xmath202
, we first substitute @xmath209 into the asymptotic expansion . by the trigonometric addition formula we have @xmath210 as a matrix decomposition we can write this as the following sum of @xmath211 and @xmath212 : @xmath213 where @xmath214 is a zero column vector , @xmath215^\intercal$ ] , @xmath216 , and @xmath217 .
similarly , for sine we have @xmath218 thus , by substituting and into with @xmath219 we obtain a matrix decomposition for @xmath202 , @xmath220 where we used the fact that @xmath55 is a rank @xmath56 matrix .
we can write more conveniently as @xmath221 one could now compute @xmath222 by first computing the vector @xmath223 and then correcting it by @xmath224 .
the matrix - vector product @xmath223 can be computed in @xmath23 operations since the expression in decomposes @xmath225 as a weighted sum of diagonal matrices and @xmath73 dcts and dsts . while this does lead to an @xmath23 evaluation scheme for schlmilch expansions , it is numerically unstable because the asymptotic expansion is employed for entries @xmath226 for which @xmath227 is small ( see section [ sec : asymptoticexpansion ] ) .
numerical overflow issues and severe cancellation errors plague this approach . figure [ fig : partition ] illustrates the entries that cause the most numerical problems .
we must be more careful and only employ the asymptotic expansion for the @xmath41 entry of @xmath228 if @xmath229 .
( 0,0 ) ; ( 5,0 )
( 8,0 ) ( 8,3 )
( 5,0 ) ; at ( .8,1.5 ) ( l1 ) @xmath202 ; at ( 5.6,1.5 ) ( l1 ) @xmath225 ; at ( 9.6,1.5 ) ( l2 ) @xmath230 ; at ( 8.5,1.7 ) ( l2 ) @xmath231 ; at ( 4,1.7 ) ( l2 ) @xmath232 ; ( 9,0)(0.200 * 3/5 + 9,0.000 * 3/5)(0.216 * 3/5 + 9,0.366 * 3/5)(0.221 * 3/5 + 9,0.476 * 3/5)(0.226 * 3/5 + 9,0.581 * 3/5)(0.232 * 3/5 + 9,0.682 * 3/5)(0.237 * 3/5 + 9,0.778 * 3/5)(0.242 * 3/5 + 9,0.870 * 3/5)(0.247 * 3/5 + 9,0.957 * 3/5)(0.253 * 3/5 + 9,1.042 * 3/5)(0.258 * 3/5 + 9,1.122 * 3/5)(0.263 * 3/5 + 9,1.200 * 3/5)(0.268 * 3/5 + 9,1.275 * 3/5)(0.274 * 3/5 + 9,1.346 * 3/5)(0.279 * 3/5 + 9,1.415 * 3/5)(0.284 * 3/5 + 9,1.481 * 3/5)(0.289 * 3/5 + 9,1.545 * 3/5)(0.295 * 3/5 + 9,1.607 * 3/5)(0.300 * 3/5 + 9,1.667 * 3/5 ) (0.385 * 3/5 + 9,2.400 * 3/5)(0.513 * 3/5 + 9,3.050 * 3/5)(0.641 * 3/5 + 9,3.440 * 3/5)(0.769 * 3/5 + 9,3.700 * 3/5)(0.897 * 3/5 + 9,3.886 * 3/5)(1.026 * 3/5 + 9,4.025 * 3/5)(1.154 * 3/5 + 9,4.133 * 3/5)(1.282 * 3/5 + 9,4.220 * 3/5)(1.410 * 3/5 + 9,4.291 * 3/5)(1.538 * 3/5 + 9,4.350 * 3/5)(1.667 * 3/5 + 9,4.400 * 3/5)(1.795 * 3/5 + 9,4.443 * 3/5)(1.923 * 3/5 + 9,4.480 * 3/5)(2.051 * 3/5 + 9,4.513 * 3/5)(2.179 * 3/5 + 9,4.541 * 3/5)(2.308 * 3/5 + 9,4.567 * 3/5)(2.436 * 3/5 + 9,4.589 * 3/5)(2.564 * 3/5 + 9,4.610 * 3/5)(2.692 * 3/5 + 9,4.629 * 3/5)(2.821
* 3/5 + 9,4.645 * 3/5)(2.949 * 3/5 + 9,4.661 * 3/5)(3.077 * 3/5 + 9,4.675 * 3/5)(3.205 * 3/5 + 9,4.688 * 3/5)(3.333 * 3/5 + 9,4.700 * 3/5)(3.462 * 3/5 + 9,4.711 * 3/5)(3.590 * 3/5 + 9,4.721 * 3/5)(3.718 * 3/5 + 9,4.731 * 3/5)(3.846 * 3/5 + 9,4.740 * 3/5)(3.974 * 3/5 + 9,4.748 * 3/5)(4.103 * 3/5 + 9,4.756 * 3/5)(4.231 * 3/5 + 9,4.764 * 3/5)(4.359 *
3/5 + 9,4.771 * 3/5)(4.487 * 3/5 + 9,4.777 * 3/5)(4.615 * 3/5 + 9,4.783 * 3/5)(4.744 * 3/5 + 9,4.789 * 3/5)(4.872 * 3/5 + 9,4.795 * 3/5)(5.000 * 3/5 + 9,4.800 * 3/5)(12,3)(9,3)(9,0 ) ; ( 0.200 * 3/5 + 9,0.000 * 3/5)(0.216 * 3/5 + 9,0.366 * 3/5)(0.221 * 3/5 + 9,0.476 * 3/5)(0.226 * 3/5 + 9,0.581 * 3/5)(0.232 * 3/5 + 9,0.682 * 3/5)(0.237
* 3/5 + 9,0.778 * 3/5)(0.242 * 3/5 + 9,0.870 * 3/5)(0.247 * 3/5 + 9,0.957 * 3/5)(0.253 * 3/5 + 9,1.042 * 3/5)(0.258 * 3/5 + 9,1.122 * 3/5)(0.263 * 3/5 + 9,1.200 * 3/5)(0.268 * 3/5 + 9,1.275 * 3/5)(0.274 * 3/5 + 9,1.346 * 3/5)(0.279
* 3/5 + 9,1.415 * 3/5)(0.284 * 3/5 + 9,1.481 * 3/5)(0.289 * 3/5 + 9,1.545 * 3/5)(0.295 * 3/5 + 9,1.607 * 3/5)(0.300 * 3/5 + 9,1.667 * 3/5 ) (0.385 * 3/5 + 9,2.400 * 3/5)(0.513 * 3/5 + 9,3.050 * 3/5)(0.641 * 3/5 + 9,3.440 * 3/5)(0.769 * 3/5 + 9,3.700 * 3/5)(0.897 * 3/5 + 9,3.886 * 3/5)(1.026 * 3/5 + 9,4.025 * 3/5)(1.154 * 3/5 + 9,4.133 * 3/5)(1.282 * 3/5 + 9,4.220 * 3/5)(1.410 * 3/5 + 9,4.291 * 3/5)(1.538 * 3/5 + 9,4.350 * 3/5)(1.667 * 3/5 + 9,4.400 * 3/5)(1.795 * 3/5 + 9,4.443 * 3/5)(1.923 * 3/5 + 9,4.480 * 3/5)(2.051 * 3/5 + 9,4.513 * 3/5)(2.179 * 3/5 + 9,4.541 * 3/5)(2.308 * 3/5 + 9,4.567 * 3/5)(2.436 * 3/5 + 9,4.589 * 3/5)(2.564 * 3/5 + 9,4.610 * 3/5)(2.692
* 3/5 + 9,4.629 * 3/5)(2.821 * 3/5 + 9,4.645 * 3/5)(2.949 * 3/5 + 9,4.661 * 3/5)(3.077 * 3/5 + 9,4.675 * 3/5)(3.205 * 3/5 + 9,4.688 * 3/5)(3.333
* 3/5 + 9,4.700 * 3/5)(3.462 * 3/5 + 9,4.711 * 3/5)(3.590 * 3/5 + 9,4.721 * 3/5)(3.718 * 3/5 + 9,4.731 * 3/5)(3.846 * 3/5 + 9,4.740 * 3/5)(3.974 * 3/5 + 9,4.748 * 3/5)(4.103 * 3/5 + 9,4.756 * 3/5)(4.231 * 3/5 + 9,4.764 * 3/5)(4.359 * 3/5 + 9,4.771 * 3/5)(4.487 * 3/5 + 9,4.777 * 3/5)(4.615 * 3/5 + 9,4.783 * 3/5)(4.744 * 3/5 + 9,4.789 * 3/5)(4.872 * 3/5 + 9,4.795 * 3/5)(5.000 * 3/5 + 9,4.800 * 3/5 ) ; ( 5,3 * 3/5+.3)(0.5 * 3/5 + 4.7,3 * 3/5+.3)(0.513 * 3/5 + 4.7,3.050 * 3/5+.3)(0.641 * 3/5 + 4.7,3.440 * 3/5+.3)(0.769 * 3/5 + 4.7,3.700 * 3/5+.3)(0.897 * 3/5 + 4.7,3.886 * 3/5+.3)(1.026 * 3/5 + 4.7,4.025 * 3/5+.3)(1.154 * 3/5 + 4.7,4.133 * 3/5+.3)(1.282
* 3/5 + 4.7,4.220 * 3/5+.3)(1.410 * 3/5 + 4.7,4.291 * 3/5+.3)(1.538
* 3/5 + 4.7,4.350 * 3/5+.3)(1.667 * 3/5 + 4.7,4.400 * 3/5+.3)(1.795 * 3/5 + 4.7,4.443 * 3/5+.3)(1.923 * 3/5 + 4.7,4.480 * 3/5+.3)(2 * 3/5 + 4.7,4.5 * 3/5+.3)(2 * 3/5 + 4.7,3)(5,3)(5,3 * 3/5+.3 ) ; ( 5 + 4,3 * 3/5+.3)(0.5 * 3/5 + 4.7 + 4,3 * 3/5+.3)(0.513 * 3/5 + 4.7 + 4,3.050 * 3/5+.3)(0.641 * 3/5 + 4.7 + 4,3.440 * 3/5+.3)(0.769 * 3/5 + 4.7 + 4,3.700 * 3/5+.3)(0.897 * 3/5 + 4.7 + 4,3.886 * 3/5+.3)(1.026 * 3/5 + 4.7 + 4,4.025 * 3/5+.3)(1.154 * 3/5 + 4.7 + 4,4.133 * 3/5+.3)(1.282 * 3/5 + 4.7 + 4,4.220 * 3/5+.3)(1.410 * 3/5 + 4.7 + 4,4.291 * 3/5+.3)(1.538 * 3/5 + 4.7 + 4,4.350 * 3/5+.3)(1.667 * 3/5 + 4.7 + 4,4.400 * 3/5+.3)(1.795 * 3/5 + 4.7 + 4,4.443 * 3/5+.3)(1.923 * 3/5 + 4.7 + 4,4.480 * 3/5+.3)(2 * 3/5 + 4.7 + 4,4.5 * 3/5+.3)(2 * 3/5 + 4.7 + 4,3)(5 + 4,3)(5 + 4,3 * 3/5+.3 ) ; ( 9,0 ) ( 12,0 ) ( 12,3 ) ( 9,3 ) ( 9,0 ) ; by section [ sec : asymptoticexpansion ] we know that if @xmath233 then @xmath87 .
therefore , we can use for the @xmath41 entry of @xmath202 provided that @xmath234 .
this is guaranteed , for instance , when @xmath235 where @xmath236 .
therefore , we can be more careful by taking the @xmath40 diagonal matrix @xmath237 with @xmath238 for @xmath239 and @xmath0 otherwise , and compute @xmath240 using the following matrix decomposition : @xmath241 where @xmath242 is the matrix @xmath243 if the @xmath41 entry of @xmath244 is nonzero , then @xmath245 and hence , only approximates @xmath46 by a weighted sum of trigonometric functions when it is safe to do so . figure
[ fig : decomposition ] shows how the criterion partitions the entries of @xmath202 .
a stable @xmath45 algorithm for evaluating schlmilch expansions follows , as we will now explain .
( 1,0 ) ( 1,4 )
( 0,0 ) ; ( 0,0 )
( 0,0 ) ; ( 0.200,0.000)(0.216,0.366)(0.221,0.476)(0.226,0.581)(0.232,0.682)(0.237,0.778)(0.242,0.870)(0.247,0.957)(0.253,1.042)(0.258,1.122)(0.263,1.200)(0.268,1.275)(0.274,1.346)(0.279,1.415)(0.284,1.481)(0.289,1.545)(0.295,1.607)(0.300,1.667 ) (0.385,2.400)(0.513,3.050)(0.641,3.440)(0.769,3.700)(0.897,3.886)(1.026,4.025)(1.154,4.133)(1.282,4.220)(1.410,4.291)(1.538,4.350)(1.667,4.400)(1.795,4.443)(1.923,4.480)(2.051,4.513)(2.179,4.541)(2.308,4.567)(2.436,4.589)(2.564,4.610)(2.692,4.629)(2.821,4.645)(2.949,4.661)(3.077,4.675)(3.205,4.688)(3.333,4.700)(3.462,4.711)(3.590,4.721)(3.718,4.731)(3.846,4.740)(3.974,4.748)(4.103,4.756)(4.231,4.764)(4.359,4.771)(4.487,4.777)(4.615,4.783)(4.744,4.789)(4.872,4.795)(5.000,4.800 ) ; at ( 1.5,2 ) ( l1 ) @xmath246 ; at ( -.1,4.2 ) ( l2 ) ; at ( 5,4.9 ) ( l2 ) ; at ( 0.1,0 ) ( l2 ) @xmath52 ; at ( 0.1,4 ) ( l2 ) @xmath247 ; at ( 1,4.9 ) ( l2 ) ; by construction all the entries of @xmath248 are less than @xmath35 in magnitude so its contribution to the matrix - vector product @xmath240 can be ignored , i.e. , we make the approximation @xmath249 the matrix - vector product @xmath250 can be computed in @xmath23 operations using dcts and dsts by .
moreover , the number of nonzero entries in @xmath242 is @xmath251 and hence , the matrix - vector product @xmath252 can be computed in @xmath45 operations using direct summation .
this algorithm successfully stabilizes the @xmath23 matrix - vector product that results from as it only employs the asymptotic expansion for entries of @xmath226 for which @xmath227 is sufficiently large . in practice , this algorithm is faster than direct summation for @xmath253 ( see figure [ fig : schlomilchresults ] ) .
however , the algorithm has an @xmath45 complexity because the criterion is in hindsight too cautious .
we can do better .
note that the asymptotic expansion is accurate for all @xmath51 entries of @xmath202 except for @xmath254 of them , where @xmath255 is the indicator function and the last equality is from the leading asymptotic behavior of the @xmath52th harmonic number as @xmath256 .
therefore , a lower complexity than @xmath45 seems possible if we use the asymptotic expansion for more entries of @xmath202 .
roughly speaking , we need to refine the partitioning of @xmath202 so that the black curve , @xmath257 , in figure [ fig : decomposition ] is better approximated . first , we are only allowed to partition @xmath202 with rectangular matrices since for those entries that we do employ the asymptotic expansion we need to be able to use dcts and dsts to compute matrix - vector products .
second , there is a balance to be found since each new rectangular partition requires @xmath73 more dcts and dsts .
too many partitions and the cost of computing dcts and dsts will be overwhelming , but too few and the cost of computing @xmath252 will dominate . these two competing costs must be balanced . to find the balance we introduce a parameter @xmath258 . roughly speaking , if @xmath259 then we partition as much as possible and the asymptotic expansion is used for every entry satisfying @xmath260 .
if @xmath261 then we do not refine and we keep the @xmath45 algorithm from section [ sec : singlepartition ] . an intermediate value of @xmath262 will balance the two competing costs . figure [ fig : hdecomposition ] shows the partition of @xmath202 that we consider and it is worth carefully examining that diagram .
the partition corresponds to the following matrix decomposition : @xmath263 where @xmath264 is the number of partitions , @xmath237 is the diagonal matrix in section [ sec : singlepartition ] , and the @xmath40 matrices @xmath265 and @xmath266 are diagonal matrices with @xmath267 note that if the @xmath41 entry of the matrix @xmath268 is nonzero , then @xmath269 and @xmath270 .
hence , @xmath271 and @xmath272 .
similarly , if the @xmath41 entry of @xmath273 is nonzero , then @xmath260 . therefore , we are only employing the asymptotic expansion on entries of @xmath274 for which it is accurate and numerically stable to do so .
( 0.23255,0 )
( 0.23255,.7 )
( 0.43478,.7 )
( 0.43478,2.7 )
( 1,2.7 )
( 2.3,4 )
( 2.3,4.56522 )
( 4.3,4.56522 )
( 4.3,4.76744 )
( 5,4.76744 )
( 0,0 ) ; ( 0,0 )
( 0,0 ) ; ( 0.200,0.000)(0.216,0.366)(0.221,0.476)(0.226,0.581)(0.232,0.682)(0.237,0.778)(0.242,0.870)(0.247,0.957)(0.253,1.042)(0.258,1.122)(0.263,1.200)(0.268,1.275)(0.274,1.346)(0.279,1.415)(0.284,1.481)(0.289,1.545)(0.295,1.607)(0.300,1.667 ) (0.385,2.400)(0.513,3.050)(0.641,3.440)(0.769,3.700)(0.897,3.886)(1.026,4.025)(1.154,4.133)(1.282,4.220)(1.410,4.291)(1.538,4.350)(1.667,4.400)(1.795,4.443)(1.923,4.480)(2.051,4.513)(2.179,4.541)(2.308,4.567)(2.436,4.589)(2.564,4.610)(2.692,4.629)(2.821,4.645)(2.949,4.661)(3.077,4.675)(3.205,4.688)(3.333,4.700)(3.462,4.711)(3.590,4.721)(3.718,4.731)(3.846,4.740)(3.974,4.748)(4.103,4.756)(4.231,4.764)(4.359,4.771)(4.487,4.777)(4.615,4.783)(4.744,4.789)(4.872,4.795)(5.000,4.800 ) ; at ( 1.5,2 ) ( l1 ) @xmath246 ; at ( 2.4,4.25 ) ( l11 ) ; at ( .42,1.34 ) ( l11 ) ; at ( -.1,4.2 ) ( l2 ) ; at ( 0.1,4.5 ) ( l2 ) @xmath275 ; at ( 0.1,4 ) ( l2 ) @xmath276 ; at ( 0.1,2.7 ) ( l2 ) @xmath277 ; at ( 0.1,.7 ) ( l2 ) @xmath278 ; at ( 0.1,0 ) ( l2 ) @xmath52 ; at ( 5,4.9 ) ( l2 ) ; at ( .9,4.9 ) ( l2 ) ; at ( .35,4.9 ) ( l2 ) ; at ( 5 - 2.7,4.9 ) ( l2 ) ; at ( 4.3,4.9 ) ( l2 ) ; ( 0.43478,0)(0.43478,1 ) ; ( 1,0)(1,3 ) ; ( 4,4.56522)(5,4.56522 ) ; ( 2,4)(5,4 ) ; each matrix - vector product of the form @xmath279 requires @xmath73 dcts and dsts from and hence , costs @xmath23 operations .
there are a total of @xmath280 matrices of this form in so exploiting the asymptotic expansion requires @xmath281 operations .
the cost of the matrix - vector product @xmath252 is proportional to the number of nonzero entries in @xmath242 . ignoring the constant @xmath282 ,
this is approximately ( by counting the entries in rectangular submatrices of @xmath242 , see figure [ fig : hdecomposition ] ) @xmath283 where the last equality follows from the assumption that @xmath284 . to balance the competing @xmath285 cost of exploiting the asymptotic expansion with the @xmath286 cost of computing @xmath252 , we set @xmath287 and find that @xmath288 .
moreover , to ensure that the assumption @xmath289 holds we take @xmath290 .
thus , the number of partitions should slowly grow with @xmath52 to balance the competing costs . with these asymptotically optimal choices of @xmath262 and
@xmath264 the algorithm for computing the matrix - vector product @xmath240 via requires @xmath291 operations . more specifically , the complexity of the algorithm is @xmath292 since @xmath293 , see figure [ fig : schlomilchresults ] .
though , the implicit constants in @xmath288 and @xmath294 do not change the complexity of the algorithm , they must still be decided on . after numerical experiments we set @xmath295 and @xmath296 for computational efficiency reasons .
table [ tab : algorithmicparameters ] summarizes the algorithmic parameters that we have carefully selected .
the user is required to specify the problem with an integer @xmath6 , a vector of expansion coefficients @xmath297 in , and then provide a working accuracy @xmath28 .
all other algorithmic parameters are selected in a near - optimal manner based on analysis .
ccc ' '' '' parameter ' '' '' & short description & formula + ' '' '' @xmath89 ' '' '' & number of terms in & @xmath298 + @xmath104 & @xmath299 & see table [ tab : smtable ] + @xmath282 & partitioning parameter & @xmath300 + @xmath262 & refining parameter & @xmath301 + @xmath264 & number of partitions & @xmath302 + [ rmk : gammashift ] a fast evaluation scheme for @xmath303 , where @xmath304 is a constant , immediately follows from this section .
the constants @xmath305 and @xmath306 in and should be replaced by diagonal matrices @xmath39 and @xmath307 , where @xmath308 and @xmath309 , respectively .
we now compare three algorithms for evaluating schlmilch expansions : ( 1 ) direct summation ( see section [ sec : existing ] ) ; ( 2 ) our @xmath45 algorithm ( see section [ sec : singlepartition ] ) , and ; ( 3 ) our @xmath1 algorithm ( see section [ sec : recursivepartition ] ) .
the three algorithms have been implemented in matlab and are publicly available from @xcite .
figure [ fig : schlomilchresults ] ( left ) shows the execution time for the three algorithms for evaluating schlmilch expansions for @xmath7 . here
, we select the working accuracy as @xmath170 and use table [ tab : algorithmicparameters ] to determine the other algorithmic parameters .
we find that our @xmath1 algorithm in section [ sec : recursivepartition ] is faster than direct summation for @xmath253 and faster than our @xmath45 algorithm for @xmath310 .
in fact , for @xmath311 the number of partitions , @xmath264 , is selected to be @xmath0 and the @xmath1 algorithm in section [ sec : recursivepartition ] is identical to the @xmath45 algorithm in section [ sec : singlepartition ] . for a relaxed working accuracy of @xmath312 or @xmath313
our @xmath1 algorithm becomes even more computationally advantageous over direct summation .
figure [ fig : schlomilchresults ] ( right ) shows the execution time for our @xmath1 algorithm with @xmath314 , @xmath315 , and working accuracies @xmath316 . for each @xmath28
there is a choice of @xmath89 that minimizes the execution time .
numerical experiments like these motivate the choice @xmath317 in .
schlomilchtimings ( 41,53 ) ( 53,48 ) ( 51,27 ) ( 50,0 ) @xmath52 ( 0,20 ) optimalm ( 50,0 ) @xmath89 ( 0,20 ) ( 32,44 ) ( 22,28.2 ) ( 20,11 )
in this section we adapt the @xmath1 algorithm in section [ sec : recursivepartition ] to an evaluation scheme for expansions of the form : @xmath318 , \label{eq : fourierbessel}\ ] ] where @xmath6 is an integer and @xmath18 is the @xmath19th positive root of @xmath319 . for @xmath7 ,
is the fourier
bessel expansion of order @xmath0 of the function @xmath194\rightarrow \mathbb{c}$ ] .
the algorithm that we describe evaluates at @xmath198 for @xmath320 , which is equivalent to computing the matrix - vector product @xmath321 , where @xmath322 is a column vector with entries @xmath181 . developing a fast algorithm for computing the matrix - vector product @xmath321 is a challenge for fft - based algorithms because the positive roots of @xmath319 are not equally - spaced . here , we are able to derive a fast evaluation scheme by considering @xmath181 as a perturbed equally - spaced grid ( see section [ sec : besselrootsperturbed ] ) .
first , we derive a matrix decomposition for @xmath323 , where @xmath324 , @xmath37 , and @xmath325 are column vectors .
afterwards , we will consider @xmath326 as a perturbation of @xmath37 . for integers @xmath6 and @xmath327 and for column vectors
@xmath324 , @xmath37 , and @xmath325 of length @xmath52 , we have the following matrix decomposition : @xmath328 [ thm : besseldecomposition ] we prove by showing that the @xmath41 entry of the matrix on the left - hand side and right - hand side are equal for @xmath47 .
pick any @xmath47 . by the neumann addition formula we have @xmath329 moreover , by applying the taylor series expansion of bessel functions we have @xmath330 by substituting into we obtain @xmath331 and
the result follows .
we now wish to write down a particular instance of that is useful for computing @xmath321 . by lemma [
lem : besselinequalities ] we can write @xmath332 , where @xmath333 and @xmath185 is a number such that @xmath334 .
in vector notation we have @xmath335 .
hence , by proposition [ thm : besseldecomposition ] we have @xmath336 this is a useful matrix decomposition since each term in the double sum can be applied to a vector in @xmath337 operations .
the diagonal matrices @xmath338 and @xmath339 have @xmath340 matrix - vector products and @xmath341 has @xmath337 matrix - vector products ( see section [ sec : recursivepartition ] and remark [ rmk : gammashift ] ) . however , for to be practical we must truncate the inner and outer sums .
let @xmath150 be an integer and truncate the outer sum in to @xmath342 .
by lemma [ lem : neumannbound ] , with @xmath343 and @xmath344 , the truncated neumann addition formula is still accurate , up to an error of @xmath35 , for the @xmath41 entry of @xmath345 provided that @xmath346 solving for @xmath19 we find that @xmath347 that is , once we truncate the outer sum in to @xmath342 the matrix decomposition is still accurate for all the columns of @xmath345 except the first @xmath348 .
for example when @xmath349 , @xmath350 and hence , once the outer sum is truncated we must not use on the first @xmath351 columns of @xmath345 . based on numerical experiments we pick @xmath150 to be the smallest integer so that @xmath352 and hence , @xmath353 .
next , we let @xmath354 be an integer and truncate the inner sum to @xmath355 . by section [ sec : besseltaylor ]
the truncated taylor series expansion is accurate , up to an error of @xmath35 , for the @xmath41 entry of @xmath345 provided that @xmath356 where the minimization is taken over @xmath357 , instead of @xmath358 , because of the relation @xmath359 ( * ? ? ? * ( 10.4.1 ) ) . solving for @xmath19
we find that @xmath360 thus , once the inner sum is truncated in to @xmath361 the decomposition is accurate for all but the first @xmath362 columns of @xmath345 .
for example when @xmath363 , @xmath364 and we must not use on the first @xmath365 columns of @xmath345 . in practice , we select @xmath354 to be the smallest integer so that @xmath366 . to avoid employing the decomposition with truncated sums @xmath367 on the first @xmath368 columns of @xmath345
, we partition the vector @xmath297 .
that is , we write @xmath369 , where @xmath370 first , we compute @xmath371 using with truncated sums @xmath367 in @xmath1 operations and then compute @xmath372 using direct summation in @xmath340 operations . at first
it seems that computing @xmath321 is @xmath373 times the cost of evaluating a schlmilch expansion , since there are @xmath373 terms in @xmath367 .
however , the computation can be rearranged so the evaluation of is only @xmath374 times the cost . for @xmath349 and @xmath363 , the values
that we take when @xmath375 , we have @xmath376 and @xmath377 so there is a significant computational saving here .
since @xmath378 ( * ? ? ?
* ( 10.4.1 ) ) we have @xmath379 d_{\underline{b}}^{2t+s}\underline{c}\\ & = \!\!\!\sum_{u=0}^{2t+k-3}\!\!\!\!\ !
d_{\underline{r}}^{u}\!\ ! \left[\!\sum_{t=\max(\lceil \frac{u - k+1}{2}\rceil,0)}^{\min(\lfloor \frac{u}{2}\rfloor , t-1)}{}^{\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!''\,\,\,\,\,\ , } \frac{(-1)^t2^{-u}}{t!(u - t)!}\!\ ! \left[\mathbf{j}_{\nu - u+2t}(\ , \underline{r}\,\tilde{\underline{\omega}}^\intercal ) + ( -1)^{u-2t}\mathbf{j}_{\nu+u-2t}(\ , \underline{r}\,\tilde{\underline{\omega}}^\intercal ) \right ] \
! \right ] \!\ !
d_{\underline{b}}^{u}\underline{c},\\ \end{aligned}\ ] ] where the single prime on the sum indicates that the first term is halved and the double prime indicates that the last term is halved . here
, the last equality follows by letting @xmath380 and performing a change of variables .
now , for each @xmath381 , the inner sum can be computed at the cost of one application of the @xmath1 algorithm in section [ sec : recursivepartition ] . since each evaluation of a schlmilch expansion costs @xmath382 operations and @xmath383 , a fourier
bessel evaluation costs @xmath384 operations .
figure [ fig : fbresults1 ] ( left ) shows the execution time for evaluating fourier
bessel expansions of order @xmath0 with direct summation and our @xmath1 algorithm with working accuracies of @xmath385 .
our algorithm is faster than direct summation for @xmath386 when @xmath170 .
relaxing the working accuracy not only decreases @xmath62 and @xmath135 , but also speeds up the evaluation of each schlmilch expansion . when @xmath387 , our algorithm is faster than direct summation for @xmath253 .
the algorithmic parameters are selected based on careful analysis so that the evaluation scheme is numerically stable and the computed vector @xmath63 has a componentwise accuracy of @xmath388 , where @xmath389 is the vector 1-norm . in figure
[ fig : fbresults1 ] ( right ) we verify this for @xmath385 by considering the error @xmath390 , where @xmath391 is the absolute maximum vector norm and @xmath392 is the vector of values computed in quadruple precision using direct summation . for these experiments
we take the entries of @xmath297 as realizations of independent gaussians so that @xmath393 .
fourierbesseltimings ( 47,53 ) ( 43.5,40 ) ( 50,-1 ) @xmath52 ( -.5,20 ) fourierbesselerrors ( 45,29.5 ) ( 45,46.5 ) ( 45,61.5 ) ( 50,-1 ) @xmath52 ( -4,23 )
we now adapt the algorithm in section [ sec : fourierbessel ] to the computation of the discrete hankel transform ( dht ) of order @xmath0 , which is defined as @xcite @xmath394 where @xmath18 is the @xmath19th positive root of @xmath95 .
it is equivalent to the matrix - vector product @xmath395 and is more difficult for fft - based algorithms than the task in section [ sec : fourierbessel ] because it may be regarded as evaluating a fourier bessel expansion at points that are not equally - spaced . to extend our @xmath1 algorithm to this setting we consider the ratios @xmath182 as a perturbed equally - spaced grid . by lemma [ lem : besselinequalities ]
we have @xmath396 we can write this in vector notation as @xmath397 , where @xmath398 . since @xmath399 and @xmath400 , we have by theorem [ thm : besseldecomposition ] the following matrix decomposition : @xmath401 each term in can be applied to a vector in @xmath337 operations .
the diagonal matrices @xmath402 and @xmath403 have @xmath340 matrix - vector products and by section [ sec : fourierbessel ] the matrix - vector product @xmath404 can be computed in @xmath337 operations .
so to compute @xmath404 pad the vector @xmath297 with @xmath405 zeros , use the algorithm in section [ sec : fourierbessel ] with @xmath52 replaced by @xmath406 , and then only keep @xmath407 for @xmath408 .
] however , for to be practical we must truncate the double sum .
let @xmath150 and @xmath354 be integers and consider truncating the sum in to @xmath409 . using similar reasoning to that in section [ sec : fourierbessel ] , if we truncate then the matrix decomposition is accurate , up to an error of @xmath35 , provided that @xmath410 and @xmath411 where we used the bound @xmath412 for @xmath413 .
equivalently , provided that @xmath414 where @xmath415 and @xmath416 are defined in and , respectively .
that is , after truncating the sums in we can use the decomposition on all the entries of @xmath417 except for those in the first @xmath418 rows . in practice , we take @xmath150 and @xmath354 to be the smallest integers so that @xmath419 . to avoid employing a truncated version of on the first few rows we compute @xmath420 using with sums @xmath421 in @xmath1 operations ,
discard the first @xmath418 entries of @xmath63 , and then compute the discarded entries using direct summation in @xmath340 operations . in the same way as section [ sec :
rearrange ] , we can rearrange the computation so that only @xmath374 , instead of @xmath373 , fourier
bessel evaluations are required .
each fourier bessel evaluation requires @xmath374 schlmilch evaluations .
hence , a discrete hankel transform requires @xmath422 schlmilch evaluations .
since @xmath423 , our algorithm for the dht requires @xmath424 operations .
the exposition in this section may suggest that the same algorithmic ideas continue to work for @xmath425 . while this is correct , the computational performance rapidly deteriorates as @xmath6 increases .
this is because the vector of bessel roots @xmath426 and the vector of ratios @xmath427 are distributed less like an equally - spaced grid for @xmath425 .
this significantly increases the number of terms required in the taylor expansion and neumann addition formula . for large @xmath425 ,
different algorithmic ideas should be employed , for example , one should take into account that @xmath428 can be very small when @xmath429 .
it is expected that the matched asymptotic ( transition ) region of @xmath22 , when @xmath430 , will be the most difficult to exploit when deriving a fast algorithm .
we have implemented our @xmath1 algorithm for computing the dht in matlab .
it is publicly available from @xcite .
figure [ fig : hankelresults1 ] ( left ) shows the execution time for computing the dht of order @xmath0 with direct summation and our @xmath1 algorithm with working tolerances of @xmath385 . our algorithm is faster than direct summation for @xmath431 when @xmath170 , for @xmath432 when @xmath433 , and for @xmath253 when @xmath434 .
figure [ fig : hankelresults1 ] ( right ) verifies that the selected algorithmic parameters do achieve the desired error bound of @xmath435 , where @xmath63 is the vector of the computed values of and @xmath392 is the vector of values computed in quadruple precision using direct summation . in this experiment
we take the entries of @xmath297 to be realizations of independent gaussians ( with mean @xmath0 and variance @xmath56 ) so that @xmath436 . dhttimings ( 60,55 ) ( 46,30 ) ( 50,-1 ) @xmath52 ( 0,18 ) dhterrors ( 45,30 ) ( 45,47 ) ( 45,62 ) ( 50,-1 ) @xmath52 ( -5,22 ) the dht is , up to a diagonal scaling , its own inverse as @xmath256 .
in particular , by @xcite we have , as @xmath256 , @xmath437 where @xmath39 is a diagonal matrix with entries @xmath438 , and @xmath439 is the @xmath40 identity matrix .
we can use the relation to verify our algorithm for the dht .
figure [ fig : hankelresults2 ] shows the error @xmath440 , where @xmath297 is a column vector with entries drawn from independent gaussians and @xmath441 for small @xmath52 , @xmath440 will be large because only holds as @xmath256 .
however , even for large @xmath52 , we observe that @xmath440 grows like @xmath45 .
this can be explained by tracking the accumulation of numerical errors .
since the entries of @xmath297 are drawn from independent gaussians and @xmath442 we have @xmath443 and hence , we expect @xmath444 . by the same reasoning
we expect @xmath445 .
finally , this gets multiplied by @xmath446 in so we expect @xmath447 . if in practice we desire @xmath448 , then it is sufficient to have @xmath449 .
dhtinverse ( 52,32 ) ( 23,36 ) ( 50,-1.5 ) @xmath52 ( -4,27 )
an asymptotic expansion of bessel functions for large arguments is carefully employed to derive a numerically stable @xmath1 algorithm for evaluating schlmilch and fourier
bessel expansions as well as computing the discrete hankel transform .
all algorithmic parameters are selected based on error bounds to achieve a near - optimal computational cost for any accuracy goal . for a working accuracy of @xmath450 ,
the algorithm is faster than direct summation for evaluating schlmilch expansions when @xmath253 , fourier
bessel expansions when @xmath386 , and the discrete hankel transform when @xmath431 .
i thank the authors and editors of the digital library of mathematical functions ( dlmf ) . without it
this paper would have taken longer .
i also thank the referees for their time and consideration .
6 , _ a fast algorithm for the evaluation of legendre expansions _ , siam j. sci .
, 12 ( 1991 ) , pp .
158179 . ,
_ algorithm 644 : a portable package for bessel functions of a complex argument and nonnegative order _ , acm trans . math .
, 12 ( 1986 ) , pp .
265273 . , _
the fast hankel transform as a tool in the solution of the time dependent schrdinger equation _ , j. comput .
phys . , 59 ( 1985 ) , pp .
136151 . , _ dual algorithms for fast calculation of the fourier - bessel transform _ , ieee trans .
. , 29 ( 1981 ) , pp .
963972 . , _ algorithms to numerically evaluate the hankel transform _ , computers and mathematics with applications , 26 ( 1993 ) , pp
. 112 . , _ fast computation of zero order hankel transform _
, j. franklin institute , 316 ( 1983 ) , pp .
317326 . , _ computation of quasi - discrete hankel transforms of integer order for propagating optical wave fields _ , j. opt .
a , 21 ( 2004 ) , pp .
5358 . , _ a fast , simple , and stable chebyshev
legendre transform using an asymptotic formula _
, siam j. sci .
comput . , 36 ( 2014 ) ,
, _ an algorithm for the convolution of legendre series _ , siam j. sci . comput . , 36 ( 2014 ) , a1207a1220 .
, _ a fast fft - based discrete legendre transform _ , submitted .
, _ die cylinderfunctionen erster und zweiter art _
ann . , 1 ( 1869 ) , pp .
467501 . , _
bounds for zeros of some special functions _ , proc .
soc . , 25 .
1970 , pp . 7274 .
, _ a hankel transform approach to tomographic image reconstruction _
, ieee transactions on medical imaging , 7 ( 1988 ) , pp .
5972 . , _ an algorithm for the fast hankel transform _ , yale technical report , august 1995 . , _ an improved method for computing a discrete hankel transform _ ,
computer physics communications , 43 ( 1987 ) , pp .
181202 . , _ an improvement on orszag s fast algorithm for legendre polynomial transform _ , trans .
processing soc .
japan , 40 ( 1999 ) , pp .
, _ a fourier bessel transform method for efficiently calculating the magnetic field of solenoids _
, j. comput .
phys . , 37 ( 1980 ) , pp .
4155 . , _ nist handbook of mathematical functions _ , cambridge university press , 2010 .
, _ an algorithm for the rapid evaluation of special function transforms _ , appl .
comput . harm .
, 28 ( 2010 ) , pp .
203226 . , _
approximation for bessel functions and their application in the computation of hankel transforms _ , comput .
8 ( 1982 ) , pp . 305311 . , _
transforms and applications handbook , third edition _ , crc press , 2010 . , _ fast algorithms for spherical harmonic expansions _ , siam j. sci .
, 27 ( 2006 ) , pp .
19031928 . , _ note sur la variation des constantes arbitraires dune intgrale dfinie _ , journal fr die reine und angewandte mathematik , 33 ( 1846 ) , pp .
268280 . ,
_ numerical evaluation of spherical bessel transforms via fast fourier transforms _ , j. comput .
phys . , 100 ( 1992 ) , pp .
294296 . , fastasytransforms , http://github.com/ajt60gaibb/fastasytransforms . , _ fast algorithms for spherical harmonic expansions , ii _ , j. comput .
phys . , 227 ( 2008 ) , pp .
42604279 . , http://functions.wolfram.com/03.01.06.0037.01 . | a fast and numerically stable algorithm is described for computing the discrete hankel transform of order @xmath0 as well as evaluating schlmilch and fourier
bessel expansions in @xmath1 operations .
the algorithm is based on an asymptotic expansion for bessel functions of large arguments , the fast fourier transform , and the neumann addition formula .
all the algorithmic parameters are selected from error bounds to achieve a near - optimal computational cost for any accuracy goal .
numerical results demonstrate the efficiency of the resulting algorithm .
hankel transform , bessel functions , asymptotic expansions , fast fourier transform 65r10 , 33c10 |
let @xmath1 be a unital algebra over the complex numbers .
a @xmath1-valued noncommutative probability space is a pair @xmath2 , where @xmath3 is a unital algebra containing a unitally embedded copy of @xmath1 and @xmath4 is a conditional expectation , namely , an idempotent linear mapping @xmath5 that restricts to the identity on @xmath1 and satisfies @xmath6 for every @xmath7 and @xmath8 .
elements of @xmath3 are called random variables or @xmath1-valued random variables .
when @xmath1 and @xmath3 are @xmath0-algebras and @xmath4 is @xmath0-preserving , then the pair @xmath2 is called a @xmath1-valued @xmath0-noncommutative probability space .
the @xmath1-valued @xmath0-distribution of an element @xmath7 is , loosely speaking , the collection of @xmath0-moments of the form @xmath9 for @xmath10 , @xmath11 and @xmath12 .
see definition [ def:*distr ] for a formal version .
we study @xmath1-valued r - diagonal random variables in @xmath1-valued @xmath0-noncommutative probability spaces .
our motivation is the results of @xcite , where certain random matrices are proved to be asymptotically @xmath1-valued r - diagonal . in the scalar - valued case ( @xmath13 ,
r - diagonal random variables were introduced by nica and speicher @xcite and these include many natural and important examples in free probability theory . in a subsequent paper @xcite , nica , shlyakhtenko and speicher found several equivalent characterizations of scalar - valued r - diagonal elements . in @xcite , niady and speicher introduced @xmath1-valued r - diagonal elements and proved some equivalent characterizations of them . in this paper
we firstly prove some further characterizations of @xmath1-valued r - diagonal elements , which are analogous to those of @xcite .
secondly , we prove a result involving power series for @xmath1-valued r - diagonal elements that is similar to the power series relation between the r - transform and the the moment series of a single random variable .
thirdly , we examine polar decompositions of @xmath1-valued r - diagonal elements , which is a more delicate topic than in the scalar - valued case .
we also study @xmath1-valued circular elements ( a special case of @xmath1-valued r - diagonal elements ) and we prove a new result about the polar decomposition of a quasinilpotent dt - operator .
[ def : maxaltptn ] given @xmath14 and @xmath15 , we define the _ maximal alternating interval partition _
@xmath16 to be the interval partition of @xmath17 whose blocks @xmath1 are the maximal interval subsets of @xmath17 such that if @xmath18 and @xmath19 , then @xmath20 .
the following definition is a reformulation of one of the characterizations of r - diagonality found in @xcite and ( in the scalar - valued case ) in @xcite .
[ def : rdiag ] we say that an element @xmath21 in a @xmath1-valued @xmath0-noncommutative probability space @xmath2 is _ @xmath1-valued r - diagonal _ if for every integer @xmath22 and every @xmath23 we have @xmath24 ( namely , odd alternating moments vanish ) and , for every integer @xmath10 , every @xmath25 and every choice of @xmath26 , we have @xmath27 where in each of the three products above , the terms are taken in order of increasing indices .
[ rem : a*rdiag ] clearly @xmath21 is @xmath1-valued r - diagonal if and only if @xmath28 is @xmath1-valued r - diagonal .
[ rem : l2.1 ] it is not difficult to see , by expansion of the left - hand - side of and induction on @xmath29 , that the @xmath1-valued @xmath0-distribution of a @xmath1-valued r - diagonal element is completely determined by the collection of even , alternating moments , namely , those of of the form @xmath30 in theorem [ thm : rdiageq ] , we prove equivalence of seven conditions for a @xmath1-valued random variable @xmath21 , including the condition for r - diagonalality in definition [ def : rdiag ] .
( for some of these equivalences we refer to @xcite . )
the six others are , in fact , @xmath1-valued analogues of those found in @xcite for the case @xmath31 .
one of these , condition [ it : cum ] , is in terms of speicher s noncrossing cumulants for the pair @xmath32 , namely , that only those associated with even alternating sequences in @xmath21 and @xmath28 may be nonvanishing .
another , condition [ it : m2 ] of theorem [ thm : rdiageq ] , is that the matrix @xmath33 is free from @xmath34 with amalgamation over the diagonal matrices with entries in @xmath1 , while still another , condition [ it : pmoms ] , is an easy reformulation of definition [ def : rdiag ] .
the proofs of the various equivalences are similar to those found in @xcite .
a unitary element of a @xmath1-valued @xmath0-noncommutative probability space @xmath2 is , of course , an element @xmath35 such that @xmath36 .
a _ haar unitary element _ is a unitary element satisfying @xmath37 for all integers @xmath38 . in the tracial , scalar - valued case , it is well known ( see proposition 2.6 of @xcite ) that being @xmath39-diagonal is equivalent to having the same @xmath0-distribution as an element of the form @xmath40 , where @xmath41 is haar unitary and @xmath42 is self - adjoint and such that @xmath43 and @xmath44 are free .
( in the case of a c@xmath45-noncommutative probability space , one can also take @xmath46 this is well known , or see , for example , corollary [ cor : scalarpolar ] for a proof . )
the analogous statement is not true in the tracial , algebra - valued case .
example [ ex : nofreepolar ] provides a counter example . however , in proposition [ prop : hutheta ] and theorem [ thm : c*pos ] , we do characterize , in terms of cumulants , when an algebra - valued r - diagonal element has the same @xmath1-valued @xmath0-distribution as a product @xmath40 with @xmath42 self - adjoint , with @xmath41 a @xmath1-normalizing haar unitary element and with @xmath43 and @xmath44 free over @xmath1 . as an application , corollary [ cor : dtpolar ] shows that the polar decomposition of the quasinilpotent dt - operator has this form .
the contents of the rest of the paper are as follows : section [ sec : cums ] briefly recalls the formulation from @xcite of speicher s @xmath1-valued cumlants , introduces notation and proves some straightforward results about traces and about self - adjointness of @xmath1-valued distributions and cumulants .
section [ sec : rdiag ] deals with the equivalence of the seven conditions that characterize algebra - valued r - diagonal elements .
section [ sec : series ] establishes formal power series relations involving @xmath1-valued alternating moments and @xmath1-valued alternating cumulants of an r - diagonal element .
section [ sec : polar ] proves conditions for traciality of @xmath0distributions of algebra - valued r - diagonal elements and examines polar decompositions and the like for r - diagonal elements .
section [ sec : circ ] examines algebra - valued circular elements , which comprise a special case of algebra - valued r - diagonal elements ; the notion of these has appeared before , notably in work of niady @xcite .
this section also contains example [ ex : nofreepolar ] concerning the polar decomposition , mentioned above .
finally , in an appendix , we investigate the distribution ( with respect to a trace ) of the positive part of the operator in this example .
in this section , we briefly recall a formulation ( from @xcite ) of speicher s theory @xcite of @xmath1-valued cumulants and describe the notation we will use . we also prove some straightforward results about traciality and self - adjointness . given a family @xmath47 of random variables in a @xmath1-valued noncommutative probability space @xmath2 , and given @xmath48 ,
the corresponding _ cumulant map _ is a @xmath49-multilinear map @xmath50 .
these are defined , recursively , by the moment - cumulant formula @xmath51,\ ] ] where @xmath52 is the set of all noncrossing partitions of @xmath17 and where for @xmath53 , @xmath54 is a multilinear map defined in terms of the cumulant maps @xmath55 for the @xmath56 obtained by restricting @xmath57 to the blocks of @xmath58 . in detail , @xmath54 can be specified ( recursively ) as follows .
if @xmath59 , then @xmath60=\alpha_j(b_1,\ldots , b_{n-1}),\ ] ] while if @xmath61 then , selecting an interval block @xmath62 with @xmath63 and @xmath64 , letting @xmath65 be obtained by restricting @xmath58 to @xmath66 and renumbering to preserve order , and letting @xmath67 and @xmath68 be @xmath69 we have @xmath70= \\ = \begin{cases } \begin{aligned}[b ] { { \hat{\alpha}}}_{j'}(\pi')[b_1,\ldots,&b_{p-2 } , \\ b_{p-1}&\alpha_{j''}(b_p,\ldots , b_{p+q-2})b_{p+q-1 } , \\ & \hspace{7em } b_{p+q},\ldots , b_{n-1 } ] , \end{aligned } & p\ge2,\,p+q-1<n \\ { { \hat{\alpha}}}_{j'}(\pi')[b_1,\ldots , b_{p-2}]b_{p-1}\alpha_{j''}(b_p,\ldots , b_{n-1}),&p\ge2,\,p+q-1=n , \\
\alpha_{j''}(b_1,\ldots , b_{q-1})b_q{{\hat{\alpha}}}_{j'}(\pi')[b_{q+1},\ldots , b_{n-1}],&p=1,\,q < n . \end{cases}\end{gathered}\ ] ] we will use the notation @xmath71 for the multilinear moment map @xmath72 given a tracial linear functional on @xmath1 and a @xmath1-valued noncommutative probability space @xmath2 , we will now characterize , in terms of @xmath1-valued cumulants , when the composition @xmath73 is tracial . in the following lemma and its proof , we will use the notation @xmath74 for cyclic left permutations . in particular , @xmath75 and if @xmath53 for some @xmath29 , then @xmath76 is the permutation obtained from @xmath58 by applying the mapping @xmath77 to the underlying set @xmath17 ; so , for example , if @xmath78 , then @xmath79 .
[ lem : taualphapi ] let @xmath80 be the @xmath1-valued distribution of a family @xmath47 with corresponding moment maps @xmath71 and cumulant maps @xmath81 .
suppose @xmath82 is a tracial linear functional on @xmath1 .
fix @xmath83 and suppose that for all @xmath84 , all @xmath85 and all @xmath86 , we have @xmath87 then for every @xmath88 , @xmath89 and @xmath90 , we have @xmath91b_n)=\tau({{\hat{\alpha}}}_{c(j)}(c(\pi))[b_2,\ldots , b_n]b_1).\ ] ] we proceed by induction on @xmath29 . to begin , if @xmath92 , then @xmath93 and using we have @xmath94b_2)=\tau(\alpha_{(j(1))}b_1\alpha_{(j(2))}b_2 ) = \tau(b_1\alpha_{(j(2))}b_2\alpha_{(j(1))})=\tau(b_1{{\hat{\alpha}}}_{c(j)}(\pi)[b_2]),\ ] ] as required .
assume @xmath95 . by the induction hypothesis , for every @xmath96 , @xmath85 and @xmath97 , including , by the original hypothesis , the case @xmath98
, we have @xmath99b_m)=\tau({{\hat{\alpha}}}_{c(j')}(c(\pi'))[b_2,\ldots , b_m]b_1).\ ] ] since @xmath61 , there is an interval block @xmath62 with @xmath100 and @xmath64 .
let @xmath56 and @xmath101 be as in and let @xmath65 be as described above those equations .
if @xmath102 and @xmath103 , then @xmath104b_n ) \\ & = \tau({{\hat{\alpha}}}_{j'}(\pi')[b_1,\ldots , b_{p-2},b_{p-1}\alpha_{j''}(b_p,\ldots , b_{p+q-2})b_{p+q-1},b_{p+q},\ldots , b_{n-1}]b_n ) \\ & = \tau({{\hat{\alpha}}}_{c(j')}(c(\pi'))[b_2,\ldots , b_{p-2},b_{p-1}\alpha_{j''}(b_p,\ldots , b_{p+q-2})b_{p+q-1},b_{p+q},\ldots , b_n]b_1 ) \\ & = \tau({{\hat{\alpha}}}_{c(j)}(c(\pi))[b_2,\ldots , b_n]b_1),\end{aligned}\ ] ] where we have used , respectively , the first case on the right - hand - side of , and again the first case on the right - hand - side of .
if @xmath102 and @xmath105 , then @xmath106b_n)&=\tau({{\hat{\alpha}}}_{j'}(\pi')[b_1\alpha_{j''}(b_2,\ldots , b_{q})b_{q+1},b_{q+2},\ldots , b_{n-1}]b_n ) \\ & = \tau({{\hat{\alpha}}}_{c(j')}(c(\pi'))[b_{q+2},\ldots , b_n]b_1\alpha_{j''}(b_2,\ldots , b_{q})b_{q+1 } ) \displaybreak[2 ] \\ & = \tau(\alpha_{j''}(b_2,\ldots , b_{q})b_{q+1}{{\hat{\alpha}}}_{c(j')}(c(\pi'))[b_{q+2},\ldots , b_n]b_1 ) \\ & = \tau({{\hat{\alpha}}}_{c(j)}(c(\pi))[b_2,\ldots , b_n]b_1),\end{aligned}\ ] ] where we have used , respectively , the first case on the right - hand - side of , , the trace property of @xmath82 and the third case on the right - hand - side of .
the case of @xmath103 and @xmath107 is done similarly , while if @xmath105 and @xmath108 , then @xmath106b_n)&=\tau(\alpha_{j(1)}b_1\alpha_{j''}(b_2,\ldots , b_{n-1})b_n ) \\ & = \tau(\alpha_{j''}(b_2,\ldots , b_{n-1})b_n\alpha_{j(1)}b_1 ) \\ & = \tau({{\hat{\alpha}}}_{c(j)}(c(\pi))[b_2,\ldots , b_n]b_1),\end{aligned}\ ] ] where we have used , respecively , the second case on the right - hand - side of , the trace property of @xmath82 and the third case on the right - hand - side of .
thus , holds and the lemma is proved .
[ prop : trace ] let @xmath80 be the @xmath1-valued distribution of a family @xmath47 with corresponding moment maps @xmath71 and cumulant maps @xmath81 .
suppose @xmath82 is a tracial linear functional on @xmath1
. then the following are equivalent : 1 .
[ it : tracial ] the linear functional @xmath109 on @xmath110 is tracial , 2 .
[ it : psitrace ] @xmath111 @xmath112 @xmath113 , we have @xmath114 3 .
[ it : alphatrace ] @xmath115 @xmath112 @xmath113 , we have @xmath116 the equivalence of [ it : tracial ] and [ it : psitrace ] is easily seen from the definition of @xmath71 .
the proof of [ it : alphatrace ] @xmath117 [ it : psitrace ] follows from the moment - cumulant formula and lemma [ lem : taualphapi ] .
we will prove [ it : psitrace ] @xmath117 [ it : alphatrace ] using the moment - cumulant formula and lemma [ lem : taualphapi ] .
suppose [ it : psitrace ] holds and let us show holds by induction on @xmath10 .
the case @xmath118 is from the tracial property of @xmath82 .
fix @xmath119 and suppose holds for all @xmath120 .
then , by lemma [ lem : taualphapi ] , for all @xmath121 , all @xmath122 and all @xmath123 , we have @xmath124b_{n_0})=\tau({{\hat{\alpha}}}_{c(j)}(c(\pi))[b_2,\ldots , b_{n_0}]b_1).\ ] ] combining this with the moment - cumulant formula and using , we get @xmath125b_{n_0 } ) \\ & = \tau(\psi_{c(j)}(b_2,\ldots , b_{n_0})b_1 ) -\sum_{\pi\in{\operatorname{nc}}(n_0)\backslash\{1_{n_0}\}}\tau({{\hat{\alpha}}}_{c(j)}(c(\pi))[b_2,\ldots , b_{n_0}]b_1 ) \\ & = \tau(\alpha_{c(j)}(b_2,\ldots , b_{n_0})b_1),\end{aligned}\ ] ] as required .
we now turn to questions of self - adjointness .
suppose @xmath1 is a @xmath0-algebra and consider a family @xmath47 of @xmath1-valued random variables in a @xmath1-valued noncommutative probability space @xmath2 .
consider an involution @xmath126 .
let @xmath110 be endowed with the @xmath0-algebra structure coming from the @xmath0-operation on @xmath1 and by setting @xmath127 for all @xmath128 .
for each @xmath10 , let @xmath129 be defined by @xmath130 [ lem : alphapi * ] let @xmath81 be the cumulant maps of the family @xmath47 .
fix @xmath83 and suppose that for all @xmath84 , all @xmath85 and all @xmath131 , we have @xmath132 then for all @xmath88 , all @xmath89 and all @xmath12 , we have @xmath60^*=\alpha_j(r(\pi))[b_{n-1}^*,\ldots , b_1^*],\ ] ] where @xmath133 is the noncrossing partition obtained from @xmath58 by applying the reflection @xmath134 to all elements in the underlying set @xmath135 .
this follows in a straightforward manner by induction on the number of blocks in @xmath58 , from the recursion formula for cumulants .
the following proposition gives a criterion for self - adjointness of the distribution of a family of @xmath1-valued random variables in terms of properties of the @xmath1-valued cumulant maps .
[ prop : alpha * ] the following are equivalent : 1 .
[ it : thetasa ] the linear mapping @xmath136 is self - adjoint .
[ it : psi * ] @xmath115 @xmath112 @xmath137 , we have @xmath138 3 .
[ it : alpha * ] @xmath115 @xmath112 @xmath113 , we have @xmath139 moreover , if @xmath2 is a @xmath0-noncommutative probability space and if @xmath140 for all @xmath128 , then the above conditions are satisfied .
[ def:*distr ] if the conditions in the last sentence of proposition [ prop : alpha * ] hold , then we say @xmath80 is the _
@xmath1-valued @xmath0-distribution _ of the family @xmath47 .
the equivalence of [ it : thetasa ] and [ it : psi * ] follows directly from the definitions .
the equivalence of [ it : psi * ] and [ it : alpha * ] is a straightforward application of the moment - cumulant formula and lemma [ lem : alphapi * ] .
we now suppose that @xmath1 is a c@xmath45-algebra and @xmath80 is a self - adjoint @xmath1-valued distribution as in [ it : thetasa ] of proposition [ prop : alpha * ] .
we say that @xmath80 is _ positive _ if @xmath141 for every @xmath142 .
it can be difficult to verify positivity of a @xmath0-distribution @xmath80 only from knowing the cumulants , though there are special cases that are exceptions to this statement ( for example , the @xmath1-valued semicircular and @xmath1-valued circular elements , discussed in section [ sec : circ ] ) .
let @xmath1 be a unital @xmath0-algebra and let @xmath2 be a @xmath1-valued @xmath0-noncommutative probability space .
it is convenient to introduce here some notation we will use below . by an _ enlargement _ of @xmath2 ,
we will mean a @xmath1-valued @xmath0-noncommutative probability space @xmath143 with an embedding @xmath144 so that the diagram @xmath145 commutes and @xmath146 .
for an element @xmath7 , consider the following sets of words and their centerings , formed from alternating @xmath21 and @xmath28 , with elements of @xmath1 between : @xmath147\bigl\{&b_0ab_1a^*b_2ab_3a^*b_4\cdots ab_{2k-1}a^*b_{2k } \\ & \quad-{{\mathcal{e}}}(b_0ab_1a^*b_2ab_3a^*b_4\cdots ab_{2k-1}a^*b_{2k})\mid k\ge1,\,b_0,\ldots , b_{2k}\in b\bigr\ } \end{aligned } \label{eq : p12 } \\ { { \mathcal{p}}}_{21}&=\begin{aligned}[t]\bigl\{&b_0a^*b_1ab_2a^*b_3ab_4\cdots a^*b_{2k-1}ab_{2k } \\ & \quad-{{\mathcal{e}}}(b_0a^*b_1ab_2a^*b_3ab_4\cdots a^*b_{2k-1}ab_{2k})\mid k\ge1,\,b_0,\ldots , b_{2k}\in b\bigr\}. \end{aligned } \label{eq : p21}\end{aligned}\ ] ] note that @xmath148 means `` starting with @xmath21 '' and @xmath149 means `` starting with @xmath28 '' , while @xmath150 means `` ending with @xmath21 '' and @xmath151 means `` ending with @xmath28 . '' when we write that a unitary @xmath41 _ normalizes _ @xmath1 in item [ it : u0p ] below , we mean that @xmath152 and @xmath153 for all @xmath154 .
the next result is essentially , a @xmath1-valued version of theorem 1.2 of @xcite .
the equivalence of [ it : pmoms ] , [ it : u0 ] and [ it : cum ] was proved in @xcite ; here we will prove the equivalence of [ it : rdiag][it : m2 ] .
[ thm : rdiageq ] let @xmath7
. then the following are equivalent : 1 .
[ it : rdiag ] @xmath21 is @xmath1-valued r - diagonal .
[ it : pmoms ] we have @xmath155 whenever @xmath14 , @xmath156 and @xmath157 .
[ it : u0p ] there is a @xmath1-valued @xmath0-noncommuative probability space @xmath143 , an element @xmath158 and a unitary @xmath159 such that 1 .
@xmath41 normalizes @xmath1 , 2 .
[ it : pword0thm ] @xmath160 and , for all @xmath38 and all @xmath23 , we have @xmath161 3 . @xmath43 is free from @xmath162 with respect to @xmath163 , 4 .
@xmath164 , 5 .
@xmath21 and @xmath40 have the same @xmath1-valued @xmath0-distribution .
[ it : u0 ] there is an enlargement @xmath143 of @xmath2 and a unitary @xmath159 such that 1 .
[ it : u0ucommb ] @xmath41 commutes with every element of @xmath1 , 2 .
[ it : u0free ] @xmath43 is free from @xmath165 with respect to @xmath163 , 3 .
[ it : u0u0 ] @xmath166 for all @xmath167 ( namely , @xmath41 is haar unitary ) , 4 .
[ it : u0distr ] @xmath21 and @xmath168 have the same @xmath1-valued @xmath0-distribution .
[ it : u ] if @xmath143 is an enlargement of @xmath2 and @xmath159 is a unitary such that 1 .
[ it : ucommb ] @xmath41 commutes with every element of @xmath1 , 2 .
[ it : ufreefroma ] @xmath43 is free from @xmath165 with respect to @xmath163 , + then @xmath21 and @xmath168 have the same @xmath1-valued @xmath0-distribution .
[ it : m2 ] consider the subalgebra @xmath169 the conditional expectation @xmath170 given by @xmath171 the subalgebra @xmath172 and the operator @xmath173 then @xmath174 and @xmath34 are free with respect to @xmath175 ( i.e. , with amalgamation over @xmath176 ) .
[ it : cum ] letting @xmath177 and @xmath178 , for every @xmath14 and @xmath179 , the @xmath1-valued cumulant map @xmath81 for the pair @xmath180 is equal to zero if @xmath57 is either not alternating or is not of even length , i.e. , if @xmath57 is not of the form @xmath181 or @xmath182 .
the rest of the section is devoted to the proof of [ it : rdiag][it : m2 ] .
this is clear , because every expectation on the left - hand - side of in definition [ def : rdiag ] is one of the expectations on the left - hand - side of , and vice - versa and because the vanishing of odd , alternating moments of @xmath21 is equivalent to @xmath183 .
next is a straighforward computation that we will use twice , so we state it as a separate lemma .
[ lem : uafr ] let @xmath2 be a @xmath1-valued @xmath0-noncommutative probability space .
assume @xmath184 are such that 1 .
[ it : unormalizes ] @xmath41 a unitary that normalizes @xmath1 2 .
@xmath43 and @xmath162 are free with respect to @xmath4 .
let @xmath185 and let @xmath186 be as in . let @xmath187 and @xmath188 .
then @xmath189 for @xmath154 , we use the notational convention @xmath190 .
we examine @xmath191 .
consider , for @xmath192 , @xmath193 letting @xmath194 , using freeness and condition [ it : unormalizes ] , we have @xmath195 so an arbitrary element of @xmath191 can be written in the form @xmath196 we have shown . proving is even easier .
indeed , we have @xmath197 so an arbitrary element of @xmath198 can be written @xmath199 .
similarly , an arbitrary element of @xmath200 is of the form @xmath201 proving .
taking conjugates proves .
the next lemma is an analogue of proposition 2.3 of @xcite .
[ lem:2.3 ] let @xmath2 be a @xmath1-valued @xmath0-noncommutative probability space .
assume @xmath184 satisfy 1 .
@xmath41 is a unitary that normalizes @xmath1 , 2 .
@xmath43 is free from @xmath162 with respect to @xmath4 , 3 .
@xmath202 , 4 .
[ it : pword0 ] @xmath203 and , for all @xmath38 and all @xmath23 , we have @xmath204 then the element @xmath185 satisfies the condition [ it : pmoms ] of theorem [ thm : rdiageq ] . from lemma [ lem : uafr ] we have and . examining in the proof of lemma [ lem : uafr ] and invoking condition [ it : pword0 ] , ( then also taking conjugates ) , we find @xmath205 using and , we see that any product of the form @xmath206 with @xmath207 for some @xmath208 can be rewritten as a word with letters belonging to the sets @xmath209 and @xmath43 , in alternating fashion . by freeness and the hypothesis @xmath202
, each such word evaluates to @xmath210 under @xmath4 .
next is an analogue of lemma 2.5 of @xcite .
[ lem:2.5 ] suppose @xmath35 is a @xmath1-normalizing haar unitary element in a @xmath1-valued @xmath0-noncommutative probabiltiy space @xmath2 .
suppose @xmath211 is a @xmath0-subalgebra that is free from @xmath43 with respect to @xmath4 .
let @xmath10 , @xmath212 and @xmath213 be such that 1 .
[ it : h1 ] @xmath214 if @xmath215 2 .
[ it : h2 ] @xmath216 and at least one of @xmath217 and @xmath218 is nonzero whenever @xmath219 and @xmath220 .
then @xmath221 the proof is very similar to the proof found in @xcite , only slightly different to take @xmath1 into account .
we use induction on the cardinality @xmath222 , of @xmath223 . if @xmath224 , then by freeness , holds . if @xmath225 , then letting @xmath226 be least such that @xmath227 , we write @xmath228 by the induction hypothesis , the second term on the right - hand - side equals @xmath210 . letting @xmath229 we have @xmath154 and the first term equals @xmath230 we will show that , by induction hypothesis , the above quantity equals @xmath210 .
indeed , we have @xmath216 and at most one of @xmath217 and @xmath218 can be zero , so @xmath231 . if @xmath232 and @xmath233 , then @xmath234 and we conclude @xmath235 . if @xmath236 and @xmath237 , then @xmath238 so @xmath239 .
consequently , @xmath240 .
thus , @xmath241 .
if @xmath236 then we see that the word to which @xmath4 is applied in satisfies the requirements [ it : h1]-[it : h2 ] and , applying the induction hypothesis , we conclude that this moment is @xmath210 .
if @xmath242 , then the moment becomes @xmath243 and again , by the induction hypothesis , this is zero .
the next result is an analogue of proposition 2.4 of @xcite .
[ lem:2.4 ] suppose that @xmath2 is a @xmath1-valued @xmath0-noncommutative probability space and @xmath184 are such that 1 .
@xmath41 is a @xmath1-normalizing haar unitary element , 2 .
@xmath43 is free from @xmath162 with respect to @xmath4 .
let @xmath185 .
then @xmath21 satisfies condition [ it : pmoms ] of theorem [ thm : rdiageq ] .
lemma [ lem : uafr ] applies and we may use the inclusions , where @xmath244 is the algebra generated by @xmath245 .
thus , for arbitrary @xmath14 , @xmath208 and @xmath246 , the product @xmath206 can be re - written in the form @xmath247 for some @xmath248 , some @xmath249 and some @xmath250 , where whenever @xmath251 for some @xmath252 , we have @xmath253 and in the case @xmath254 , we have @xmath255
. now lemma [ lem:2.5 ] applies and we conclude @xmath256 .
we first show [ it : u]@xmath257[it : u0 ] .
there is a @xmath1-valued @xmath0-noncommutative probability space @xmath258 with a unitary @xmath259 such that @xmath260 commutes with every element of @xmath1 and @xmath261 for all @xmath14 .
for example , we could take @xmath262 to be the algebra @xmath263 $ ] , where @xmath264 $ ] is the group @xmath0-algebra of the cyclic group of infinite order and where @xmath265 , where , for @xmath266 $ ] , @xmath267 equals the coefficient in @xmath268 of the identity element @xmath269 ; let @xmath74 be a generator of @xmath270 and denote also by @xmath271 $ ] the corresponding element of the group algebra ; then @xmath272 is a unitary with the desired properties .
then we let @xmath273 be the algebraic ( amalgamated ) free product of @xmath1-valued @xmath0-noncommutative probability spaces and let @xmath159 be the copy of @xmath259 arising from the free product construction .
then @xmath41 commutes with every element of @xmath1 , @xmath43 and @xmath165 are free and @xmath164 . by
[ it : u ] , the elements @xmath21 and @xmath168 have the same @xmath1-valued @xmath0-distribution .
therefore , the requirements of [ it : u0 ] are fulfilled .
we now show [ it : u0]@xmath257[it : u0p ] , taking @xmath274 .
we need only show that condition [ it : pword0thm ] of [ it : u0p ] holds , namely , that @xmath275 and @xmath276 since , by hypothesis , @xmath41 commutes with every element @xmath1 and @xmath21 has the same @xmath1-valued @xmath0-distribution as @xmath168 , we have @xmath277 where the second equality is due to freeness namely , condition [ it : u0free ] of [ it : u0 ] , and the last equality is because @xmath164 , namely , condition [ it : u0u0 ] of [ it : u0 ] .
similarly , we have @xmath278 with the penultimate equality due to freeness .
thus , [ it : u0p ] holds . the implication [ it : u0p]@xmath257[it : pmoms ] follows immediately by appeal to lemma [ lem:2.3 ] .
we now show [ it : pmoms]@xmath257[it : u ] .
we suppose @xmath143 is an enlargement of @xmath2 and @xmath35 is a unitary satisfying [ it : ucommb ] and [ it : ufreefroma ] of [ it : u ] .
we must show that @xmath21 and @xmath168 have the same @xmath1-valued @xmath0-distribution .
taking @xmath258 as in the proof of [ it : u]@xmath257[it : u0 ] above and replacing @xmath143 by the algebraic free product with amalgamation @xmath279 , we may assume there exists a haar unitary @xmath280 such that @xmath260 commutes with every element of @xmath1 and @xmath281 is free from @xmath282 .
thus , the triple @xmath283 forms a free family of sets . we make the following claims , which we will prove one after the other : 1 . [ cl : a , va ] @xmath21 and @xmath284 have the same @xmath1-valued @xmath0-distribution , 2 . [ cl : ua , uva ] @xmath168 and
@xmath285 have the same @xmath1-valued @xmath0-distribution , 3 .
[ cl : uacond ] each of @xmath285 and @xmath168 satisfies condition [ it : pmoms ] of theorem [ thm : rdiageq ] , 4 .
[ cl : a , ua ] @xmath21 and @xmath168 have the same @xmath1-valued @xmath0-distribution . to prove [ cl : a , va ] , we note that both @xmath21 and @xmath284 satisfy condition [ it : pmoms ] of theorem [ thm : rdiageq ] ; the operator @xmath21 does so by hypothesis , and the operator @xmath284 does so by lemma [ lem:2.4 ] .
moreover , for each @xmath38 and @xmath286 , we have @xmath287 where the last inequality is due to freeness , and @xmath288 thus , by remark [ rem : l2.1 ] , @xmath21 and @xmath284 have the same @xmath1-valued @xmath0-distribution , and the claim is proved . to prove [ cl : ua , uva ] , note that @xmath43 is free from @xmath165 and from @xmath289 and , by [ cl : a , va ] , @xmath21 and @xmath284 are identically @xmath0-distributed .
this implies [ cl : ua , uva ] . to prove [ cl : uacond ]
, we argue from lemma [ lem:2.4 ] that @xmath285 satisfies condition [ it : pmoms ] of theorem [ thm : rdiageq ] , because @xmath290 is a @xmath1-normalizing haar unitary .
that @xmath168 satisfies condition [ it : pmoms ] of theorem [ thm : rdiageq ] now follows from [ cl : ua , uva ] . to prove [ cl : a , ua ] , note that the analogues of and hold when @xmath260 is replaced by @xmath41 , by the same arguments as given above . also , both @xmath21 and @xmath168 satisfy condition [ it : pmoms ] of theorem [ thm : rdiageq ] ; the operator @xmath21 does so by hypothesis , and the operator @xmath168 does so by [ cl : uacond ] . by remark [ rem :
l2.1 ] , @xmath21 and @xmath168 are identically distributed .
this finishes the proof of [ it : pmoms]@xmath257[it : u ] .
this completes the proof of the equivalence of [ it : pmoms ] , [ it : u0p ] , [ it : u0 ] and [ it : u ] of theorem [ thm : rdiageq ] . for the next lemma
, we use the sets @xmath200 and @xmath291 as in and , and we use additionally the notation @xmath292 so that we have @xmath293 now we turn to condition [ it : m2 ] of theorem [ thm : rdiageq ] and take @xmath294 and @xmath176 as defined there . [
lem : m2alg ] let @xmath295 be the subalgebra generated by @xmath296 .
then @xmath297 for @xmath298 , we have @xmath299 using induction on @xmath29 , we easily see that every word of the form @xmath300 belongs to the right - hand - side of , from which we easily deduce that the inclusion @xmath301 in holds .
the reverse inclusion is easily proved by judicious choice of @xmath302 and @xmath303 in . from lemma
[ lem : m2alg ] and , we have @xmath304 and , clearly , we have @xmath305 to show [ it : pmoms ] @xmath117 [ it : m2 ] , take @xmath306 and @xmath307 for @xmath308 and consider the product @xmath309 since each @xmath310 is a @xmath311 matrix whose @xmath312-th entry belonging to @xmath313 , for each @xmath314 , we see that every entry of the @xmath311 matrix @xmath309 is a sum of products of the form of the form that condition [ it : pmoms ] of theorem [ thm : rdiageq ] guarantees has expectation zero .
thus , every entry of the @xmath311 matrix @xmath309 evaluates to zero under @xmath4 .
the same remains true if we left multiply by @xmath315 or if we choose @xmath316 and then right multiply by @xmath317 , or if we do both .
this shows that each of the products @xmath318 expects to zero under @xmath175 .
this is [ it : m2 ] of theorem [ thm : rdiageq ] . to show [ it : m2 ] @xmath117 [ it : pmoms ] , using
, it is straightforward to arrange , given any @xmath319 i_{n+1}\in\{1,2\}$ ] and any @xmath246 , that the product @xmath206 arise as either the @xmath320 or @xmath321 entry of a product of one of the forms for suitable @xmath322 and for each @xmath323 .
thus , @xmath256 . since the equivalence [ it : cum ] @xmath324
[ it : pmoms ] from theorem [ thm : rdiageq ] was proved in @xcite , the proof of theorem [ thm : rdiageq ] is complete .
in this section , we derive some formal power series relations involving moments and cumulants of r - diagonal elements . the main result is essentially a variation on the combinatorial proof of the power series relation for the r - transform , here modified to handle r - diagonal elements .
the result is of intrinsic interest and will be used in the appendix to investigate the particular operator considered in example [ ex : nofreepolar ] . throughout this section
, @xmath1 will be a unital @xmath0-algebra , @xmath2 will be a @xmath1-valued @xmath0-noncommutative probability space and @xmath7 will be a @xmath1-valued r - diagonal element .
we let @xmath177 and @xmath178 and let @xmath81 for sequences @xmath57 in @xmath325 denote the @xmath1-valued cumulant maps of the pair @xmath180 . for @xmath167 ,
for the @xmath326-multilinear cumulant maps that are the only ones that need not be zero , we write @xmath327 [ prop : mrecurse ] for @xmath167 and @xmath23 , let us write @xmath328 and @xmath329 then , for every @xmath10 and @xmath330 , we have @xmath331 \alpha_\ell^{(1)}\bigg ( & b_{s(1)+1}m_{k(1)}^{(2)}(b_{s(1)+2},\ldots , b_{s(1)+2k(1)+1 } ) , \notag\\ & b_{s(2)+2}m_{k(2)}^{(1)}(b_{s(2)+3},\ldots , b_{s(2)+2k(2)+2 } ) , \displaybreak[2]\notag\\ & b_{s(3)+3}m_{k(3)}^{(2)}(b_{s(3)+4},\ldots , b_{s(3)+2k(3)+3 } ) , \notag\\ & \hspace{7em}\vdots \displaybreak[2]\notag\\ & b_{s(2\ell-2)+2\ell-2}m_{k(2\ell-2)}^{(1)}(b_{s(2\ell-2)+2\ell-1},\ldots , b_{s(2\ell-2)+2k(2\ell-2)+2\ell-2 } ) , \notag\\ & b_{s(2\ell-1)+2\ell-1}m_{k(2\ell-1)}^{(2)}(b_{s(2\ell-1)+2\ell},\ldots , b_{s(2\ell-1)+2k(2\ell-1)+2\ell-1 } ) \notag\\ \bigg)&b_{s(2\ell)+2\ell}m_{k(2\ell)}^{(1)}(b_{s(2\ell)+2\ell+1},\ldots , b_{s(2\ell)+2k(2\ell)+2\ell } ) \notag\end{aligned}\ ] ] and @xmath332 \alpha_\ell^{(2)}\bigg ( & b_{s(1)+1}m_{k(1)}^{(1)}(b_{s(1)+2},\ldots , b_{s(1)+2k(1)+1 } ) , \\ & b_{s(2)+2}m_{k(2)}^{(2)}(b_{s(2)+3},\ldots , b_{s(2)+2k(2)+2 } ) , \displaybreak[2]\\ & b_{s(3)+3}m_{k(3)}^{(1)}(b_{s(3)+4},\ldots , b_{s(3)+2k(3)+3 } ) , \\ & \hspace{7em}\vdots \displaybreak[2]\\ & b_{s(2\ell-2)+2\ell-2}m_{k(2\ell-2)}^{(2)}(b_{s(2\ell-2)+2\ell-1},\ldots , b_{s(2\ell-2)+2k(2\ell-2)+2\ell-2 } ) , \\ & b_{s(2\ell-1)+2\ell-1}m_{k(2\ell-1)}^{(1)}(b_{s(2\ell-1)+2\ell},\ldots , b_{s(2\ell-1)+2k(2\ell-1)+2\ell-1 } ) \\ \bigg)&b_{s(2\ell)+2\ell}m_{k(2\ell)}^{(2)}(b_{s(2\ell)+2\ell+1},\ldots , b_{s(2\ell)+2k(2\ell)+2\ell}),\end{aligned}\ ] ] where @xmath333 , which includes the case @xmath334 . using the moment - cumulant formula , we have @xmath335b_{2n}.\ ] ]
if @xmath336 yields a nonzero term in the above sum , then the block of @xmath58 containing @xmath337 must contain alternating odd and even numbers increasing from left to right .
such a block is indicated in in figure [ fig : ncp ] .
( 430,70)(0,0 ) ( 10,50)(1,0)189.5 ( 220.5,50)(1,0)109.5 ( 199.5,50)(1,1)7 ( 206.5,57)(1,-2)7 ( 213.5,43)(1,1)7 ( 10,10)(0,1)40 ( 90,10)(0,1)40 ( 170,10)(0,1)40 ( 250,10)(0,1)40 ( 330,10)(0,1)40 ( 5,0)@xmath21 ( 85,0)@xmath28 ( 165,0)@xmath21 ( 245,0)@xmath21 ( 325,0)@xmath28 ( 50,25)(60,20 ) ( 38,22)@xmath338 ( 130,25)(60,20 ) ( 118,22)@xmath339 ( 290,25)(60,20 ) ( 265,22)@xmath340 ( 380,25)(80,20 ) ( 365,22)@xmath341 ( 190,25)(20,0)3 the ovals represent the locations of the other blocks of the partition , and the quantities @xmath342 in the ovals are the lengths of the respective gaps , which may be zero .
if we sum the quantity @xmath343b_{2n}\ ] ] over all partitions @xmath336 whose block containing @xmath337 is the given one shown in the figure , then by the moment - cumulant formula applied to each of the @xmath344 ovals , we obtain precisely the summand of the summation appearing in .
now fixing @xmath345 and summing this value over all possible values of @xmath346 equals the sum of the quantity over all @xmath336 whose first block contains @xmath344 elements . finally , summing over all values of @xmath345 yields the equality .
the other equality is proved in the same way , by changing indices .
[ thm : powerseries ] consider the formal power series @xmath347 then @xmath348 = 1+\sum_{\ell=1}^\infty\alpha^{(1)}_\ell\big(b_1g(b_2,b_1),&b_2f(b_1,b_2),b_1g(b_2,b_1 ) , \\ & \ldots , b_2f(b_1,b_2),b_1g(b_2,b_1)\big)b_2f(b_1,b_2 ) , \end{aligned } \label{eq : frecurse } \\
g(b_1,b_2 ) & \begin{aligned}[t ] = 1+\sum_{\ell=1}^\infty\alpha^{(2)}_\ell\big(b_1f(b_2,b_1),&b_2g(b_1,b_2),b_1f(b_2,b_1 ) , \\ & \ldots , b_2g(b_1,b_2),b_1f(b_2,b_1)\big)b_2g(b_1,b_2 ) .
\end{aligned } \label{eq : grecurse}\end{aligned}\ ] ] the meaning of the above formulas should be clear . in and ,
the series are formal power series in variables @xmath349 and @xmath350 , and for a given @xmath29 , the corresponding terms should be thought of as being of degree @xmath351 . in the formulas and , the formulas on the right - hand - sides mean the formal power series obtained by substituting and formally expanding .
since the degree @xmath351 term in is precisely @xmath352 , and in it is similarly @xmath353 , the assertions
follow from proposition [ prop : mrecurse ] .
the expressions in can be treated more precisely and in a more general framework using the notion of multilinear function series that was described in @xcite .
formally , a _ multilinear function series _ over @xmath1 is a family @xmath354 where @xmath355 is the constant term and where , when @xmath10 , @xmath356 is a @xmath29-fold multilinear function @xmath357 .
the set @xmath358 $ ] of all multilinear function series over @xmath1 was shown in @xcite to be an associative ring and a @xmath1-bimodule under natural operations .
for example , the product is given by @xmath359 , for @xmath360 as above and @xmath361 $ ] , where the @xmath210-th term of @xmath359 is @xmath362 and the @xmath29-th term , for @xmath10 , is the multilinear map @xmath363 moreover ,
@xmath358 $ ] is equipped with a composition operation , ( which will be generalized below )
. the identity element of @xmath358 $ ] with respect to composition is denoted @xmath364 ; it is the multilinear function series whose @xmath29-th term is @xmath210 unless @xmath118 , in which case it is the identity map on @xmath1 .
let @xmath365 be a sequence in @xmath1 .
let us say that a _ variable assignement _ for a multlinear function series @xmath366 is a choice @xmath367 .
then the _ evaluation _ of @xmath360 at @xmath260 is just the formal sum @xmath368 note that , though this is ostensibly a summation of elements of @xmath1 , no sort of convergence is assumed .
formally , it may be thought of as the sequence @xmath369 in @xmath1 , and equality of such formal sums amounts to equality of the corresponding sequences .
if @xmath370 $ ] is the multilinear function series whose @xmath210-th term is @xmath337 and whose @xmath351-th term for @xmath10 is @xmath371 as described in proposition [ prop : mrecurse ] , and if @xmath372 then in , @xmath373 is just @xmath374 evaluated at @xmath260 and @xmath375 is just @xmath376 evaluated at @xmath260 .
[ def : multicomp ] suppose that @xmath377 is a family in @xmath358 $ ] , where each @xmath378 has zero constant term and suppose that @xmath379 $ ] .
let @xmath380 be a function .
then the _ multivariate composition _
@xmath381 is obtained by , heuristically , replacing the @xmath57-th argument of @xmath356 by @xmath382 for each @xmath29 and @xmath57 .
more precisely , it is the element of @xmath358 $ ] whose @xmath29-th term is @xmath383 when @xmath384 and when @xmath10 it is the @xmath29-fold multilinear function that sends @xmath385 to @xmath386 \chi_p\big ( \psi_{k(1)}^{(f(p,1))}(b_1,&\ldots , b_{k(1 ) } ) , \psi_{k(2)}^{(f(p,2))}(b_{k(1)+1},\ldots , b_{k(1)+k(2)}),\cdots , \\ &
\psi_{k(p)}^{(f(p , p))}(b_{k(1)+\cdots+k(p-1)+1},\ldots , b_{k(1)+\cdots+k(p-1)+k(p ) } ) \big ) .
\end{aligned}\ ] ] the following result is a rephrasing of proposition [ prop : mrecurse ] .
note that the product @xmath387 is given by @xmath388 if @xmath29 is even and @xmath389 for @xmath390 , let @xmath391 be the multilinear function series with zero constant term , whose even terms vanish and whose @xmath392-th term is @xmath393 .
let @xmath394 be given by @xmath395 and @xmath396 .
then @xmath397 now result from evaluating both sides of at @xmath260 , as given in .
this , then , is a formal interpretation of theorem [ thm : powerseries ] .
in this section , we suppose that @xmath1 is a unital @xmath0-algebra and @xmath2 is a @xmath1-valued @xmath0-noncommutative probability space .
we investigate questions of traces and polar decomposition for @xmath1-valued r - diagonal elements . first some notation we will use in this section : if @xmath7 , then we write @xmath177 and @xmath178 and for @xmath167 , let @xmath398 and @xmath399 be the multilinear maps @xmath400 that are the cumulant maps for the pair @xmath180 corresponding to the alternating sequences @xmath401 and @xmath402 , respectively , each of length @xmath403 . note that , from proposition [ prop : alpha * ] , we have @xmath404 for all @xmath390 , @xmath38 and @xmath286 .
the following result is an immediate consequence of proposition [ prop : trace ] .
[ prop : tracerdiag ] suppose @xmath7 is @xmath1-valued r - diagonal .
suppose @xmath82 is a tracial linear functional on @xmath1 .
then @xmath73 restricted to @xmath405 is tracial if and only if , for all @xmath38 and all @xmath23 , we have @xmath406 we now turn to results about @xmath1-valued r - diagonal elements writen in the form @xmath40 where @xmath41 is unitary and where @xmath407 .
first , we make an observation about the polar decompositions of r - diagonal elements in w@xmath45-noncommutative probability spaces .
[ prop : w*polar ] suppose @xmath1 is a von neumann algebra and @xmath2 is a @xmath1-valued w@xmath45-noncommutative probability space .
suppose @xmath7 is @xmath1-valued r - diagonal and @xmath408 is the polar decomposition of @xmath21 .
then @xmath409 for all integers @xmath10 .
thus , if @xmath260 is unitary , then it is haar unitary . using the amalgamated free product construction for von neumann algebras , we get a larger w@xmath45-noncommutative probability space @xmath143 containing a haar unitary @xmath41 that commutes with every element of @xmath1 and such that @xmath165 and @xmath43 are free with respect to @xmath163 .
thus , @xmath410 and @xmath43 are free with respect to @xmath163 . by condition
[ it : u ] of theorem [ thm : rdiageq ] , @xmath168 has the same @xmath1-valued @xmath0-distribution with respect to @xmath163 as does @xmath21 with respect to @xmath4 .
the polar decomposition of @xmath168 is @xmath411 .
since we are in a w@xmath45-noncommutative probability space , given any element @xmath268 with polar decomposition @xmath412 , the joint @xmath1-valued @xmath0-distribution of the pair @xmath413 is determined by the @xmath1-valued @xmath0-distribution of @xmath268 .
therefore , the @xmath1-valued @xmath0-distribution of @xmath290 equals that of @xmath260 .
now , using that @xmath41 is a haar unitary that commutes with @xmath1 and that @xmath41 and @xmath260 are @xmath0-free , it is straightforward to show @xmath414 for every integer @xmath10 . thus , also @xmath415 .
proposition 2.6 of @xcite shows that in the scalar - valued case , given an r - diagonal element in a tracial @xmath0-noncommutative probability space , it has the same @xmath0-distribution as some element @xmath40 , where @xmath41 is a haar unitary , @xmath42 is self - adjoint and @xmath41 is @xmath0-free from @xmath42 .
the following observation will be used in example [ ex : nofreepolar ] below , to show that the analogous statement need not hold in the algebra - valued case .
[ lem : notfree ] suppose in a @xmath1-valued @xmath0-noncommutative probability space @xmath2 , @xmath416 is @xmath1-valued @xmath39-diagonal , where @xmath35 is unitary , where @xmath417 and where @xmath43 and @xmath44 are free ( over @xmath1 ) with respect to @xmath4 . if @xmath418 , then @xmath419 .
we have @xmath420 , while using freeness of @xmath43 and @xmath44 , we find @xmath421 we now study r - diagonal elements that can be written in the form @xmath40 where @xmath41 is a @xmath1-normalizing haar unitary , @xmath407 and the sets @xmath43 and @xmath44 are free over @xmath1 .
it turns out these can be characterized in terms of cumulants .
our study culminates in theorem [ thm : c*pos ] and corollary [ cor : polarfree ] , which apply to c@xmath45- and w@xmath45-noncommutative probability spaces , but along the way we prove algebraic versions too .
the following proposition includes an analogue of proposition 2.6 of @xcite .
note that the equivalence of parts [ it : eaa*][it : up ] of it are special cases of proposition [ prop : hutheta ] , below .
however , we include the statment and proof of it , because both are easier than the more general case , and serve as templates . [
prop : hu ] let @xmath7
. then the following are equivalent : 1 .
[ it : eaa * ] @xmath21 is @xmath1-valued r - diagonal and for all @xmath38 and all @xmath286 , we have @xmath422 2 .
[ it : beta12 ] @xmath21 is @xmath1-valued r - diagonal and for all @xmath38 , we have agreement @xmath423 of the @xmath403-th order @xmath1-valued cumulant maps 3 . [
it : up0 ] there exists a @xmath1-valued @xmath0-noncommutative probability space @xmath143 with elements @xmath424 such that 1 .
[ it : psa ] @xmath407 2 .
[ it : uhu ] @xmath41 is a haar unitary that commutes with every element of @xmath1 3 .
[ it : upfree ] @xmath43 and @xmath44 are free ( over @xmath1 ) with respect to @xmath4 4 . [
it : updista ] the element @xmath40 has the same @xmath1-valued @xmath0-distribution as @xmath21 5 . [ it : podds ] the odd moments of @xmath42 vanish , i.e. , for every @xmath22 and all @xmath23 , we have @xmath425 4 .
[ it : up ] there exists a @xmath1-valued @xmath0-noncommutative probability space @xmath143 with elements @xmath424 such that parts [ it : psa][it : updista ] of [ it : up0 ] hold .
[ it : adista * ] @xmath21 is @xmath1-valued r - diagonal and has the same @xmath1-valued @xmath0-distribution as @xmath28 . in part [ it : up0 ] or [ it : up ] , the even moments of @xmath42 are given by , for every @xmath38 and all @xmath286 , @xmath426 the equivalence of [ it : adista * ] , [ it : eaa * ] and [ it : beta12 ] is clear from the moment - cumulant formula and remark [ rem : l2.1 ] . indeed , using remark [ rem : a*rdiag ] , condition [ it : beta12 ] says that @xmath21 and @xmath28 have the same @xmath1-valued cumulants , since all other cumulants of @xmath1-valued r - diagonal elements must vanish .
we now show [ it : beta12]@xmath257[it : up0 ] .
endow @xmath427 with the @xmath0-operation that extends the given one on @xmath1 and satisfies @xmath428 .
let @xmath429 be the @xmath1-valued distribution of an element whose odd order cumulant maps are all zero and whose @xmath1-valued cumulant map of order @xmath403 is @xmath398 , for all @xmath38 ; indeed , @xmath80 is constructed using the moment - cumulant formula .
write @xmath42 for the element @xmath360 of @xmath427 . by proposition [ prop :
alpha * ] and the identity , @xmath80 is a self - adjoint map . by the amalgamated free product construction , as in
, there is a @xmath1-valued @xmath0-noncommutative probability space @xmath430 ) containing elements @xmath424 satisfying conditions [ it : psa ] , [ it : uhu ] and [ it : upfree ] of [ it : up0 ] .
the moment - cumulant formula and the fact that @xmath423 for all @xmath226 imply that the identities and hold ; in particular , condition [ it : podds ] of [ it : up0 ] holds .
let @xmath431 .
by lemma [ lem:2.3 ] , @xmath432 is @xmath1-valued r - diagonal . in order to prove that @xmath21 and @xmath432 have the same @xmath1-valued @xmath0-distribution , using remark [ rem : l2.1 ] , it will suffice to show that we always have @xmath433 to show
, we use @xmath434 and . to show , we use @xmath435 by freeness and the fact that @xmath41 commutes with every element of @xmath1
, we get @xmath436 so , by appeal to and the fact that @xmath21 and @xmath28 have the same @xmath1-valued @xmath0-distribution , we get . thus , @xmath21 and
@xmath437 have the same @xmath1-valued @xmath0-distribution ; namely , [ it : up0 ] holds . clearly , [ it : up0]@xmath257[it : up ]
now we will show [ it : up ] implies [ it : eaa * ] and .
let @xmath437 .
that @xmath432 is @xmath1-valued r - diagonal follows from lemma [ lem:2.4 ] . but and hold by the same arguments as above .
thus , we have .
here is a more general version of proposition [ prop : hu ] .
[ prop : hutheta ] let @xmath7 and let @xmath438 be an automorphism of the @xmath0-algebra @xmath1
. then the following are equivalent : 1 .
[ it : eaa*theta ] @xmath21 is @xmath1-valued r - diagonal and for all @xmath38 and all @xmath286 , we have @xmath439 2 .
[ it : beta12theta ] @xmath21 is @xmath1-valued r - diagonal and for all @xmath38 , the @xmath403-th order @xmath1-valued cumulant maps satisfy , for all @xmath286 , @xmath440 3 .
[ it : up0theta ] there exists a @xmath1-valued @xmath0-noncommutative probability space @xmath143 with elements @xmath424 such that 1 . [ it : psatheta ] @xmath407 2 .
[ it : uhutheta ] @xmath41 is a haar unitary that normalizes @xmath1 and satisfies @xmath441 for all @xmath154 3 .
[ it : upfreetheta ] @xmath43 and @xmath44 are free ( over @xmath1 ) with respect to @xmath4 4 . [ it : updistatheta ] the element @xmath40 has the same @xmath1-valued @xmath0-distribution as @xmath21 5 .
[ it : poddstheta ] the odd moments of @xmath42 vanish , i.e. , for every @xmath22 and all @xmath23 , we have @xmath442 4 .
[ it : uptheta ] there exists a @xmath1-valued @xmath0-noncommutative probability space @xmath143 with elements @xmath424 such that parts [ it : psatheta][it : updistatheta ] of [ it : up0theta ] hold . in part
[ it : up0theta ] or [ it : uptheta ] , the even moments of @xmath42 are given by , for every @xmath38 and all @xmath286 , @xmath443 in the proof , we will use the following lemma .
[ lem : alphahhu ] suppose @xmath21 is @xmath1-valued r - diagonal .
fix @xmath10 and suppose that for all @xmath444 and all @xmath445 the equality holds .
then for all @xmath446 and all @xmath447 , we have @xmath448 \\
= \theta\big({{\hat{\alpha}}}_{(1,2,\ldots,1,2)}(\pi)[\theta(b_1),b_2,\theta(b_3),\ldots , b_{2n-2},\theta(b_{2n-1})]\big).\end{gathered}\ ] ] this follows by straightforward computation using the recursive formula for the maps @xmath54 by induction on @xmath449 , namely , on the number of blocks in @xmath58 .
suppose @xmath450 .
then both sides of vanish unless both blocks are of even length . if both blocks are intervals , then @xmath451 for some @xmath452 and using we have @xmath453 \\ & = \begin{aligned}[t ] & \alpha_{(2,1,\ldots,2,1)}(b_1,\theta(b_2),b_3,\ldots,\theta(b_{2p-2}),b_{2p-1})\theta(b_{2p } ) \\ & \quad\cdot\alpha_{(2,1,\ldots,2,1)}(b_{2p+1},\theta(b_{2p+2}),b_{2p+3},\ldots,\theta(b_{2n-2}),b_{2n-1 } ) \end{aligned } \\ & = \begin{aligned}[t ] & \theta\big(\alpha_{(1,2,\ldots,1,2)}(\theta(b_1),b_2,\theta(b_3),\ldots , b_{2p-2},\theta(b_{2p-1}))b_{2p } \\ & \quad\cdot\alpha_{(1,2,\ldots,1,2)}(\theta(b_{2p+1}),b_{2p+2},\theta(b_{2p+3}),\ldots , b_{2n-2},\theta(b_{2n-1}))\big ) \end{aligned } \\ & = \theta\big({{\hat{\alpha}}}_{(1,2,\ldots,1,2)}(\pi)[\theta(b_1),b_2,\theta(b_3),\ldots , b_{2n-2},\theta(b_{2n-1})]\big).\end{aligned}\ ] ]
if @xmath450 and @xmath58 has an internal interval of the form @xmath454 for some @xmath455 , then again using we have @xmath456 \label{eq : ii1 - 1 } \\ & = \begin{aligned}[t ] \alpha_{(2,1,\ldots,2,1)}\big(&b_1,\theta(b_2),b_3,\ldots,\theta(b_{2p-2}),b_{2p-1 } , \\ & \;\theta(b_{2p})\alpha_{(2,1,\ldots,2,1)}\big(b_{2p+1},\theta(b_{2p+2}),\ldots , b_{2p+2q-1}\big)\theta(b_{2p+2q } ) , \\ & \ ; b_{2p+2q+1},\theta(b_{2p+2p+2}),b_{2p+2q+3},\ldots,\theta(b_{2n-2}),b_{2n-1}\big ) \end{aligned } \label{eq : ii1 - 2 } \displaybreak[2 ] \\ & = \begin{aligned}[t ] \alpha_{(2,1,\ldots,2,1)}\big(&b_1,\theta(b_2),b_3,\ldots,\theta(b_{2p-2}),b_{2p-1 } , \\ & \;\theta\big(b_{2p}\alpha_{(1,2,\ldots,1,2)}\big(\theta(b_{2p+1}),b_{2p+2},\ldots,\theta(b_{2p+2q-1})\big)b_{2p+2q}\big ) , \\ & \ ; b_{2p+2q+1},\theta(b_{2p+2p+2}),b_{2p+2q+3},\ldots,\theta(b_{2n-2}),b_{2n-1}\big ) \end{aligned } \label{eq : ii1 - 3 } \displaybreak[2 ] \\ & = \begin{aligned}[t ] \theta\big(\alpha_{(1,2,\ldots,1,2)}\big(&\theta(b_1),b_2,\theta(b_3),\ldots , b_{2p-2},\theta(b_{2p-1 } ) , \\ & \ ; b_{2p}\alpha_{(1,2,\ldots,1,2)}\big(\theta(b_{2p+1}),b_{2p+2},\ldots,\theta(b_{2p+2q-1})\big)b_{2p+2q } , \\ & \ ; \theta(b_{2p+2q+1}),b_{2p+2p+2},\theta(b_{2p+2q+3}),\ldots , b_{2n-2},\theta(b_{2n-1})\big)\big ) \end{aligned } \label{eq : ii1 - 4 } \\ & = \theta\big({{\hat{\alpha}}}_{(1,2,\ldots,1,2)}(\pi)\big[\theta(b_1),b_2,\theta(b_3),\ldots , b_{2n-2},\theta(b_{2n-1})\big]\big ) .
\label{eq : ii1 - 5}\end{aligned}\ ] ] if @xmath450 and @xmath58 has an internal interval of the form @xmath457 for some @xmath458 , then again using we have @xmath459 \label{eq : ii2 - 1 } \\ & \quad=\begin{aligned}[t ] \alpha_{(2,1,\ldots,2,1)}\big(&b_1,\theta(b_2),\ldots , b_{2p-3},\theta(b_{2p-2 } ) , \\ & \ ; b_{2p-1}\alpha_{(1,2,\ldots,1,2)}\big(\theta(b_{2p}),b_{2p+1},\ldots,\theta(b_{2p+2q-2})\big)b_{2p+2q-1 } , \\ & \ ; \theta(b_{2p+2q}),b_{2p+2p+1},\ldots,\theta(b_{2n-2}),b_{2n-1}\big ) \end{aligned } \label{eq : ii2 - 2 } \displaybreak[2 ] \\ & \quad=\begin{aligned}[t ] \alpha_{(2,1,\ldots,2,1)}\big(&b_1,\theta(b_2),\ldots , b_{2p-3},\theta(b_{2p-2 } ) , \\ & \ ; \theta^{-1}\big(\theta(b_{2p-1})\alpha_{(2,1,\ldots,2,1)}\big(b_{2p},\theta(b_{2p+1}),\ldots , b_{2p+2q-2}\big)\theta(b_{2p+2q-1})\big ) , \\ & \ ; \theta(b_{2p+2q}),b_{2p+2p+1},\ldots,\theta(b_{2n-2}),b_{2n-1}\big ) \end{aligned } \label{eq : ii2 - 3 } \displaybreak[2 ] \\ & \quad=\begin{aligned}[t ] \theta\big(\alpha_{(1,2,\ldots,1,2)}\big(&\theta(b_1),b_2,\ldots,\theta(b_{2p-3}),b_{2p-2 } , \\ & \ ; \theta(b_{2p-1})\alpha_{(2,1,\ldots,2,1)}\big(b_{2p},\theta(b_{2p+1}),\ldots , b_{2p+2q-2}\big)\theta(b_{2p+2q-1 } ) , \\ & \ ; b_{2p+2q},\theta(b_{2p+2p+1}),\ldots , b_{2n-2},\theta(b_{2n-1})\big)\big ) \end{aligned } \label{eq : ii2 - 4 } \\ & \quad=\theta\big({{\hat{\alpha}}}_{(1,2,\ldots,1,2)}(\pi)\big[\theta(b_1),b_2,\theta(b_3),\ldots , b_{2n-2},\theta(b_{2n-1})\big]\big ) .
\label{eq : ii2 - 5}\end{aligned}\ ] ] this finishes the proof of the case @xmath450 . the induction step when @xmath460 is very similar .
we see that both sides of vanish unless @xmath58 has an internal interval of even length , and then , using the induction hypothesis , one gets recursive formulas like in , except that in - and - , each @xmath461 appearing immediately after an equality sign is replaced by @xmath462 $ ] and each @xmath463 is replaced by @xmath464 $ ] , where @xmath465 is obtained from @xmath58 by removing the corresponding internal interval and renumbering . the proof is patterned after the proof of proposition [ prop : hu ] .
the equivalence [ it : eaa*theta]@xmath324[it : beta12theta ] follows easily from the moment - cumulant formula and lemma [ lem : alphahhu ] , by induction on @xmath226 , keeping in mind that all of the cumulant maps vanish except the @xmath403-th order ones @xmath466 and @xmath467 , ( @xmath38 ) .
we now show [ it :
beta12theta]@xmath257[it : up0theta ] .
endow @xmath427 with the @xmath0-operation that extends the given one on @xmath1 and satisfies @xmath428 .
let @xmath429 be the @xmath1-valued distribution of an element whose odd order @xmath1-valued cumulant maps are all zero and whose @xmath1-valued cumulant map of order @xmath403 is the map @xmath468 given by @xmath469 for all @xmath38 , where the second equality above is from .
write @xmath42 for the element @xmath360 of @xmath427 . by proposition [ prop :
alpha * ] and the identity , @xmath80 is a self - adjoint map
. we claim ( letting @xmath468 be shorthand for @xmath470 with @xmath337 repeated @xmath403 times and similarly letting @xmath471 be shorthand for @xmath472 ) , that for every @xmath38 , every @xmath473 and every @xmath286 , we have @xmath474 = { { \hat{\alpha}}}_{(2,1,\ldots,2,1)}(\pi)\big[\theta^{-1}(b_1),b_2,\theta^{-1}(b_3),\ldots , b_{2k-2},\theta^{-1}(b_{2k-1})\big].\ ] ] if @xmath475 then this holds by definition , and the general case is proved by induction on the number @xmath449 of blocks in @xmath58 . for the induction step , if @xmath58 has an internal interval block of the form @xmath476 , for some integers @xmath477 , then letting @xmath478 be obtained from @xmath58 by removing the block @xmath1 and renumbering , we have @xmath479 \\ & \;={{\hat{\gamma}}}_{k - s}(\pi')\big[b_1,\ldots , b_{2r-2},b_{2r-1}\gamma_s(b_{2r},\ldots , b_{2r+2s-2})b_{2r+2s-1},b_{2r+2s},\ldots , b_{2k-1}\big ] \\ & \;=\begin{aligned}[t ] { { \hat{\alpha}}}_{(2,1,\ldots,2,1)}(\pi')\big[&\theta^{-1}(b_1),b_2,\ldots,\theta^{-1}(b_{2r-3}),b_{2r-2 } , \\ & \theta^{-1}\big(b_{2r-1}\alpha_{(2,1,\ldots,2,1)}\big(\theta^{-1}(b_{2r}),b_{2r+1},\ldots,\theta^{-1}(b_{2r+2s-2})\big)b_{2r+2s-1}\big ) , \\
& b_{2r+2s},\theta^{-1}(b_{2r+2s+1}),\ldots , b_{2k-2},\theta^{-1}(b_{2k-1})\big ] \end{aligned } \\ & \;=\begin{aligned}[t ] { { \hat{\alpha}}}_{(2,1,\ldots,2,1)}(\pi')\big[&\theta^{-1}(b_1),b_2,\ldots,\theta^{-1}(b_{2r-3}),b_{2r-2 } , \\ & \theta^{-1}(b_{2r-1})\alpha_{(1,2,\ldots,1,2)}\big(b_{2r},\theta^{-1}(b_{2r+1}),\ldots , b_{2r+2s-2}\big)\theta^{-1}(b_{2r+2s-1 } ) , \\ & b_{2r+2s},\theta^{-1}(b_{2r+2s+1}),\ldots , b_{2k-2},\theta^{-1}(b_{2k-1})\big ] \end{aligned } \\ & \;={{\hat{\alpha}}}_{(2,1,\ldots,2,1)}(\pi)\big[\theta^{-1}(b_1),b_2,\theta^{-1}(b_3),\ldots , b_{2k-2},\theta^{-1}(b_{2k-1})\big],\end{aligned}\ ] ] where the first and last equalities are by the recursive formula for @xmath480 ( see ) the second equality is by the induction hypothesis and , and the third equality is from .
the cases when @xmath58 has an internal interval block of the form @xmath481 for some integers @xmath482 or when @xmath58 has only two blocks , both of them intervals , are treated similarly , to prove . from the equality , the fact that all odd @xmath1-valued moments vanish and the moment - cumulant formula , we deduce the equality for all @xmath38 and all @xmath445 . using a crossed product construction and the amalgamated free product construction ,
there is a @xmath1-valued @xmath0-noncommutative probability space @xmath430 ) containing elements @xmath424 satisfying conditions [ it : psatheta ] , [ it : uhutheta ] and [ it : upfreetheta ] of [ it : up0theta ] .
( this should be reasonably clear , but see the proof of lemma [ lem : c * ] , below for more details . ) by construction , [ it : poddstheta ] holds .
let @xmath431 .
by lemma [ lem:2.3 ] , @xmath432 is @xmath1-valued r - diagonal . in order to prove that @xmath21 and @xmath432 have the same @xmath1-valued @xmath0-distribution , using remark [ rem : l2.1 ] , it will suffice to show that the equations and always hold . to show
, we use @xmath483 and . to show
, we use @xmath484 thus , we get @xmath485 where the first equality is due to freeness , the second equality is by [ it : upfreetheta ] of [ it : up0theta ] , the third equality is from and the fourth equality is .
this proves we get .
thus , @xmath21 and @xmath437 have the same @xmath0-distribution ; namely , [ it : up0theta ] holds . the implication [ it : up0theta]@xmath257[it : uptheta ] is trivially true . assuming [ it : uptheta ]
, the equality follows by writing @xmath486 we now show [ it : uptheta]@xmath257[it : eaa*theta ] .
that @xmath21 is @xmath1-valued r - diagonal follows from lemma [ lem:2.4 ] . to show ,
let @xmath437 . by hypothesis ,
@xmath21 and @xmath432 have the same @xmath1-valued @xmath0-distribution .
of course , the equalities and hold .
therefore , we have @xmath487 \\ & = \theta\big({{\widetilde{{{\mathcal{e}}}}}}\big(u\big(p\theta(b_1)p\theta(b_2)p\theta(b_3)p\cdots
\theta(b_{2k-2})p\theta(b_{2k-1})p\big)u^*\big)\big ) \displaybreak[2 ] \\ & = \theta\big({{\widetilde{{{\mathcal{e}}}}}}\big({{\tilde a}}\theta(b_1){{\tilde a}}^*b_2{{\tilde a}}\theta(b_3)p\cdots b_{2k-2}{{\tilde a}}\theta(b_{2k-1}){{\tilde a}}^*\big)\big ) \\ & = \theta\big({{\mathcal{e}}}\big(a\theta(b_1)a^*b_2a\theta(b_3)p\cdots b_{2k-2}a\theta(b_{2k-1})a^*\big)\big).\end{aligned}\ ] ] where the second equality is , the third is by condition [ it : uhutheta ] of [ it : up0theta ] and the fourth is .
[ lem : c * ] in proposition [ prop : hutheta ] , if @xmath1 is a c@xmath45-algebra and if @xmath2 is a @xmath1-valued c@xmath45-noncommutative probability space , then in part [ it : up0theta ] , @xmath143 can be realized as a @xmath1-valued c@xmath45-noncommutative probability space . in the proof of proposition [ prop : hutheta ] ,
the @xmath0-noncommutative probability space @xmath143 was constructed as an algebraic amalgamated free product @xmath488 here @xmath489 is an algebraic crossed product of @xmath1 by the automorphism @xmath438 and @xmath490 is the canonical conditional expectation , and the unitary @xmath41 of proposition [ prop : hutheta][it : up0theta ] is from the element @xmath491 ; recall , @xmath80 is defined abstractly by specifying cumulants , and is the @xmath0-distribution of the element we called @xmath42 . if @xmath1 is a c@xmath45-algebra , then we can take , instead , the crossed product c@xmath45-algebra @xmath492 and the canonical conditional expectation @xmath493 .
we also want a @xmath1-valued c@xmath45-noncommutative probability space to replace @xmath494 .
endow @xmath495 with the @xmath0-operation determined by that of @xmath1 and by setting @xmath496 and let @xmath497 denote the @xmath0-distribution of the pair @xmath32 . since @xmath2 is a @xmath1-valued c@xmath45-noncommutative probability space , we have ( see , for example , @xcite ) , the hilbert @xmath1-module @xmath498 , on which @xmath3 acts by left multiplication as bounded , adjointable operators .
we may without loss of generality assume @xmath3 is generated as a c@xmath45-algebra by @xmath499
. it will be convenient to identify @xmath500 with @xmath501 , which is obtained by separation and completion , after endowing @xmath495 with the inner product @xmath502 , on which @xmath495 acts by left multiplication as bounded , adjointable operators on @xmath500 .
let @xmath58 denote this action .
letting @xmath503 denote the defining mapping @xmath504 , the set @xmath505 is a complemented subspace of @xmath500 isomorphic to the hilbert @xmath1-module @xmath1 , with a self - adjoint projection @xmath506 .
we have the conditional expectation @xmath507 given by @xmath508 , and we have @xmath509 for all @xmath510 .
consider the closed right @xmath1-submodules of @xmath500 , @xmath511 then @xmath1 ( the image of @xmath512 ) is a submodule of @xmath513 , the spaces @xmath513 and @xmath514 are orthogonal to each other with respect to the @xmath1-valued inner product and their sum @xmath515 , which we write @xmath516 , is also a closed submodule of @xmath500 .
we have the representation @xmath517 given by @xmath518 we also have the bounded operator @xmath519 on @xmath516 given by @xmath520 since @xmath521 , we easily see that @xmath519 is self - adjoint .
thus , the representation @xmath522 of @xmath1 extends to a representation @xmath523 by setting @xmath524 . of course
, @xmath525 and , if we identify @xmath1 with @xmath526 , then we have the conditional expectation @xmath527 given by @xmath528 .
now we see that , for every @xmath529 , we have @xmath530 . indeed , for a monomial @xmath531 ,
if @xmath29 is odd , then @xmath532 and @xmath533 , while for @xmath29 even , we have @xmath534 and @xmath535 where for the third equality we used . thus , we may replace @xmath494 with the @xmath1-valued c@xmath45-noncommutative probability space @xmath536 , and instead of the algebraic amalgamated free product , we let @xmath537 identifying @xmath42 with @xmath538 and @xmath539 with the unitary corresponding to the group element @xmath337 in the crossed product construction , we are done . [
thm : c*pos ] let @xmath1 be a c@xmath45-algebra and suppose @xmath2 is a @xmath1-valued c@xmath45-noncommutative probability space .
suppose @xmath7 and let @xmath438 be a @xmath0-automorphism of @xmath1 . then the following are equivalent : 1 .
[ it : eaa*c*pos ] @xmath21 is @xmath1-valued r - diagonal and for all @xmath38 and all @xmath286 , we have @xmath540 2 .
[ it : beta12c*pos ] @xmath21 is @xmath1-valued r - diagonal and for all @xmath38 , the @xmath403-th order @xmath1-valued cumulant maps @xmath541 and @xmath542 ( notation defined near ) satisfy , for all @xmath286 , @xmath543 3 .
[ it : c*pos ] there exists a @xmath1-valued c@xmath45-noncommutative probability space @xmath143 with elements @xmath544 satisfying 1 .
[ it : uc*pos ] @xmath41 is a haar unitary that normalizes @xmath1 and satisfies @xmath441 for all @xmath154 2 .
[ it : pc*pos ] @xmath46 , 3 .
[ it : c*posfree ] @xmath545 and @xmath44 are free ( over @xmath1 ) with respect to @xmath163 4 .
[ it : c*posdistr ] @xmath21 and @xmath40 have the same @xmath1-valued @xmath0-distribution
. moreover , the even moments of @xmath42 in [ it : c*pos ] are given by @xmath546 for all @xmath38 and all @xmath286 .
the equivalence of [ it : eaa*c*pos ] and [ it : beta12c*pos ] , as well as the implication [ it : c*pos]@xmath257[it : eaa*c*pos ] follow from proposition [ prop : hutheta ] .
it remains to show [ it : eaa*c*pos]@xmath257[it : c*pos ] .
assuming [ it : eaa*c*pos ] , by proposition [ prop : hutheta ] and lemma [ lem : c * ] , there is a @xmath1-valued c@xmath45-noncommutative probability space @xmath143 , and elements @xmath41 and @xmath407 of @xmath547 such that [ it : uc*pos ] , [ it : c*posfree ] and [ it : c*posdistr ] of [ it : c*pos ] hold , the even moments of @xmath42 are given by and the odd moments of @xmath42 all vanish . by modifying @xmath143 , if necessary
, we may assume that @xmath548 , where @xmath549 is a symmetry ( i.e. , a self - adjoint unitary element ) that commutes with @xmath550 and with every element of @xmath1 , satisfies @xmath551 and such that @xmath552 is free from @xmath43 .
indeed , we may without loss of generality assume @xmath547 is the c@xmath45-algebra generated by @xmath553 and the gns representation of @xmath163 is faithful , in which case @xmath547 is the amalgamated free product of @xmath554 and @xmath555 over @xmath1
. we may enlarge @xmath143 to be the amalgamated free product of @xmath554 and @xmath556 , where the conditional expectation of the latter onto @xmath1 is @xmath557 , where @xmath558 is the state sending @xmath559 to @xmath210 .
we let @xmath560 and note that the pair @xmath561 has the same @xmath1-valued @xmath0-distribution as @xmath562 .
thus , @xmath21 has the same @xmath0-distribution as @xmath563 .
we observe that , with respect to @xmath163 , @xmath564 is a @xmath1-normalizing haar unitary with @xmath565 .
we need only show that @xmath566 is free from @xmath567 with respect to @xmath163 .
this is straightforward to verify .
indeed , we need only show that every alternating product in the sets @xmath568 and @xmath569 evaluates to zero under @xmath163 .
however , keeping in mind that @xmath549 commutes with @xmath550 , we may rewrite each such product as an alternating product in the sets @xmath43 and @xmath570 and such an alternating product evaluates to @xmath210 under @xmath163 by freeness of @xmath571 and @xmath43 . in the case of @xmath31 ,
theorem [ thm : c*pos ] reduces to the following , which is certainly well known , though we did nt find a good reference .
it would follow straighforwardly from proposition 2.6 of @xcite .
[ cor : scalarpolar ] let @xmath572 be a tracial ( scalar - valued ) c@xmath45-noncommutative probability space and let @xmath7
. then the following are equivalent : 1 .
[ it : scalarrdiag ] @xmath21 is r - diagonal .
[ it : scalarup ] there exists a tracial ( scalar - valued ) c@xmath45-noncommutative probability space @xmath573 and elements @xmath544 such that 1 .
[ it : uscalar ] @xmath41 is a haar unitary , 2 .
[ it : pscalar ] @xmath46 , 3 .
[ it : scalarfree ] @xmath545 and @xmath44 are free with respect to @xmath574 4 . [
it : scalardistr ] @xmath21 and @xmath40 have the same @xmath0-distribution .
we apply theorem [ thm : c*pos ] in the case @xmath31 .
then of course each cumulant map @xmath575 is just a real number and @xmath438 is the identity map .
the condition just becomes @xmath576 for every @xmath226 , and this holds , by proposition [ prop : tracerdiag ] , because we assume @xmath82 is a trace .
now the equivalence of [ it : scalarrdiag ] and [ it : scalarup ] above follows from the equivalence of [ it : beta12c*pos ] and [ it : c*pos ] in theorem [ thm : c*pos ] . for an element @xmath21 in a @xmath1-valued w@xmath45-noncommutative probability space
whose polar decomposition is @xmath577 , the joint @xmath1-valued @xmath0-disctribution of @xmath578 is completely determined by the @xmath1-valued @xmath0-distribution of @xmath21 , and vice - versa .
thus , the following corollary follows from the theorem [ thm : c*pos ] .
[ cor : polarfree ] let @xmath1 be a von neumann algebra and suppose @xmath2 is a @xmath1-valued w@xmath45-noncommutative probability space whose gns - representation is faithful .
suppose @xmath7 has zero kernel and dense range and let @xmath438 be a normal @xmath0-automorphism of @xmath1
. then the following are equivalent : 1 .
[ it : eaa*w*pos ] @xmath21 is @xmath1-valued r - diagonal and for all @xmath38 and all @xmath286 , holds .
[ it : beta12w*pos ] @xmath21 is @xmath1-valued r - diagonal and for all @xmath38 , the @xmath403-th order @xmath1-valued cumulant maps @xmath541 and @xmath542 satisfy , for all @xmath286 .
[ it : w*polar ] letting @xmath577 be the polar decomposition of @xmath21 , 1 . [ it : uw * ] @xmath41 is a haar unitary that normalizes @xmath1 and satisfies @xmath441 for all @xmath154 , 2 .
[ it : w*polarfree ] @xmath545 and @xmath579 are free ( over @xmath1 ) with respect to @xmath163 .
[ ex : ub ] perhaps the easiest example of a @xmath1-valued r - diagonal element satisfying the conditions of theorem [ thm : c*pos ] is when @xmath580 . in this case , the operator @xmath21 can be realized in the crossed product c@xmath45-algebra @xmath492 with respect to the canonoical conditional expectation onto @xmath1 . in the case
when @xmath1 is a commutative von neumann algebra with a specified normal , faithful tracial state and @xmath438 is a trace - preserving normal automorphism that is ergodic , such @xmath1-valued r - diagonal operators were studied in @xcite and , among other results , their brown measures were computed .
in this section , we examine algebra - valued circular elements , which are a very special class of algebra - valued r - diagonal elements . as before ,
let @xmath1 be a unital @xmath0-algebra and let @xmath2 be a @xmath1-valued @xmath0-noncommutative probability space ; let @xmath7 ; for convenience we label @xmath177 and @xmath178 and use the involution @xmath581 with @xmath582 .
let @xmath583 be the @xmath1-valued @xmath0-distribution of @xmath180 , where we set @xmath496 .
let @xmath584 and let @xmath585 be the family of cumulant maps for the pair @xmath180 .
[ def : circ ] we say that @xmath21 is _ @xmath1-valued circular _ if @xmath586 whenever @xmath587 and @xmath588 .
clearly , @xmath1-valued circular elements are @xmath1-valued r - diagonal elements .
the notion of @xmath1-valued circular first appeared in @xcite , where niady proved interesting combinatorial results about a certain @xmath1-valued circular operator for @xmath589 $ ] that he noted was equal to the quasinilpotent dt - opertor @xmath590 from @xcite ( and a proof that it is , in fact , @xmath1-valued circular can be found in @xcite ) see example [ ex : circdt ] for more details .
the notion of @xmath1-valued circular operators is closely connected to that of @xmath1-valued semicircular ( also called @xmath1-gaussian ) operators that were introduced by speicher @xcite and were studied and constructed on fock spaces by shlyakhtenko @xcite . at the purely algebraic level , consider a family @xmath591 of @xmath1-valued random variables in a @xmath1-valued noncommutative probability space .
let @xmath592 and denote by @xmath593 the @xmath1-valued cumulant maps associated to this family .
the family is said to be _ centered @xmath1-valued semicircular _ if @xmath594 for every @xmath57 of length not equal to @xmath595 .
in fact , the following result is immediate from the definitions : [ prop : circsemi ] suppose @xmath21 is an element of a @xmath1-valued @xmath0-noncommutative probability space and let @xmath596 let @xmath584 and let @xmath585 be the family of cumulant maps for the pair @xmath180 .
let @xmath593 be the family of cumulant maps for the pair @xmath597
. then the following are equivalent : 1 .
@xmath21 is @xmath1-valued circular 2 .
the pair @xmath597 is centered @xmath1-valued semicircular , @xmath598 and @xmath599 .
furthermore , when the above conditions hold , we have @xmath600 using speicher s result that freeness is equivalent to vanishing of mixed cumulants ( see section 3.3 of @xcite ) , we have : if @xmath21 is a @xmath1-valued circular element with associated cumulant maps @xmath601 and @xmath602 , then @xmath603 and @xmath604 are free with respect to the conditional expectation onto @xmath1 if and only if @xmath605 suppose @xmath1 is a c@xmath45-algebra and @xmath80 is the @xmath1-valued distribution of a family of @xmath1-valued random variables indexed by @xmath364 .
we say that @xmath80 is positive if @xmath141 for every @xmath142 .
in general , it can be difficult to decide whether @xmath80 is positive knowing only the @xmath1-valued cumulant maps of the family .
however , in the case that the family is @xmath1-valued semicircular , an answer is provided by theorem 4.3.1 of @xcite . in the case that @xmath364 is finite , this condition is equivalent to complete positivity of the covariance @xmath606 , which is defined by @xmath607 in the case of the family @xmath597 of real and imaginary parts of a @xmath1-valued circular system from proposition [ prop : circsemi ] , the covariance @xmath608 is given by @xmath609 thus , we have the following corollary of speicher s theorem mentioned above and of shlyakhtenko s construction @xcite of @xmath1-valued semicircular elements . to recall the set - up , let @xmath1 be a @xmath610-algebra as before , consider a @xmath1-valued circular element @xmath21 , let @xmath611 and let @xmath80 be the distribution and @xmath601 , @xmath602 the cumulant maps of @xmath180 , as usual .
endow the algebra @xmath495 with the @xmath0-operation that extends the given one on @xmath1 by setting @xmath496 .
[ cor : poscirc ] the distribution @xmath80 is positive if and only if @xmath601 and @xmath602 are completely positive maps from @xmath1 into itself . in this case , @xmath80 is the @xmath0-distribution of a @xmath1-valued circular element in a @xmath1-valued c@xmath45-noncommutative probability space . turning back to the case of @xmath1 a general @xmath0-algebra , without assuming anything about boundedness or positivity , from proposition [ prop : tracerdiag ] , we immediately see the following : [ prop : tracecirc ] if @xmath21 is a @xmath1-valued circular element with associated cumulant maps @xmath601 and @xmath602 and if @xmath82 is a tracial linear functional on @xmath1 , then @xmath73 is tracial on @xmath405 if and only if , for all @xmath8 , we have @xmath612 we will now examine the power series considered in section [ sec : series ] for a @xmath1-valued circular element @xmath21 . theorem [ thm : powerseries ] yields [ prop : recur ] consider the formal power series ( in the sense described in section [ sec : series ] ) @xmath613 for a @xmath1-valued circular element @xmath21
. then @xmath614 g(b_1,b_2)&=1+\alpha_{(2,1)}(b_1f(b_2,b_1))b_2g(b_1,b_2 ) .
\label{eq : grecur } \end{aligned}\ ] ] in the case of a @xmath1-valued c@xmath45-noncommutative probability space , the series define @xmath1-valued holomorphic functions with domain equal to the subset of @xmath615 consisting of all pairs @xmath616 such that @xmath617 . from these functions
one can recover only some of the information about the @xmath1-valued distribution of @xmath21 .
later , in the appendix , we will study a particular case of these functions evaluated at @xmath618 .
[ ex : circdt ] as observed by niady @xcite , ( see @xcite for a proof ) , the quasinilpotent dt - operator @xmath590 from @xcite is a @xmath619$]-valued circular operator , with cumulant maps given by @xmath620 we note that if @xmath438 is the automorphism of @xmath619 $ ] corresponding to the homeomophism @xmath621 of @xmath622 $ ] , then @xmath623 thus , corollary [ cor : polarfree ] applies and we have the following result .
[ cor : dtpolar ] let @xmath590 be a quasinilpotent dt - operator in an @xmath619$]-valued @xmath624-noncommutative probability space @xmath2 with @xmath4 faithful and let @xmath625 be its polar decomposition
. then 1 .
@xmath626 is an @xmath619$]-normalizing haar unitary element , 2 . for every @xmath627
$ ] we have @xmath628 , 3 .
@xmath629 and @xmath630 are free with respect to @xmath4 . as advertised , here is an application of lemma [ lem : notfree ] , yielding a @xmath1-valued r - diagonal element @xmath21 with @xmath1 two - dimensional and such that in the polar decomposition @xmath577 of @xmath21 , @xmath41 is unitary but @xmath43 and @xmath579 are not free over @xmath1 .
[ ex : nofreepolar ] take @xmath631 , @xmath632 and define @xmath633 then @xmath601 and @xmath602 are completely positive and satisfy the condition of proposition [ prop : tracecirc ] for traciality .
thus , by corollary [ cor : poscirc ] , there is a @xmath1-valued c@xmath45-noncommutative probability space @xmath2 containing a @xmath1-valued circular element @xmath21 with corresponding cumulant maps @xmath601 and @xmath602 ; we assume @xmath3 is generated as a c@xmath45-algebra by @xmath499 , and that the gns representation of @xmath4 is faithful ; by proposition [ prop : tracecirc ] , @xmath634 is a positive trace on @xmath3 ; since it has faithful gns representation , it is faithful .
we also have @xmath635 and @xmath636 .
thus by lemma [ lem : notfree ] , @xmath21 does not have the same @xmath1-valued @xmath0-distribution as any element @xmath40 in any @xmath1-valued @xmath0-noncommutative probability space with @xmath41 unitary , with @xmath42 self - adjoint and with @xmath43 and @xmath44 free over @xmath1 .
in particular , we may take the von neumann algebra generated by the image of the gns representation of @xmath82 we get a larger @xmath1-valued @xmath0-noncommutative probability space @xmath143 and in @xmath547 we have the polar decomposition @xmath577 of @xmath21 . by proposition
[ prop : afacts ] below , @xmath21 has zero kernel , so @xmath41 is unitary . by proposition [ prop : w*polar ] ,
@xmath41 is a haar unitary . but @xmath43 and @xmath579 are not free with respect to @xmath163 .
in this appendix , we describe investigations of the distribution @xmath637 of the element @xmath638 with respect to the trace @xmath82 , where @xmath21 is the @xmath1-valued circular element described in example [ ex : nofreepolar ] .
a mathematica notebook file containing the detailed calculations will be made available with this paper . using the equations and with @xmath639 for @xmath549 in some neighborhood of @xmath210 in @xmath49 and letting @xmath640 ,
we have @xmath641 we are interested in the moment generating function for @xmath638 with respect to the trace @xmath82 , and this is the function @xmath642 the recursive relations in proposition [ prop : recur ] yield @xmath643 since @xmath631 and @xmath644 and @xmath645 are @xmath1-valued functions , we write @xmath646 and @xmath647 . using the definitions of @xmath601 and @xmath602 from example [ ex : nofreepolar ] , we have the recursive relations @xmath648 we substitute @xmath649 and , using simple elimination , arrive at a polynomial identity for @xmath650 : @xmath651 this allows computation of arbitrarily many terms of the series expansion for @xmath650 around @xmath210 , and we find @xmath652 the stieltjes transform @xmath653 of the measure @xmath654 is , for @xmath655 in the complement of the closed support of @xmath654 , @xmath656 note that @xmath657 is real when @xmath655 is real and @xmath658 is large . from
we get the identity @xmath659 stieltjes inversion can be used to find the measure @xmath654 from knowledge of the algebraic function @xmath375 : @xmath654 has an atom at a point @xmath660 if and only if , @xmath661 and then the limit exists and the value of this limit equals @xmath662 while elsewhere on the real line , @xmath654 has density given by @xmath663 note that , since @xmath664 is positive and bounded , @xmath654 is compactly supported in @xmath665 .
of course , an explicit formula for @xmath375 can be found involving radicals , ( solving the quartic ) and care must be taken to find the correct branch of @xmath375 near the real axis ; the correct branch of @xmath375 is the one that near @xmath666 has asymptotics @xmath667 .
however , the following facts can be seen by a somewhat less arduous analysis : using the method of newton s polytope to find puiseux series asymptotic expantions for the algebraic function roots of the polynomial equation , we find that near @xmath670 , the four algebraic functions are @xmath671 where the three roots @xmath672 for @xmath673 result from chosing the different branches of @xmath674 .
we set @xmath675 to be the one that makes @xmath674 negative real when @xmath655 is negative real .
already we see that the stieltjes transform @xmath375 fails the condition at @xmath676 , so @xmath677 has no atom at @xmath210 and , thus , @xmath21 has zero kernel and zero co - kernel .
since @xmath654 has no support in @xmath678 , we see that the stieltjes transform must be equal to either @xmath679 or @xmath675 near @xmath670 ( and in the domain of @xmath375 ) , since as @xmath655 approaches negative numbers near zero from above , both @xmath680 and @xmath681 approach nonzero numbers . to see that @xmath21 is not invertible
, it will suffice to see that @xmath682 , since it will yield nonzero density for @xmath654 on some interval @xmath683 .
the discriminant of the polynomial in , with respect to the variable @xmath375 , is @xmath684 whose real roots , other than @xmath210 , are approximately @xmath685 , and @xmath686 .
the coefficient of @xmath687 in is @xmath688 , which has roots only at @xmath210 .
thus , the only place on the real axis where @xmath375 can diverge to @xmath689 is @xmath210 , and the only places where two or more of the four algebraic roots of can agree is where @xmath655 is one of the roots of the polynomial .
the polynomial has real coefficients , so roots come in complex conjugate pairs .
thus @xmath657 , which is real and negative for @xmath690 , must remain real as @xmath655 approaches @xmath210 through negative numbers .
moreover , setting @xmath375 equal to @xmath210 in yields @xmath670 , so @xmath375 is nonvanishing on the negative real axis .
therefore , @xmath657 is real and strictly negative for all @xmath691 .
however , this is clearly not possible for the function @xmath679 , according to the asymptotics .
thus , @xmath682 and @xmath21 is not invertible .
similar considerations show that the maximum of the support of @xmath654 , which is equal to @xmath692 , can only be one of the roots of .
however , the coefficients in the moment series yield lower bounds on @xmath692 ; already the coefficient of @xmath294 yields @xmath693 .
so @xmath692 equals the largest of the real roots of , as required for the assertion in the proposition about @xmath694 . from the asymptotics
, we easily see that the density of the measure @xmath654 behaves asymptotically like @xmath695 as @xmath696 approaches @xmath210 from the right .
thus , the distribution @xmath697 has density whose asymptotic expansion is @xmath698 as @xmath549 tends to @xmath210 from the right .
we calculated the density of @xmath697 numerically ; a plot is in figure [ fig : densplot ] ; for comparison purposes , the density @xmath699 of the quarter - circular element @xmath700 is plotted on the same grid , with the dashed line . for details
see the mathematica notebook file that is available with this paper . | for an element in an algebra - valued @xmath0-noncommutative probability space , equivalent conditions for algebra - valued r - diagonality ( a notion introduced by niady and speicher ) are proved .
formal power series relations involving the moments and cumulants of such r - diagonal elements are proved .
decompositions of algebra - valued r - diagonal elements into products of the form unitary times self - adjoint are investigated ; sufficient conditions , in terms of cumulants , for @xmath0-freeness of the unitary and the self - adjoint part are proved , and a tracial example is given where @xmath0-freeness fails .
the particular case of algebra - valued circular elements is considered ; as an application , the polar decompostion of the quasinilpotent dt - operator is described .
research supported in part by nsf grant dms-1301604 . ] research supported in part by nsf grant dms-1202660 . ] |
cold dark matter ( cdm ) has had considerable success in accounting for the observed large - scale structure of our universe . on galactic and sub - galactic scales , however , the abundance of low - mass structure and the degree of concentration of dark matter halos predicted by cdm simulations have provoked a considerable amount of discussion . to clarify how cdm behaves on the smallest scales , diemand and collaborators recently performed the highest resolution numerical simulations of dark matter clustering to date ( diemand et al . 2005 ) .
these simulations start with an initial spectrum of density fluctuations extending down to the free - streaming mass of the dark matter candidate , which for a generic weakly interacting 100 gev particle ( such as a supersymmetric neutralino ) is roughly @xmath0 , where freeze - out occurs at an epoch @xmath1 .
as expected from scaling the results of simulations on larger scales , fluctuations in this initial distribution collapse and virialize to form structures with roughly 200 times the background density at a redshift of @xmath2 this first generation of microhalos survives to some degree as substructure in the larger halos that form subsequently in the simulations . extrapolating to the present - day , diemand et al .
2005 suggest that @xmath3 earth - mass clumps should survive in the halo of the milky - way , amounting to about 0.1% of its total mass .
if this were the case , the nearest earth - mass clump would be on the order of 0.1 pc distant from the earth and would have a typical size of 0.01 pc .
such nearby clumps might be observable as sources of gamma - rays , produced in dark matter annihilations .
they might also contribute to diffuse cosmic ray fluxes of positrons or anti - protons . in order to motivate experimental searches , however , it is important to determine whether the predicted small clumps of dark matter would actually survive the grainy tidal field due to stars while crossing the galactic bulge and disk @xmath4 times .
here we estimate the rate at which earth - mass clumps are disrupted in stellar encounters . we show that the nearest clumps in a galaxy like the milky way are likely to be tidally disrupted by repeated encounters over a hubble time . nevertheless , tidal debris from individual clumps may produce distinct microstreams that are potentially observable in dark matter detection experiments .
extended bodies generally have a complex response to external tidal heating ( gnedin & ostriker 1999 ) .
we can estimate the net effect of this process by applying the basic scaling from the semi - analytic mass - loss model of taylor & babul ( 2001 ) , which has been shown to match tidal heating and mass loss rates in high - resolution numerical simulations . according to this model , whenever a system on a general orbit spends time @xmath5 in a strong tidal field , that is a field in which the instantaneous tidal limit for the object @xmath6 ( calculated as in the circular - orbit case cf .
binney & tremaine 1987 ) is smaller than its size @xmath7 , it loses a fraction @xmath8 of the mass outside @xmath6 , where @xmath9 is the instantaneous orbital period . to estimate
when material at radius @xmath7 in a microhalo will experience strong shocks , we need to calculate the minimum impact parameter such that @xmath10 .
this is equivalent to the condition that the tidal force generated during an encounter with a star of mass @xmath11 at an impact parameter @xmath12 , @xmath13 , exceeds the restoration force for a small displacement @xmath14 , @xmath15 , where @xmath16 is the mean density of the microhalo interior to @xmath7 . for a typical microhalo in the simulations by diemand et al .
( 2005 ) , the density at the half - mass radius is @xmath17 , while the highest resolved density ( at 0.1 @xmath18 ) is @xmath19 times higher .
we do not expect a divergent ( phase space ) density in these smallest halos .
even if a cusp existed below the simulation resolution , it would contain less than 3% of the microhalo s mass , extrapolating from the profile given by diemand et al .
thus for a rapid encounter with a solar - mass star , impact parameters of @xmath20 or less will produce a strong shock over a range of radii containing 97% or more of the mass in the microhalo .
@xmath21 is the average density within 10% of the virial radius @xmath22 and can be thought of as a constant until the microhalo is virtually destroyed as shown in angus & zhao ( 2006 , hereafter az , figures 7c and 12b ) .
a microhalo moving at a relative speed of @xmath23 through a star field with a number density @xmath24 will encounter stars with an impact parameter @xmath25 at a rate of @xmath26 per million years .
each encounter shocks the microhalo for a time @xmath27 years , and liberates a fraction of the microhalo s total mass .
this fraction , from the scaling in taylor and babul ( 2001 ) , is roughly @xmath28 . here , @xmath29 myr is the period of a circular orbit of speed @xmath30 m/s around the solar - mass perturber at a distance of @xmath31 .
thus , the mass of a microhalo decreases exponentially , with the e - folding time given by @xmath32 , where we have expressed @xmath33 in terms of the parameters of the satellite . averaged over a hubble time , we find that a microhalo s mass decays exponentially , @xmath34 , where the e - folding time and the present mass are given by [ tau ] = 10 gyr ( 0.004 ^-3 ) ( _ mic 10 ^ -3 ) ^-1/2 , where @xmath35 is the mass density of galactic ( disk , bulge and halo ) stars averaged along the orbit of a microhalo over the past 10 gyrs ( at earlier times
, few stars might have formed ) . in general , @xmath35 is a function of the pericenter , apocenter and vertical height of the microhalo orbit .
e.g. , a microhalo on a thick disk orbit of height 1000 pc near the solar neighborhood would have made @xmath4 disk crossings and gone through star fields of an accumulated column density @xmath36 by today .
it goes through the galactic disk with a mean surface density of 46@xmath37pc@xmath38 or an average volume density @xmath39pc@xmath40 , thus this microhalo s mass is predicted to have decayed by about 6 e - foldings . another way to think of
the semi - analytical disruption rate is to note that the logarithm of the bound fraction @xmath41 , where @xmath42 is the stellar mass density encountered . to get an indication of how robust the semi - analytical massloss rate is , we have also run n - body simulations of a simplistic single encounter between a star and a microhalo for comparison .
we use the tree - code of vine and sigurdsson 1998 .
we use a total particle number of @xmath43 and softening of @xmath44pc consistent with interparticle separations of 3.5@xmath45pc .
the tree code tolerance parameter , @xmath46 is the cell size divided by particle separation and is set to 0.6 in these simulations .
initially the microhalo particles were distributed in a spherical nfw density profile with a @xmath47pc and a total mass of @xmath48 .
two sets of velocity space initial conditions were used ; tangentially anisotropic ( ta ) microhalos with randomly inclined circular orbits @xmath49 and radially anisotropic ( ra ) microhalos with @xmath50 , where @xmath51 and @xmath52 are the microhalo s internal density and circular velocity respectively ( cf .
angus and zhao 2006 ) .
two - body relaxation time , measured by the mean squared normalised energy drift after @xmath53 time steps using @xmath43 particles for an isolated microhalo , is about 3gyr .
so a microhalo resolved with @xmath43 particles should be free from numerical relaxation on time scales of 15myr , i.e. , the dynamical time scale of the microhalo . to measure the effect of encounters with stars ,
we shoot a single star of mass @xmath54 with a relative velocity @xmath55 towards a microhalo with a closest approach @xmath56 for penetrative encounters and @xmath57 for non - penetrative encounters .
such an impact is typical for a microhalo passing through a random star field with an equivalent star column density @xmath58 .
the duration of a star passage , say , along the x - axis from @xmath59 to @xmath60 is about @xmath61 dynamical time of the microhalo , hence microhalo particles are virtually stationary ( moving less than 0.01 percent of its orbital radius ) during a star passage , so the perturbations are indeed in the impulse regime .
it appears that our semi - analytical model captures the essential feature in numerical simulations of the impulse on @xmath48 microhalos if assuming @xmath62 ( cf .
however , there are microhalos of a few times earth mass , and these can be , e.g. , 5 times more massive and denser than the @xmath48 microhalos experimented here ( cf . fig.2 of diemand et al .
allowing for more robust microhalos and to be conservative in estimating the disruption rate of microhalos , we shall adopt the rate from the semi - analytical model assuming @xmath63 hereafter .
note that the scaling of our semi - analytical model eq .
[ tau ] also agrees with the literature on impulse - induced massloss in the context of diffuse star clusters , e.g. , moore ( 1996 ) , and eqs .
6 - 8 of wielen ( 1985 ) or eq . 8 of wasserman & salpeter ( 1994 ) . [ oldfig ]
to study the dependence of mass loss on orbital parameters , we have integrated orbits of microhalos in a flattened , axisymmetric galaxy potential with a nearly flat rotation curve : @xmath64 we launch 500 orbits from the solar neighborhood , @xmath65 , with a gaussian velocity distribution ( @xmath66 dispersion in each direction ) .
we read off the stellar density along the orbit using the besancon model of the galaxy ( robin et al .
2003 ) , and then take a time average of the star density .
the besancon model is a widely accepted detailed prescription of the number density of stars and stellar remnants of different ages and masses in the galactic spheroids ( bulge and halo ) and disks ( thin and thick ) .
a few examples of the microhalo orbits are given together with the star count model in fig .
[ angus1 ] .
these simulations suggest that the population of microhalos with apocenters less than 12 kpc has @xmath67 , and are hence past their half - life ( purple orbit in fig .
[ angus1 ] ) .
the more planar orbits , with @xmath68 and apocenter @xmath69 ( blue orbit in fig .
[ angus1 ] ) , have decayed by approximately 1 e - fold with @xmath70 .
complete destruction should happen to disky orbits penetrating into the bulge ; e.g. , the red orbit in fig . [ angus1 ] goes through a typical column density of @xmath71 with each disk crossing at a small angle , and has on average @xmath72 @xmath73 , corresponding to about 19 e - folds . in general
, we observe a strong correlation between the orbital shape and the disruption rate .
orbits which go through dense regions of the galaxy are likely disrupted due to tides ( cf .
[ oldfig ] ) .
also plotted in the inset of fig .
[ angus1 ] is the zone of destruction in the velocity space using 500 microhalos launched from the solar neighborhood .
there is a clear trend for halos with more planar ( smaller @xmath74 ) orbits , especially those co - rotating with the disk , and orbits with lower energy ( hence smaller apocenters ) to enter regions of higher density .
the sample averaged density is @xmath75 or @xmath76 for median or mean density , so microhalos on earth - crossing orbits are likely to have been severely stripped .
about 10% of the microhalos are in low - density regions ( below @xmath77 ) , and might survive more - or - less intact . the chance of survival , however , is likely even more severely reduced in the triaxial , barred and evolving potential of the milky way , where most orbits are stochastic box orbits which pass through the dense center of the milky way at some point over a hubble time .
additionally , it has been shown in az that disk shocking and tidal stripping may combine with stellar encounters to destroy microhalos even more efficiently than shown here . of course , these orbits are mearly a sample of all orbits passing the sun .
in fact , the fraction of halos disrupted would be much lower for high - energy orbits beyond the solar neighbourhood as the tides at larger pericentre are weaker along with the rapidly decreasing fraction of stars . what was also keenly noted by berezinsky et al .
( 2005 ) was that apocentric disk crossings beyond the extent of the thin disk are likely to boost the lifetime of microhalos .
the destruction rate is certainly much higher for low - energy orbits contained inside the bulge or those that have pericentres within the solar radius .
these michohalos would be exposed to stronger tides and higher stellar densities added to the fact they will cross the disk twice every orbit . however , these orbits are irrelavent for direct experiments on earth , and less important for indirect experiments because microhalos are barely identifiable from the background beyond about 0.1pc - 1pc distance from us . additionally , the calculations done by diemand et al .
( 2005 ) did not find self - consistent orbits and their comments on the current situation in the milky way are major extrapolations from dark - matter only simulations that stop at redshift 26 , and only ever simulate a much smaller total volume . since halo orbits in the mw would be strongly affected by the formation of the galaxy .
so what we have here is desruption as a function of orbital parameters .
we note that the long term fate of the cusp at the very center of each microhalo is unclear .
simulations have shown that systems with a universal density profile may lose all but 0.11% of their mass and still retain a bound central region hayashi et al .
such objects may indeed survive to the present - day in the solar neighborhood , but they will have little effect on dark matter detection experiments . the tidal debris stripped out of microhalos may actually be of greater interest , as discussed below .
nevertheless , az ( figures 7c and 12b ) showed that the inner density of a microhalo ( @xmath21 ) is largely unaffected until it is stripped down to the inner shell .
the existence of dark substructure in our galaxy s halo has important implications for the prospects of both direct and indirect dark matter detection ( green et al . 2005 ; berezinsky et al . 2003 ; pieri et al . 2005 ;
koushiappas et al . 2004 ) .
the presence of substructure affects direct detection rates in a straight forward way : if the solar system happens to be located inside of an overdense region of dark matter , then the rate observed will be enhanced proportionally to the density of the clump .
if , as is much more likely , our solar system is not inside a clump , then the rate will be modestly reduced by the fact that some fraction of the overall density is contained in substructures and thus does not contribute to the density of the smooth component .
zhao and silk 2005 estimated that the nearest minihalo of @xmath78 is 1 kpc away with a size of 50pc .
diemand et al .
2005 estimated that unstripped earth - mass halos have a filling factor of @xmath79 . hence in either case
it is _ unlikely _ for the sun to be inside a substructure .
tidal debris from encounters with stars changes the picture substantially . while the probability to find our solar system inside of a @xmath80-sized bound remnant of a microhalo is on the order of @xmath81 ,
the probability is of order unity for the solar system to be moving presently inside an ( unbound ) substructure .
disrupted streams are inhomogeneous regions with at least 10 times the volume filling factor of the original unstripped clump .
this is because at each disk crossing a burst of tidal debris is released due to encounters with disk stars , and the tails would lead / trail at least @xmath82pc ( assuming @xmath83 m/s speed of escape ) along the system s trajectory in 100 myrs , the time between disk crossings .
the time for the solar system to cross a stream at a general angle will be longer than 50 years , the time to cross a bound microhalo ( diemand et al . 2005 ) .
moving inside the inhomogeneous debris of the microhalos could cause transient enhancements of the event rate in direct detection experiments .
furthermore clumps on polar orbits survive longer than planar orbits ( cf .
[ angus1 ] ) , this makes the velocity distribution of the micro - streams non - gaussian , and could imprint an interesting annual modulation on the direct detection signal ( morgan et al .
2005 ; freese et al . 2005 ) .
fig.12 of angus & zhao ( 2006 ) shows the morphology of a microstream : even after the microhalo is fully disrupted , the microstream is still a distinct entity with a negligable tangential size compared to its length . as a result
, these filamentary streams may be the fate of a large proportion of the milky way s microhalos and the enhanced background from these structures may impose interesting features on direct detection experiments as well as having interesting annihilation flux properties .
the implications of dark microhalos and microstreams are rather different for the case of indirect detection .
indirect measurements sample the distribution of dark matter , through its annihilation rate , over large regions of the halo .
furthermore , dark matter annihilation rates , and thus indirect detection signals , are proportional not to the density of dark matter but to the dark matter density squared .
substructure within the galactic halo thus may be capable of boosting the dark matter annihilation rate and enhancing the prospects of detecting dark matter indirectly ( taylor & silk 2003 ) .
techniques employed for the indirect detection of dark matter include gamma - ray , anti - matter and neutrino detectors ( silk & srednicki 1984 ) .
gamma - rays annihilating in nearby and dense substructures , if present , could provide point sources potentially observable by gamma - ray telescopes ( pieri et al . 2005 ; koushiappas et al . 2004 ) .
anti - matter produced in dark matter annihilations are deflected by galactic magnetic fields and thus only the diffuse spectrum can be studied .
neutrinos are not as useful for identifying annihilations in the galactic halo , but instead are used to search for dark matter particles annihilating in the core of the sun .
the neutrino flux produced through dark matter matter annihilations in the sun is tied to the local density averaged over very long periods of time and therefore the presence of substructure is of little importance .
nearby microhalos , being tidally disrupted , are not particularly bright point sources of gamma - rays . while diemand et al .
2005 predict that the nearest intact microhalo will be much brighter than known dwarf spheroidals in gamma - rays , we find that the situation reverses for a microhalo left with only a one percent bound remnant . anti - matter fluxes produced through dark matter annihilations do not depend on the nature or location of individual dark substructures , but rather on the distribution of dark matter averaged over large volumes .
the value of the quantity @xmath84 , averaged over the contributing volume ( a few kiloparces for positrons or tens of kiloparsecs for anti - protons ) , determines the overall boost factor for the anti - matter fluxes generated through dark matter annihilations . with a distribution of substructure of the form @xmath85
, each decade of mass contributes almost equally to the boost factor , so it is important to determine over what range of masses substructures can survive .
previous works ( helmi et al . 2002 ; stoehr et al . 2003 ) emphasized the role of largers substructures , with the mass of draco dwarf spheroidal .
diemand et al .
2005 argue that the nearest earth - mass halo is factor of a few brighter than draco .
we argue , however , that the nearest earth - mass halo is likely to be tidally dispersed , and hence fainter than draco .
we have proposed ( zhao & silk 2005 ) that the nurseries of the first stars in the universe are dense minihalos of @xmath78 formed at @xmath86 .
unlike the microhalos such minihalos can survive the galactic tides until the present day .
the nearest minihalo is only 1 - 2 kpc from the sun , and is an order of magnitude brighter than the combined annihilation from the draco galaxy . in light of the excess reported by the heat collaboration ( barwick et al .
1997 ; coutu et al .
1999 ; baltz & edsjo 1999 ) , substructure is of particular importance for dark matter searches involving positrons . to produce such a signal with thermally generated neutralinos or other type of wimps , however , boost factors of @xmath8750 or higher are required ( baltz et al . 2005 ; hooper & kribs 2004 ; hooper & servant 2005 ) . although it has been shown that more massive substructures do not naturally provide such a large boost factor ( hooper et al .
2004 ) , effects due to baryons , or the presence of intermediate mass black holes could perhaps make them brighter ( prada et al . 2004 ) , although probably not enough to account for the heat excess .
after the effects of tidal disruption are considered , the same is found to be true for earth - mass microhalos .
we agree with previous studies ( berezinsky et al . 2003 ) which determine that positron boost factors larger than @xmath872 - 5 are unlikely .
given this conclusion , another explanation for the excess observed by heat appears to be required .
our understanding of these issues will be dramatically improved with data from the up - coming cosmic anti - matter experiments pamela and ams-02 ( profumo & ullio 2004 ) .
using a semi - analytic estimate of the tidal mass - loss rate , together with numerical integration of a fair sample of microhalo orbits , we find that most microhalos present in the solar neighborhood will have been heavily stripped by stellar encounters , producing microstreams of tidal debris . more generally , in environments with very low stellar densities such as the outer parts of the galactic disk or in a sextans - like dwarf galaxy , microhalos are only mildly heated .
but in high - density environments such as the galactic bulge or an m32-like elliptical ( @xmath88 to @xmath89 in inner 1kpc of m32 ( mateo 1998 ) ) microhalos are likely fully destroyed . in disk galaxies such as the milky way
, the fraction of microhalos which are disrupted depends strongly on the orbital inclination and pericentre .
microhalos on orbits coplanar with the disk are very quickly disrupted .
microhalos in the outer part of galaxy halos will encounter far fewer stars , however , and thus their annihilation rates will be less dramatically reduced .
a possible consequence of this might be to detect little gamma - ray emission from the central part of external galaxies , where stars have disrupted most of the dark substructure , and more flux from the outer regions .
a ring of gamma - rays surrounding a galaxy , if detected , would provide a strong confirmation of the existence of dark substructure .
the tidal effects discussed in this paper and in several preprints appeared during the marathon referee process of this paper ( berezinsky , dokuchaev , eroshenko 2005 , astro - ph/0511494 , green & goodwin 2006 , astro - ph/0604142 , goerdt , gnedin , moore , diemand , stadel 2006 , astro - ph/0608495 , angus & zhao 2006 , astro - ph/0608580 ) have important consequences for the direct and indirect detection of dark matter .
in particular , the probability of the earth intersecting a tidal stream at any given time is considerably larger than the probability of it being inside a microhalo s bound remnant , and thus tidal streams may potentially increase the reach of direct detection experiments .
indirect detection rates are not enhanced , however , as once stellar encounters are considered , earth mass halos are not a significant contributor to the overall dark matter annihilation rate .
hsz acknowledges support from pparc advanced fellowship and outstanding oversea young scholarship from chinese academy of science .
gwa acknowledges an overseas fieldwork grant from pparc and hospitality from beijing university .
jet acknowledges financial support from the nsf ( grant ast-0307859 ) and the doe ( contract de - fg02 - 04er41316 ) .
angus g.w . &
zhao h. 2006 , astro - ph/0608580 , mnras , submitted ( az ) baltz e.a .
, & edsjo j. , 1999 , phys .
d , 59 , 023511 baltz e.a .
, edsjo j. , freese k. , & gondolo p. , 2005
, preprint(astro - ph/0211239 ) ; barwick s.w .
_ et al . _
[ heat collaboration ] , 1997 , apj , 482 , l191 berezinsky v. , dokuchaev v. , & eroshenko y. , 2003 , phys .
d , 68 , 103003 berezinsky v. , dokuchaev v. , & eroshenko y. , 2006 , phys . rev .
d , accepted , ( astro - ph/0511494 ) binney j.j .
, & tremaine s. , 1987 galactic dynamics , priceton university press coutu s. , _ et al . _ , 1999 ,
app , 11 , 429 diemand j. , moore b. , & stadel j. , 2005 , nature , 433 , 389 freese k. , gondolo p. , & newberg h.j . , 2005 , phys . rev .
d , 71 , 043516 gnedin o.y .
, & ostriker j.p . , 1999 , apj , 513 , 626 goerdt t. & gnedin o. , moore b. , diemand j. , stadel j. 2006 , mnras , submitted , ( astro - ph/0608495 ) green a. & goodwin s. 2006 , mnras , submitted , ( astro - ph/0604142 ) green a.m. , hofmann s. , & schwarz d.j . , 2005 ,
jcap , 8 , 3 hayashi e. , navarro j.f . , taylor j.e . , stadel j. & quinn t.m . , 2003 ,
apj , 584 , 541 helmi a. , et al , 2002 , phys .
d , 66 , 3502 hooper d. , taylor j.e .
, & silk j. , 2004 , phys .
d , 69 , 103509 hooper d. , & kribs g.d . , 2004 , phys .
d , 70 , 115004 hooper d. , & silk j. , 2005 , phys .
d , 71 , 083503 hooper d. , & servant g. , 2005 , preprint(hep - ph/0502247 ) koushiappas s.m .
, zentner a.r . , & walker t.p . , 2004 ,
phys . rev .
d , 69 , 043501 mateo m. , 1998 , ann .
rev . of astron . and astrophys . , 36 , 435 moore b. , 1996 , apj , 461 , l13 morgan b. , green a.m. , & spooner n.j.c . , 2005 , phys .
d , 71 , 103507 ; pieri l. , branchini e. , & hofmann s. , 2005 , phys .
, 95 , 211301 ; prada f. et al . , 2004 ,
lett . , 93 , 1301 profumo s. , & ullio p. , 2004
, jcap , 0407 , 006 ; robin a.c .
, reyl c. , derrier s. , & picaud s. , 2003 , a & a , 409 , 523 silk j. , & srednicki m. , 1984 , phys . rev .
lett . , 53 , 624 stoehr f. , et al .
2003 , mnras , 345 , 1313 .
taylor j.e . , & babul a. , 2001 , apj , 559 , 716 taylor j.e . , & silk j. , 2003 , mnras , 339 , 505 wasserman i. , & salpeter e. , 1994 , apj , 433 , 670 wielen r. , 1985 dynamics of star clusters , ed .
by j. goodman and piet hut , iau , p449 | we point out that the usual self - similarity in cold dark matter models is broken by encounters with individual normal galactic stars on sub - pc scale .
tidal heating and stripping must have redefined the density and velocity structures of the population of the earth - mass dark matter halos , which are likely to have been the first bound structures to form in the universe .
the disruption rate depends strongly on _ galaxy types _ and the orbital distribution of the microhalos ; in the milky way , stochastic radial orbits are destroyed first by stars in the triaxial bulge , microhalos on non - planar retrograde orbits with large pericenters and/or apocenters survive the longest . the final microhalo distribution in the _ solar neighborhood _
is better described as a superposition of filamentry microstreams rather than as a set of discrete spherical clumps in an otherwise homogeneous medium .
this has important consequences to our detections of microhalos by direct recoil signal and indirect annihilation signal . |
before we come to identify the origin of the appearance of quantum structures in nature , we want to expose a macroscopic machine that produces quantum structure .
we shall make use intensively of the internal functioning of this machine to demonstrate our general explanation .
the machine that we consider consists of a physical entity @xmath1 that is a point particle @xmath2 that can move on the surface of a sphere , denoted @xmath3 , with center @xmath4 and radius @xmath5 .
the unit - vector @xmath6 where the particle is located on @xmath3 represents the state @xmath7 of the particle ( see fig .
for each point @xmath8 , we introduce the following experiment @xmath9 .
we consider the diametrically opposite point @xmath10 , and install a piece of elastic of length 2 , such that it is fixed with one of its end - points in @xmath11 and the other end - point in @xmath10 .
once the elastic is installed , the particle @xmath2 falls from its original place @xmath6 orthogonally onto the elastic , and sticks on it ( fig 1,b ) . 0.7 cm 2.2 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 9 pt fig . 1 : a representation of the quantum machine . in ( a ) the physical entity @xmath12
is in state @xmath13 in the point @xmath14 , and the elastic corresponding to the experiment @xmath15 is installed between the two diametrically opposed points @xmath16 and @xmath17 . in (
b ) the particle @xmath12 falls orthogonally onto the elastic and sticks to it . in ( c )
the elastic breaks and the particle @xmath12 is pulled towards the point @xmath16 , such that ( d ) it arrives at the point @xmath16 , and the experiment @xmath18 gets the outcome @xmath19 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ then the elastic breaks and the particle @xmath2 , attached to one of the two pieces of the elastic ( fig .
1,c ) , moves to one of the two end - points @xmath11 or @xmath10 ( fig . 1,d ) . depending on whether the particle @xmath2 arrives in @xmath11 ( as in fig .
1 ) or in @xmath10 , we give the outcome @xmath20 or @xmath21 to @xmath9 . in figure 2
we represent the experimental process connected to @xmath9 in the plane where it takes place , and we can easily calculate the probabilities corresponding to the two possible outcomes . in order to do so we remark that the particle @xmath2 arrives in @xmath11 when the elastic breaks in a point of the interval @xmath22 , and arrives in @xmath10 when it breaks in a point of the interval @xmath23 ( see fig .
we make the hypothesis that the elastic breaks uniformly , which means that the probability that the particle , being in state @xmath7 , arrives in @xmath11 , is given by the length of @xmath22 ( which is @xmath24 ) divided by the length of the total elastic ( which is 2 ) .
the probability that the particle in state @xmath7 arrives in @xmath10 is the length of @xmath23 ( which is @xmath25 ) divided by the length of the total elastic .
if we denote these probabilities respectively by @xmath26 and @xmath27 we have : @xmath28 the probabilities that we find in this way are exactly the quantum probabilities for the spin measurement of a spin 1/2 quantum entity , which means that we can describe this macroscopic machine by the ordinary quantum formalism with a two dimensional complex hilbert space as the carrier for the set of states of the entity . 0.7 cm 3.8 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 9 pt fig .
2 : a representation of the experimental process in the plane where it takes place . the elastic of length 2 , corresponding to the experiment @xmath18 , is installed between @xmath16 and @xmath17 .
the probability , @xmath29 , that the particle @xmath12 ends up in point @xmath16 is given by the length of the piece of elastic @xmath30 divided by the length of the total elastic .
the probability , @xmath31 , that the particle @xmath12 ends up in point @xmath32 is given by the length of the piece of elastic @xmath33 divided by the length of the total elastic .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
already from the advent of quantum mechanics it was known that the structure of quantum theory is very different from the structure of the earlier existing classical theories .
this structural difference has been expressed and studied in different mathematical categories , and we mention here some of the most important ones : ( 1 ) if one considers the collection of properties ( experimental propositions ) of a physical entity , then it has the structure of a boolean lattice for the case of a classical entity , while it is non - boolean for the case of a quantum entity ( birkhoff and von neumann 1936 , jauch 1968 , piron 1976 ) , ( 2 ) for the probability model , it can be shown that for a classical entity it is kolmogorovian , while for a quantum entity it is not ( foulis and randall 1972 , randall and foulis 1979 , 1983 , gudder 1988 , accardi 1982 , pitovski 1989 ) , ( 3 ) if the collection of observables is considered , a classical entity gives rise to a commutative algebra , while a quantum entity does nt ( segal 1947 , emch 1984 ) .
the presence of these deep structural differences between classical theories and quantum theory has contributed strongly to the already earlier existing belief that classical theories describe the ordinary understandable part of reality , while quantum theory confronts us with a part of reality ( the micro - world ) that is impossible to understand . therefore still now there is the strong paradigm that _ quantum mechanics can not be understood_. the example of our macroscopic machine with a quantum structure challenges this paradigm , because obviously the functioning of this machine can be understood .
the aim of this paper is to show that the main part of the quantum structures can indeed be explained in this way and the reason why they appear in nature can be identified . in this paper
we shall analyze this explanation , that we have named the hidden measurement approach , in the category of the probability models .
we refer to ( aerts and van bogaert 1992 , aerts , durt and van bogaert 1993 , aerts , durt , grib , van bogaert and zapatrin 1993 , aerts 1994 , aerts and durt 1994a ) for an analysis of this explanation in other categories .
the original development of probability theory was aiming at a formalization of the description of a probability that appears as the consequence of _ a lack of knowledge_. the probability structure appearing in situations of lack of knowledge was axiomatized by kolmogorov and such a probability model is now called kolmogorovian . since the quantum probability model is not kolmogorovian ,
it has now generally been accepted that the quantum probabilities are _ not _ a description of a _ lack of knowledge_. sometimes this conclusion is formulated by stating that the quantum probabilities are _ ontological _ probabilities , as if they would be present in reality itself . in the hidden measurement approach we show that the quantum probabilities can be explained as being due to a _ lack of knowledge _ , and we prove that what distinguishes quantum probabilities from classical kolmogorovian probabilities is the _ nature of this lack of knowledge_. let us go back to the quantum machine to illustrate what we mean . if we consider again our quantum machine ( fig . 1 and fig .
2 ) , and look for the origin of the probabilities as they appear in this example , we can remark that the probability is entirely due to a _ lack of knowledge _ about the measurement process .
namely the lack of knowledge of where exactly the elastic breaks during a measurement .
more specifically , we can identify two main aspects of the experiment @xmath9 as it appears in the quantum machine . 1 .
the experiment @xmath9 effects a real change on the state @xmath7 of the entity @xmath1 .
indeed , the state @xmath7 changes into one of the states @xmath34 or @xmath35 by the experiment @xmath9 .
2 . the probabilities appearing are due to a _ lack of knowledge _ about a deeper reality of the individual measurement process itself , namely where the elastic breaks . these two effects give rise to quantum - like structures .
the lack of knowledge about a deeper reality of the individual measurement process we have referred to as the presence of hidden measurements that operate deterministically in this deeper reality ( aerts 1986 , 1987 , 1991 ) , and that is the origin of the name that we gave to this approach .
a consequence of this explanation is that quantum structures shall turn out to be present in many other regions of reality where the two mentioned effects appear .
we think of the many situations in the human sciences , where generally the measurement disturbs profoundly the entity under study , and where almost always a lack of knowledge about the deeper reality of what is going on during this measurement process exists .
in the final part of this paper we give some examples of quantum structures appearing in such situations .
if the quantum structure can be explained by the presence of a lack of knowledge on the measurement process , we can go a step further , and wonder what types of structure arise when we consider the original models , with a lack of knowledge on the measurement process , and introduce a variation of the magnitude of this lack of knowledge .
we have studied the quantum machine under varying lack of knowledge , parameterizing this variation by a number @xmath36 $ ] , such that @xmath37 corresponds to the situation of maximal lack of knowledge , giving rise to a quantum structure , and @xmath38 corresponds to the situation of zero lack of knowledge , generating a classical structure , and other values of @xmath0 correspond to intermediate situations , giving rise to a structure that is neither quantum nor classical ( aerts , durt and van bogaert 1992 , 1993 ) .
it is this model that we have called the @xmath0-model , and we want to introduce it again in this paper .
we start from the quantum machine , but introduce now different types of elastic .
an @xmath39-elastic consists of three different parts : one lower part where it is unbreakable , a middle part where it breaks uniformly , and an upper part where it is again unbreakable . by means of the two parameters
@xmath36 $ ] and @xmath40 $ ] , we fix the sizes of the three parts in the following way .
suppose that we have installed the @xmath39-elastic between the points @xmath10 and @xmath11 of the sphere .
then the elastic is unbreakable in the lower part from @xmath41 to @xmath42 , it breaks uniformly in the part from @xmath43 to @xmath44 , and it is again unbreakable in the upper part from @xmath45 to @xmath11 ( see fig .
3 ) . 0.7 cm 3.8 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 9 pt fig .
3 : a representation of the experiment @xmath46 .
the elastic breaks uniformly between the points @xmath47 and @xmath48 , and is unbreakable in other points . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ an @xmath9 experiment performed by means of an @xmath49-elastic shall be denoted by @xmath50 . we have the following cases : _ ( 1 )
_ @xmath51 .
the particle sticks to the lower part of the @xmath39-elastic , and any breaking of the elastic pulls it down to the point @xmath10 .
we have @xmath52 and @xmath53 . _ ( 2 ) _ @xmath54 .
the particle falls onto the breakable part of the @xmath39-elastic .
we can easily calculate the transition probabilities and find : @xmath55 _ ( 3 ) _
the particle falls onto the upper part of the @xmath57-elastic , and any breaking of the elastic pulls it upwards , such that it arrives in @xmath11 .
we have @xmath58 and @xmath59 .
if we want to analyze the structure of the quantum probability model in the light of axioms that have been formulated for classical probability theory , we first have to be very specific about the situation that we consider . in physics
( and hence also in quantum mechanics and classical mechanics ) we consider a situation where we have a physical entity @xmath1 that can be in different states @xmath60 , and we ll denote the set of states by @xmath61 . on this physical entity @xmath1 , in a certain state @xmath62 , we perform experiments @xmath63 , that respectively have sets of possible outcomes @xmath64 , @xmath65 , @xmath66 , .... let us denote the collection of all relevant experiments by @xmath67 .
there are different places in which probability appears in this scheme . in many occasions
it is not possible to prepare the entity @xmath1 in such a way that we know in which state it is before we start an experiment .
we can only prepare it such that we are left with a situation of lack of knowledge about the state of the entity .
this situation of lack of knowledge is described by means of a probability measure @xmath68 $ ] on the set of states , such that @xmath69 is a @xmath70-algebra of measurable subsets of @xmath61 , and for @xmath71 we have that @xmath72 is the probability that the state of the entity @xmath1 is in the subset @xmath73 .
we have ( a ) @xmath74 , and ( b ) @xmath75 for sets @xmath76 such that @xmath77 for @xmath78 . what we call states @xmath79 are often called pure states , and what we call situations of lack of knowledge on the states @xmath80 are often called mixed states. even when the entity @xmath1 is in a state @xmath62 , and an experiment @xmath81 is performed , probability , defined as the limit of the relative frequency connected to an outcome @xmath82 , appears . for a fixed state @xmath62 ,
the probability that an experiment @xmath81 gives an outcome in a subset @xmath83 , denoted by @xmath84 , can be described as a probability measure on the outcome set @xmath64 of the experiment @xmath81 .
hence a map @xmath85 $ ] such that @xmath86 is a collection of measurable subsets of @xmath64 and ( a ) @xmath87 , and ( b ) @xmath88 for sets @xmath89 such that @xmath90 for @xmath91 .
most of the time we measure a probability in the laboratory that contains both just mentioned probabilities .
it is the probability that in a situation @xmath92 of lack of knowledge on the states , an experiment @xmath81 gives an outcome in a subset @xmath83 , and we ll denote it by @xmath93 .
this probability is , for a given experiment @xmath81 , a map @xmath94 $ ] , where @xmath95 is the set of probability measures on @xmath61 .
as we have done in ( aerts 1994 ) we introduce for an experiment @xmath81 the eigenstate sets , as maps @xmath96 , where for @xmath83 we have : @xmath97 we also introduce the possibility - state sets , as maps @xmath98 , where for @xmath99 we have : @xmath100 clearly we always have : @xmath101 [ theorem1 ] we consider an entity @xmath1 in a situation with lack of knowledge about the states described by the probability measure @xmath92 on the state space .
for an arbitrary experiment @xmath81 and set of outcomes @xmath102 we have : @xmath103
_ proof : _ as defined we have that @xmath104 is the probability that the state of the entity is in the subset @xmath105 .
if the state is in @xmath105 , the experiment @xmath81 gives with certainty an outcome in @xmath106 , and therefore @xmath107 .
as defined we have that @xmath108 is the probability that the state of the entity is in @xmath109 .
if the state is in @xmath109 the experiment @xmath81 has a possible outcome in @xmath106 , and therefore @xmath110 .
we shall show in the next section that for a classical probability model , the two inequalities become equalities .
but first we want to illustrate all these concepts on the quantum machine .
it is easy to see how these concepts are defined for the quantum machine ( and also for the @xmath0-model ) . for a considered experiment @xmath9 ( or @xmath50 in the @xmath0-model )
, we have an outcome set @xmath111 .
the set of states is @xmath112 .
the situations with lack of knowledge about the states are described by probability measures on the surface of the sphere .
we have also described @xmath113 for an arbitrary @xmath114 ( see ( [ equation3 ] ) and ( [ equation4 ] ) ) . for the eigenstate sets and the possibility- state sets we have ( see fig .
4 . ) : @xmath115 0.7 cm 2.2 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 9 pt fig . 4 :
we have represented the eigenstate sets @xmath116 and @xmath117 .
if the state of the entity ( the position of the particle @xmath12 ) is in @xmath116 [ or in @xmath118 , then the experiment @xmath50 gives with certainty the outcome @xmath20 [ or with certainty the outcome @xmath21 ] .
we also have represented the possibility state sets , @xmath120 [ @xmath121 , which is the collection of states where the entity gives a possible outcome @xmath20 [ @xmath21 ] . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we can consider the following specific cases : for @xmath0 = 1 we always have @xmath122 = 0 , and the @xmath0-model reduces to the original quantum machine that we introduced in section 1 .
it is a model for the spin of a spin 1/2 quantum entity .
the transition probabilities are the same as the ones related to the outcomes of a stern - gerlach spin experiment on a spin 1/2 quantum particle , of which the quantum - spin - state in direction @xmath123 @xmath124 , denoted by @xmath125 , and the experiment @xmath9 corresponding to the spin experiment in direction @xmath126 @xmath127 , is described respectively by the vector and the self adjoint operator @xmath128 of a two - dimensional complex hilbert space . for the eigenstate sets and possibility - state sets we find : @xmath129 suppose that we consider a situation with lack of knowledge about the state , described by a uniform probability distribution @xmath92 on the sphere , which corresponds to a random distribution of the point on the sphere
. then we can easily calculate the following probabilities : @xmath130 on the other hand , we have : @xmath131 which shows that the inequalities of theorem 1 ( see ( 8) ) are very strong in this quantum case .
the classical situation is the situation without fluctuations .
if @xmath0 = 0 , then @xmath122 can take any value in the interval @xmath133 $ ] , and we have : @xmath134 we again consider the situation of an at random distribution of the point particle on the sphere described by the probability distribution @xmath92 .
we then have in this case : @xmath135 on the other hand we have : @xmath136 which shows that the inequalities of theorem [ theorem1 ] ( see ( [ equation1 ] ) ) have become equalities in this case .
to give a clear picture of the general situation , we introduce additional concepts .
first we remark that the regions of eigenstates @xmath137 and @xmath138 , and the regions of possibility states @xmath139 and @xmath140 , are determined by the points of spherical sectors of @xmath3 centered around the points @xmath11 and @xmath10 ( see fig . 4 and fig . 5 ) .
0.7 cm 3.2 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 9 pt fig . 5 : we have represented the different angles @xmath141 and @xmath142 characterizing the spherical sectors of the eigenstate sets and possibility state sets . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we denote a closed spherical sector centered around the point @xmath8 with angle @xmath143 by @xmath144 , and a open spherical sector with the same angle by @xmath145 .
we call @xmath146 the angle of the spherical sectors corresponding to @xmath137 for all @xmath11 , hence for @xmath147 we have @xmath148 , and @xmath149 for @xmath132 ( see fig . 5 ) .
we can verify easily that @xmath138 is determined by a spherical sector centered around the point @xmath10 .
we call @xmath150 the angle of this spherical sector , hence for @xmath151 we have @xmath152 and @xmath153 for @xmath132 . for @xmath151
we have @xmath154 , and @xmath155 for @xmath132 .
for @xmath151 we have @xmath156 and @xmath157 for @xmath132 .
we have : @xmath158
we want to formulate now the classical situation in this general scheme .
we shall see that in the characterization of a classical probability model , the inequalities of theorem 1 play an important role .
[ definition1 ] suppose that we are in the situation @xmath92 of lack of knowledge on the states .
we say that an experiment @xmath81 is a classical experiment iff @xmath159 for all subsets @xmath83 . with other words ,
a classical experiment is an experiment where all states , except a collection of measure zero , give rise to predetermined outcomes for this experiment .
classical experiments are experiments with predetermined outcomes ( except for a set of states of measure zero that as a consequence do not contribute to the statistics ) . for these classical experiments
we can show that probability always originates in a lack of knowledge on the states .
if we are in the situation @xmath92 of lack of knowledge on the states , and @xmath81 is a classical experiment , then for each @xmath160 we have : @xmath161 _ proof : _ a direct consequence of definition [ definition1 ] and theorem [ theorem1 ] .
we shall show now that for classical experiments bayes formula for the conditional probability is valid . to be able to analyze the validity of bayes formula in the scheme that we have presented here
, we must give an operational definition for the concept of conditional probability .
here we are confronted with a conceptual problem , since in most textbooks , the conditional probability is defined by means of bayes formula .
since the conditional probability is a primary physical quantity that is measured in the laboratory , we should define it operationally and without the use of bayes formula .
we want to make clear that there is a distinction between the occurrence of an outcome when an experiment is performed , and the conditioning on an outcome corresponding to an experiment .
given a situation @xmath92 of lack of knowledge on the states of an entity @xmath1 , described by a probability measure on this set of states @xmath61 .
we condition the entity @xmath1 on a subset @xmath162 for an experiment @xmath163 , if we consider during the performance of the experiment @xmath81 only those trials , where the situation of the entity before the experiment @xmath81 is such that we can predict the outcome for the experiment @xmath163 to occur with certainty in @xmath164 , if we would decide to perform the experiment @xmath163 . from this definition
follows that conditioning is equivalent to a change of the situation @xmath92 before the experiment in such a way that the experiment @xmath163 would give with certainty an outcome in @xmath164 , if it would be executed .
the new situation of lack of knowledge is described by the probability measure that we shall denote by @xmath165 $ ] .
it is defined for an arbitrary subset @xmath166 as follows : @xmath167 now that we have introduced this concept of ` conditioning ' on an experiment , we can introduce the general concept of conditional probability .
given a situation @xmath92 of lack of knowledge on the states of an entity , described by the probability measure @xmath92 , and given two experiments @xmath81 and @xmath163 , then we want to consider the conditional probability @xmath168 .
this is the probability that the experiment @xmath81 makes occur an outcome in the set @xmath169 , when the situation is conditioned on the set @xmath164 for the experiment @xmath163 .
the conditional probability is a map @xmath170 $ ] .
given a situation @xmath92 of lack of knowledge on the states , and two experiments @xmath81 and @xmath163 .
if the experiment @xmath81 is a classical experiment , then the conditional probability @xmath171 satisfies bayes formula .
more specifically we have @xmath172 . _
proof : _ we have ( see [ equationconditional ] ) that @xmath173 . if @xmath81 is a classical experiment it follows from theorem 2 that @xmath174 .
this shows that bayes formula is valid . from this theorem
can intuitively be seen that bayes formula for the conditional probability shall not be valid for the case of two experiments that are both non - classical .
we shall now analyze all these situations in the @xmath0-model and show how the conditional probability in the @xmath0-model evolves continuously from the quantum transition probability , for the case of @xmath37 , to a classical kolmogorovian probability satisfying bayes formula , for the case @xmath132 .
we shall also show that for values of @xmath0 strictly between @xmath5 and @xmath175 , the conditional probability is neither quantum nor classical .
given a situation @xmath92 of lack of knowledge about the state of the point particle , described by a uniform probability measure on the sphere .
this corresponds to the situation where the particle @xmath2 is distributed at random on the sphere . for a fixed @xmath0 , and two parameters @xmath122 and @xmath176 both in the interval @xmath177 $ ] ,
there are also given the two experiments @xmath50 and @xmath178 .
0.7 cm 3.2 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 9 pt fig . 6 : an illustration of the situation corresponding to the conditional probability for the @xmath179- model .
the lack of knowledge about the state of the particle is described by a uniform probability distribution @xmath180 on the sphere . for a fixed @xmath179 and two parameters @xmath181 and @xmath182 in the the interval @xmath183
$ ] , we consider two experiments @xmath184 and @xmath185 .
we want to calculate the conditional probability @xmath186 that the experiment @xmath187 gives the outcome @xmath188 , when the entity is conditioned on the outcome @xmath189 for the experiment @xmath185 .
this means that before the start of the experiment @xmath184 , the situation is such that if we would perform the experiment @xmath185 , the outcome @xmath189 would come out with certainty .
this conditioning is expressed by a new probability measure on the sphere , which is zero outside the grey area , and equal to the old one , except for a renormalization factor , in the grey area . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in general we consider the conditional probability for arbitrary elements of the set of measurable subsets of the outcome sets of the two experiments .
since in the @xmath0-model we only have experiments @xmath50 , with two outcomes @xmath20 and @xmath21 , we want to alleviate somewhat the notation . therefore we shall denote the conditional probability that the experiment @xmath50 gives the outcome @xmath20 ( respectively @xmath21 ) , when the entity is conditioned for the outcome @xmath190 of the experiment @xmath178 by @xmath191 ( respectively @xmath192 ) . in a similar way we denote the conditional probability that the experiment @xmath50 gives the outcome @xmath20 ( respectively @xmath21 ) , when the entity is conditioned for the outcome @xmath193 of the experiment @xmath178 by @xmath194 ( respectively @xmath195 ) ( see fig .
we repeat again : the conditional probability @xmath191 is the probability that the experiment @xmath50 gives the outcome @xmath20 , if the entity is conditioned on the outcome @xmath190 for the experiment @xmath178 .
this means that the lack of knowledge on the states is such that if we would decide to perform the experiment @xmath178 , the outcome @xmath190 would come out with certainty . with other words ( see fig .
6 ) , the state of the entity is such that the particle is distributed uniformly inside the spherical sector @xmath196 , the grey area on figure 6 .
it is easy to formulate in a similar way the four other conditional probabilities @xmath197 and @xmath195 . the explicit calculation of these conditional probabilities is a long exercise of classical calculus , and therefore we refer to ( aerts d. and aerts s. 1994 b ) for a detailed exposition of this calculation .
here we only give the result .
let us call @xmath198 the angle between the two vectors @xmath11 and @xmath199 , then we have : @xmath200 where @xmath201 is the function of heaviside , and : @xmath202 @xmath203 @xmath204 where : @xmath205 and : @xmath206 let us first consider the two extremal cases , the classical case , where @xmath207 , and the quantum case , where @xmath208 . _
( 1 ) the classical case ( @xmath132 ) _ in this case we find : @xmath209 which is a linear function in the angle . _
( 2 ) the quantum case ( @xmath37 ) _ for this case
we only have to take into account the contribution @xmath210 of ( [ equationp ] ) , and hence we find : @xmath211 which is the well known quantum transition probability between the states @xmath34 and @xmath212 .
the conditional probability @xmath191 evolves continuously from the quantum transition probability between the states @xmath34 and @xmath212 to a linear function of the angle between the two vectors @xmath11 and @xmath199 .
a complete probabilistic analysis of the intermediate situation of the @xmath0-model is presented in ( aerts d. and aerts s. 1994 b ) . here
we only show that for a specific value of @xmath213 the conditional probabilities of the @xmath0-model can not be fitted neither into a quantum probability model nor into a classical probability model . for @xmath214
, the conditional probabilities of the @xmath0- model can not be fitted neither into a classical nor into a quantum probability model . _
proof : _ we shall give a proof ex absurdum , and suppose that there does exist a kolmogorovian model satisfying bayes formula for the conditional probability for this value of @xmath0 .
we consider three experiments @xmath50 , @xmath215 and @xmath216 , such that @xmath217 and @xmath11 , @xmath6 and @xmath199 are in the same plane , with an angle of @xmath218 between @xmath11 and @xmath6 , between @xmath6 and @xmath199 , and between @xmath199 and @xmath11 .
if we use the general expression for the conditional probability ( [ equationp ] ) we find @xmath219 , @xmath220 and @xmath221 . as we have defined , @xmath222 is the probability that the experiment @xmath215 gives the outcome @xmath223 , if we have conditioned the entity in such a way that if we would perform the experiment @xmath216 , it would give with certainty an outcome @xmath190 . to be able to express more clearly the hypothesis of the existence of a kolmogorovian probability model , we write these conditional probabilities in a more standard notation .
hence @xmath224 , @xmath225 , and @xmath226 under the preparation @xmath92 .
if there does exist a kolmogorovian probability model , satisfying the bayes formula , there exists a set @xmath227 ( playing the role of the sample space ) , and a @xmath70- algebra @xmath228 ( playing the role of the set of events ) , and a probability measure @xmath229 $ ] , such that the conditional probabilities @xmath230 , @xmath231 and @xmath232 can be written under the appropriate form .
this means that there exists elements @xmath233 such that : @xmath234 we also have @xmath235 and @xmath236 .
using the fact that @xmath237 is a probability measure , and ( [ equation30 ] ) and ( [ equation31 ] ) , and substituting the values of the conditional probabilities , we get : @xmath238 if we subtract ( [ equation33 ] ) from ( [ equation32 ] ) we find : @xmath239 which implies that @xmath240 on the other hand from ( [ equation34 ] ) follows @xmath241 the inequalities ( [ equation42 ] ) and ( [ equation43 ] ) deliver us the contradiction that we were looking for .
the conclusion is that for these values of the conditional probabilities there does not exists a kolmogorovian probability model satisfying the bayes formula .
also for the proof of the non - existence of a hilbertian model we proceed ex absurdum .
if there exists a two dimensional complex hilbert space model , where the conditional probabilities are described by the transition probabilities , we can find three orthonormal basis @xmath242 , @xmath243 , @xmath244 such that : @xmath245 this means that there exists five angles @xmath246 and @xmath247 such that : @xmath248 if @xmath242 is an orthonormal basis , we have @xmath249 , and hence @xmath250 and also the complex conjugate @xmath251 if we multiply these last equations term by term we find : @xmath252 this we can write as @xmath253 but then we must have @xmath254 if we fill in the values @xmath255 and @xmath256 we find : @xmath257 from this contradiction we can conclude that there does not exist a two dimensional hilbert space model , such that the conditional probabilities can be described by transition probabilities in this hilbert space .
this theorem shows that we really have identified a new region of probabilistic structure in this intermediate domain .
in ( aerts d. and aerts s. 1994 b ) we show that for any value of @xmath0 different from @xmath175 , the probability structure of the @xmath0-model is non kolmogorovian ( not satisfying bayes formula for the conditional probabilities ) .
we also show that there is a domain of @xmath258 where a hilbert space model can be found , but another domain where this is not the case .
we come now to the last section of this paper , where we like to give a sketch of how these non - kolmogorovian probabilities appear in other regions of nature .
as follows from the foregoing analysis , non - classical experiments , giving rise to a non - classical structure , are characterized by the presence of non - predetermined outcomes .
this makes it rather easy to recognize the non - classical aspects of experiments in other regions of reality .
let us consider the situation of a decision process developing in the mind of a human being , and we refer to ( aerts d. and aerts .
s. 1994 a , b ) for a more detailed description .
hence our entity is a person , its states being the possible states of this person .
experiments are questions that can be asked to the person , and on which she or he has to respond with yes or no. the typical situation of an opinion poll can be thought of as a concrete example .
let us consider three different questions : 1 .
@xmath11 : are you in favor of the use of nuclear energy ? 2 .
@xmath6 : do you think it would be a good idea to legalize soft - drugs ? 3 .
@xmath199 : do you think capitalism is better than social - democracy ?
we have chosen such type of questions that many persons shall not have predetermined opinions about them .
since the person has to respond with yes or no , she or he , without opinion before the questioning , shall form her or his opinion during this process of questioning .
we can use the @xmath0- model to represent this situation
. 0.7 cm 0.7 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 9 pt fig . 7 : a representation of the question @xmath259 by means of the @xmath179-model .
we have indicated the three regions corresponding to predetermined answer yes , without predetermined answer , and predetermined answer no. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ to simplify the situation , but without touching the essence , we make the following assumptions about the probabilities that are involved . we suppose that in all cases 50% of the persons have answered the question @xmath11 with yes , but only 15% of the persons had a predetermined opinion .
this means that 70% of the persons formed their answer during the process of questioning . for simplicity
we make the same assumptions for @xmath6 and @xmath199 .
we can represent this situation in the @xmath0-model as shown in figure 7 .
we also make some assumptions of the way in which the different opinions related to the three questions influence each other .
we can represent an example of a possible interaction by means of the @xmath0-model ( figure 9 ) .
one can see how a person can be a strong proponent for the use of nuclear energy , while having no predetermined opinion about the legalization of soft drugs ( area 1 in figure 9 ) .
area ( 4 ) corresponds to a sample of persons that have predetermined opinion in favor of legalization of soft drugs and in favor of capitalism . for area ( 10 )
we have persons that have predetermined opinion against the legalization of soft drugs and against capitalism .
all the 13 areas of figure 9 can be described in such a simple way .
0.7 cm 3 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 9 pt fig . 8 : a representation of the three questions @xmath259 , @xmath260 , and @xmath261 by means of the @xmath179-model .
we have numbered the 13 different regions .
for example : ( 1 ) corresponds to a sample of persons that have predetermined opinion in favor of nuclear energy , but have no predetermined opinion for both other questions , ( 4 ) corresponds to a sample of persons that have predetermined opinion in favor of legalization of soft drugs and in favor of capitalism , ( 10 ) corresponds to a sample of persons that have predetermined opinion against the legalization of soft drugs and against capitalism , ( 13 ) corresponds to the sample of persons that have no predetermined opinion about non of the three questions , etc ... _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ deliberately we have chosen the different fractions of people in such a way that the conditional probabilities fit into the @xmath0-model for a value of @xmath214 .
this means that we can apply theorem 4 , and conclude that the collection of conditional probabilities corresponding to these questions @xmath11 , @xmath6 and @xmath199 can neither be fitted into a kolmogorovian probability model nor into a quantum probability model .
we are developing now in brussels a statistics for such new situations , that we have called interactive statistics. by means of this statistics it should be possible to make models for situations where part of the properties to be tested are created during the process of testing .
the further development of an intermediate ( between classical and quantum ) probability theory and an interactive statistics could be very fruitful as well for physics as for other sciences .
we work now at the construction of a general theory , where also the intermediate structures can be incoorporated ( aerts 1986 , aerts 1987 , aerts , durt and van bogaert 1993 , aerts 1994 , aerts and durt 1994a , 1994b , coecke 1994a , 1994b , 1994c ) .
this theory can probably be used to describe the region of reality between microscopic and macroscopic , often referred to as the mesoscopic region .
actually , physicists use a very complicated heuristic mixture of quantum and classical theories to construct models for mesoscopic entities .
there is however no consistent theory , and a general intermediate theory could perhaps fill this gap .
we try to find examples of simple physical phenomena in the mesoscopic region that could eventually be modeled by an @xmath0-like model ( aerts and durt 1994 b ) .
if we succeed in building this intermediate theory , we would not only have a new theory for the mesoscopic region , but the existence of such a theory would also shed light on old problems of quantum mechanics ( the quantum classical relation , the classical limit , the measurement problem , etc ... ) . in till now we have only been developing the kinematics of this intermediate theory ( aerts 1994 and aerts and durt 1994 a ) , but , once the kinematics is fully developed , the way to construct a dynamics for the intermediate region is straightforward
. we can study the imprimitivity system related to the galilei group and look for representations of this galilei group in the group of automorphisms of the kinematical structure of the intermediate theory .
if we can derive an evolution equation in this way , it should continuously transform with varying @xmath0 from the schrdinger equation ( @xmath37 ) , to the hamilton equations ( @xmath207 ) .
as we have shown in the last section of this paper , the development of an interactive statistics could be of great importance for the human sciences , where often non - predetermined outcome situations appear .
it could lead to a new methodology for these sciences .
actually one is aware of the problem of the interaction between subject and object , but it is generally thought that this problems can not be taken into account in the theory
. we also want to remark that the hidden measurement approach defines a new quantization procedure . starting from a classical mechanical entity , and adding lack of knowledge ( or fluctuations) on the measurement process
, out of the classical entity appears a quantum - like entity .
another problem that we are investigating at this moment is an attempt to describe quantum chaos by means of this new quantization procedure .
it can be shown that the sensitive dependence on the initial conditions that can be found in the @xmath0-model for the classical situation in the set of unstable equilibrium states disappears when the fluctuations on the measurement process increase .
this could be the explanation for the absence of quantum chaos .
accardi , l. ( 1982 ) , _ on the statistical meaning of the complex numbers in quantum mechanics , _
nuovo cimento , * 34 * , 161 .
aerts , d. ( 1987 ) , _ the origin of the non- classical character of the quantum probability model , _ in _ information , complexity , and control in quantum physics _ , a. blanquiere , et al .
, eds . , springer - verlag .
aerts , d. ( 1991 ) , _ a macroscopic classical laboratory situation with only macroscopic classical entities giving rise to a quantum mechanical probability model , _ in _ quantum probability and related topics , vol .
vi , _ l. accardi , ed . ,
world scientific , singapore .
aerts , d. , durt , t. and van bogaert , b. ( 1992 ) , _ a physical example of quantum fuzzy sets , and the classical limit , _ in _ proceedings of the first international conference on fuzzy sets and their applications , tatra montains math .
publ . _ * 1 * , 5
. aerts , d. , durt , t. and van bogaert , b. ( 1993 ) , _ quantum probability , the classical limit and non - locality , _ in _ symposium on the foundations of modern physics _ , t. hyvonen , ed . , world scientific , singapore .
aerts , d and durt , t. ( 1994 b ) , _ quantum , classical and intermediate ; a measurement model , _ , in the proceedings of the _ symposium on the foundations of modern physics , _ helsinki , 1994 , ed .
laurikainen , c. montonen , k. sunnar borg , editions frontieres , gives sur yvettes , france . | we explain the quantum structure as due to the presence of two effects , ( a ) a real change of state of the entity under influence of the measurement and , ( b ) a lack of knowledge about a deeper deterministic reality of the measurement process .
we present a quantum machine , where we can illustrate in a simple way how the quantum structure arises as a consequence of the two mentioned effects .
we introduce a parameter @xmath0 that measures the size of the lack of knowledge on the measurement process , and by varying this parameter , we describe a continuous evolution from a quantum structure ( maximal lack of knowledge ) to a classical structure ( zero lack of knowledge ) .
we show that for intermediate values of @xmath0 we find a new type of structure , that is neither quantum nor classical .
we apply the model that we have introduced to situations of lack of knowledge about the measurement process appearing in other regions of reality .
more specifically we investigate the quantum - like structures that appear in the situation of psychological decision processes , where the subject is influenced during the testing , and forms some of his opinions during the testing process .
our conclusion is that in the light of this explanation , the quantum probabilities are epistemic and not ontological , which means that quantum mechanics is compatible with a determinism of the whole .
fund and clea , brussels free university , krijgskundestraat 33 , 1160 brussels , belgium , e - mail : [email protected] |
over the last years , significant progress has been made in the modelization of strongly correlated materials .
such systems typically contain partially filled @xmath2 or @xmath3 orbitals@xcite , which lie relatively close to the nuclei .
electronic coulomb interactions can then induce substantial corrections to a bloch one - particle picture , ranging from renormalizations of effective parameters in the sense of landau to full localization of the @xmath2- or @xmath3-degrees of freedom in the mott insulator .
one of the difficulties in describing such correlation effects is to separate the usually rather small energy range of the correlated @xmath2 and @xmath3 orbitals from the larger energy scale of the itinerant degrees of freedom , e.g. the @xmath4-orbitals of ligand atoms , but also of higher or lower lying states of the transition metal , rare earth or actinide atoms themselves .
the former orbitals are mainly responsible for the low - energy physical properties of the compounds , while the latter act as a screening medium , setting in particular the actual value of the coulomb repulsion of the correlated degrees of freedom . as discussed in ref .
, screening is a dynamic process which leads in general to a frequency dependent @xmath5 . in a realistic approach to strongly correlated materials
, @xmath1 can be determined at the random phase approximation ( rpa ) level@xcite , once the bands and their eigenstates are computed by an ab - initio calculation .
the static value @xmath6 , evaluated in this way@xcite , has been recently used in the dynamical mean field theory ( dmft ) calculations of materials@xcite , and the effect of the frequency dependent screening has been either neglected or empirically taken into account by adjusting the effective static @xmath7 .
the dmft approach , combined with the density functional theory ( dft ) , is an extremely powerful tool to treat ab - initio strongly correlated systems , once the low - energy model is determined@xcite .
however , very little is known on the impact of the frequency dependence of the interaction in the low - energy part of the spectrum . the hardest obstacle in order to include the dynamic @xmath1 into the dmft framework has been the lack of a reliable solver for the quantum impurity problem with a frequency dependent hubbard interaction .
the @xmath8 computed so far for various materials@xcite shows some common features .
the unscreened @xmath1 ( @xmath9 ) is up to an order of magnitude larger than the screened one ( @xmath7 ) . the frequency dependence ,
although complicated , can be represented roughly by a single plasmon frequency @xmath10 , whose value is usually much larger than the bandwidth ( antiadiabatic plasmon ) .
if one wants to deal with its precise form , @xmath11 can be resolved in many plasmon contributions , which characterize the screening process .
these features make the problem difficult : the large @xmath9 rules out the application of traditional weak coupling expansion methods , while the presence of many plasmons prevents the direct use of methods developed in the context of the hubbard - holstein hamiltonian @xcite , as many bosonic baths coupled to the fermion degrees of freedom will be necessary to fully resolve @xmath8 .
recently , this problem has been overcome by a continuous time quantum monte carlo ( ctqmc ) solver proposed by werner and millis@xcite , where a multi - plasmon lang - firsov transformation@xcite is treated exactly in the context of a hybridization expansion algorithm for the dmft impurity hamiltonian@xcite .
also the weak coupling ctqmc algorithm@xcite by rubtsov can treat generic retarded interactions , but it is limited to a not - so - large dynamic @xmath1 and not - so - large screening frequencies , and therefore it becomes prohibitively costly for realistic applications of dynamic screening interactions in a multi - orbital context .
another major problem still left ( even if realiable monte carlo data are available ) is the possibility of computing spectral properties , such as high - energy plasmon satellites .
indeed , the presence of screening modes leads usually to a quite complicated spectrum with a series of peaks located at high energies ( at multiples of the plasma frequencies ) .
those features are hard to get by the usual maximum entropy ( me ) methods@xcite , used to invert the noisy qmc data in the imaginary time domain into the spectral properties at real frequencies .
the me methods are quite reliable at low energy , but usually are not capable to deal with high frequency features . in this paper
, we present a dmft approach based on a bose factor ansatz ( bfa ) for the green s function which is able to handle a generic @xmath8 interaction in a strong coupling antiadiabatic regime , a typical situation in strongly correlated materials , and provides a robust and general way to compute the full spectrum of the frequency dependent ( retarded ) @xmath1 , with an accuracy capable to resolve the high energy satellites .
our method is based on the separation between the energy scales set by the screened value @xmath7 and treated using well established solvers@xcite , and the dynamic part treated with various levels of approximation , the simplest and most insightful one taken from the dynamic atomic limit .
the paper is organized as follows . in sec .
[ ansatz ] we specify the anderson impurity model we would like to solve and the green s function ansatz used in our method , in sec . [ atomic ] we present the dynamic atomic limit approximation ( dala ) to our approach , in sec .
[ beyond ] we show various ways to improve upon the dala , and in sec . [ overview ]
we describe their performances . in secs .
[ results ] and [ srvo3 ] we report our results for a single - band lattice model at half - filling and a three - band model with the dft density of state ( dos ) of 3 , respectively . finally , sec . [ conclusions ] summarizes our findings .
we discuss here the case of a multi - orbital hubbard - holstein model . in the context of the dmft approach ,
one maps the full lattice problem into a single - site anderson impurity problem coupled to an effective bath .
the bath is determined self - consistently by requiring that the impurity green s function equals the on - site green s function on the lattice.@xcite therefore , computing in the most effective way the green s function of the anderson model ( am ) is of key importance to have a feasible dmft scheme . here
, we have the additional complication that the effect of screening makes the on - site hubbard interaction retarded . in the matsubara imaginary time action formalism
the dynamic anderson model reads : @xmath12 where @xmath13 is the inverse temperature , @xmath14 is the screened interaction , @xmath15 is the set of orbital and spin indexes , and @xmath16 are the creation and annihilation operators satisfying the antisymmetric commutation relations .
@xmath17 is the effective hopping term coupled with the bath via the hybridization @xmath18 , and @xmath19 is the chemical potential .
let us assume that the dynamic part of @xmath1 is orbital - independent and couples only to the total charge .
@xmath20 with @xmath21 .
therefore , we assume that all details of the interaction due to effective spin and hund s couplings are embedded in the instantaneous part of the full @xmath1 . here and thereafter we are going to take the convention that the instantaneous @xmath22 is the static ( fully screened ) limit of @xmath1 with @xmath23 being a _ repulsive _ contribution .
we are mainly interested in evaluating the green s function @xmath24 and its spectral properties for the model in eqs .
[ s_eff ] and [ generic_interaction ] . we are going to rewrite it in the form : @xmath25 where @xmath26 is the green s function for the model in eq .
[ s_eff ] , but with a static on - site repulsion , namely @xmath27 .
we highlight that the above factorization is defined in the _ time domain _ , a feature which is borrowed from the dynamic atomic solution of the problem whose form is known analytically , as explained in sec .
[ atomic ] . in that limit the green s function assumes exactly the form in eq .
[ eq : ansatz ] , with @xmath28 the instantaneous @xmath7 atomic green s function .
the static model is much easier than the dynamic one , since it contains only the energy scales set by the screened @xmath29 and the kondo resonance with the bath , and it can be solved by means of various techniques@xcite , which are usually very robust and efficient in this case . on the other hand , @xmath30 is a bose factor , which is a functional of @xmath31 , and it is not known a priori .
however , we will present various approximations where the function @xmath30 is derived .
it contains the information of the plasmon ( or phonons ) excitations , and the plasmon ( or phonons ) satellites . a great advantage of dealing with the green s function ansatz in eq .
[ eq : ansatz ] is the possibility to compute very accurate spectral functions over the whole energy range , including the intermediate - high energy plasmon satellites . indeed , since the bose factor @xmath30 can be estimated analytically by means of some approximation , its numerical value is known at machine precision , and its spectral function @xmath32 can be obtained via a pad approximant,@xcite in an accurate and robust way . on the other hand , a me
approach has to be used to find the spectral function @xmath33 of the static green s function @xmath34 .
however , this does not pose any particular problem , since there are no high - energy features in @xmath33 , and its energy range is set by @xmath7 , where the me is reliable in presence of data with good statistics.@xcite the spectral function @xmath35 of the full green s function @xmath36 expressed as a functional of @xmath37 and @xmath38 reads : @xmath39 the spectral functions obtained in this way are reported for instance in fig .
[ spectral ] , which corresponds to the green s functions plotted in fig .
[ benchmark ] .
the quality of the satellite resolution is striking , much higher than the one usually obtained with me methods@xcite , particularly at energies far away from the fermi level .
we would like to stress that the spectral convolution in eq .
[ spectral_conv ] is general , and can be used not only for the approximated green s functions we are going to derive in secs .
[ atomic ] and [ beyond ] .
for instance , our approach to compute the spectral properties can be applied to the green s function obtained by means of the algorithm in ref . .
given the full green s function @xmath40 of the dynamic impurity problem of eq .
[ s_eff ] , one defines an auxiliary green s function @xmath41 as @xmath42 , with @xmath43 taken from the atomic limit as described in sec .
[ atomic ] .
this is an effective way to exploit the separation of the low energy properties , kept in @xmath44 , from the high frequency features correctly reproduced by the dala . at this point
, one computes@xmath38 , the spectral representation of @xmath44 , by using me , and @xmath37 , the spectrum of @xmath43 , by means of the pad approximant , and evaluates the full spectral function in eq .
[ spectral_conv ] .
in order to find a way to determine @xmath30 in eq .
[ eq : ansatz ] , for the moment let us take into account the single - orbital symmetric case .
thus , we can drop all the orbital and spin indexes , and simplify considerably the notation . by inverting eq .
[ eq : ansatz ] , one gets : @xmath45 where the rightmost - hand side of the above equation is the approximation for @xmath46 taken in the dynamic atomic limit ( dal ) , when the hybridization @xmath47 is zero .
it turns out that the green s function in the dal is analytically solvable by means of a hubbard - stratonovich transformation@xcite , and therefore @xmath48 is exactly known in a close analytic form , such that : @xmath49 where @xmath50 are bosonic matsubara frequencies , with @xmath51 relative integer . beside the atomic limit , this approximation is exact in the static and the non - interacting limits ( in both cases @xmath52 ) .
notice that it retains all the non - perturbative character of @xmath53 . to have a better idea on the quality of this approximation
, we are going to test it for the dynamic @xmath1 with a single plasmon mode @xmath10 , which is equivalent to the anderson - holstein model with @xmath54 and the electron-``phonon '' coupling given by @xmath55 .
we use the ctqmc algorithm by rubtsov,@xcite which can handle retarded interactions and yield the exact green s function in a weak coupling regime , to benchmark our approximation for the particle - hole symmetric system with @xmath56 , @xmath57 , and few values of @xmath10 and @xmath9 .
the energy units are expressed in terms of the half bandwidth ( @xmath58 ) of the semicircular dos .
( color online ) spectral function of the @xmath59 green s functions reported in fig .
[ benchmark ] . ] as one can see from fig .
[ benchmark ] , the dala works very well for the cases analyzed , since it gives a @xmath60 which almost coincides with the numerically exact @xmath40 given by the ctqmc algorithm .
the accuracy is particularly impressive in the case @xmath57 , @xmath56 , @xmath61 , and @xmath62 , where the impact of the dynamic part is reduced by the larger @xmath63 , and a smaller @xmath9 , namely when the the energy scales of the static part set by @xmath7 are well separated form the dynamic contributions in @xmath64 .
moreover , from fig . [ benchmark ]
it is apparent that the low energy properties of the system are strongly renormalized by the effect of the high energy components of @xmath1 .
this is quite insightful on the importance of the dynamic screening effects in the treatment of more realistic models that we are going to tackle in sec .
[ results ] .
this can be noted also from the spectral properties reported in fig .
[ spectral ] , which correspond to the green s functions @xmath60 plotted in fig .
[ benchmark ] .
a spectral weight transfer from the low frequency spectrum to the high energy satellites is clearly visible in the figure .
since the dala is obtained in the @xmath65 limit , it works well in the intermediate - strong coupling regime , with @xmath7 and the dynamic part large , as we have seen in the cases analyzed in fig .
[ benchmark ] , where @xmath7 was quite close to the critical @xmath66 ( @xmath67 , see ref . ) for the mott transition of the static hubbard model .
however , it deteriorates as @xmath7 is getting smaller and the hybridization @xmath47 becomes important to set the low - energy properties of the system .
to show this , let us take into account the anderson - holstein model with @xmath56 , @xmath62 , and @xmath61 at @xmath57 , for which the dala gives a result very close to the exact one .
now , let us keep @xmath10 and @xmath68 fixed , such that the dala bose factor ( eq .
[ eq : dala ] ) is unchanged , while we vary @xmath7 from strongly to more weakly correlated values .
the resulting green s functions are plotted in fig .
[ weaker ] .
it is apparent that the dala accuracy reduces as @xmath7 decreases , and the resulting bias is more pronounced in the low - energy part of the green s function .
indeed , the discrepancy is larger around @xmath69 in @xmath40 ( fig .
[ weaker]a ) , which corresponds to a larger difference at small matsubara frequencies in @xmath70 ( fig .
[ weaker]b ) . on the other hand ,
the high - energy tails of @xmath40 are very well reproduced by the dala , as it is confirmed also by the inspection of @xmath71 at large @xmath72 .
( color online ) half - filled anderson impurity model with retarded screened interaction : @xmath62 , @xmath73 , @xmath57 , for various values of @xmath7 ( a ) numerically exact rubtsov s ctqmc and dala @xmath40 ; ( b ) ctqmc and dala imaginary part of @xmath70 . ]
the correct high - energy asymptotics of the dala is a nontrivial property of this approximation , which is borrowed from the atomic limit exactly built - in . in order to further analyze this important feature , we take into account the temperature dependence of the dala in the symmetric anderson model with a dynamic interaction given by @xmath74 , @xmath75 , and @xmath76 . in this not - so - correlated case
, the model can be solved exactly down to low temperatures ( @xmath77 ) even by the ctqmc algorithm , to benchmark the temperature dependence of our approximation .
results are plotted in fig . [ pinning ] . as we already found in the previous analysis , at large matsubara frequencies the dependence of the green s function
is correctly given by the dala , which in this case becomes almost indistinguishable from the exact ctqmc result for @xmath78 .
we note that the dala is capable to reproduce the decay of the imaginary part of @xmath71 well beyond the @xmath79 term , as it is apparent from fig .
[ pinning](b ) .
it is also worth noting that the relative accuracy of the approximation increases with the temperature , as it is shown in fig .
[ pinning](d ) .
indeed , the exact bose factor @xmath46 is getting closer to @xmath80 as the temperature increases .
at low temperatures , it is the @xmath81 value which is poorly reproduced by the dala .
again , this is related to the roughness of the approximation at low - frequency , which does not describe accurately the low - energy excitations around and below the coherent temperature .
indeed , the friedel sum rule is clearly violated , as one can see in fig .
[ pinning](a ) , where the condition @xmath82 valid at half filling is not met by the dala .
therefore , going beyond the dala is needed to capture the low - energy low - temperature features of the spectral function , while its high - energy properties , as the plasmon satellites , can be successfully taken into account at this level of approximation .
a way to improve the dynamic atomic limit in both the intermediate - low energy correlations and low temperature regimes is provided by the lang - firsov approach .
this approximation has been widely used in the literature to tackle electron - phonon models in the antiadiabatic limit , when the electron - phonon coupling is @xmath83 , with @xmath10 the phonon frequency .
the same applies to models where the interaction is retarded by the charge screening plasmons . in the latter case ,
the antiadiabatic regime is met more often , since the plasmon frequences are larger than the phonon ones , usually by an order of magnitude .
however , it should be noted that in the realistic retarded hubbard @xmath1 also the electron - plasmon coupling is stronger , as @xmath84 , and @xmath85 is an order of magnitude larger than @xmath7 . in any case , if @xmath86 , the lang - firsov approach describes well the low energy properties of the system .
here , we use the lang - firsov approximation in a new and original way , as a low frequency correction to the dynamic factor of our green s function ansatz .
the factorization in the @xmath87 space implies a convolution in the @xmath88 space , @xmath89 where @xmath90 are the matsubara components of the bose factor . a way to improve upon
the dala is to choose the @xmath91 factor such that the lang - firsov behavior is obtained at low frequency .
this can be done , by introducing an enhanced @xmath91 defined as : @xmath92 where @xmath93 , @xmath94 , and @xmath95 are parameters determined by the following conditions : @xmath96 , which keeps the correct number of particles provided by @xmath97 ; @xmath98 for @xmath99 ( the first matsubara frequency ) , which gives the correct friedel sum rule fulfilled by the lang - firsov approximation ( and broken by the dala ) ; @xmath95 is the crossover frequency between the lf behavior at low energy and the atomic dala behavior at high frequency , and its optimal value is @xmath100 , with @xmath10 the lowest holstein frequency .
the lf green s function is given by the usual expression : @xmath101 ( i \omega ) } , } \label{lf_green_function}\ ] ] which clearly implies that an anderson model with static hubbard @xmath7 must be solved , hybridized via a renormalized bath @xmath102 .
therefore , to get the dala+lf green s function one has to solve two static models ( with regular and renormalized bath ) and mix them together by using our definition in eq . [ mixing ] . one should note that the lang - firsov green s function in eq .
[ lf_green_function ] has been written for a single plasmon ( or phonon ) anderson - holstein model , which gives a rough representation of the dynamically screened @xmath64 present in _ ab - initio _ models .
for instance the c - rpa approximation for @xmath64 usually leads to a quite broad spectrum of screening plasmons , which is difficult or impossible to fit accurately by a single frequency model .
however , a generalization of the eq
. [ lf_green_function ] can be easily done by following the same lines as in the ref . .
if the lf renormalization factor @xmath103 is replaced by @xmath104 where @xmath105 , and @xmath106 are real frequencies , all the screening plasmons are treated on the same footing . although the lf approximation is accurate only for large @xmath106 , the eq .
[ generalized_lf ] is a good approximation if @xmath107 goes rapidly to zero for small @xmath106 . for the holstein single - mode interaction @xmath108 , and one
recovers the standard lf renormalization factor .
the improvement of the dala+lf correction with respect to the simple dala factor is apparent in fig .
[ lang_firsov_low_t](a ) , to be compared with fig .
[ pinning](a ) .
in particular , at low frequency the lf is capable to recover the pinning condition ( @xmath109 for the semicircular density of states at half filling ) , violated by the dala at low temperatures .
this is reflected in the behavior of the bose factor @xmath110 drawn in fig .
[ lang_firsov_low_t](b ) , whose value at @xmath69 goes correctly to 1 for @xmath111 , in contrast to the @xmath43 which does not change with temperature . indeed , the exact @xmath91 factor shows a quite strong temperature dependence , as reported in fig .
[ pinning](d ) . in fig .
[ lang_firsov_low_t](c ) , we plot the spectral representation of the dala+lf green s function at @xmath77 obtained by the dala factor assisted analytic continuation described in sec .
[ ansatz ] . another way to improve upon
the dala is to rely on the diagrammatic expansion of the interaction .
this can be done in various ways .
here we took two routes : making a cumulant green s function expansion in the full @xmath64 and the retarded part @xmath112 only . in the former approach ,
dubbed `` gaussian cumulants '' ( gc ) and described in this subsection , the perturbation theory is built on the gaussian action containing the hybridized @xmath113 as the bare green s function . in the latter method , named `` instantaneous bold cumulants '' ( ibc ) and introduced in subsection [ diagrammatic_deltau ] ,
the perturbation expansion requires the calculation of density - density correlators within the static @xmath7 model by means of the ctqmc algorithm ( or other algorithms suitable for static interacting models).@xcite the first - order expansion in the full @xmath64 leads to the following expression for the green s function : @xmath114 where @xmath115 , and @xmath116 is the connected part of the density - density correlator easily computed for the gaussian @xmath113 propagator .
@xmath117 is the spin ( and orbital ) integrated density .
note the exponential form in eq .
[ fullu ] , which comes from the first order cumulant expansion in @xmath1 .
according to the definition in eq .
[ eq : ansatz ] , the factor @xmath91 obtained in this way is given by the ratio @xmath118 , which reads @xmath119 with @xmath120 .
the above expression for the dynamic factor has the advantage to be very accurate at both small @xmath1 , since the interaction is treated perturbatively , and strong coupling , as the density - density correlator will factor up by giving the exact dynamic atomic limit ( see the appendix ) .
incidentally , this is the reason why the cumulant expansion is more effective than the regular perturbation theory for this case .
one can go beyond the approximation for @xmath91 in eq .
[ first_order ] and compute the density - density correlator @xmath121 by using the interacting static @xmath7 green s function @xmath53 as propagator instead of the hybridized green s function @xmath122 .
this requires a correlated method , as the corresponding action is no longer gaussian and the wick theorem can not be applied .
the qmc methods can compute @xmath121 directly in the interacting systems.@xcite this development represents a consistent diagrammatic first - order expansion in @xmath123 where the reference theory is the static @xmath7 model . the main difference with respect to the previous method reported in subsec .
[ diagrammatic_fullu ] stems from the fact that not only hybridization effects but also the impact of the instantaneous @xmath124 interaction are included in @xmath121 .
the cumulant expansion of the full green s function in the retarded part @xmath123 with @xmath53 taken as the instantaneous bold propagator gives directly a factorized form of the type in eq .
[ eq : ansatz ] . at the first order
, it takes the expression in eq .
[ fullu ] with @xmath122 replaced by @xmath53 and the thermal averages @xmath125 computed for the static interacting model , such that @xmath126 .
it turns out that in this case the @xmath91 factor reads : @xmath127 in the following we report a comparison of the different approaches proposed in this work . as common benchmark
, we chose to compute the green s function of the anderson - holstein impurity problem both in the symmetric and asymmetric cases , and check the bfa solutions against the numerically exact one provided by rubtsov s ctqmc algorithm .
the factorization introduced in eq .
[ eq : ansatz ] is an extremely useful green s function ansatz to compute thermal and spectral properties coming from a generic retarded interaction in the multiband anderson model , and also the hubbard model at the dmft level .
the product in the time domain between a bose factor @xmath91 embedding the dynamic properties of the interaction and an auxiliary green s function @xmath28 including instantaneous interactions and low energy features , leads to a deconvolution of the spectrum into low and high frequency contributions .
its low frequency part , depending on @xmath28 , can be easily obtained by available and well developed me methods , while the high energy features , as the plasmon satellites , difficult or impossible to obtain by standard analytical continuation , are directly given by the analytically known bose factor @xmath91 , which can be inverted in an accurate way by the pad approximants .
the green s function bose factor is a very general ansatz .
however , the factor @xmath91 is not exactly known in the generic case and needs some approximations .
the most practical and physically insightful one is borrowed from the dynamic atomic limit , where the factorization is exactly given by a @xmath28 depending only on the instantaneous @xmath7 times a factor which depends only on the retarded part @xmath128 .
the dala proposed in sec .
[ atomic ] consists of keeping the atomic dynamic factor @xmath91 and taking the @xmath28 from the exact numerical solution of the instantaneous @xmath7 model . as reported in figs .
[ benchmark ] and [ all_half_filling](b ) , the dala performs well in the antiadiabatic regime when @xmath129 , and in the strong coupling regime , with a large @xmath7 .
the quality of the low energy part of the dala green s function worsens as the interaction becomes weaker and @xmath10 gets smaller , while its high energy tails are well reproduced even in the intermediate coupling .
the major failure of the dala is the breaking of the friedel sum rule , which is apparent at low temperature ( below @xmath130 ) or away from the antiadiabatic regime .
that is not surprising since the dala is built upon the atomic limit . in general , the dala works when the high energy ( unscreened ) part of the retarded interaction is well separated ( in frequency ) from the low energy ( screened ) part , which is the most common situation in the case of realistic hamiltonians , with @xmath131 determined _ ab - initio _ by the c - rpa approach@xcite .
( color online ) symmetric anderson model with retarded screened interaction as in fig .
[ weaker ] : @xmath62 , @xmath73 , @xmath57 , for @xmath132 and two values of @xmath7 , ( a ) @xmath133 ; ( b ) @xmath56 .
the numerically exact rubtsov s ctqmc , dala , dala+lf , gc , and ibc green s functions are reported . ]
( color online ) asymmetric spin unpolarized anderson model with retarded screened interaction as in fig .
[ all_half_filling ] : @xmath62 , @xmath73 , for @xmath132 and two values of @xmath7 , ( a ) @xmath133 ; ( b ) @xmath56 . in this case
the number of particles @xmath134 was set to be around the quarter filling .
the numerically exact rubtsov s ctqmc , dala , dala+lf , gc , and ibc green s functions are reported . ]
an improvement upon the dala is represented by the lang - firsov approximation , which fulfills the friedel sum rule and cures the low - energy low - temperature behavior .
the way to incorporate the lf into the dala factor , described in subsec .
[ dalalf ] , clearly improves the low - energy features of the dala green s function , as one can see in figs .
[ all_half_filling ] and [ all_quarter_filling ] , where the value of @xmath135 yielded by the dala+lf approach is closer to the exact numerical result in both the half - filling and quarter - filling anderson - holstein impurity models .
the dala+lf factor provides an overall better agreement with the numerically exact green s function computed by the ctqmc method .
the lf correction turns out to be important particularly in the weak - intermediate coupling and at low temperature , while in other cases the simple dala approximation is already good enough . for instance , in fig . [ all_half_filling](b ) , when @xmath7 is strong and the temperature not - so - low ( @xmath57 ) , the dala green s function coincides with the dala+lf one , and both are on the top of the exact numerical solution .
a common limitation of the dala and dala+lf approximations is that they become inaccurate when @xmath136 , while the work very well in the antiadiabatic regime .
a different route to determine @xmath91 is built on diagrammatic expansion techniques .
we have proven that the cumulant expansion of the green s function up to the first order in the interaction provides the exact connection to the dynamic atomic limit ( see appendix ) .
indeed , the @xmath91 factor computed in this way ( see eq . [ first_order ] ) has the exact atomic behavior in the zero hybridization limit .
we propose two flavors for the cumulant expansion . in both cases one has to evaluate the density - density correlator @xmath137 . in the first one ( the gaussian cumulants or `` gc '' )
the thermal average is computed with a gaussian action based on the hybridized non - interacting green s function , while in the other one ( the instantaneous bold cumulants or `` ibc '' ) the brakets are computed with the instantaneous @xmath7 green s function .
the gc works well in the weak interacting regime ( small @xmath7 ) and it worsens as @xmath7 is getting larger , as shown in fig . [ all_half_filling ] .
away from half - filling , the gc green s function is reliable in the `` empty '' part of the spectrum ( for @xmath138 ) , and where the dynamic part is more relevant in setting the tails of the green s function . on the other hand ,
the ibc performs well everywhere in all cases when @xmath10 is large , as seen in figs .
[ all_half_filling ] and [ all_quarter_filling ] , and it is supposed to work quite well even away from the antiadiabatic regime , since it retains the feedback of the instantaneous interaction on the dynamic part via the @xmath121 correlator .
the price to pay in the latter case is the statistical uncertainty of @xmath139 , as @xmath121 must be evaluated by means of monte carlo techniques , in contrast to the gc approach where @xmath121 is known up to machine precision thanks to wick s theorem .
this could lead to some inaccuracy in the analytical continuation of the ibc bose factor .
if the goal is to compute the spectral representation of a green s function with retarded interactions , we found that the most effective way is to use the bose factor taken from the dynamic atomic limit .
as already reported in sec .
[ ansatz ] , one has to define an auxiliary green s function @xmath41 as @xmath42 , and then use eq .
[ spectral_conv ] to compute the spectrum .
the dala factor is the simplest to invert , and its bosonic spectral representation is the `` physical '' density of plasmonic ( or phononic ) modes , whereas the other factors beyond the dala include also some low energy contributions .
therefore , the dala factor is the most recommended in the assisted analytical continuation .
[ half_filled_bethe ] in the following , we are going to present applications where our bfa approach is used as solver of the anderson impurity problem resulting from the dmft self - consistency conditions .
therefore , in contrast to what has been shown in the methodological sections , the following results are the converged solutions of the dmft equations for the full lattice hamiltonian , which incorporates the retarded @xmath64 as the on - site interaction .
the first application is on the single band half - filled hubbard - holstein model solved by the dmft on the bethe lattice . to study the impact of the retarded screened interaction on the mott transition , we analyze the half - filled model at @xmath130 with different @xmath7 , @xmath9 , and @xmath10 parameters .
we choose to work always with @xmath140 , a `` physical '' condition which states that the unscreened @xmath1 is larger than the screened one .
the screening frequency is taken such that we are in the antiadiabatic regime , and we use our various bose factor approximations to predict the critical value of @xmath9 for the mott transition , once the other parameters are fixed , such that @xmath141 and @xmath56 . as reported in literature,@xcite the metal - mott insulator transition in the hubbard - holstein model
is first order . in tab .
[ table ] we present our lower and upper critical values obtained for @xmath130 by means of the bfa at various levels of approximation .
gc & 12.3 & 12.5 + dala & 8.2 & 8.4 + dala+lf & 4.9 & 5.2 + as one can see , the actual values at criticality depend quite strongly on the approximation used , with the dala+lf giving the results aligned with the numerically exact ctqmc method by werner and millis@xcite ( see fig . [ critical_u ] ) . in order to have a close comparison to the data published in ref .
, we carried out some calculations also for @xmath142 .
the agreement between the dala+lf and the monte carlo predictions is quite remarkable .
this highlights again the importance of the lf correction at low energy in order to accurately predict the physical properties .
[ critical_u ] summarizes the results for the upper critical line for different sets of @xmath10 and @xmath7 , whose values are taken not so far from the critical @xmath143 of the static model computed at @xmath130@xcite .
it is clear that the fully retarded model with @xmath140 is more correlated than the static one with the same instantaneous @xmath7 .
the mott transition happens at values of screened @xmath7 _ lower _ than the critical static @xmath144 for any finite @xmath10 .
the dependence of the critical parameters on @xmath10 is also clear . at
fixed screened @xmath7 , a smaller @xmath10 corresponds to a smaller unscreened @xmath9 at which the mott transition is reached .
indeed , if the frequency of the screening plasmon is closer to the fermi level , it is easier for the unscreened part to induce the mott transition at low energy .
the same conclusions were reached by werner and millis by means of their numerically exact ctqmc algorithm .
this shows the impact of the dynamic screening features on the low energy properties of the model . in order to get the same effective low energy parameters
, the `` effective '' instantaneous @xmath7 can be up to @xmath145 larger than the true screened value even for plasmon frequencies @xmath146 in the antiadiabatic regime . in fig .
[ spectral_function_critical_u ] we report the spectral representation of a half - filled holstein - hubbard model with @xmath141 , @xmath56 , @xmath147 , and temperature @xmath142 , quite close to the mott transition . the spectral function has been obtained with the help of the dala factor by the method in sec .
[ ansatz ] .
it describes with high accuracy not only the low - energy features but also the plasmon satellites in the anti - adiabatic regime , typical of realistic materials .
( color online ) critical @xmath144 line of the mott transition calculated at @xmath130 and half - filling on a bethe lattice for the screened instantaneous interaction @xmath7 as a function of @xmath148 and @xmath149 ( green diamonds and red circles respectively ) .
the critical values are obtained from the green s function computed by the dala+lf ansatz .
the behavior of the critical screened @xmath1 is in agreement with the one computed by werner and millis with their ctqmc algorithm at @xmath141 and @xmath142 ( black upper triangles ) .
indeed , the critical line calculated by means of our dala+lf approach ( blue lower triangles ) is on top of their ctqmc points@xcite .
this highlights the accuracy of our approximated dala+lf method if compared to the exact numerical result .
note the quite strong temperature dependence of the critical @xmath144 in going from @xmath130 to @xmath142 .
this is a quite interesting effect which deserves further analysis . ]
spectral representation of the green s function for a holstein - hubbard model with @xmath141 , @xmath56 , @xmath147 and @xmath142 , close to the blue mott transition line of fig .
[ critical_u ] .
the analytic continuation has been assisted by the dala bose factor , which allows one to describe accuratley the plasmon satellites centered around frequences multiple of @xmath10 . ]
our second application is on a realistic hamiltonian for 3 , a very well studied material which represents a benchmark for theories describing strongly correlated compounds .
3 is the prototype of a correlated metal where the many - body treatment of correlation in the @xmath2 manyfold is important to explain the spectral properties@xcite .
therefore , it has been the subject of intensive studies@xcite applying dmft in the context of realistic strongly correlated hamiltonians . indeed
, its band structure is relatively simple due to its undistorted perovskite structure , resulting in the occupation of one electron in three degenerate @xmath151 bands crossing the fermi level .
the @xmath4 oxygen ligands are quite well separated from the @xmath2 levels , such that the definition of a low - energy 2g - hamiltonian is unambiguous .
thus , 3 has been the testing case for many new dft - dmft implementations.@xcite on the other hand , 3 has been the subject of intensive experimental activity , with magnetic , electrical , optical measurements @xcite , and by means of photoemission spectroscopy ( pes)@xcite and angled resolved pes ( arpes ) @xcite . here , we consider a model where only the @xmath151 electron is retained , and all the others contribute to screen the local lattice interaction .
the dft band structure has been calculated with the linear muffin - tin orbital ( lmto ) framework in the atomic sphere approximation ( asa ) , which allows one to work with a native @xmath2-projected and localized orbital representation .
the realistic retarded @xmath1 for this compound has been computed in ref .
based on the c - rpa construction .
the low - energy hamiltonian we are going to work with consists of the lda 2g - hamiltonian , the plasmonic part giving rise to the dynamical screening and the following static interaction hamiltonian : @xmath152 where @xmath153=@xmath154 is the usual density operator , with @xmath155,@xmath156 denoting orbital and spin indexes , and @xmath157 , @xmath158 representing the @xmath159 localized orbitals . for the 2g - orbitals of srvo@xmath160 , @xmath161 and @xmath162 , with the screened value of the interaction @xmath163 ev , and the hund s coupling @xmath164 ev .
the additional retarded interaction which is included in our model , couples to the total charge of the system , as described by eq .
[ generic_interaction ] .
there is no need for an explicit double - counting term , since such a correction is absorbed into the effective chemical potential fixing the particle number to one .
to study the impact of @xmath165 on the low energy properties of the model , we took into account different holstein single - plasmon @xmath64 s , and compared them with the corresponding static model .
we used our bfa approach in its simplest dala formulation , and computed the spectrum at @xmath57 and @xmath132 ev@xmath166 .
indeed , it turns out that at those temperatures the lf correction is irrelevant , and so the dala performs here at best even in terms of efficiency . in fig .
[ static_vs_dynamic ] , we report the spectra computed with the instantaneous @xmath167 interaction , and two retarded interactions with the same unscreened @xmath168 , but different screening @xmath10 s ( one at 5 ev , the other at 15 ev ) . the static model has a quasiparticle peak at the fermi level typical of strongly correlated compounds , with a lower and upper shoulders reminiscent of the lower and upper hubbard bands . in the dynamic model , the effect of the 15 ev plasmon is to renormalize the quasiparticle spectral weight , transferred at higher energies , while the shape of the low energy spectrum is almost unchanged .
the effect of the 5 ev plasmon , closer to the low energy sector , is more remarkable . beside a stronger spectral weight reduction
, there is a shift of the upper hubbard shoulder to lower energies than in the corresponding static model .
this effect is certainly due to the interplay between the plasmon satellite at 5 ev , visible in fig .
[ static_vs_dynamic ] , and the low energy features of the spectrum .
( color online ) spectral function of one electron in 3 @xmath151 bands for the 3 obtained by the dft in the local density approximation .
the dmft calculations are done at @xmath57 ev@xmath166 with the dala bose factor solver for the retarded @xmath1 and the hirsch - fye qmc solver for the instantaneous interaction .
three types of interactions are reported : the purely static one , with @xmath163 ev , a dynamic holstein one with the same instantaneous interaction but with the unscreened @xmath169 ev and a plasmon frequency @xmath62 ev , another holstein interaction with the same @xmath7 and @xmath9 but different @xmath170 .
note the displacement of the upper hubbard shoulder shifted at lower energy by the presence of the plasmon at @xmath62 ev . ]
( color online ) spectral function at @xmath57 ev@xmath166 of one electron in 3 @xmath151 bands for the 3 obtained by the dft in the local density approximation .
the dmft calculations are carried out with the dala bose factor solver for the retarded @xmath1 and the hirsch - fye qmc solver for the instantaneous interaction .
the result of the realistic green s function is reported , showing a more correlated behavior than the corresponding static model with the same instantaneous @xmath171 .
the quasiparticle peak is smaller than the corresponding static model , with a spectral weight transfer at higher energies . ]
the realistic retarded @xmath64 is characterized by two main screening frequencies at 5 and 15 ev@xcite , while the unscreened value of @xmath1 is 16 ev , much larger than the one in the models considered so far . by applying our bose factor ansatz to this problem
, we found the spectral function in fig .
[ static_vs_realistic ] , reported together with the instantaneous interaction with the same screened @xmath167 . also here
we note the quasiparticle weight reduction , and the shift of the upper hubbard shoulder to lower frequencies , as in the model @xmath1 analyzed before , while the lower hubbard band is almost unchanged with respect of the static model .
its maximum corresponds to the position found in pes measurements , and well documented in previous dmft studies on 3 .
the weight reduction of the quasiparticle peak coincides with a smaller value of @xmath172 , namely a larger value of the effective mass @xmath173 .
in particular , for the realistic dynamic model we found a @xmath174 , which gives an effective mass renormalized by 2 with respect to the dft band structure , while for the corresponding static model we obtained a value of @xmath175 .
recent arpes data yielded an effective mass @xmath176@xcite , which is in a good agreement with our findings for the realistic retarded interaction . on the other hand , the static model with
the same screened @xmath7 underestimates the correlation by a factor of @xmath177 , and so the value of @xmath173 . in the literature ,
a static model with a larger instantaneous @xmath178 has been used to find the experimental mass renormalization@xcite .
such a larger value of @xmath1 could be justified by the constrained lda method@xcite used to determine _ a priori _ the on - site interaction , but known to overestimate its strength .
the difficulty here is to reproduce by the same model both the effective mass and the position of the lower hubbard band , which turns out to be shifted at lower energies ( @xmath179 ev ) by a stronger @xmath7 .
this is the reason why in ref .
the authors made the choice to work with @xmath180 ev , a slightly weaker effective static interaction which gives the lower hubbard band correctly peaked at @xmath181 ev .
some cluster calculations give the same position for the lower hubbard band , with the interpretation that its peak depends on the hybridization and screening properties provided by the ligands@xcite . with our
dynamically screened model we describe correctly both the mass renormalization ( @xmath182 ) and the lower hubbard band position peaked at @xmath181 ev , as apparent in fig .
[ static_vs_realistic ] .
this analysis highlights the importance of including the proper _ retarded _ interaction , to have a reliable and fully _ ab - initio _ description of the correlation in these materials .
we note however , that the present description includes only the 2 g orbitals , so that the above conclusions are valid in an energy range where no other orbital contributions are present .
we introduced a factorized form @xmath183 for the green s function of the anderson model with generic retarded ( dynamic ) interaction , dubbed `` bose factor ansatz '' .
we proposed various approximations for the green s function bose factor @xmath91 , the most practical and effective one borrowed directly from the dynamic atomic limit ( dala ) , whose form is analytically known .
the dala provides an accurate way to compute the green s function in the antiadiabatic limit _ and _ evaluate the spectral properties in the full frequency range by means of an improved analytical continuation method . in practice ,
the inversion from the imaginary to the real frequency domain is assisted by the factor @xmath91 which retains the main information on the position and strength of the plasmon satellites , and enters in the analytical continuation as a convolution with the instantaneous part @xmath28 .
we carefully analyzed the pros and cons of the dala , and found various ways to improve @xmath91 , by either using the lang - firsov approximation at low - frequency , or resorting to diagrammatic techniques .
finally , we applied our approach to lattice problems in the context of the dmft formalism .
we took into account the hubbard - holstein hamiltonian at half - filling and on the bethe lattice , and we studied the mott transition in the spirit of looking at the retarded interaction as resulting from `` realistic screening '' of the bare @xmath1 by a single plasmon .
the second application has been for the realistic 3 hamiltonian , where the @xmath159 electrons interact via a retarded on - site @xmath1 previously determined _ ab - initio _ at the c - rpa level@xcite . in both cases , it turns out that it is important to retain the retarded features of the local interaction resulting from the dynamic screening in order to have a reliable _ ab - intio _ description of correlated materials
. also , our approach could be useful to determine whether some spectral signatures at intermediate energy ( @xmath184 ev ) seen in a broad class of correlated materials@xcite come from a bulk dynamic screening . in perspective , more work has to be done theoretically to rationalize the effects of the screening on both the ground and excited states properties of correlated compounds .
morever , by means of the same formalism , one can study jan - teller models to describe the impact of the electron - phonon coupling to the spectral function of the distorted compound .
last but not least , dealing with a frequency dependent @xmath1 is an essential step toward the implementation of the gw+dmft framework@xcite , where the screening resulting from the gw polarization has to be included consistently in the low - energy correlated model solved at the dmft level .
therefore , behind this work there are important experimental , theoretical and methodological implications that one can now start taking into account .
in this appendix we prove that the bose factor @xmath91 reported in eq .
[ first_order ] fulfills the exact atomic limit , i.e. it equals the one in eq .
[ eq : dala ] for @xmath65 .
this is a non trivial property , which guarantees the gc approximation to provide a reasonable description of the insulating phase ( and the metal - to - insulator transition ) in the dmft framework , where the hybridization function of the anderson impurity assumes a crucial frequency dependence in order to represent the coupling with the self - consistent bath . in the strong coupling case
, the hybridization @xmath185 goes to zero for small @xmath88 .
thus , in the dmft language , this phase is mapped into an anderson impurity problem close to the atomic limit , and its accurate solution is required around @xmath65 .
we start by noting that the density - density correlator @xmath121 factorizes in the atomic limit , as @xmath186 where @xmath187 . in the most general case
, @xmath121 can be resolved into its spin and orbital components , by defining @xmath188 , with @xmath189 .
the factorization in eq .
[ atomic_chi1chi1 ] holds also for the spin resolved quantities : @xmath190 . once the wick theorem is applied and the former correlators are written in terms of the @xmath122 s , it is straightforward to prove the spin resolved identity and consequently the integrated one of eq .
[ atomic_chi1chi1 ] , by using the atomic limit expression for @xmath122 .
for instance , in the atomic limit and for a generic @xmath19 , the connected part of @xmath191 reads : @xmath192 now we use a hubbard - stratonovich ( hs ) transformation to rewrite the @xmath193 factor in a form which includes @xmath191 in linear terms only .
we obtain the following identity : @xmath194 where @xmath195 is the functional integral over the complex hs field @xmath196 . by exploiting the atomic limit expression of @xmath191 in eq .
[ atomic_chi1 ] , one can compute the integral involving the product of @xmath197 and @xmath191 , which gives the result : @xmath198 with @xmath199 the fourier components of the hs field . by integrating the functional integral in eq .
[ hubbard_stratonovich ] in the fourier space , one gets the final expression for the exponent of the gc factor : @xmath200 we note here that @xmath23 vanishes in the static limit , as @xmath201 .
therefore , the first term in the right - hand side of the above equation vanishes as well .
we are left with the last term in eq . [ semifinal ] , which is exactly equal to the exponent of the dala factor .
thus , we have proven that @xmath202 for an arbitrary @xmath19 and @xmath23 . to conclude , the fact that the first - order cumulant expansion fulfills the exact dynamic atomic limit justifies the use of cumulants instead of the standard first - order developements , and validates also the cumulant of the instantaneous bold factor in eq .
[ ibc ] .
this work was supported by idris / genci under grant number 101393 , and by the french anr under project surmott .
we thank ferdi aryasetiawan , alexander i. lichtenstein , andrew j. millis , takashi miyake , jan m. tomczak and philipp werner for stimulating discussions .
99 a. georges , g. kotliar , w. krauth , and m. j. rozenberg , rev .
phys . * 68 * , 13 ( 1996 ) . m. imada , a. fujimori , and y. tokura , rev . mod
. phys . * 70 * , 1039 ( 1998 ) .
d. c. langreth , phys .
b * 1 * , 471 ( 1970 ) .
f. aryasetiawan , m. imada , a. georges , g. kotliar , s. biermann , and a. i. lichtenstein , phys .
rev . b*70 * , 195104 ( 2004 ) .
f. aryasetiawan , k. karlsson , o. jepsen , and u. schnberger , phys .
b * 74 * , 125106 ( 2006 ) . t. miyake and f. aryasetiawan , phys . rev .
b * 77 * , 085122 ( 2008 ) .
m. aichhorn , l. pouroskii , v. vildosola , m. ferrero , o. parcollet , t. miyake , a. georges , and s. biermann , phys . rev .
b * 80 * , 085101 ( 2009 ) . t. miyake , f. aryasetiawan , m. imada , phys . rev .
b * 80 * , 155134 ( 2009 ) . a. c. hewson , the kondo problem to heavy fermions , cambdridge studies in magnetism , cambrdige university press ( 1993 ) .
a. c. hewson , and d. meyer , j. phys . : condens .
matter * 14 * , 427 ( 2002 ) .
g. sangiovanni , m. capone , c. castellani , and m. grilli , phys .
lett . * 94 * , 026401 ( 2005 ) .
g. sangiovanni , m. capone , and c. castellani , phys .
b * 73 * , 165123 ( 2006 ) .
p. barone , r. raimondi , m. capone , c. castellani , and m. fabrizio , european phys .
lett . * 79 * , 47003 ( 2007 ) .
m. hohenadler and w. von der linden , phys .
b * 71 * , 184309 ( 2005 ) .
w. koller , d. meyer , and a. c. hewson , phys .
b * 70 * , 155103 ( 2004 ) .
g. s. jeon , t .- h . park , j. h. han , h. c. lee , and h .- y .
choi , phys .
b * 70 * , 125114 ( 2004 ) . p. werner and a. j. millis , phys . rev .
b * 74 * , 155107 ( 2006 ) .
p. werner , a. comanac , l. d medici , m. troyer , and a. j. millis , phys .
* 97 * , 076405 ( 2006 ) .
i. g. lang , and y. a. firsov , sov .
jetp * 16 * , 1301 ( 1963 ) .
p. werner and a. j. millis , phys .
99 * , 146404 ( 2007 ) .
p. werner and a. j. millis , phys .
lett . * 104 * , 146401 ( 2010 ) .
a. n. rubtsov , and a. i. lichtenstein , jetp letters * 80 * , 61 ( 2004 ) .
a. n. rubtsov , v. v. savkin , and a. i. lichtenstein , phys .
b * 72 * , 035122 ( 2005 ) .
j. e. gubernatis , m. jarrell , r. n. silver , and d. s. sivia , phys .
b * 44 * , 6011 ( 1991 ) .
m. jarrell and j. e. gubernatis , phys .
* 269 * , 133 ( 1996 ) .
j. e. hirsch and r. m. fye , phys .
56 * , 2521 ( 1986 ) .
h. j. vidberg , and j. w. serene , j. low temp .
* 29 * , 179 ( 1977 ) .
serge florens , phd thesis , ecole normale superieure , paris ( 2003 ) .
m. j. rozenberg , g. kotliar , and x. y. zhang , phys .
b * 49 * , 10181 ( 1994 ) .
a. georges , and w. krauth , phys .
b * 48 * , 7167 ( 1993 ) .
a. liebsch , phys .
lett . * 90 * , 096401 ( 2003 ) .
e. pavarini , s. biermann , a. poteryaev , a. i. lichtenstein , a. georges , and o. k. andersen , phys .
lett . * 92 * , 176403 ( 2004 ) .
e. pavarini , a. yamasaki , j. nuss , and o. k. andersen , new j. phys .
* 7 * , 188 ( 2005 ) .
i. a. nekrasov , g. keller , d. e. kondakov , a. v. kozhevnikov , th .
pruschke , k. held , d. vollhardt , and v. i. anisimov , phys .
b * 72 * , 155106 ( 2005 ) .
i. a. nekrasov , k. held , g. keller , d. e. kondakov , th .
pruschke , m. kollar , o. k. andersen , v. i. anisimov , and d. vollhardt , phys .
b * 73 * , 155112 ( 2006 ) .
v. i. anisimov , d. e. kondakov , a. v. kozhevnikov , i. a. nekrasov , z. v. pchelkina , j. w. allen , s .- k .
kim , p. metcalf , s. suga , a. sekiyama , g. keller , i. leonov , x. ren , and d. vollhardt , phys .
b * 71 * , 125119 ( 2005 ) .
f. lechermann , a. georges , a. poteryaev , s. biermann , m. posternak , a. yamasaki , and o.k .
andersen , phys .
b * 74 * , 125120 ( 2006 ) .
b. amadon , f. lechermann , a. georges , f. jollet , t. o. wehling , and a. i. lichtenstein , phys .
b * 77 * , 205112 ( 2008 ) .
g. trimarchi , i. leonov , n. binggeli , dm .
korotin , v. i. anisimov , j. phys . :
matter * 20 * , 135227 ( 2008 ) .
m. onoda , h. ohta , and h. nagasawa , solid state comm .
* 79 * , 281 ( 1991 ) .
i. h. inoue , o. goto , h. makino , n. e. hussey , m. ishikawa , phys .
b * 58 * , 4372 ( 1998 ) . h. makino , i. h. inoue , m. j. rozenberg , i. hase , y. aiura , s. onari , phys . rev .
b * 58 * , 4384 ( 1998 ) .
a. fujimori , i. hase , h. namatame , y. fujishima , y. tokura , h. eisaki , s. uchida , k. takegahara , and f. m. f. de groot , phys .
69 * , 1796 ( 1992 ) .
i. h. inoue , i. hase , y. aiura , a. fujimori , y. haruyama , t. maruyama , and y. nishihara , phys . rev .
lett . * 74 * , 2539 ( 1995 ) .
k. maiti , d. d. sarma , m. rozenberg , i. inoue , h. makino , o. goto , m. pedio , and r. cimino , europhys . lett . * 55 * , 246 ( 2001 ) .
a. sekiyama , h. fujiwara , s. imada , s. suga , h. eisaki , s. i. uchida , k. takegahara , h. harima , y. saitoh , i. a. nekrasov , g. keller , d. e. kondakov , a. v. kozhevnikov , th .
pruschke , k. held , d. vollhardt , and v. i. anisimov , phys .
lett . * 93 * , 156402 ( 2004 ) .
t. yoshida , k. tanaka , h. yagi , a. ino , h. eisaki , a. fujimori , and z .- x .
shen , phys .
lett . * 95 * , 146404 ( 2005 ) .
h. wadati , t. yoshida , a. chikamatsu , h. kumigashira , m. oshima , h. eisaki , z. x. shen , t. mizokawa , and a. fujimori , phase transitions * 79 * , 617 ( 2006 ) .
r. eguchi , t. kiss , s. tsuda , t. shimojima , t. mizokami , t. yokoya , a. chainani , s. shin , i. h. inoue , t. togashi , s. watanabe , c. q. zhang , c. t. chen , m. arita , k. shimada , h. namatame , and m. taniguchi , phys .
lett * 96 * , 076402 ( 2006 ) .
m. takizawa , m. minohara , h. kumigashira , d. toyota , m. oshima , h. wadati , t. yoshida , a. fujimori , m. lippmaa , m. kawasaki , h. koinuma , g. sordi , and m. rozenberg , phys .
b * 80 * , 235104 ( 2009 ) .
t. yoshida , m. hashimoto , t. takizawa , a. fujimori , m. kubota , k. ono , h. eisaki , arxiv:1003.2269 ( 2010 ) .
c. castellani , c. r. natoli , and j. ranninger , phys .
b * 18 * , 4945 ( 1978 ) .
o. gunnarsson , o. k. andersen , o. jepsen , and j. zaanen , phys .
b * 39 * , 1708 ( 1989 ) . m. cococcioni and s. de gironcoli , phys . rev .
b * 71 * , 035105 ( 2005 ) .
r. j. o. mossanek , m. abbate , and a. fujimori , phys .
b * 74 * , 155127 ( 2006 ) .
r. j. o. mossanek , m. abbate , t. yoshida , a. fujimori , y. yoshida , n. shirakawa , h. eisaki , s. kohno , p. t. fonseca , and f. c. vicentin , phys .
b * 79 * , 033104 ( 2009 ) .
m. taguchi , a. chainani , n. kamakura , k. horiba , y. takata , m. yabashi , k. tamasaku , y. nishino , d. miwa , t. ishikawa , s. shin , e. ikenaga , t. yokoya , k. kobayashi , t. mochiku , k. hirata , and k. motoya , phys .
b * 71 * , 155102 ( 2005 ) .
s. biermann , f. aryasetiawan , a. georges , phys .
* 90 * , 086402 ( 2003 ) . | dynamical screening of the coulomb interactions in correlated electron systems results in a low - energy effective problem with a dynamical hubbard interaction @xmath0 .
we propose a green s function ansatz for the anderson impurity problem with retarded interactions , in which the green s function factorizes into a contribution stemming from an effective static - u problem and a bosonic high - energy part introducing collective plasmon excitations .
our approach relies on the scale separation of the low - energy properties , related to the instantaneous static @xmath1 , from the intermediate to high energy features originating from the retarded part of the interaction .
we argue that for correlated materials where retarded interactions arise from downfolding higher - energy degrees of freedom , the characteristic frequencies are typically in the antiadiabatic regime . in this case ,
accurate approximations to the bosonic factor are relatively easy to construct , with the most simple being the boson factor of the dynamical atomic limit problem .
we benchmark the quality of our method against numerically exact continuous time quantum monte carlo results for the anderson - holstein model both , at half- and quarter - filling .
furthermore we study the mott transition within the hubbard - holstein model within extended dynamical mean field theory .
finally , we apply our technique to a realistic three - band hamiltonian for 3 .
we show that our approach reproduces both , the effective mass renormalization and the position of the lower hubbard band by means of a dynamically screened @xmath1 , previously determined _ ab - initio _ within the constrained random phase approximation .
our approach could also be used within schemes beyond dynamical mean field theory , opening a quite general way of describing satellites and plasmon excitations in correlated materials . |
let us consider the action of a very general class of cosmological models with a non - minimal coupling @xmath0 the multiplet of scalar fields @xmath1 has a rigid internal @xmath2 symmetry .
internal indices are raised and lowered by the constant metric @xmath3 .
representatives of this class are parametrized by different choices of the functions @xmath4 , @xmath5 and @xmath6 and the number of scalar components @xmath7 . in order not to spoil the internal @xmath2 symmetry ,
@xmath4 , @xmath5 and @xmath6 must be ultra - local functions of the modulus @xmath8 . in @xcite ,
the divergent part of the one - loop effective action for this class of models has been calculated in a closed form . due to its universality , ( [ onaction ] )
covers many important cosmological models , so that the results of @xcite immediately yield the one - loop corrections to the desired model . in the following ,
we will make use of these results several times .
the basic idea of higgs inflation is that the standard model ( sm ) higgs boson and the inflaton are one and the same scalar particle . in order to ensure consistency with observational data ,
it is necessary to assume a non - minimal coupling @xmath9 to gravity with a large coupling constant @xmath10 .
the tree - level graviton - higgs sector of the non - minimal higgs inflation model is described by ( [ onaction ] ) for the choices @xmath11 the matter sector of the model is given by the interaction part of the sm and can be described schematically as @xmath12 with the sums extending over scalar fields @xmath13 , vector gauge fields @xmath14 and dirac spinors @xmath15 .
the @xmath16 , @xmath17 and @xmath18 are the corresponding scalar , gauge and yukawa couplings .
this sector is dominated by the heavy masses @xmath19 of the @xmath20 and @xmath21 bosons and the yukawa top - quark @xmath22 .
a key feature of this model is a suppression mechanism that is induced by the non - minimal coupling and can be parametrized by the suppression function @xmath23 with `` prime '' meaning differentiation with respect to @xmath24 . for high energies
@xmath25 , the suppression function ( [ suppressionfunction ] ) behaves as @xmath26 .
the predictions derived in this model depend sensitively on the quantum corrections . using the one - loop results of @xcite with the choices
( [ treelevelnmhcouplings ] ) , the one - loop corrections to ( [ treelevelnmhcouplings ] ) are given by @xcite @xmath27 it is important to note that only the propagator of the radial higgs mode is suppressed by ( [ suppressionfunction ] ) .
the angular goldstone modes that run in the loop do not feel this suppression . for high energies , the logarithmic prefactors
@xmath28 , @xmath29 and @xmath30 therefore only receive goldstone but no higgs contributions .
neglecting graviton loops ] and expanding in inverse powers of @xmath31 , leads to @xmath32 in the expression for the anomalous scaling @xmath33 , the contribution of ordinary matter ( [ masses ] ) to the effective potential has also been included . in order to establish contact to the usual slow - roll formalism of inflation
, we perform a transformation from the original field variables - denoted jordan frame ( jf ) parametrization - to the so called einstein frame ( ef ) .
the name ef derives from the fact that expressed in this field variables , the action formally resembles the situation of general relativity ( gr ) minimally coupled to a scalar field .
the field transformation from the jf to the ef involves a conformal transformation of the metric field , a non - linear transformation of the scalar field and a re - definition of the potential ( ef quantities are denoted by a hat ) @xmath34 all cosmological parameters can be expressed in terms of the ef effective potential and derivatives thereof @xmath35 moreover , it turns out that their behaviour is essentially determined by one single quantity - the inflationary anomalous scaling @xmath36 .
the extra term @xmath37 is due to the quantum goldstone contributions contained in @xmath38 .
however , calculating the quantum corrections is not sufficient .
since we intend to connect two energy scale that are separated by many orders of magnitude , we have to take into account the dependence of the coupling constants on the energy scale .
thus , we must calculate the beta functions and evaluate the renormalization group ( rg ) flow of the couplings from the electroweak scale @xmath39 up to the high energies @xmath25 during inflation @xcite .
the beta functions form a complicated system of coupled ordinary differential equations that can only be solved numerically @xcite .
once this is done , we can express all cosmological parameters in terms of these couplings evaluated at inflationary energy scales .
the most important result is the spectral index as a function of the higgs mass - shown in fig .
[ spectralindex ] - that leads to a constraint of the higgs mass @xcite @xmath40 as a function of the higgs mass @xmath41 for different values of the top - quark mass @xmath42 . ]
the announcement of the lhc suggests a higgs mass around @xmath43 gev .
although the allowed window ( [ higgsmassrange ] ) does not contain this value , it still comes remarkably close to it .. only the combination of many features including the rg running leads to values of @xmath41 close to the measured one .
] moreover , the numerical result can be improved by including higher loops and ultimately leads to a shift of about @xmath44 gev bringing the lower bound of @xmath41 numerically very close to the measured value of @xmath41 , see e.g. @xcite .. as can be seen e.g. in fig .
( [ spectralindex ] ) , already the uncertainty in the measured value of the top - quark mass alone can shift the lower bound about @xmath45 gev . ]
despite the appealing nature of this unified scenario and the encouraging numerical results , there still remain conceptual problems .
some of these problems were addressed in @xcite , but probably the most fundamental and conceptually difficult problem is associated with the frame dependence of quantum corrections . while it is rather easy to see that the two formulations in the jf and ef are _ mathematically _ equivalent at the level of the classical action and the equations of motion , we are interested in the question whether this equivalence still holds at the quantum level .
the problem is more general and arises in all theories with a non - minimal coupling , but it has in particular consequences for the non - minimal higgs inflation model as the beta functions that determine the rg flow are derived by the effective action . considering only the first quantum corrections , the question of frame dependence
can be paraphrased as the question whether the following diagram does commute or not .
( 0,0)(6,0 ) ; ( 0,-3)(6,-3 ) ; ( -0.2,-2.8)(-0.20,-0.2 ) ; ( 6.2,-2.8)(6.2,-0.2 ) ; ( -1,-1.5)(7,-1.5 ) ; ( a ) at(-0.3,0.2)@xmath46 ; ( a ) at(6.6,0.2)@xmath47 ; ( a ) at(-1,-3.2)@xmath48 ; ( a ) at(7,-3.2)@xmath49 ; ( a ) at(3,-1.2)``quantization '' ; ( a ) at(3,0.3 ) ; ( a ) at(3,-3.3)field parametrization ; ( 1.1,-1.2)(1.1,-1.8 ) ; ( 4.9,-1.2)(4.9,-1.8 ) ; in order to answer this question , we explicitly calculated the one - loop effective action in both parametrizations .
then we expressed the ef effective action in the jf parametrization and compared it with the jf effective action .
we found that both results do not coincide @xcite .
in order to understand the origin of this frame dependence , we have to embed our cosmological model into the more general setup of field theory .
already in the mid - eighties , gregori a. vilkovisky proposed a solution to the problem of gauge and parametrization dependence of the effective action @xcite .
we will roughly sketch the main idea and then discuss the relevance for our cosmological problem .
let us adopt the condensed dewitt notation , where discrete internal and space - time indices as well as the continuous space - time point are collected in one single index @xmath50 .
we can then describe a model by a single generalized field @xmath51 with the understanding that the einstein summation convention is extended to include integration over space - time .
when we think of the configuration space of fields @xmath52 as differentiable manifold , we can think of @xmath53 as a point in @xmath52 . in this picture ,
different parametrizations @xmath51 and @xmath54 that describe the same point simply correspond to different choices of coordinate systems in @xmath52 .
we would like to calculate the effective action and follow the standard procedure .
we start with the feynman path integral @xmath55=\int{\cal d}\phi\,e^{i(s[\phi]+j_{i}\phi^{i})}$ ] which is a functional of the source @xmath56 .
the functional @xmath57 $ ] that generates all _ connected _ greens functions is defined by @xmath58}$ ] .
mean field _ is defined as @xmath59}{\delta j^{k}}$ ] .
the effective action can then be obtained as functional legendre transformation @xmath60=w[j]-j_{k}\langle\phi^{k}\rangle$ ] and leads to the exact equation for the full effective action @xmath61}=\int{\cal d}\phi\,\text{e}^{i\,\left\{s[\phi]-\frac{\delta\gamma[\langle\phi\rangle]}{\delta\langle\phi\rangle^{k}}\,\left(\phi^{k}-\langle\phi\rangle^{k}\right)\right\}}\label{effactfull}\end{aligned}\ ] ] this equation can be solved iteratively in powers of @xmath62 and yields at first order the one - loop contribution @xmath63 vilkovisky observed that the term @xmath64 in ( [ effactfull ] ) is a _ coordinate difference _ with respect to configuration space and has no _ geometrical _ meaning .
he therefore proposed to replace it by a geometrically meaningful quantity .
he borrowed the concept of the `` world function ''
@xmath65 that was originally introduced by synge @xcite in the space - time context and lifted it up to the configuration space : @xmath66=(\text{geod .
between $ \phi$ and $ \langle\phi\rangle$})^2\;.\end{aligned}\ ] ] ( 1,0 ) .. controls + ( 40:0.5 cm ) and + ( 120:0.75 cm ) .. ( 3,+0.25 ) ; ( 1,0.4)(1.2,0.2 ) ; ( a ) at(0.5,0.55)@xmath67 $ ] ; ( a ) at(1,-0.25)@xmath68 ; ( a ) at(3,0.0)@xmath53 ; ( 3,0.25)(3.3,-0.15 ) ; ( 3.2,0.1)(3.4,0.15 ) ; ( a ) at(4.1,0.15)@xmath69 $ ] ; ( 1,0 ) circle ( 0.75pt ) ; ( 3,+0.25 ) circle ( 0.75pt ) ; in order to render the formalism covariant , the coordinate difference is replaced by the geometrical meaningful covariant derivative of the world function @xmath70\;.\end{aligned}\ ] ] at the one loop - level ( [ effactoneloop ] ) , the covariant reformulation leads to a replacement of the `` partial '' functional derivatives by covariant ones @xmath71 vilkovisky proposed some physical assumptions that would fix the configuration space metric @xmath72 and the configuration space connection @xmath73 and would lead to a `` unique '' effective action .
regardless of whether we are willing to accept these assumptions , there are several conclusions - independent of these assumptions - that can be drawn from the covariant reformulation and help to shed some light onto the cosmological debate .
first of all , we notice that jf and ef are just two parametrizations among infinitely many others .
it is also clear that the cosmological debate is just a very special case of the more general problem of parametrization dependence of the effective action .
the analysis of the previous section shows why the off - shell results of the naive formulation of the effective action calculated in two different parametrizations do not coincide .
moreover , from ( [ covdiffaction ] ) we also see why the one - loop on - shell effective action is independent of the parametrization . )
we have at no point made use of the equations of motion . ] on - shell , the extra term in ( [ covdiffaction ] ) vanishes _ independently _ of @xmath73 , which is in agreement with the on - shell theorems that prove the parametrization independence of the s - matrix .
having detected the origin of the frame dependence as the lack of covariance of the _ mathematical _ formalism , we can now turn to the discussion of the _ physical _ consequences . in the aforementioned cosmological debate often _
physical _ arguments are given in favour or against one or the other frame . a popular argument that should serve to support the jf as the `` real '' or `` physical '' frame is the claim that we measure real physical distances and time intervals with the jordan frame metric .
, this is only possible with respect to another dimensionful reference quantity @xmath74 .
this reference quantity @xmath74 is build from units like kilogram , meter , seconds etc . in the cosmological context of the jordan frame the effective gravitational constant depends on @xmath75 . since @xmath24 is a scalar function that depends on space @xmath76 and time @xmath77 , it leads to a varying @xmath78 . consider for simplicity a homogeneous scalar field @xmath79 and imagine we would perform a cavendish - type experiment at two different times @xmath77 and @xmath80 in order to determine whether @xmath78 is time dependent or not .
suppose that we would find @xmath81 .
how can we say whether the gravitational constant @xmath82 has changed or whether the units ( or both ) have changed ?
we simply can not !
all we can say is that the _ ratio _ @xmath83 has changed with time . among the infinitely many parametrizations , jf and ef are distinguished in the sense that they correspond to the extreme cases where either @xmath78 or @xmath74 are constant . in the jf @xmath78
is varying , while @xmath74 is constant and vice versa in the ef .
] however , from the above considerations it should be clear once and for all that _ there is no distinguished physical frame_. it is pointless to talk about the _ physical _ meaning of one or the other frame in the same sense as it is meaningless to ask whether it is more physical to describe a mechanical system in spherical or cartesian coordinates . if we describe the same physical theory , the result will of course not depend on the parametrization as long as the formalism is covariant .
some parametrizations may be preferred in the sense that they are more adapted to the symmetries of the underlying theory , but nature certainly does not care about how we parametrize her .
we thank a. o. barvinsky , c. kiefer and a. a. starobinsky for numerous fruitful discussions . c. s. was partially supported by the erc grant 277570-digt .
a. k. acknowledges support by the rfbr grant 11 - 02 - 00643 . | we investigate a very general class of cosmological models with scalar fields non - minimally coupled to gravity .
a particular representative in this class is given by the non - minimal higgs inflation model in which the standard model higgs boson and the inflaton are described by one and the same scalar particle .
while the predictions of the non - minimal higgs inflation scenario come numerically remarkably close to the recently discovered mass of the higgs boson , there remains a conceptual problem in this model that is associated with the choice of the cosmological frame . while the classical theory is independent of this choice , we find by an explicit calculation that already the first quantum corrections induce a frame dependence . we give a geometrical explanation of this frame dependence by embedding it into a more general field theoretical context . from this analysis ,
some conceptional points in the long lasting cosmological debate : `` jordan frame vs. einstein frame '' become more transparent and in principle can be resolved in a natural way .
address = school of mathematical sciences , university of nottingham university park , nottingham , ng7 2rd , uk address = dipartimento di fisica e astronomia and infn , via irnerio 46 , 40126 bologna , italy , altaddress = l.d .
landau institute for theoretical physics of the russian academy of sciences , kosygin str . 2 , 119334 moscow , russia |
local changes in the physical conditions in the solar atmosphere can cause substantial changes in the parameters of observed absorption lines .
in such a situation , the spectral line is said to respond to the structural inhomogeneities which have appeared in the line formation region , and the stronger is the response , the more sensitive is the line to changes in the medium .
comprehensive information on the sensitivity of lines to different atmospheric parameters is essential for solving many problems in the spectrum analysis .
caccin et al . made a fundamental contribution to the problem using the sensitivity function @xcite and the sensitivity coefficients @xcite , they evaluated the changes in the central depth , halfwidth , and equivalent width of several spectral lines caused by changes in the temperature and microturbulent velocity .
unfortunately , the authors have not analyzed the line sensitivity measure introduced by them , and the sensitivity coefficients did not find wide use because of apparently cumbersome calculations .
when the sensitivity coefficients are calculated , a variation in a model atmosphere parameter requires that the whole model be recalculated . besides that it is difficult to measure the line sensitivity , since the atmospheric parameters have a combined effect on the line profile and it is not simple to determine the sensitivity to one of them .
studying the contribution and response functions , we have concluded that it is possible to determine the line sensitivity through the use of the response functions by calculating a set of sensitivity parameters .
this way is much more handy as regards calculations though it also needs much machine time .
the prime objective of this study is to develop a technique for calculating the sensitivity parameters based on the response functions , investigate the sensitivity of lines to different atmospheric parameters , and find a way to separating those lines which are highly sensitive to one of the atmospheric parameters and have a low sensitivity to other parameters .
the first section describes the algorithm for calculating the sensitivity parameters for absorption lines which form under the lte conditions ; the second section presents the results for fraunhofer lines ; the third section deals with the scope for application of the sensitivity parameters to the diagnostics of the photosphere ; and in conclusion we give principal inferences and recommendations for choosing the necessary lines .
the calculation of line sensitivity parameters is based on the idea of response functions , which has been developed well enough by many authors @xcite . the literature on the response functions was briefly reviewed by demidov @xcite .
recall that the response function describes the magnitude of response of a line to a local change in an atmospheric parameter at each point in the atmosphere along the line of sight .
it depends on the kind of disturbance and is a function of the wavelength and optical depth .
it is called the emission or the depression response function depending on whether it is calculated for an emission or a depression . the function maxima point to the region in the atmosphere where the emerging emission or depression in the line is the most sensitive to variations in the atmospheric parameters . to estimate the sensitivity of absorption lines
, we have chosen the depression response functions proposed for the first time in @xcite .
a variation in the observed line depression , @xmath1 , which results from a local disturbance of some atmospheric parameter is determined as @xcite @xmath2 here @xmath3 is the line depression ; @xmath4 is the response function of the depression ; @xmath5 is the atmospheric parameter which undergoes the disturbance @xmath6 ; @xmath7 is the optical depth in the logarithmic scale ( @xmath8 where @xmath9 is the optical depth at the wavelength @xmath10 nm ) ; @xmath11 is the distance to the line center .
if @xmath6 is not zero in a thin layer @xmath12 in the atmosphere close to the depth @xmath7 , the following approximate equality is true for the depression response function : @xmath13 it follows from ( 1 ) and ( 2 ) that the response function defines the rate of variation in the line depression @xmath14 at every point @xmath7 in the atmosphere relative to the rate of variation in the parameter @xmath5 which undergoes the local disturbance @xmath6 .
if we integrate the response function with respect to @xmath7 and divide it by the value of the `` emerging '' depression @xmath15 , we obtain a dimensionless quantity which is a variation in the observed depression , @xmath16 , relative to the local variation @xmath6 .
we denote this quantity by @xmath17 and call it the parameter , or the indicator , of the line depression sensitivity .
it is calculated with the expression @xmath18 as @xmath19 in ( 1 ) is always taken positive , the sign of the sensitivity parameter corresponds to the sign of @xmath20 .
it should be remembered therefore : if the sensitivity parameter @xmath21 is positive , the depression @xmath22 is growing when @xmath5 increases by @xmath23 , and if @xmath21 is negative , @xmath22 is diminishing . by analogy with ( 3 ) , we may define the sensitivity indicator for the line equivalent width : @xmath24 where @xmath25 is the integral response function . according to [ 13 ] , it has the form @xmath26 to calculate the sensitivity parameters with ( 3 ) , ( 4 ) , we have first to calculate reliably the depression response functions . for the lte photospheric lines ,
the expression for the depression response function proposed in @xcite may be used : @xmath27 \exp \left ( -{\frac { \tau _ r } { \mu } } \right ) , \ ] ] where @xmath28
@xmath29 here @xmath30 ; @xmath31 , @xmath32 and @xmath33 , @xmath34 are the absorption coefficients and optical depths in the line ( @xmath35 ) and in the continuum ( @xmath36 ) ; @xmath37 is the planck function .
since the response function depends on the kind of disturbance , i.e. , on @xmath5 , it is necessary to calculate the response function for each atmospheric parameter whose effect on the line profile we want to estimate .
let the temperature variation @xmath38 occur in the solar atmosphere , then the depression function of the response to this temperature variation , according to ( 6 ) , is @xmath39 \exp \left ( { -\frac{\tau _ r } { \mu } } \right ) .\ ] ] similarly we get the expressions for the depression functions of the response to the variations in the gas pressure , @xmath40 , @xmath41 \exp \left ( { -\frac{\tau_r } { \mu } } \right ) , \ ] ] and in the microturbulent velocity , @xmath42 , @xmath43 } { dv}+1 } \right)\times\ ] ] @xmath44 \exp \left ( -\frac{\tau_r } { \mu } \right ) .\ ] ] the algorithm for calculating the sensitivity parameters of the fraunhofer lines is based on formulae ( 7)(9 ) .
the calculation program is based on the spansat program for the fraunhofer line calculations @xcite .
it is well known that the temperature sensitivity of a line depends on the line excitation potential , the sensitivity decreasing with increasing potential .
just this rule is commonly used by observers when they choose lines .
however no investigations have thus far been carried out concerning the line sensitivity to other atmospheric parameters and the dependence of the sensitivity on the line atomic parameters such as the wavelength @xmath45 , ionization potential , atomic weight , as well as on the abundance and on the line central depth , equivalent width , and half - width .
the employment of the above technique for calculating the line sensitivity parameters allowed us to elucidate the problem . [ cols="^,<,>,>,>,>,>,>,>,>,>",options="header " , ] to find lines which are the most sensitive to one atmospheric parameter and at the same time have low sensitivity to other parameters , we calculated and compared the relative variations of line parameters , @xmath46 , @xmath47 , @xmath48 , rather than sensitivity indicators , for actually observed fluctuations in the temperature ( 2% ) , gas pressure ( 30% ) , and microturbulent velocity ( 50% ) .
this way proved to be appropriate for finding the necessary lines . here
we give as an example the wavelengths of typical highly sensitive lines in respect to their relative sensitivity .
the most responsive to temperature are the lines 670.778 li i , 657.279 ca i , 623.936 sc i , 546.049 ti i , 482.745 v i , 511.248 cr i , 543.253 mn i , 512.768 fe i , 662.502 fe i , 673.952 fe i , 612.026 fe i , 635.384 fe i , 671.032 fe i , 680.186 fe i , 685.164 fe i , 401.109 co i , 412.830 y i ; the most responsive to gas pressure are the lines 477.588 c i , 505.214 c i , 658.762 c i , 711.317 c i , 833.516 c i , 824.252 n i , 777.196 o i , 777.539 o i , 926.590 o i , 871.783 mg i , 789.637 mg ii , 875.202 si i , 634.709 si ii , 891.207 ca ii , 989.070 ca ii , 463.407 cr ii , 463.531 fe ii , 624.756 fe ii ; and to microturbulent velocity : 408.295 mn i , 438.925 fe i , 448.974 fe i , 508.334 fe i , 512.735 fe i , 514.174 fe i , 522.553 fe i , 550.678 fe i , 625.256 fe i , 633.534 fe i , 450.828 fe ii , 402.090 co i ,
508.111 ni i , 464.866 ni i , 510.554 cu i , 481.053 zn i , 460.733 sr i , 488.369 y ii , 420.898 zr ii , 455.404 ba ii , 649.691 ba ii . analyzing the starting data of these lines and the calculated values of @xmath46 , @xmath47 , @xmath49
, we can conclude that the most responsive to temperature are the lines with excitation potentials from 0 to 2 ev , central depths up to 0.35 , equivalent widths to 3 pm .
relative variations of the equivalent widths of these lines are 2530% when the temperature changes by 2% , while they are 27% when the gas pressure changes by 30% and microturbulent velocity by 50% . moderate - strong lines of light atoms with @xmath50 ev are highly responsive to gas pressure . for them
, @xmath51 is as much as 2548% when @xmath52 changes by 30% and is only 26% when @xmath53 changes by 2% and @xmath54 by 50% .
the equivalent widths and half - width depths of strong lines of heavy atoms with @xmath5512 pm and @xmath566 pm are very sensitive to the microturbulent velocity .
for these lines , @xmath47 amounts to 1118% and @xmath57 runs as high as 4570% when @xmath54 changes by 50% , and it is 2% when @xmath52 changes by 30% and @xmath53 by 2% .
as we could make sure , the sensitivity of every line depends on numerous atom and line parameters as well as on the model atmosphere parameters .
it is measured by a diversity of sensitivity indicators with which it is possible to determine the atmospheric parameter producing the strongest response of the line and to estimate the magnitude of the response .
it should be taken into consideration which sensitivity , the absolute or relative one , is more suitable for the particular problem .
our calculations reveal that the lines which form under the lte conditions in the solar atmosphere are the most responsive to temperature variations .
evidence of this is found , for example , in the values of parameters of relative sensitivity to temperature , gas pressure , and microturbulent velocity which attain 15 , 1.5 , 1.5 , respectively ( see table 1 ) .
even minor temperature fluctuations , of the order of 1% , can be observed in the most responsive lines .
fluctuations in the density of matter or in the gas pressure as well as variations in the microturbulent velocity change the line only slightly as compared to temperature . using the lines most responsive to the gas pressure , it is possible to detect only 1015% fluctuations of the pressure in the line formation region .
variations in the microturbulent velocity can be measured from variations in the half - widths and equivalent widths of highly sensitive lines if @xmath58 is about 30% .
direct calculations of the response of line depression to variations in atmospheric parameters for disturbed and undisturbed model atmospheres , confirm that the sensitivity indicators can be used for the solar atmosphere diagnostics in those cases when fluctuations in @xmath53 , @xmath52 , and @xmath54 do not exceed 8% , 50% , and 100% , respectively .
it is better to use the response functions themselves when fine spectral analysis problems are solved .
a detailed analysis of the shape of response functions can give an additional valuable inftyormation .
the results of our analysis show that the sensitivity indicators proposed for measuring the fraunhofer line response do not lack a physical meaning .
we think it reasonable to include the sensitivity parameters in the initial data for spectral lines together with the central depth , equivalent width , and other principal characteristics when the data bases such as `` the fraunhofer spectrum '' @xcite are compiled .
o. g. badalyan and v. n. obridko , `` on intensity variations in the photospheric lines and continuous spectrum during small variations in physical parameters , '' astron .
3 , pp . 561567 , 1975 a.s .
gadun and v. a. sheminova , spansat : program for calculating absorption line profiles in stellar atmospheres in the lte approximation [ in russian ] , kiev , 1988 ( inst .
theoretical physics , as ukraine preprint no itf-88 - 87p ) .
a. s. gadun , m. m. sosonkina , and v. a. sheminova , `` the fraunhofer solar spectrum data bank , '' kinematika i fizika nebes .
tel [ kinematics and physics of celestial bodies ] , vol . 8 , no .
2 , pp . 8082 , 1992 .
v. a. sheminova , `` depths of formation of magnetically sensitive absorption lines in the solar atmosphere , '' kinematika i fizika nebes .
tel [ kinematics and physics of celestial bodies ] , vol . 8 , no .
4462 , 1992 . | parameters are proposed for measuring the sensitivity of fraunhofer lines to the physical conditions in the solar atmosphere .
the parameters are calculated based on depression response functions in the lte approximation .
the sensitivity of lines to the temperature , gas pressure , and microturbulent velocity depending on the line and atomic parameters is investigated .
the greatest relative temperature sensitivity is shown by weak lines , while the greatest absolute sensitivity is displayed by moderate lines of abundant heavy atoms with low ionization and excitation potentials .
the excitation potential and line strength are the crucial factors for the temperature sensitivity .
the highest pressure sensitivity is observed for moderate lines of light atoms with very high excitation potentials ( exceeding 6 ev ) , and strong photospheric lines ( 8pm@xmath014pm ) of heavy atoms are the most responsive to the microturbulent velocity .
the sensitivity parameters can be also used to advantage for physical diagnostics of the photosphere when the temperature , pressure , and microturbulent velocity fluctuations are no more than 8% , 50% , and 100% , respectively .
[ multiblock footnote omitted ] main astronomical observatory , national academy of sciences of ukraine + zabolotnoho 27 , 03689 kyiv , ukraine + e - mail : [email protected] |
over the past decades , the conventional two - dimensional electron system ( 2des ) in semiconductor heterostructures in a strong magnetic field has been studied .
it is found that electron - electron interaction brings various phases [ e.g. , fractional quantum hall effect ( fqhe ) states @xcite and charge density wave ( cdw ) states @xcite ] .
similarly , graphene , @xcite a flat sheet of carbons with a honeycomb lattice exfoliated from a graphite in 2004 , @xcite is under intensive investigation as a new 2des whose electron has a valley degree of freedom ( @xmath0 ) and a linear dispersion .
exhibition of fqhe @xcite and cdw @xcite states in the new 2des has been expected ; fqhe was recently observed in a single - layer graphene ( slg ) .
@xcite furthermore , few - stacked graphenes have recently attracted attention due to their intriguing properties : a band structure tunable by the number of layers and their stacking sequence , and a band gap controllable by a perpendicular electrical field . @xcite we focus on bernal bilayer and abc - stacked multilayer graphenes , which have chiral electrons concentrated on outer layers in low - energy states . in this paper , the cdw ground state of the 2des in bilayer graphene ( blg ) and multilayer graphene ( @xmath1-lg , @xmath2 ) in a strong magnetic field
is studied with the hartree - fock approximation . as candidates of the ground state , the electron ( hole ) wigner crystal , electron ( hole ) bubble crystal , and skyrme crystal
are considered .
a skyrme crystal is the state with a topological texture of valley pseudospins .
it has the lowest energy around a filling @xmath3 in slg .
@xcite our mean - field analysis predicts that the valley skyrme crystal also has low energy in blg and @xmath1-lg ( @xmath4 ) at the first excited landau level when the interlayer distance @xmath5 is assumed to be infinitesimal .
this paper is organized as follows . in the next section
, we set up a low - energy effective hamiltonian in the presence of a magnetic field for slg , blg , and @xmath1-lg .
then , the hartree - fock approximation is applied to the system of interacting electrons and self - consistent equations are derived . in sec .
[ sec : cdw ] , cdw states are introduced , including skyrme crystals ( more properly , the meron and meron pair crystal ) . as a preliminary calculation ,
the energy of a skyrmion ( antiskyrmion ) pair excitation at a @xmath3 ferromagnetic state is evaluated and compared with that of a separated particle - hole excitation to find the condition where the skyrme crystal has low energy . in sec .
[ sec : results ] , numerical results are presented at the landau level where the skyrme crystal is expected . in sec .
[ sec : discussion ] , the validity of the results and a relation to experiments are discussed .
for a single - layer graphene ( slg ) , the low - energy effective hamiltonian around the @xmath6 valley is given by @xmath7 which has an electron with a linear dispersion .
@xcite here , @xmath8 is fermi velocity , and @xmath9 and @xmath10 are pauli matrices acting on sublattice ( a , b ) space .
the hamiltonian for the @xmath11 valley is @xmath12 . in a perpendicular magnetic field @xmath13 ,
using a magnetic ladder operator @xmath14 ( @xmath15 ) , the single - particle hamiltonian is written as @xmath16 where @xmath17 is magnetic length .
the eigenenergies are @xmath18 where a positive ( negative ) sign is taken as the electron ( hole ) state .
the corresponding eigenstates have the form @xmath19 for the @xmath6 valley . in the landau gauge , @xmath20 is given by @xmath21 where @xmath22 is the length of the system , @xmath23 is the hermite polynomial , and @xmath24 is the guiding center coordinate
. a macroscopic number of states with different @xmath25 are degenerated at one landau level .
a single - particle density on each sublattice @xmath26 becomes broader as index @xmath27 increases .
the states with broad density induce a bubble cdw and lose the profit to form skyrmions ( see sec . [
sec : cdw ] ) .
bernal - stacked bilayer graphene ( blg ) and abc - stacked multilayer graphene @xcite ( @xmath1-lg ) have low - energy quasiparticles with a dispersion @xmath28 , where @xmath1 ( @xmath29 ) is the number of layers .
@xcite in two adjacent layers , blg and @xmath1-lg have interlayer hoppings @xmath30 between the sublattice @xmath31 in the lower layer and the sublattice @xmath32 in the upper layer . in a magnetic field , the effective single - particle hamiltonian for blg and @xmath1-lg , which act on the outermost layer ( top , bottom ) space ,
have the same form @xmath33 where off - diagonal components represent @xmath1-times hopping via high - energy states in the inner layers .
its eigenenergy is @xmath34 where @xmath35 . at zero energy , @xmath1 landau levels are degenerated .
the corresponding eigenstate has the form @xmath36 for the @xmath6 valley .
notice that in the first excited state realized at @xmath37 , the upper component of the wave function is given by @xmath38 , which is the ground - state wave function of the ordinary 2@xmath5 electrons . at landau level @xmath39 ,
the valley degree of freedom coincides with the layer degrees of freedom .
if the layer degree of freedom in the @xmath1-lg model is regarded as the sublattice degree of freedom , eq .
( [ eq : mlgwf ] ) is formally identical to that of the slg model when @xmath40 . in the following
, the case of @xmath40 ( @xmath41 ) represents the model of slg ( blg ) .
although the electron state is mainly considered in the following , the hole state can be treated in the same way . in the present case of strong magnetic fields ,
electronic spins are completely polarized and the gap around the neutrality point is sufficiently large due to exchange enhancement ; thus we take only one spin component and ignore the effect of landau - level transitions . the interaction between the electrons in different layers
is weakened by an interlayer distance @xmath5 ( @xmath42 for blg ) .
when @xmath43 t , @xmath5 is quite small compared to magnetic length @xmath44 , so we consider the vanishing limit of interlayer distance as an approximate model . apart from a constant kinetic term , the hamiltonian for interacting electrons in @xmath1-lg ( @xmath45 ) is given by @xmath46 where the field operator @xmath47 is represented by @xmath48 and coulomb interaction is written as @xmath49 here , @xmath50 is the dielectric constant , @xmath51 represents valley @xmath6 and @xmath11 , respectively , @xmath52 is an annihilation operator of the electron at valley @xmath53 with guiding center @xmath25 .
a valley scattering term is relatively small when the cutoff @xmath54 is used , @xcite since @xmath55 ( @xmath56 is the lattice constant ) , so we can ignore it as far as the cdw with magnetic length scale is concerned .
then , the hamiltonian in @xmath57 space is written as @xmath58 @xmath59 @xmath60 where @xmath61 is the laguerre polynomial , and @xmath62 is defined by @xmath63 and @xmath64 .
when landau - level index @xmath27 is smaller than the number of layers @xmath1 , the valley degree of freedom coincides with that of the layers ( sublattices for slg ) , and @xmath65 and @xmath66 and @xmath67 are twice the amount of eq .
( [ eq : density ] ) . for the hartree - fock decoupling of @xmath68
, we assume the following order parameters : @xmath69 where @xmath70 is a reciprocal lattice vector of the cdw state
. then the hartree - fock hamiltonian is given by @xcite @xmath71 where @xmath72 .
the hartree - fock potential consists of a direct term @xmath73 @xmath74 and an exchange term @xmath75 @xmath76 here , @xmath77 is a bessel function of the first kind .
the real - space density of electrons at valley @xmath53 and layer ( sublattice ) @xmath78 is given by @xmath79 the filling factor at landau level @xmath27 is defined as @xmath80,\ ] ] where @xmath81 ( @xmath82 is a area of the system ) is the degeneracy of landau orbitals and @xmath83 is the number of electrons .
the factor 2 comes from the valley degree of freedom . to determine the order parameters self - consistently ,
the green s function method is used .
@xcite the single - particle matsubara green s function is defined by @xmath84 the fourier transformation @xmath85 relates to the order parameters @xmath86 by @xmath87 the equation of motion for @xmath88 is derived as @xmath89 from the heisenberg equation . here
, @xmath90 is the matsubara frequency for fermions , and the self - energy @xmath91 is given by @xmath92 to solve this self - consistent equation , we diagonalize the self - energy matrix @xmath93 where @xmath94 is the @xmath95th eigenvector with eigenvalue @xmath96 .
the order parameters are obtained from the eigenvectors and eigenvalues @xmath97 where @xmath98 is the fermi - dirac distribution function .
the chemical potential @xmath99 is determined from @xmath100 self - consistent equations are numerically calculated to yield the order parameters and the hartree - fock energy per particle for several cdw states introduced in sec .
[ sec : cdw ] .
the order parameter sum rule at zero temperature , @xcite extended to the case of valley degeneracy @xmath101 is easily derived from eq .
( [ eq : odprmt ] ) . here , @xmath102 is the contribution from valley @xmath53 electrons to the partial filling factor @xmath103 .
this relation is used to check convergence of the results .
it is useful to map the valley degrees of freedom to a pseudospin .
@xcite in this language , the components of the pseudospin vector density @xmath104 are defined by @xmath105 @xmath106 @xmath107 in this paper , states with a topological pseudospin texture , which is called a skyrmion , @xcite are considered .
a skyrmion is a kind of spin texture usually used to describe a magnetic order and was first introduced in hadron physics .
@xcite for conventional and slg 2des , a state with aligned skyrmions , which is called a skyrme crystal , has been shown theoretically to be the ground state around @xmath3 .
@xcite the excitation energy of a skyrmion in a ferromagnetic uniform state at @xmath108 is evaluated to find the conditions on which the skyrme crystal is preferred .
@xcite when the interlayer distance @xmath109 , the energy of a skyrmion ( antiskyrmion ) pair excitation @xmath110 and the energy of a widely separated particle - hole pair excitation @xmath111 are given by @xmath112 ^2 e^{-q^2 l_b^2 /2 } dq,\ ] ] @xmath113 ^2
e^{-q^2 l_b^2 /2 } dq , \ ] ] where the form factor @xmath114 has the form @xmath115 & ( n \ge m ) \\ l_n \left ( \frac{q^2l_b^2}{2 } \right ) & ( n <
m ) \end{cases}.\ ] ] the ratio of the two energies of the pair excitations is the same as that of single - particle excitations , from particle - hole symmetry . for conventional , slg , blg , and @xmath1-lg ( @xmath116 ) 2dess , these energies at landau level @xmath117 are presented in table [ tableexcitation ] .
it shows that skyrmion excitation is favored ( 1 ) at @xmath118 in all systems as single - particle wave functions being identical to conventional one , ( 2 ) at @xmath119 in slg , and ( 3 ) at @xmath120 in @xmath1-lg ( @xmath4 ) .
thus we can expect that the skyrme crystal becomes the ground state in these situations around @xmath121 .
.hartree - fock quasiparticle and skyrmion ( antiskyrmion ) particle - hole excitation gaps ( in unit of @xmath122 ) at landau level @xmath27 for 2des in a conventional semiconductor structure ( @xmath123 and @xmath124 ) , slg ( @xmath125 and @xmath126 ) , blg ( @xmath127 and @xmath128 ) , tri - lg ( @xmath129 and @xmath130 ) , tetra - lg ( @xmath131 and @xmath132 ) , and penta - lg ( @xmath133 and @xmath134 ) . for conventional and slg 2des ,
the energies of the two excitations were compared by yang _
_ @xcite the situation where the skyrmion is favored is emphasized by thick `` @xmath135 '' .
[ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] the reason why a skyrmion can be a low - energy excitation is the following .
@xcite the ground state at @xmath108 is a pseudospin ferromagnetic liquid state .
when a hole is introduced in this state without flipping pseudospin of other electrons , the charge density of the hole is given by an eigenstate of the angular momentum , and is concentrated . on the other hand ,
when introduction of a hole is accompanied by pseudospin - flip of other electrons , many states with the same angular momentum are connected by the coulomb interaction , and the charge density of the hole has a wider distribution .
this connected quantum state corresponds to a skyrmion . when the charge is confined like a wave function @xmath136 given by eq .
( [ eq : singlewf ] ) , the formation of a skyrmion reduces the charge locality . at high landau level @xmath27 ,
however , a wave function @xmath137 is intrinsically broad , so the benefit to form skyrmions is lacking . in @xmath1-layered graphene ,
the spinor wave function has the localized component @xmath138 at landau level @xmath37 , so it can drive the system to form skyrmions .
when zeeman energy for valley pseudospins does not exist , a skyrmion splits into two merons ( half - skyrmions ) .
four textures of a meron are possible from the two direction at center and the two vorticities .
charge density wave states where electrons , holes , or merons form a triangular or square lattice structure are considered .
the order parameters @xmath139 are defined at points @xmath140,\
\ \text{for a triangular lattice},\ ] ] @xmath141 where @xmath95 and @xmath142 are integers .
the following states are assumed : 1 .
electron wigner crystal ( ewc ) and @xmath143-electron bubble crystal ( ebc@xmath143 ) : a triangular or square lattice with one or @xmath143 electrons per unit cell .
the fundamental length in @xmath57 space @xmath144 is determined from the condition that the cdw has @xmath143 electrons in a unit cell : @xmath145 . here , @xmath146 is the filling factor of electrons and @xmath147 is the area of a unit cell . the low - energy state is pseudospin ferromagnetic because of the pauli principle .
2 . hole wigner crystal ( hwc ) and @xmath143-hole bubble crystal ( hbc@xmath143 ) : a triangular or square lattice with one or @xmath143 holes per unit cell .
the fundamental length in @xmath57 space @xmath144 is determined from the condition that the cdw has @xmath143 holes in a unit cell : @xmath148 . here
, @xmath149 is the filling factor of holes and @xmath147 is the area of a unit cell .
the low - energy state is pseudospin ferromagnetic because of the pauli principle .
3 . meron crystal ( mc ) : a square lattice with four merons of charge @xmath150
( @xmath151 ) or @xmath152 ( @xmath153 ) per unit cell , equally spaced .
a meron pair is equivalent to one skyrmion , so mc can be seen as a state with two skyrmions per unit cell .
thus @xmath144 is determined from the condition that the cdw has @xmath143 ( @xmath154 ) skyrmions in a unit cell : @xmath148 .
the @xmath155 component of pseudospin density in real space and the vorticity alternate from one site to the next ( figs .
[ fig : mcvpsz ] and [ fig : mcvps2d ] ) .
the @xmath156 orientation of pseudospins has @xmath157 symmetry ( fig .
[ fig : mcvps2d ] ) .
the density distribution in real space is bipartite in layers .
meron pair crystal ( mpc ) : a triangular lattice with four merons per unit cell .
the merons are not equally spaced and bound into pairs .
the @xmath144 is determined from the same way as mc .
the energy of mpc is similar to mc , and has slightly lower energy in the low quasiparticle density regime ( close to @xmath108 ) in general .
we use the partial filling factor @xmath103 at landau level @xmath27 ; thus the total filling factor is given by @xmath158 for any - layered graphene .
assume that zeeman splitting is sufficiently large , so the phase diagram for @xmath159 $ ] is identical to that for @xmath160 $ ] .
furthermore , the hamiltonian has electron - hole symmetry around @xmath108 , so the phase diagram for @xmath161 is caught by alternating particles for @xmath162 to antiparticles . numerical calculation for mc and mpc
are done at @xmath163 , since too many wave vectors are needed to get well - converged solutions at @xmath164 . in the following results ,
the energies with an accuracy of @xmath165 are presented .
wigner and bubble crystals are calculated only in triangular symmetry ; a square lattice generally has higher energy than a triangular one in a low quasiparticle density regime .
it is difficult to get mc solutions close to @xmath166 , so we extrapolate the order parameters of the mc solutions in @xmath167 . to execute the extrapolation , the quantity @xmath168 is fitted by a quadratic curve .
the energies of the extrapolated mc states are represented in the following figures as a dotted line . it is noted that uncertainty remains in the extrapolation especially in the 4-lg case .
it comes from the relatively low validity of fitting the order parameters which have an inflection point near @xmath169 .
the energies of the valley - concentrated hole wigner crystal states which are quite accurately approximated by gaussian form order parameters @xmath170 are represented as `` ghwc '' in the following figures . in the vanishing interlayer distance limit , the ghwc state with @xmath171 is degenerated to the hwc solutions with @xmath172 ( figs .
[ g2l2high ] , [ g3l3high ] , [ g4l4high ] ) . in the following we show phase diagrams obtained by the present hf approximation in the whole range of @xmath103 .
it should be remarked that the true phase diagram should contain regions of incompressible liquid states that can not be obtained by the hf approximation .
thus the phase diagrams are partly incorrect
. however , hf calculation gives qualitatively correct results when the fractional quantum hall states do not appear .
this is established from comparisons with the results by the exact diagonalization method @xcite or the density matrix renormalization group method .
@xcite in this paper we focus on the possibility of meron crystals near @xmath121 , where the liquid states are not expected , but only the charge - ordered states compete .
therefore , the following discussion as to the realization of the meron crystal is reliable .
the hartree - fock ( hf ) phase diagram of the 2des in slg has been obtained and compared with that of the conventional one .
@xcite it is shown that skyrme crystals ( mc and mpc ) become the ground state around @xmath121 at landau levels @xmath118 and @xmath173 .
@xcite considering the excitation energies for slg ( table [ tableexcitation ] ) , it is also possible for the skyrme crystal phase to occur at @xmath174 and @xmath175 , but this has not been found yet in a mean - field calculation . although the same hf calculation had been done for slg , @xcite we executed additional checks and investigated the higher filling regime . at landau level @xmath118 ,
the phase diagram is the following : ewc for @xmath176 $ ] , hwc for @xmath177 $ ] , mc for @xmath178 $ ] , and mpc for @xmath179.$ ] @xcite at landau level @xmath180 , the phase diagram is the following : ewc for @xmath181 $ ] , hwc for @xmath182 $ ] , mc for @xmath183 $ ] , and mpc for @xmath184 $ ] .
@xcite the range of a skyrmionic ( mc or mpc ) phase is narrower than that of the @xmath118 case . at landau level @xmath174 ,
the phase diagram is the following : ewc for @xmath185 $ ] , ebc2 for @xmath186 $ ] , hbc2 for @xmath187 $ ] , and hwc for @xmath188 $ ] .
it is characteristic that 2-electron ( hole ) bubble crystals exist around @xmath189 .
the skyrmionic ground state is not seen in the range @xmath190 .
the extrapolating analysis , however , suggests that the hwc and mc state are almost degenerated in @xmath191 $ ] . at landau level @xmath192 ,
the phase diagram is the following : ewc for @xmath193 $ ] , ebc2 for @xmath194 $ ] , ebc3 for @xmath195 $ ] , hbc3 for @xmath196 $ ] , hbc2 for @xmath197 $ ] , and hwc for @xmath198 $ ] .
the skyrmionic state does not appear in @xmath199 .
although the mc and mpc solutions are not found in @xmath167 for slg at @xmath174 and @xmath175 , the skyrme crystal is expected to have lower energy in the immediate vicinity of @xmath200 from the analysis in sec .
[ subsec : excitation ] .
the hf calculation for degenerated zero - energy landau levels @xmath201 in blg suggests that the skyrme crystal states of real spin or orbital pseudospin occur .
@xcite in what follows , the results for the first excited landau level @xmath202 in blg are presented .
figure [ g2l2 ] shows the energies per electron for several crystal structures .
it shows the following sequence of ground states : ewc for @xmath203 $ ] , hwc for @xmath204 $ ] .
the bubble state does not appear unlike the phase diagram at landau level @xmath174 for slg .
figure [ g2l2high ] shows the energies in the area close to @xmath205 .
the extrapolated energy of the mc solutions have lower value than hwc for @xmath206 $ ] .
figure [ g3l3 ] shows the energies per electron for several crystal structures at landau level @xmath192 in tri - lg .
it shows the following sequence of ground states : ewc for @xmath207 $ ] , ebc2 for @xmath208 $ ] , hbc2 for @xmath209 $ ] , hwc for @xmath210 $ ] .
although the bubble states are found around @xmath211 , its range is narrower than that of the @xmath192 case in slg .
the skyrmionic ground state is not seen in the range @xmath199 .
figure [ g3l3high ] shows the energies near @xmath212 .
the extrapolated states of the mc solutions have an energy close to that of hwcs in the vicinity of @xmath213 .
figure [ g4l4 ] shows the energies per electron for various crystal structures at landau level @xmath214 for tetra - lg .
it shows the following sequence of ground states : ewc for @xmath215 $ ] , ebc2 for @xmath216 $ ] , hbc2 for @xmath217 $ ] , and hwc for @xmath218 $ ] .
the bubble states are found in the broad range @xmath219 $ ] .
the skyrmionic ground state is not seen in the range @xmath220 .
figure [ g4l4high ] shows the energies near @xmath221 .
the extrapolated states of the mc solutions have higher energy than that of hwcs in the vicinity of @xmath222 . as previously mentioned , however , the extrapolation method is no longer valid in tetra - lg . according to the analysis in sec .
[ subsec : excitation ] , the skyrme crystal is expected to have the lowest energy in the vicinity of @xmath223 .
the hartree - fock ( hf ) calculation suggests that the meron crystal ( mc ) phase appears at landau level @xmath174 in blg , and hole wigner crystal ( hwc ) and mc states are degenerated around @xmath108 at landau level @xmath174 in slg , @xmath192 in tri - lg in the case of vanishing interlayer distance .
our calculation strongly suggests that meron pair crystals ( mpcs ) will have lower energy in the immediate vicinity of @xmath121 for its triangular symmetry .
skyrme crystals ( mc and mpc ) have a charge distribution and collective mode different from that of hwc , so these states can be distinguished by transport properties @xcite and a microwave absorption spectrum .
@xcite furthermore , the cdws exist in outer layers , so the local density of states ( ldos ) can be measured in a spectroscopic manner .
@xcite in this paper , the unidirectional stripe phase is not considered .
the stripe phase , however , also will appear around @xmath224 in blg and @xmath1-lg , as slg @xcite and conventional 2des if fractional quantum hall states are not realized .
such a state , if exists , will be identified by anisotropic conduction . @xcite although we ignored the effects of disorder , a finite valley zeeman energy , and landau - level transitions , it is unclear how these affect degeneracy of hwc and mc around @xmath108 . in particular ,
landau - level transitions in blg and @xmath1-lg are larger than that of slg under a magnetic field @xmath225 t. the gap near a charge neutrality point is @xmath226 for slg , @xmath227 for blg , @xmath228 for tri - lg , and @xmath229 for tetra - lg .
the typical coulomb energy is @xmath230 . for slg ,
the ratio @xmath231 is independent of field @xmath232 . in this case
it is shown that the landau - level mixing does not change the cdw phase diagram ( except the skyrmionic crystal ) .
@xcite for blg and @xmath1-lg , high magnetic fields are needed to achieve a comparable ratio : @xmath233 for blg , @xmath234 for tri - lg , and @xmath235 for tetra - lg .
it has been pointed out that anisotropy in the pseudospin arises in the order of @xmath236 .
@xcite in the multilayer graphene , finite layer separation also brings anisotropy .
this anisotropy is quite small , since @xmath237 , but may have some effect when the energies of two phases are quite close .
we have done a calculation taking into account only the effect of @xmath238 as a preliminary investigation , and found that the finite @xmath5 is slightly unfavorable for the skyrme crystal .
however , to obtain a definite conclusion for the effect of finite @xmath5 , we need to take into account the effect of @xmath239 also .
such calculation is left for future investigation .
wallace , phys . rev . *
71 * , 622 ( 1947 ) .
j. c. slonczewski and p. r. weiss , phys .
rev . * 109 * , 272 ( 1958 ) .
g. w. semenoff , phys .
lett . * 53 * , 2449 ( 1984 ) .
f. d. m. haldane , phys .
lett . * 61 * , 2015 ( 1988 ) .
t. ando , j. phys .
jpn , * 74 * , 777 ( 2005 ) . n. m. r. peres , f. guinea , and a. h. castro neto , phys .
b * 73 * , 125411 ( 2006 ) .
r. ct and a. h. macdonald , phys .
lett . * 65 * , 2662 ( 1990 ) .
r. ct and a. h. macdonald , phys .
b * 44 * , 8759 ( 1991 ) .
r. ct , l. brey and a. h. macdonald , phys .
b * 46 * , 10239 ( 1992 ) . x. m. chen and j. j. quinn , phys . rev .
b * 45 * , 11054 ( 1992 ) . | the ground state of the two - dimensional electron systems in bernal bilayer and abc - stacked multilayer graphenes in the presence of a strong magnetic field is investigated with the hartree - fock approximation .
phase diagrams of the systems are obtained , focusing on charge density wave states including states with vortices of valley pseudospins ( called a skyrme crystal ) . the single - electron states in these stacked graphenes
are given by two - component wave functions .
that of the first excited landau level has the same component as the lowest landau level of the ordinary two - dimensional electrons . because of this
localized wave functions , the skyrme crystal has low energy in this first excited level up to four layers of graphene , when the inter - layer distance is assumed to be infinitesimal . at the same time , bubble crystals are suppressed , so the phase diagram is different from that of a single - layer graphene . |
most local moment magnets undergo a transition from a high temperature paramagnetic state to a magnetically long range ordered state below some critical temperature .
the ordered state is like a solid while the paramagnetic state is like the gaseous state .
the state analogous to a liquid is elusive in magnets .
that is because most magnetic solids have a unique state with the lowest energy and are able to freeze into that solid - like long range ordered state at sufficiently low temperatures .
if one is able to suppress this tendency to order , one could attain a liquid like state of spins where they would be strongly entangled and yet dynamically fluctuating down to @xmath12 .
this can be achieved by enhancing quantum fluctuations which can melt the magnetic solid .
one way of constructing such a quantum spin - liquid ( qsl ) is by arranging the magnetic moments on low dimensional or geometrically frustrated lattices .
indeed , qsl s were first demonstrated to exist for quasi - one - dimensional spin chains like kcuf@xmath13 and sr@xmath14cuo@xmath13 ( see refs . for experimental reviews ) .
the quest for spin liquid realizations in higher dimensions has led to a flurry of activity in the last two decades resulting in the discovery of quite a few candidate spin liquid materials @xcite .
the best established candidates are the quasi - two - dimensional kagome lattice quantum magnet herbertsmithite zncu@xmath13(oh)@xmath15cl@xmath14,@xcite the triangular lattice organic magnets @xmath16(bedt - ttf)@xmath14cu@xmath14(cn)@xmath13 ( refs . ) and etme@xmath13sb[pd(dmit)@xmath14]@xmath14,@xcite and the recently discovered yb based triangular lattice magnet ybmggao@xmath18 @xcite . there are some other materials which showed promise either because of their geometrically frustrated lattice , like the 3-dimensional hyper - kagome iridate na@xmath18ir@xmath13o@xmath19 , or because of novel frustration mechanisms like the kitaev qsl candidates @xmath20iro@xmath13 . both these material families however , showed ordering or freezing at low temperatures @xcite .
recently a new quasi two - dimensional quantum magnet ca@xmath0cr@xmath1o@xmath2 has been discovered with a kagome bilayer structure @xcite . using bulk measurements like magnetic susceptibility and heat capacity , and microscopic measurements like @xmath21sr and neutron scattering
this material has been reported to show all the expected signatures of a gapless quantum spin liquid ( qsl ) @xcite . in this work
we report synthesis and structure of polycrystalline samples of the recently discovered qsl ca@xmath0cr@xmath1o@xmath2 .
we present a detailed study of magnetic susceptibility @xmath3 versus temperature @xmath4 , magnetization @xmath5 versus magnetic field @xmath6 at various @xmath4 , and heat capacity @xmath7 versus @xmath4 at various @xmath6 .
polycrystalline samples of ca@xmath0cr@xmath1o@xmath2 were prepared by solid state synthesis .
the starting materials caco@xmath13 ( 99.99% , alfa aesar ) and cr@xmath14o@xmath13 ( 99.99% , alfa aesar ) were taken to make the ca : cr ratio @xmath22 and mixed thoroughly in an agate mortar and pelletized .
the pellet was placed in a covered al@xmath14o@xmath13 crucible , heated in air at @xmath23c for @xmath24 hrs for calcination and then heated to @xmath25c for @xmath26 hrs , and then quenched in argon to room temperature .
after the initial heat treatment , the material was reground and pressed into a pellet and given two heat treatments of @xmath24 h each at @xmath27c , with an intermediate grinding and pelletizing step .
the pellet was always brought to room temperature by quenching in argon .
hard well sintered pellets were obtained which are dark green in color .
powder x - ray diffraction ( pxrd ) patterns were obtained at room temperature using a rigaku diffractometer with cu k@xmath28 radiation , in the @xmath29 range from @xmath30 to @xmath31 with a step size of @xmath32 .
intensity data were accumulated for @xmath33 s / step .
the magnetization @xmath5 versus temperature @xmath4 and magnetic field @xmath6 were measured using a vsm option of a quantum design physical property measurement system ( qd - ppms ) , and the heat capacity @xmath7 as a function of @xmath4 and @xmath6 was measured using a qd - ppms .
the powder x - ray diffraction ( pxrd ) data is shown in fig .
[ fig - xrd ] .
the pxrd data could be refined starting with the structural model recently reported for ca@xmath0cr@xmath1o@xmath2 @xcite .
the rietveld refinement results are also plotted in fig .
[ fig - xrd ] .
the lattice parameters obtained from the fit are @xmath34 , @xmath35 , @xmath36 , @xmath37 , and the cell volume @xmath38 @xmath39 .
the wyckoff atomic positions , site occupancies , and thermal parameters obtained from the fit are given in table [ structure ] .
these parameters match well with recently reported data on single crystals and confirm the single phase synthesis of polycrystalline ca@xmath0cr@xmath1o@xmath2 .
@xcite six out of the seven cr ions per formula unit are in the cr@xmath40 valence state while one is in the cr@xmath41 valence state @xcite .
the cr@xmath40 ions are magnetic with @xmath42 while the cr@xmath41 ion is non - magnetic .
the magnetic results discussed below are consistent with this distribution .
figure [ fig - structure ] shows the arrangements of the magnetic cr ions .
these ions sit in oxygen tetrahedra .
there are @xmath43 inequivalent tetrahedra shown as the @xmath43 different colors in fig .
[ fig - structure ] .
the structure is made up of layers of these different tetrahedra alternating along the @xmath44-axis as shown in the top panel of fig .
[ fig - structure ] . within the layers
the tetrahedra are arranged on a kagome lattice as can be seen in the lower panel in fig .
[ fig - structure ] .
the magnetic connectivity between the kagome layers is such that the blue and orange kagome layers interact with each other but are isolated from the other layers .
similarly , the yellow and purple kagome layers interact but are isolated from the others .
thus , magnetically , ca@xmath0cr@xmath1o@xmath2 can be viewed as a layered bi - layer kagome system .
|c |c| c| c| c| c| c | atom & wyck & x & y & z & occ & b ( ) + ca1 & 36f & 0.292(4 ) & 0.165(8 ) & -0.070(6 ) & 0.97 & 0.0058 + ca2 & 36f & 0.182 & -0.204(3 ) & -0.008(5 ) & 0.96 & 0.0071 + ca3 & 36f & 0.374 & 0.157(7 ) & 0.022 & 1 & 0.0086 + ca4 & 12c & 0.666(7 ) & 0.333 & 0.087 & 1 & 0.0099 + cr1 & 36f & 0.316(2 ) & 0.154(5 ) & 0.124 & 1 & 0.0012 + cr2 & 36f & 0.164(7 ) & -0.142 & -0.107 & 0.95 & 0.0143 + cr3a & 12c & 0.000 & 0.000 & -0.018(4 ) & 0.64(5 ) & 0.0031 + cr4b & 12c & 0.000 & 0.000 & -0.007(5 ) & 0.33 & 0.0064 + o3a & 12c & 0.000 & 0.000 & -0.023(2 ) & 0.65(9 ) & 0.001 + o3b & 12c & 0.000 & 0.000 & -0.364 & 0.0256(9 ) & 0.0900 + o1 & 36f & 0.300(9 ) & 0.189(8 ) & 0.054(9 ) & 0.84(8 ) & 0.0192 + o2 & 36f & 0.209 & 0.173(4 ) & 0.182(5 ) & 1.000 & 0.0110 + o3 & 36f & 0.116(9 ) & -0.0331 & 0.130(7 ) & 1.000 & 0.0395 + o4 & 36f & 0.367 & 0.147(6 ) & 0.131 & 1.000 & 0.0008 + o5 & 36f & 0.1239 & -0.155(7 ) & -0.062 & 0.95(4 ) & 0.033 + o6 & 36f & 0.231(7 ) & -0.267(6 ) & -0.124(4 ) & 1.000 & 0.0081 + o8 & 36f & -0.010(3 ) & -0.227(7 ) & -0.118(4 ) & 0.86(5 ) & 0.0094 + o9 & 36f & -0.106 & 0.011(5 ) & 0.006(6 ) & 0.91(3 ) & 0.0131 + [ structure ] figure [ fig - chi ] ( a ) shows the magnetic susceptibility @xmath45 versus temperature @xmath4 of ca@xmath0cr@xmath1o@xmath2 measured in a magnetic field @xmath46 and @xmath47 koe .
curie - weiss like local moment paramagnetism is clearly evident . at low temperatures
@xmath3 increases rapidly and reaches values which are quite large and similar to values seen in ferromagnetic materials .
however , no anomaly signalling long ranged magnetic order is observed down to @xmath48 k. the @xmath49 data at high temperatures @xmath50 k were fit by the curie - weiss expression @xmath51 , where @xmath52 is a @xmath4 independent contribution , @xmath7 is the curie constant , and @xmath53 is the weiss temperature .
the @xmath49 data and the fit are shown in fig .
[ fig - chi ] ( b ) .
the fit gave the values @xmath54
cm@xmath39/cr , @xmath56 cm@xmath39k / cr , and @xmath57 k. the value of @xmath7 is smaller than the value @xmath58
cm@xmath39k / cr expected for @xmath42 with a @xmath59-factor @xmath60 .
the observed @xmath7 leads to an effective magnetic moment @xmath61 which is smaller than the @xmath62 expected for spin @xmath42 .
however , out of the @xmath63 cr ions in each formulae unit of ca@xmath0cr@xmath1o@xmath2 , @xmath64 are in cr@xmath65 valence state with @xmath42 and @xmath66 cr ion is in cr@xmath67 valence state and is expected to be non - magnetic with @xmath68 .
the estimated reduced moment @xmath69 is close to the expectation @xmath70 .
the value of @xmath57 k is small and positive indicating weak ferromagnetic exchange interactions .
this is consistent with the huge increase in @xmath3 at low @xmath4 seen in fig .
[ fig - chi ] ( a ) .
the magnetic exchange interactions are however , more complex as can be inferred from the deviation below about @xmath71 k of @xmath49 from the curie - weiss fit as can be seen in fig .
[ fig - chi ] ( b ) inset .
the @xmath49 data deviates upwards of the curie - weiss fit which means that the @xmath72 data becomes smaller than expectation from the fit .
this indicates the presence of antiferromagnetic interactions . thus the magnetic susceptibility data suggests the presence of both ferromagnetic and antiferromagnetic exchange interactions with the ferromagnetic interactions being the dominant ones leading to a net positive weiss temperature @xmath73 k. the magnetization @xmath5 versus magnetic field @xmath6 data measured at various temperature @xmath4 are shown in fig . [ fig - mh ] . for @xmath74
k , @xmath75 isotherms are linear in @xmath6 . for the @xmath75 data at @xmath76
k one observes a slight curvature . however , the @xmath75 isotherms at @xmath77 and @xmath78 k show clear curvature with tendency of saturation .
this is consistent with net ferromagnetic interactions which are weak .
the data at @xmath48 k , although still increasing with @xmath6 , are near saturation to values close to @xmath79% of the value expected ( @xmath80cr ) for @xmath42 moments .
this is again consistent with only 6 out of 7 chromium ions being magnetic .
the fact that the magnetic moments can be saturated at magnetic fields @xmath81 t also suggests that the energy scale of the largest magnetic exchange interactions in ca@xmath0cr@xmath1o@xmath2 is small @xmath8 k. heat capacity @xmath7 versus temperature @xmath4 data for ca@xmath0cr@xmath1o@xmath2 measured between @xmath48 and @xmath82 k in various magnetic fields @xmath6 are shown in the main panel in fig .
[ fig - cp ] ( a ) .
the inset shows the same data plotted as @xmath83 versus @xmath4 .
the first thing to note is that the @xmath84 data below @xmath85 k shows an upturn and approximately saturates below @xmath86 k. this might suggest an onset of long - ranged order . on application of a magnetic field
this upturn moves up in temperature and develops into a complete peak at @xmath87 t. the peak then moves to higher temperatures for larger @xmath6 .
the broad peak is very unlike that expected for a second order phase transition where a @xmath88-like anomaly is usually seen in the heat capacity .
thus we believe that this anomaly does not signal a magnetic phase transition .
this is supported by the entropy recovered under the peak as we now describe . from fig .
[ fig - cp ] ( a ) inset it is clear that the magnetic fields have a pronounced effect at low temperatures .
however , the data above about @xmath89 k is independent of @xmath6 suggesting that it is mostly of non - magnetic origin .
we therefore use the data above @xmath89 k and extrapolate it to lower temperatures to get an approximate estimation of the lattice heat capacity .
this data is shown as the solid curve in fig .
[ fig - cp ] ( a ) . using this
we can get the magnetic contribution to the heat capacity at various magnetic fields by subtracting the approximate lattice contribution from the @xmath11 data .
this has been done and the resulting difference data @xmath90 are shown in fig .
[ fig - cp ] ( b ) for all magnetic fields .
the @xmath90 thus obtained can be used to estimate the magnetic entropy by integrating @xmath91 versus @xmath4 data .
the temperature dependence of the entropy @xmath92 estimated in this way is shown in fig .
[ fig - cp ] ( b ) inset for all magnetic fields .
the @xmath92 data is presented in units of @xmath93ln2 per magnetic cr .
we note that @xmath92 for @xmath84 reaches about @xmath94ln2 at @xmath95 k. it is therefore unlikely that a long ranged magnetic order will occur at lower temperatures .
if such a transition does occur it can involve only @xmath96ln2 entropy which would mean a much reduced moment ordering .
this is consistent with previous @xmath9 measurements on single crystalline ca@xmath0cr@xmath1o@xmath2 down to @xmath97 k in @xmath84 that have revealed the absence of long ranged magnetic ordering @xcite . in a magnetic field
, the anomaly in the magnetic contribution moves to higher temperatures although the magnitude of the peak does not change as can be seen most clearly in fig .
[ fig - cp ] ( b ) .
the temperature of the peak @xmath98 as a function of magnetic field @xmath6 is plotted in the smaller inset in fig .
[ fig - cp ] ( b ) .
we find an almost linear dependence of @xmath98 on @xmath6 as can be seen by the solid curve through the data which is the relation @xmath99 .
the entropy @xmath92 associated with the peak in the magnetic contribution to @xmath7 , shown in fig .
[ fig - cp ] ( b ) inset , is also pushed up in temperatures with increasing @xmath6 and we recover the full @xmath93ln2 at @xmath100 t. this behaviour of the entropy in a magnetic field is a hallmark of geometrically frustrated magnets .
the frustration suppresses the tendency for long ranged order leading to the accumulation of the entropy of the unordered @xmath42 moments at lower temperatures .
the magnetic field leads to partial alignment of the disordered moments at higher temperatures than at @xmath84 leading to magnetic entropy being recovered to higher temperatures .
the observation that fields of @xmath101 t affect the magnetic heat capacity and are able to move the magnetic entropy to higher temperatures again suggests that the magnetic energy scales in ca@xmath0cr@xmath1o@xmath2 are @xmath8@xmath47 k.
we have synthesized polycrystalline samples of the recently discovered kagome bilayer spin liquid material ca@xmath0cr@xmath1o@xmath2 and studied in detail its temperature dependent magnetic susceptibility , isothermal magnetization , and temperature and magnetic field dependent heat capacity measurements .
the magnetic measurements indicate the presence of both ferromagnetic ( fm ) and antiferromagnetic ( afm ) exchange interactions with the fm interactions dominating .
the net magnetic scale is about @xmath8@xmath102 k as evidenced by the near saturation of the magnetization at @xmath48 k in a magnetic field of @xmath103 t. the curie constant is consistent with one out of the 7 chromium ions per formula unit being non - magnetic .
the magnetic cr ions form a bilayer kagome lattice .
the heat capacity in @xmath84 shows an incomplete anomaly peaked around @xmath104 k. the entropy recovered between @xmath48 k and @xmath89 k in @xmath84 is close to @xmath105 rln2 suggesting absence of magnetic ordering for lower temperatures , consistent with a spin - liquid state .
this broad anomaly peaked in @xmath7 is consistent with previous zero - field measurements on single crystal ca@xmath0cr@xmath1o@xmath2 and was associated with the onset of coherent quantum fluctuations @xcite .
a similar broad anomaly has been observed for several other qsl candidates .
for example , the organic triangular lattice spin liquid etme@xmath13sb[pd(dmit)@xmath14]@xmath14,@xcite shows an anomaly in the heat capacity at @xmath106 k while the recently discovered qsl candidate ybmggao@xmath18 shows a heat capacity anomaly at @xmath107 k @xcite . this anomaly for qsls
is understood to be a crossover from a thermally disordered state to a quantum disordered state .
a low temperature anomaly in the heat capacity which moves up in temperatures on the application of magnetic field is a hallmark of frustrated magnets in general .
for example , in addition to the above materials , the pyrochlore spin - ice material pr@xmath14zr@xmath14o@xmath1 also shows an anomaly in the heat capacity at @xmath104 k which is attributed to the formation of a collective spin - ice state .
this anomaly has been shown to move approximately linearly to higher temperatures with magnetic fields @xcite .
the heat capacity anomaly for ca@xmath0cr@xmath1o@xmath2 occurs at @xmath108 k at @xmath84 and moves to higher temperatures in a magnetic field approximately linearly .
the linear dependence of the peak temperature with @xmath6 is intriguing . since the ferromagnetic exchange is dominant in ca@xmath0cr@xmath1o@xmath2 it is possible that in - plane short - ranged order develops and is strengthened in a field .
however , future measurements would be needed to understand this observation . c. balz , b. lake , j. reuther , h. luetkens , r. schnemann , t. herrmannsdrfer , y. singh , a. t. m. nazmul islam , e. m. wheeler , j. a. rodriguez - rivera , t. guidi , g. g. simeoni , c. baines , and h. ryll , nat .
* 12 * , 942 ( 2016 ) .
s. petit , e. lhote , s. guitteny , o. florea , j. robert , p. bonville , i. mirebeau , j. ollivier , h. mutka , e. ressouche , c. decorse , m. ciomaga hatnean , and g. balakrishnan , phys .
b * 94 * , 165153 ( 2016 ) . | we report synthesis of polycrystalline samples of the recently discovered spin liquid material ca@xmath0cr@xmath1o@xmath2 and present measurements of the magnetic susceptibility @xmath3 versus temperature @xmath4 , magnetization @xmath5 versus magnetic field @xmath6 at various @xmath4 , and heat capacity @xmath7 versus @xmath4 at various @xmath6 .
the magnetic measurements indicate the presence of both ferromagnetic and antiferromagnetic exchange interactions with dominant ferromagnetic interactions and with the largest magnetic energy scale @xmath8 k. @xmath9 shows a broad anomaly at @xmath10 k which moves to higher temperatures in a magnetic field .
the evolution of the low temperature @xmath11 and the magnetic entropy is consistent with frustrated magnetism in ca@xmath0cr@xmath1o@xmath2 . |
in the first paper ( paper i ) @xcite of this series of two papers , we obtained a number of analytical results concerning the continuum percolation of overlapping hyperspheres and overlapping oriented hypercubes that applied across all euclidean space dimensions . among other results ,
it was shown analytically that certain lower - order pad approximants on the mean cluster number @xmath6 are lower bounds on @xmath6 for both hyperspheres and hypercubes in @xmath0-dimensional euclidean space @xmath1 and that they become exact asymptotically as @xmath7 . in this asymptotic limit
, the dimensionless density at percolation @xmath3 tends to @xmath8 .
an important consequence of the aforementioned analysis is that this large-@xmath0 percolation value is an important contribution to the low - dimensional percolation value . in other words ,
low - dimensional results encode high - dimensional information .
percus - yevick - like approximations for the cluster number @xmath6 were also obtained that also become asymptotically exact as @xmath7 .
the analysis was aided by a striking _ duality _ between the equilibrium hard - hypersphere ( hypercube ) fluid system and the continuum percolation models of overlapping hyperspheres ( hypercubes ) , namely , @xmath9 where @xmath10 is the pair connectedness function at some radial distance @xmath11 and reduced density @xmath12 for the continuum percolation models and @xmath13 is the total correlation function for the equilibrium hard - particle models.@xcite it was shown that the large-@xmath0 percolation threshold @xmath3 of overlapping hyperspheres and hypercubes is directly related to the large-@xmath0 freezing - point density of corresponding equilibrium hard - particle models.@xcite the extension of these results for overlapping hyperspheres and hypercubes to the case of overlapping particles of general _ anisotropic _ shape in @xmath0 dimensions with a specified orientational probability distribution was also described . the bounds and approximations reported in paper i were applied to assess the accuracy of previous computer simulation results for @xmath3 that span dimensions up to @xmath14 in the case of overlapping hyperspheres @xcite and up to @xmath15 in the case of hypercubes.@xcite it is convenient to restate the best lower bound on @xmath3 that was found in ref .
, namely , @xmath16 } , \label{eta-2 - 1}\ ] ] where @xmath17 and @xmath18 are the trimer and tetramer statistics defined and computed as a function of @xmath0 in paper i. comparison of this lower bound to kr " uger s simulation data for hyperspheres revealed that the bound became progressively tighter as @xmath0 increased and became an excellent estimate for @xmath19 .
since it becomes increasingly challenging to estimate percolation thresholds from simulations in high dimensions , it was not surprising that even kr " uger s high - quality simulation data fell slightly below the lower bound ( [ eta-2 - 1 ] ) for @xmath20 .
the analytical results of paper i revealed that the simulation data reported in ref . for both hyperspheres and hypercubes were considerably more problematic .
in particular , the authors reported incorrectly that the quantity @xmath21 for these systems were nonmonotonic in dimension and that hyperspheres have lower thresholds than hypercubes in higher dimensions while the reverse is true in lower dimensions .
one of the main purposes of the present paper is to exploit the accuracy of the lower bound ( [ eta-2 - 1 ] ) to provide an efficient simulation method , called the _ rescaled - particle _ algorithm , to estimate continuum percolation thresholds across many dimensions .
another objective is to obtain an upper bound on @xmath3 as well as to provide additional analytical results for certain cluster statistics . in sec .
[ anal ] , we derive these analytical results . in sec .
[ sim ] , we describe in detail the rescaled - particle method , which is applicable for general continuum percolation models ( e.g. , spherical and nonspherical particle shapes ) . results for the thresholds of both overlapping hyperspheres and oriented hypercubes for @xmath5 are compared to upper and lower bounds on @xmath3 in sec .
[ results ] . in sec . [ conclusions ] , we summarize our conclusions and discuss future work .
a prototypical continuum percolation model consists of equal - sized overlapping ( poisson distributed ) hyperparticles in @xmath1 at number density @xmath22 ; see paper i and numerous references therein .
it is convenient to introduce the reduced number density @xmath12 , defined by the relation @xmath23 where @xmath24 is the @xmath0-dimensional volume of a hyperparticle ; see ref . for an explicit expression of this quantity for a hypersphere , for example .
a cluster statistic that has been considered by various investigators in one , two and three dimensions for overlapping spheres is @xmath25 , the average number of @xmath2-mers per unit number of particles .
@xcite this @xmath2-mer statistic obeys the following constraint @xmath26 where it is to be noted that @xmath27 is the probability that a given particle is part of a @xmath2-mer.@xcite in what follows , we will derive estimates for @xmath25 and related cluster statistics for arbitrary dimension .
it was shown in ref . that @xmath25 can be explicitly expressed for any @xmath0 for overlapping hyperspheres as certain multidimensional integrals involving exponentials whose arguments contain the union volume of @xmath2 `` exclusion spheres . ''
two spheres of radius @xmath28 are considered to be connected if they overlap , i.e. , if the center of one lies within a spherical `` exclusion '' region of radius @xmath29 centered around the other sphere ( see fig . 3 of paper i ) .
for example , for @xmath30 , @xmath31 and @xmath32 , we have @xmath33 \label{n1 } \\
n_2 & = & \frac{\rho}{2 } \int_{\mathbb{r}^d } \exp[-\rho v_2(r;d ) ] f(r ) d{\bf r } \label{n2}\\ n_3 & = & \frac{\rho^2}{6 } \int_{\mathbb{r}^d } d{\bf r}_{12 } \int_{\mathbb{r}^d } d{\bf r}_{13 } \;f(r_{12 } ) f(r_{23 } ) \exp[-\rho v_3(r_{12},r_{13},r_{23 } ; d ) ] \nonumber \\ & + & \frac{\rho^2}{3 } \int_{\mathbb{r}^d } d{\bf r}_{12 } \int_{\mathbb{r}^d } d{\bf r}_{13 } \exp[-\rho v_3(r_{12},r_{13},r_{23 } ; d ) ] , \label{n3}\end{aligned}\ ] ] where @xmath34 is the union volume of @xmath35 spheres of radius @xmath29 centered at positions @xmath36 , @xmath37 and @xmath38 ( @xmath39 ) . moreover
, the radial function @xmath40 defines the connectedness criterion , i.e. , @xmath41 and @xmath42 is the heaviside step function . note that the factor @xmath43 , appearing in eq .
( [ n1 ] ) , is the ratio of the exclusion volume @xmath44 to the volume of a sphere . by virtue of the fact that the spheres are poisson distributed in space
, it follows that the mean number of overlaps per sphere @xmath45 is given by @xmath46 the _ average number of clusters per unit volume _
@xmath47 is directly related to the sum over @xmath25 , @xcite namely @xmath48 note that the ratio @xmath49 is the _ average number of clusters per particle _ and its derivative with respect to @xmath22 ( or @xmath12 ) determines the contact value of the _ blocking function _
@xmath50 , @xcite which is directly related to the conditional probability of finding two particles belonging to different clusters separated by distance r , given that one of the particles is at the origin .
for example , in the case of hyperspheres of diameter @xmath29 , the three - dimensional expression given in ref .
generalizes as follows : @xmath51 we recall here that for overlapping hyperspheres , the blocking function @xmath50 can be obtained immediately from the pair connectedness function @xmath52 for any radial distance @xmath11 via the relation @xcite @xmath53 the fact that @xmath52 is bounded in the interval @xmath54 $ ] , implies the same bounds on @xmath50 .
the _ average cluster number _
@xmath55 is the average number of particles in a randomly chosen cluster and is the inverse of @xmath49 , @xcite namely , @xmath56 this is to be distinguished from the cluster number @xmath6 ( average number of particles in the cluster containing a randomly chosen particle ) , which is related to the second moment of @xmath25 : @xcite @xmath57 unlike @xmath6 , which can also be expressed in terms of the pair connectedness function @xmath52 , @xcite the average cluster number @xmath55 does not diverge at the percolation threshold when @xmath58 .
@xcite it was shown in ref . that the mean number of overlaps per sphere at the threshold @xmath4 [ cf .
( [ n ] ) ] tends to unity as @xmath7 , i.e. , @xmath59 which applies to spherical as well as nonspherical particles ( with specified orientational distribution ) .
this does not mean that the concentrations of monomers , dimers , trimers , etc . at the threshold are negligibly small , even if finite clusters become more ramified as the space dimension grows .
@xcite to explicitly prove this property , we first observe that the exact formula for @xmath60 [ cf .
( [ n1 ] ) ] together with the exact asymptotic result @xmath61 which applies to any oriented centrally symmetric particle ( e.g. , spheres , cubes , ellipsoids , etc . ) @xcite , implies the following asymptotic result @xmath62 we will now show that as @xmath0 becomes large , monomers , as opposed to any @xmath2-mer for @xmath63 , are dominant in so far as concentration is concerned , i.e. , @xmath60 is appreciably larger than @xmath64 and therefore is appreciably larger than @xmath25 with @xmath65 , since @xmath66 for any positive but bounded @xmath12 .
let us begin with the formula ( [ n2 ] ) for the dimer statistic @xmath64 , which we can rewrite as follows : @xmath67 \int_0^d r^{d-1 } \exp[2^d \eta \alpha(r;d ) ] , \label{n2 - 2}\ ] ] where @xmath68 here we have used the fact that @xmath69 , where the latter quantity is the intersection volume of two spheres of radius @xmath70 whose centers are separated by the distance @xmath11 .
the dimensionless intersection volume @xmath71 , which has support in the interval @xmath72 $ ] , has been explicitly given for any @xmath0 in a variety of representations @xcite and played an important role in paper i. now since @xmath73 decays to zero exponentially fast according to the asymptotic relation @xcite @xmath74 it immediately follows from ( [ n2 - 2 ] ) and ( [ eta_c ] ) that @xmath75 and hence we find @xmath76 which is what we set out to prove .
we now derive lower bounds on @xmath25 for @xmath77 as a function of @xmath0 for any @xmath12 .
let us begin with the case @xmath31 .
since @xmath78 is a monotonically decreasing function of @xmath11 , we have that @xmath79 in the interval @xmath80 $ ] and hence combined with the exact formula ( [ n2 - 2 ] ) yields the lower bound @xmath81 \exp[2^d \eta \alpha(d;d ) ] .
\label{n2-bound}\ ] ] it is noteworthy that in light of ( [ alpha - asym ] ) , the lower bound ( [ n2-bound ] ) becomes asymptotically exact in the high-@xmath0 limit , i.e. , we recover ( [ n2 - 3 ] ) . using similar arguments and the formulas for @xmath25 given in ref .
, we obtain the following generally weaker lower bounds on @xmath25 for any @xmath2 : @xmath82.\ ] ] while for @xmath30 , this bound is exact , it is weaker than ( [ n2-bound ] ) for @xmath31 . through second order in @xmath12 , formulas ( [ n1 ] ) , ( [ n2 ] ) and
( [ n3 ] ) for @xmath60 , @xmath64 and @xmath83 , respectively , yield @xmath84 similarly , using expression ( [ defq ] ) and the relations immediately above , we can obtain the corresponding density expansion for @xmath55 : @xmath85 it was noted in ref . that the pole of the [ 1,1 ] pad ' e approximant of the density expansion of @xmath55 for @xmath86 yielded an upper bound on the threshold @xmath3 for overlapping spheres . for general @xmath0 , the [ 1,1 ] pad ' e approximant for @xmath55 for either overlapping hyperspheres or oriented hypercubes
is given by @xmath87 } \approx \frac{1 - \frac{c_3}{3 \cdot 2^{d } } \eta } { 1-\left[2^{d-1}+ \frac{c_3}{3 \cdot 2^d}\right]\eta } , \label{1 - 1}\ ] ] where the trimer statistic @xmath17 for both models is given in ref . .
we now observe that the pole of ( [ 1 - 1 ] ) is an upper bound on @xmath3 for overlapping hyperspheres and oriented hypercubes for any @xmath88 , i.e. , @xmath89 } , \qquad d \ge 3 .
\label{upper}\ ] ] using the same methods described in paper i , it is straightforward to prove that for sufficiently large @xmath0 and any @xmath90 , relation ( [ 1 - 1 ] ) bounds @xmath55 from below and hence ( [ upper ] ) is a rigorous upper bound on the threshold .
this is consistent with the observation ( noted in sec .
[ sub ] ) that the actual function @xmath55 does not diverge when the mean cluster number @xmath6 diverges , i.e. , when @xmath91 .
we will show in sec .
[ results ] that the expression ( [ upper ] ) bounds the simulation data for the threshold from above for both hyperspheres and hypercubes for @xmath92 .
moreover , according to ref .
, the second term within the brackets of inequality ( [ upper ] ) goes to zero exponentially fast in the limit @xmath7 , and hence this bound asymptotically becomes @xmath93 which is the exact asymptotic result .
@xcite since the lower bound ( [ eta-2 - 1 ] ) also becomes exact in this high-@xmath0 limit , the bounds ( [ eta-2 - 1 ] ) and ( [ upper ] ) converge to the exact asymptotic value of @xmath8 .
finally , we note that combining relation ( [ defq ] ) and the approximant ( [ 1 - 1 ] ) gives the following approximation of @xmath49 : @xmath94\eta } { 1 - \frac{c_3}{3 \cdot 2^{d } } \eta}. \label{rho_c}\ ] ] substituting ( [ rho_c ] ) into ( [ block1 ] ) yields the following the approximation for the contact value of the blocking function @xmath95 in the appendix , we provide plots of @xmath96 and p(d ) versus @xmath12 for selected dimensions .
a commonly employed approach to estimate the percolation threshold @xmath3 for continuum percolation in two and three dimensions is the particle - addition method .
@xcite starting from a configuration of a small number of particles with random positions in the simulation domain subject to periodic boundary conditions , new particles are added to the domain sequentially with randomly chosen positions in the simulation domain . each time a new particle is added , the largest cluster in the system ( i.e. , the one containing the largest number of particles )
is identified using a burning - algorithm .
@xcite this process is repeated until a system - spanning cluster forms . although conceptually intuitive and used across dimensions , @xcite this method become progressively less computationally efficient as the dimension increases .
@xcite first , to obtain accurate estimates of the percolation threshold , the size of the particles should be much smaller than the linear extent of simulation domain so that adding a single particle leads to a very small increase in the reduced density @xmath12 . without _ a priori _ knowledge of the percolation threshold , one needs to start with sufficiently dilute particle configurations , i.e. , with very small @xmath12 .
therefore , an extremely large number of particles needs to be added to the simulation domain until the system percolates . for each particle addition , one needs to identify the largest cluster and check whether it spans the system , which makes the method computationally very expensive .
moreover , the dynamic nature of particle addition makes it difficult to implement efficient methods to check for local particle connectivity ( overlaps of pairs of particles , e.g. , the cell method ) .
this problem increases in severity as the dimension increases .
we note that highly efficient algorithms have been developed for investigating clustering and percolation properties of overlapping disks in two dimensions .
for example , several frontier - tracking methods have been devised to provide very precise estimates of the percolation threshold @xmath3 of overlapping disks .
@xcite however , such algorithms can not be applied in higher dimensions because there is no analogous localized boundary in the gradient percolation method for high - dimensional systems .
@xcite other efficient variations of the particle addition method have been implemented , e.g. , particles are added only to a single growing cluster , and the cluster - size distribution rather than the spanning cluster is used to obtain @xmath3 ; @xcite and rapid `` union - find '' methods have been developed to keep track of the connected clusters as particles are added .
@xcite however , these variations of the particle - addition method , although much more efficient than the original one , becomes progressively more difficult to apply as the dimension increases .
tight rigorous lower bounds on the percolation threshold , such as the ones derived in ref . 1 ,
enable us to devise a highly efficient method to estimate @xmath3 for overlapping particles with arbitrary shapes and orientations in @xmath1 .
the basic idea is to generate initial static particle configurations at a value of @xmath12 that is taken to be the best lower - bound value , allowing one to finally arrive at the critical value in a computationally efficient manner , even in high dimensions .
we note that procedures in which particles are rescaled to obtain @xmath3 have been previously proposed .
@xcite however , to the best of our knowledge , our method is the first one to combine the particle rescaling procedure with the tightest lower bound values to efficiently and accurately obtain @xmath3 .
initially , a poisson distribution of a large number of points in the simulation domain is generated . each point in configuration
is then taken to be the centroid of a particle with a specified or random orientation and a characteristic particle length scale @xmath97 ( e.g. , the diameter of a sphere ) .
the initial value of @xmath97 is chosen such that the reduced density @xmath12 of the system equals the tightest lower bound value .
for each particle @xmath98 , a near - neighbor list ( nnl ) is obtained that contains the centroids of the particles @xmath99 whose distance @xmath100 to particle @xmath98 is smaller than @xmath101 ( @xmath102 ) .
the value of @xmath103 generally depends on the continuum - percolation models of interest .
a rule of thumb for a choosing good value of @xmath103 is that the nnl list only contains particles that overlap at the percolation threshold . in our simulation for hyperspheres and hypercubes , we have used @xmath104 $ ] , depending on the space dimension
. then the particle sizes are slowly and uniformly increased by increasing @xmath97 ( i.e. , rescaling the particles ) , which leads to an increase of the reduced density by a small amount @xmath105 .
after each rescaling , the particles in the nnl are checked for overlap and the largest cluster in the system is identified .
the process is repeated until a system - spanning cluster forms . since static particle configurations and predetermined nnl
are used , the complexity for identifying clusters is significantly reduced .
in addition , as the space dimension increases , it was shown that the lower bounds derived in paper i become increasingly tighter .
@xcite thus , total number of rescaling before percolation is achieved is much smaller than the total number of particle additions , which dramatically improves the efficiency of the algorithm .
furthermore , the increase of the reduced density @xmath105 can take arbitrarily small values when @xmath3 is approached , rather than a fixed discontinuous value determined by the particle size via particle - addition methods .
this smooth approach to the critical value allows a more accurate estimate of the percolation threshold .
@xmath106{eta_size.eps } & \includegraphics[height=6.0cm , keepaspectratio]{eta_frac.eps } \\
\mbox{(a ) } & \mbox{(b)}\\ \end{array}$ ] similarly , given an upper bound on @xmath3 , such as inequality ( [ upper ] ) , one can start with a percolated system , and rescale the particles ( i.e. , decreasing the particle size @xmath97 ) to reduce @xmath12 .
figure [ fig_clusterdynamic ] shows the cluster statistics ( e.g. , the linear size of the largest cluster @xcite and the fraction of particles in the largest cluster ) associated with the particle - addition method and the rescaled - particle method for overlapping hyperspheres in @xmath107 .
the initial configurations for the rescaled - particle method include both a non - percolated configuration with @xmath12 equal to the lower - bound value @xmath108 and a percolated configuration with @xmath12 equal to the upper - bound value @xmath109 .
we note that close to percolation , the cluster containing the largest number of particles also possesses the largest linear size .
it is clear that the particle - addition method and the rescaled - particle method produce very similar cluster statistics yet the rescaled - particle method is considerably more computationally efficient in three dimensions .
the computational efficiency improves for overlapping hyperspheres and hypercubes as the space dimension increases beyond three . since the upper bound on @xmath3 derived here
is not as tight as the lower bound in relatively low dimensions , we mainly use the rescaled - particle method starting from non - percolated configurations with @xmath12 equal to the tightest lower - bound value , as predicted from ( [ eta-2 - 1 ] ) .
however , as noted in sec .
[ bounds ] , as @xmath110 , both the upper and lower bounds converge to the exact asymptotic value of @xmath8 .
.estimates of the percolation threshold @xmath3 for overlapping hyperspheres as obtained from the rescaled - particle algorithm , the lower bound ( [ eta-2 - 1 ] ) , and the upper bound ( [ upper ] ) . also included are the numerical estimates @xmath111 of the percolation threshold from a previous study @xcite that satisfy the bounds ( [ eta-2 - 1 ] ) and ( [ upper ] ) . these results are not reported for @xmath112 because they violate the lower bounds . [ cols="^,^,^,^,^",options="header " , ] [ tab_hypercube ] using the rescaled - particle method starting from a reduced density given by the lower - bound estimate ( [ eta-2 - 1 ] ) , we compute the percolation threshold for overlapping hyperspheres and oriented hypercubes in dimensions two through eleven . for each dimension , different system sizes @xmath113 are used .
specifically , we employ @xmath114 for @xmath115 , @xmath116 for @xmath117 , @xmath118 for @xmath119 , and @xmath120 for @xmath121 and the results are extrapolated by spline fitting the finite - system - size data in a log - log plot to obtain the infinite - system - size estimate of @xmath3 . for each system size , the percolation threshold is obtained by averaging over 1000 independent particle configurations for @xmath122 and 3 , 500 independent particle configurations for @xmath123 , and 100 independent particle configurations for @xmath124 .
the obtained percolation threshold values @xmath3 for overlapping hyperspheres and hypercubes in dimensions two through eleven are respectively given in table 1 and table 2 , and displayed in figs .
[ fig_bound1 ] and [ fig_bound2 ] .
we also provide in the tables and figures the corresponding values of the lower bound ( [ eta-2 - 1 ] ) and upper bound ( [ upper ] ) on @xmath3 for purposes of comparison .
note that our simulation data lie very close to the lower - bound values , and that the lower bounds and data converge quickly to one another as @xmath0 increases .
moreover , we include the numerical estimates of @xmath3 from previous simulation studies for hyperspheres @xcite and hypercubes @xcite in case they do not violate the bounds ( [ eta-2 - 1 ] ) and ( [ upper ] )
. it can be clearly seen that our rescaled - particle method yields much more accurate estimates of @xmath3 , especially in high dimensions , since we start with particle configurations that are already very close to percolation .
@xmath125{bound_sphere.eps } \end{array}$ ] @xmath125{bound_cube.eps } \end{array}$ ] following ref .
, we use the threshold estimate obtained from the @xmath126 $ ] pad approximant of @xmath6 as the basis to obtain accurate analytical approximations for @xmath3 that applies across all dimensions for hyperspheres and oriented hypercubes .
specifically , we fit the following function to the simulation data for @xmath127 : @xmath128 where @xmath129}\ ] ] is the pole associated with the [ 2 , 1 ] pad ' e approximant for the mean cluster numver @xmath6 , i.e. , the tightest lower bound for @xmath3 , explicitly given by eq .
( 119 ) of paper i. this lower bound becomes exact for sufficiently large @xmath0 .
therefore , in agreement with the conclusions of ref . , we see that the high - dimensional percolation behavior is an important contribution to the low - dimensional percolation value . in other words , low - dimensional results encode high - dimensional information .
we find that @xmath130 and @xmath131 for hyperspheres with correlation coefficient equal to @xmath132 ; and @xmath133 and @xmath134 for hypercubes with correlation coefficient equal to @xmath135 .
note that due to the quality of the available numerical data reported in paper i , the analogous analytical approximations based on those data for hyperspheres and hypercubes only used numerical threshold estimates for @xmath136 and @xmath137 , respectively .
thus , the formula ( [ approx ] ) for hyperspheres and hypercubes supersedes in accuracy the ones provided in ref . .
as pointed out in ref .
, the numerical estimates of @xmath3 for hypercubes in @xmath138 by wagner _ et al . _
@xcite violate the tightest lower bound ( [ eta-2 - 1 ] ) . in addition , these simulation data are questionable in high dimensions since the mean number of overlaps per particle @xmath45 [ defined in eq .
( [ n ] ) ] evaluated at the percolation threshold , i.e. , @xmath139 is incorrectly found to be a nonmonotonic function of @xmath0 . in particular
, these authors found that @xmath4 first decreases as @xmath0 increases for @xmath140 and then increases as @xmath0 increases for @xmath141 .
this led to the incorrect conclusion that hyperspheres have lower thresholds than hypercubes in higher dimensions while the reverse is true in lower dimensions .
@xmath142{n_c_sphere.eps } \end{array}$ ] @xmath142{n_c_cube.eps } \end{array}$ ] in figs .
[ fig_nc_sphere ] and [ fig_nc_cube ] , we show @xmath4 as a function of @xmath0 computed using the estimates of @xmath3 obtained from our simulations for overlapping hyperspheres and hypercubes , respectively .
the quantities @xmath143 and @xmath144 are also shown for purposes of comparison , where @xmath108 and @xmath109 are respectively the tightest lower bound ( [ eta-2 - 1 ] ) and upper bound ( [ upper ] ) .
it can be clearly seen that @xmath4 for both overlapping hyperspheres and hypercubes are indeed monotonic functions of @xmath0 , which quickly converge to the asymptotic value of unity as @xmath0 increases .
this indicates again that the large-@xmath0 asymptotic percolation value is an important contribution to the low - dimensional percolation value .
moreover , one can see from tables 1 and 2 that hypercubes always have a lower threshold than hyperspheres for any fixed finite dimension , and the thresholds of these two systems approach one another in the limit @xmath145 , as predicted in ref . . in fig
[ cluster ] , we plot the concentration of monomers and dimers , @xmath60 and @xmath64 , at the percolation threshold @xmath3 as a function of dimension @xmath0 for overlapping hyperspheres as obtained from the exact expressions ( [ n1 ] ) and ( [ n2 - 2 ] ) and the simulation data for @xmath3 given in table i. observe that , consistent with analysis given in sec . [ bounds ] ,
@xmath60 becomes appreciably larger than @xmath64 ( and hence @xmath146 , etc . ) as the dimension increases .
we include in the figure the lower bound ( [ n2-bound ] ) on @xmath64 , which we see becomes tighter as the space dimension increases , as the analysis of sec .
[ bounds ] predicts .
@xmath125{n2.eps } \end{array}$ ]
we have supplemented the analytical results obtained in paper i by deriving additional formulas and bounds for certain cluster statistics , such as the concentration of @xmath2-mers and related quantities , and obtained an upper bound on the percolation threshold @xmath3 .
we utilized this upper bound and the tightest lower bound on @xmath3 obtained in paper i to devise an efficient simulation method , called the _ rescaled - particle _ algorithm , to estimate continuum percolation properties across many space dimensions .
we applied this simulation procedure here to compute , with heretofore unattained accuracy , the threshold @xmath3 and associated mean number of overlaps per particle @xmath4 for both overlapping hyperspheres and oriented hypercubes for @xmath5 .
comparison of these simulations results to corresponding upper and lower bounds on these percolation properties revealed that the bounds converge to one another as the space dimension increases .
it is noteworthy that the lower bound provides an excellent estimate of @xmath3 and @xmath4 , even for relatively low dimensions .
we confirmed a prediction of paper i that low - dimensional percolation properties encode high - dimensional information .
we also showed that the concentration of monomers dominate over concentration values for higher - order clusters ( dimers , trimers , etc . )
as the space dimension becomes large .
finally , we provided accurate analytical estimates of the pair connectedness function and blocking function at their contact values for any @xmath0 as a function of density . in paper
i , the extension of the continuum percolation results obtained for overlapping hyperspheres and oriented hypercubes to cases in which the overlapping hyperparticles are nonspherical ( anisotropic in shape ) with some specified orientation distribution function ( e.g. , random orientations ) was briefly discussed .
future work will expound on this extension to overlapping anisotropically - shaped hyperparticles with random orientations ..
the exploration of the generalizations of the techniques of this series of papers to bound percolation thresholds in the lattice setting @xcite ( bond and site percolation @xcite ) represents an intriguing area for future research .
this work was supported by the materials research science and engineering center program of the national science foundation under grant no .
dmr-0820341 and and by the division of mathematical sciences at the national science foundation under award number dms-1211087 . in sec .
ii , we noted that the inverse of the cluster number @xmath55 is the of the average number of clusters per unit volume @xmath49 [ see eq .
( [ defq ] ) ] .
we derived the approximation ( [ 1 - 1 ] ) for @xmath55 , which immediately leads to the approximation ( [ rho_c ] ) for @xmath49 and thus , the approximation ( [ pb ] ) for the contact values of the blocking function @xmath96 and pair connectedness function @xmath147 . in this appendix , we compare the approximation ( [ rho_c ] ) for @xmath49 versus @xmath12 with available numerical data for overlapping spheres in @xmath107 .
@xcite these results are plotted in fig .
[ fig_rho ] . observe that the approximation agrees with the simulation data very well and bounds the data from below .
since the approximation ( [ rho_c ] ) for @xmath49 improves as the space dimension increases beyond three , we expect that results derived from it , such as the relation ( [ pb ] ) for the blocking function and pair connectedness function at contact , will provide accurate approximations across dimensions for @xmath148 . in particular , we provide plots of the contact values @xmath96 and @xmath149 [ as predicted by ( [ pb ] ) ] versus @xmath12 for overlapping hyperspheres in dimensions @xmath150 and @xmath151 up to the respective percolation thresholds in figs .
[ fig_b_3 ] , [ fig_b_7 ] and [ fig_b_11 ] . the functions @xmath96 and @xmath147 are equal to unity and zero , respectively , at @xmath152 and decrease and increase monotonically with increasing @xmath12 up to @xmath3 .
we also see that @xmath96 and @xmath147 vary less appreciably with increasing @xmath12 from the maximum value of unity and minimum value of zero , respectively , as the space dimension increases . for equilibrium hard - particle systems ,
the total correlation function @xmath13 generally takes on both negative and positive values depending on the values of the radial distance @xmath11 and reduced density @xmath12 .
one can think of the duality relation ( [ dual ] ) as the mapping that is required to convert a correlation function for a hard - particle system to a non - negative pair connectedness function ( bounded from above by unity ) for the corresponding overlapping particle system . | in the first paper of this series [ s. torquato , j. chem .
phys . * 136 * , 054106 ( 2012 ) ] , analytical results concerning the continuum percolation of overlapping hyperparticles in @xmath0-dimensional euclidean space @xmath1 were obtained , including lower bounds on the percolation threshold . in the present investigation
, we provide additional analytical results for certain cluster statistics , such as the concentration of @xmath2-mers and related quantities , and obtain an upper bound on the percolation threshold @xmath3 .
we utilize the tightest lower bound obtained in the first paper to formulate an efficient simulation method , called the _ rescaled - particle _ algorithm , to estimate continuum percolation properties across many space dimensions with heretofore unattained accuracy .
this simulation procedure is applied to compute the threshold @xmath3 and associated mean number of overlaps per particle @xmath4 for both overlapping hyperspheres and oriented hypercubes for @xmath5 .
these simulations results are compared to corresponding upper and lower bounds on these percolation properties .
we find that the bounds converge to one another as the space dimension increases , but the lower bound provides an excellent estimate of @xmath3 and @xmath4 , even for relatively low dimensions .
we confirm a prediction of the first paper in this series that low - dimensional percolation properties encode high - dimensional information .
we also show that the concentration of monomers dominate over concentration values for higher - order clusters ( dimers , trimers , etc . ) as the space dimension becomes large .
finally , we provide accurate analytical estimates of the pair connectedness function and blocking function at their contact values for any @xmath0 as a function of density . |
circumstellar material profoundly influences the star formation process .
a large - scale envelope acts as the local mass reservoir during the collapse and growth of a central protostar ; a more compact disk regulates how that material is transported onto the star itself .
a great deal of progress has been made in understanding the physical conditions present in the gas and dust surrounding sun - like stars ( @xmath6 - 3m@xmath5 ) .
however , it is unclear if those results can be extrapolated across the entire stellar mass spectrum .
the low end of this spectrum ( @xmath7m@xmath5 ) is of particular interest , as observations of the material around young low - mass objectsm@xmath5 as low - mass objects , including both very low - mass stars and brown dwarfs . ]
can help resolve an ongoing debate over their dominant formation mechanism .
the observed abundance of such low - mass objects relative to their more massive counterparts is difficult to account for with the traditional model for isolated star formation .
two distinct modifications have been proposed : ( @xmath8 ) the turbulent fragmentation of cloud cores into smaller building units ( e.g. , * ? ? ?
* ) ; and ( @xmath9 ) a dynamical alternative where one component of a multiple system is ejected and prematurely cut off from its accretion reservoir ( e.g. , * ? ? ?
* ; * ? ? ?
both scenarios make distinct and conflicting predictions about the gas and dust environments that should be associated with low - mass objects . in the former , a disk / envelope structure similar to those noted around higher mass t tauri stars
would be expected . and for the latter , only a truncated disk ( with an initial @xmath10au ) would survive the tidal stripping and ejection process @xcite .
if the remnant material is sufficiently viscous , it could spread to radii up to @xmath0100au in @xmath01myr @xcite , but would have shed a vast majority of its initial mass in the ejection process .
therefore , observational constraints on the structure of the gas and dust around low - mass objects offer one avenue to help distinguish how they form .
a variety of observations demonstrate that the signatures of circumstellar gas and dust at small radii ( up to a few au ) are indeed common for young low - mass objects ( see the recent review by * ? ? ?
more detailed individual studies confirm that such material has a geometry and composition similar to the disks around more massive t tauri stars @xcite , although typically at lower masses @xcite .
however , these _ unresolved _ observations can not unambiguously constrain some key properties of this material , most significantly its spatial structure and extent . in this article , we present new submillimeter observations of the molecular gas and dust surrounding iras 04158 + 2805 . located in the @xmath01myr - old taurus star - forming region , iras 04158 + 2805
has a cool central source with an estimated spectral type of m5-m6 @xcite .
recently , @xcite provided an initial analysis of this source and the dust structure surrounding it by modeling the broadband spectrum along with optical / infrared scattered light images .
they suggested that their observations are best explained by a large ( diameter of 2240au ) , massive ( dust mass of 1 - 2@xmath1110@xmath12m@xmath5 ; gas / dust ratio of @xmath13 ) circumstellar disk around a low - mass star ( @xmath14 - 0.2m@xmath5 ) , with no need to include any contribution from a more extended envelope .
our new data represent the first high spatial resolution view of the gas and dust environment around this and similar cool , young objects at submillimeter wavelengths .
they also provide an opportunity for a rare , albeit crude , estimate of @xmath15 from the spatio - kinematics of circumstellar gas that is independent of pre - main - sequence evolution models .
these observations are addressed in 2 , and the resulting data products are highlighted in 3 . in 4 , we discuss the structure of the gas and dust surrounding iras 04158 + 2805 and what it can tell us about the central source and the formation of low - mass objects in general .
the results are summarized in 5 .
submillimeter interferometric observations of iras 04158 + 2805 ( @xmath16 , @xmath17 [ j2000 ] ) were conducted in both the extended and compact configurations of the submillimeter array ( sma ; * ? ? ?
* ) on 2006 december 8 and 2007 january 27 , respectively . in these configurations , the 8 sma antennas ( 6 m diameter each ) spanned baselines from @xmath015 - 225 m .
double sideband receivers with a total bandwidth of 4ghz were tuned to an effective continuum frequency of 339.9ghz ( 883@xmath1 m ) .
the correlator was simultaneously configured to observe the co @xmath2=3@xmath32 transition at 345.796ghz at a spectral resolution of 0.70 km s@xmath18 .
the observations cycled between iras 04158 + 2805 and two gain calibrators ( 3c111 and 3c84 ) , with 20 minutes on target and 10 minutes on calibrator .
the data were obtained in excellent observing conditions , with zenith optical depths at 225ghz of 0.05 - 0.07 .
the raw visibilities were calibrated with the idl - based mir package .
passband calibration was conducted with bright quasars ( 3c273 , 3c279 ) , and complex gain calibration was performed with 3c111 . using 3c84 as a consistency check on the gain solution , we estimate that phase noise generates an effective seeing " of @xmath1901 .
titan and callisto were used to set the absolute flux scale , which is accurate to @xmath010% .
the miriad package was utilized for the standard tasks of fourier inversion , deconvolution , and imaging of the calibrated visibilities .
continuum and line maps from the combined datasets were made with natural weighting , providing a synthesized beam fwhm of @xmath20 at a position angle of 92 .
iras 04158 + 2805 was also mapped at 450 and 850@xmath1 m using the jiggle - mode of the scuba camera at the 15 m james clerk maxwell telescope ( jcmt ) on 2003
november 12 in dry , stable conditions ( @xmath21 at 225ghz ) .
flux calibration accurate to @xmath010% and 25% at 850 and 450@xmath1 m , respectively , was achieved with observations of neptune , and pointing was referenced to dg tau and crl 618 .
the bright emission , @xmath22jy at 850@xmath1 m and @xmath23jy at 450@xmath1 m , was not resolved at either wavelength in 15 and 9 beams , respectively .
an additional map at 350@xmath1 m was obtained with the sharc - ii camera at the 10 m caltech submillimeter observatory ( cso ) on 2004 october 3 in similarly dry weather .
those data confirm an unresolved source ( 9 beam ) with a total flux density of @xmath24jy .
all errors on flux densities include absolute calibration uncertainties ( 10% at 850@xmath1 m , and @xmath025% at shorter wavelengths ) .
optical images of iras 04158 + 2805 in the @xmath25-band were obtained with the university of hawaii 2.2 m telescope on 2004 october 6 .
after the standard reduction of ccd data , the @xmath26s images were co - added and convolved with a 3-pixel ( 07 ) fwhm gaussian kernel to match the typical seeing for the observations and improve sensitivity to faint emission .
the observations were not photometric .
images of the gas and dust surrounding iras 04158 + 2805 are shown in figure [ images ] .
the optical @xmath25-band ( @xmath00.7@xmath1 m ) image in figure [ images]a reveals a high aspect ratio dark lane of _ absorbing _ material oriented roughly e - w ( shown as grayscale - filled contours ) that can be traced nearly 8 in diameter .
first discovered by ( * ? ? ?
* see also watson et al .
2007 ) , this silhouette lane obscures the faint nebulosity associated with the nearby herbig ae star v892 tau . also pointed out by @xcite , the image in figure [ images]a shows the conical nebulae perpendicular to the dark lane that are often produced by starlight scattered off the surface of flattened dust structures .
figure [ images]b shows the first high resolution image of the thermal continuum emission from iras 04158 + 2805 , at a wavelength of 883@xmath1 m .
the morphology of that emission is remarkably similar to the optical silhouette , extending roughly 8 across , corresponding to a diameter of 1120au at the distance of the taurus clouds ( @xmath27pc ; * ? ? ?
note that @xcite trace @xmath28-band scattered light for this source to twice this diameter ( @xmath016 , or 2240au ) .
the difference compared to the @xmath25-band image in figure [ images]a is related to the comparatively poor sensitivity of our observations , as was also noted for the infrared scattered light images in their study . because the submillimeter emission was clearly unresolved in the 350 and 450@xmath1 m maps , we can infer that it has a diameter of @xmath29 ( 1260au ) .
the absence of very short antenna spacings implies that the interferometric observations have significantly diminished sensitivity to structures larger than @xmath010(e.g .
* ) . therefore , because of the diminished sensitivity to faint submillimeter emission on large scales , there is no reason to suspect that the sized inferred here is physically inconsistent with the scattered light image presented by @xcite .
the integrated flux density at 883@xmath1 m is @xmath30mjy , consistent with the scuba measurement .
a fit of the sma visibilities with an elliptical gaussian model indicates an inclination of @xmath31 ( where 90 corresponds to edge - on ) at a position angle of @xmath32 ( measured east of north ) , in excellent agreement with the constraints imposed by modeling the scattered light @xcite . fits with a ring or pedastal emission model
give the same results .
the high surface brightness at large distances from the central source and large spatial extent of the continuum emission are fairly unique .
the steep drop of the visibilities over a short range of baseline lengths , as shown in figure [ vis ] , indicates that essentially all of the emission is concentrated at large spatial scales ( radii @xmath33au ) .
more significantly , the visibilities show a pronounced null at @xmath030 - 40k@xmath34 and very little flux on longer baselines .
these are definitive signatures of a brightness distribution with a central hole " , inside of which there is significantly diminished intensity . assuming a simple ring geometry
, the location of the null suggests an inner radius of @xmath0250 - 300au ( e.g. , * ? ? ?
asymmetries in the submillimeter map confirm this geometry , with emission peaks centered roughly @xmath35 ( @xmath0265au ) from the image center .
the western lobe is approximately 50% brighter than the eastern lobe .
these asymmetries are real , as they are clearly detected in 3 separate observing runs at the same location and intensity ( the 2 detailed here , and the hybrid compact / extended observations described by * ? ? ?
a velocity - integrated intensity ( 0@xmath36 moment ) map of the co @xmath2=3@xmath32 emission from iras 04158 + 2805 is shown in figure [ images]c .
the co emission morphology is significantly different than the optical silhouette and submillimeter continuum , occupying a region @xmath04 on a side with a central , slightly elongated peak .
the integrated intensity is @xmath37jy km s@xmath18 ( using 9 channels for a total width of 6.3 km s@xmath18 ) .
fainter co emission extends above and below the plane of the continuum emission in an
x "- shaped pattern out to @xmath02 from the image center .
there is a steep drop in the co intensity along the plane of the silhouette that corresponds precisely to the positions of the submillimeter continuum peaks described above .
the kinematic structure of the co associated with iras 04158 + 2805 is exhibited in figure [ chanmaps ] .
these channel maps clearly show a co velocity gradient along the major axis of the continuum emission and silhouette ( i.e. , in the east - west direction ) , similar to the pattern expected for rotation around a central source .
estimating the masses of young stars and brown dwarfs generally requires reference to theoretical models of their structural evolution .
the quantitative reliability of such models for individual stars at ages @xmath381myr is highly uncertain , particularly for the low end of the stellar mass spectrum where complicated convection physics , dusty atmospheres , and uncertain molecular opacities and initial conditions can strongly affect the observational diagnostics ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
dynamical constraints on @xmath15 from the orbital properties of either a companion star ( e.g. , * ? ? ? * ) or gas in a circumstellar disk ( e.g. , * ? ? ?
* ) provide an extremely valuable independent check that can potentially be used to calibrate these models ( see * ? ? ?
@xcite report the only such measurements of young @xmath0m6 dwarfs to date , for an eclipsing binary system in orion . with the first spatially resolved measurements of molecular gas in apparent rotation around a young low - mass object with similar spectral type , the co observations of iras 04158 + 2805 presented here can provide another such estimate of @xmath15 , although admittedly with a great deal more uncertainty .
figure [ pv_spec ] shows the position - velocity diagram of the co @xmath2=3@xmath32 line emission for iras 04158 + 2805 , where the angular offset is the distance from the observed phase center ( such that positions to the east have positive values ) and the velocity offset is taken relative to the systemic value ( @xmath39 km s@xmath18 ) .
overlaid on the diagram is an inclined ( @xmath40 ) keplerian rotation profile that matches the general kinematic trend in the data rather well ( thick curve ) , with a central point mass of @xmath41m@xmath5 .
additional rotation profiles for @xmath42 and 0.45m@xmath5 are also shown for reference .
it should be noted that these profiles assume pure keplerian orbital motion around a central source , which may not be the case in the more complex iras 04158 + 2805 environment .
for now , we assume that the dominant kinematic trend is such rotation , but potential complications will be addressed below . a comparison of this dynamical @xmath15 estimate with predictions from stellar evolution models is a challenge because of uncertainties in the luminosity ( due to scattered light contamination at short wavelengths ) , extinction , and effective temperature for the central source
. recent analysis of the scattered light spectrum from iras 04158 + 2805 suggest that this source is cool , with initial spectral type estimates of m6@xmath43 @xcite refined to [email protected] ( optical ) and [email protected] ( infrared ; * ? ? ?
* ) .k7 to m3 @xcite .
] using the latter classifications and the @xcite empirical effective temperature scale , we estimate @xmath45k .
unfortunately , large uncertainties in the extinction , where estimates range from @xmath46 - 16 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , make a luminosity determination difficult .
however , we can use pre - main - sequence models for an assumed age in the inferred temperature range to estimate a central mass . using the @xmath47 range quoted above and an age of @xmath01myr ,
the @xcite and @xcite models indicate @xmath48 - 0.16m@xmath5 .
this mass range does not include the additional uncertainties on the adopted effective temperature scale or the unknown age of the source .
the @xmath15 values inferred from the pre - main - sequence models lie roughly a factor of 2 below the nominal value that best describes the kinematics of the co gas inferred above .
the unquoted and poorly understood uncertainties in both measurements may be the simple explanation for this apparent discrepancy . without a detailed physical model for the local gas structure around iras 04158 + 2805 ,
it is difficult to estimate a statistically acceptable range of @xmath15 from the co kinematics ; there may very well be overlap with the higher end of the @xmath15 range from the pre - main - sequence models .
the effective temperature scale adopted above is also uncertain .
a @xmath0500k increase in the effective temperature ( to a value that is typically associated with m2 stars ) would reconcile the @xmath15 estimates .
indeed , using an independent method of fitting infrared spectra , @xcite infer @xmath49k for the iras 04158 + 2805 central source , leading to @xmath50 - 0.45m@xmath5 in the aforementioned models .
moreover , @xcite have shown that pre - main - sequence models generally tend to underpredict @xmath15 compared to dynamical measurements in this low - mass range , although typically only by @xmath51% .
in their recent study that explicitly includes the effects of scattered starlight , @xcite argue for a very luminous central source , @xmath52l@xmath5 .
while this value must be somewhat degenerate with the dust properties and structure assumed in their model , it is clearly the most robust estimate available .
the pre - main - sequence models imply that such a luminous cool source would be very young , with an age significantly less than 1myr .
the model mass tracks at such ages are not well understood in light of the uncertain initial conditions .
clearly , the aforementioned uncertainties can be solely responsible for the apparently different @xmath15 estimates from the co kinematics and pre - main - sequence models .
alternatively , there is also an appealing and simple physical explanation @xmath3 the iras 04158 + 2805 central source may be a roughly equal - mass binary . in this scenario ,
the @xmath15 estimate from the co kinematics refers to the total mass of the stellar system ( i.e. , the mass interior to the gas that produces the line emission ) , while the estimate from the pre - main - sequence models assumes only a single star .
therefore we would expect to see both a high luminosity ( as claimed by * ? ? ? * ) and @xmath15(co ) @xmath53(models ) .
a variety of surveys suggest that such binaries are fairly common , @xmath035 - 45% for m dwarfs @xcite and @xmath010 - 30% for cooler objects
( e.g. , * ? ? ?
* ) . while it is difficult to definitively state the properties of the central source(s )
, we can further assess the nature of iras 04158 + 2805 by analyzing the structure of the material that surrounds it .
given the elongated structures shown in figure [ images ] , it is reasonable to suggest that the gas and dust around iras 04158 + 2805 reside in an exceptionally large circumstellar disk .
@xcite have recently demonstrated that they are able to fit high - quality scattered light images and the sed for this source with a simple disk structure , and without the need to invoke any envelope component . with the new data presented here , it is worthwhile to re - examine the structure of this material .
iras 04158 + 2805 is exceptionally bright at submillimeter wavelengths , with an 850@xmath1 m luminosity larger than @xmath085% of all other sources detected in the @xcite taurus survey .
breaking that sample down further , the iras 04158 + 2805 submillimeter emission is brighter than @xmath090% of class ii sources ( disk only ) , but is comparable to the median class i source ( disk + envelope ) . if we adopt the standard emissivity for disks ( 0.034@xmath54 g@xmath18 at 883@xmath1 m , including a gas - to - dust ratio of 100 ) and assume that the submillimeter continuum emission is optically thin and has a characteristic temperature of @xmath020k ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the integrated sma flux density suggests a large total mass of gas and dust is present around iras 04158 + 2805 , @xmath00.03m@xmath5 .
this value is in good agreement with the disk mass range inferred by @xcite , @xmath55 - 0.04m@xmath5 .
estimating masses for circumstellar material in this way is inherently uncertain ( perhaps by an order of magnitude ) , due to the challenge of observationally constraining the optical properties of dust grains and the assumed mass conversion of a trace constituent ( dust ) to the presumably dominant species ( molecular gas ) . perhaps more compelling ( and easier to interpret ) is the unusual geometry that is observed for the dust structure around iras 04158 + 2805 .
the enormous size of the scattered light nebulae noted by ( * ? ? ?
* diameter ) and the submillimeter continuum emission presented here ( 1120au diameter ) is exceptionally rare for circumstellar disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , rivaled only by the circumbinary disk around uy aur @xcite .
on the contrary , the observed size is fairly typical ( if not on the small end ) for larger - scale envelopes ( e.g. , * ? ? ? * ; * ? ? ? * ) .
moreover , our resolved observations of the submillimeter continuum reveal that a large central region ( perhaps @xmath0500au in diameter ) has significantly diminished intensity at 883@xmath1 m , suggesting a ring - like or toroidal geometry for the dust structure .
similar geometries have been inferred for a handful of circumstellar disks , and a variety of underlying causes are possible .
the so - called transition " disks have their inner regions ( out to a few tens of au in radius ) largely cleared of observable material , presumably by disk evolution processes like particle growth , photoevaporation , or dynamical interactions with a young planetary system ( e.g. , * ? ? ?
* ; * ? ? ?
however , given the large spatial scale of the central depression and the absence of a corresponding dip in the infrared part of the iras 04158 + 2805 sed , as shown in figure [ sed ] , this scenario is unlikely .
alternatively , extremely dense disks with edge - on orientations can potentially render even the submillimeter emission optically thick , resulting in a central depression in its brightness distribution ( e.g. , * ? ? ?
although the iras 04158 + 2805 dust structure is not oriented edge - on , perhaps longer pathlengths through the exceptionally large structure can provide compensating column densities . if the bulk of the submillimeter emission is optically thick , the spectrum should have a spectral index @xmath56 , where @xmath57 @xcite .
power - law fits to the submillimeter photometry are significantly steeper than the optically thick case , with @xmath58 ( 0.45 - 1.3 mm ; @xmath59 for 0.85 - 1.3 mm ) . at these wavelengths ,
such a steep spectrum is only noted for a few percent of class ii sources in taurus and @xmath60 oph , but is more common ( 25 - 35% ) in the envelopes around class i objects @xcite .
the observed steep submillimeter sed would be difficult to produce unless a large fraction of the dust was optically thin at those wavelengths . as noted by @xcite ,
a clear test of this possibility can be made with resolved continuum observations at @xmath61@xmath1 m . in such data , the material should become optically thick only at significantly smaller radii than at 883@xmath1 m , resulting in a more centrally concentrated emission profile ( or , equivalently , the null in the visibilities should be detected at a comparatively longer deprojected baseline ) .
note that if the material inside this central depression is optically thick at 883@xmath1 m , an enormous reservoir of mass is not accounted for in the above estimate . a third possibility for explaining the submillimeter continuum depression is that the emitting dust resides in a circumbinary ring .
the material around close binary stars is expected to be dynamically cleared on scales similar to the physical separation of the stellar components ( e.g. , * ? ? ?
* ; * ? ? ?
numerical simulations of this dynamical interaction suggest that a circumbinary disk gap would be cleared out to a radius @xmath01.5 - 3@xmath11 the projected semimajor separation of the stellar components , and the individual circumstellar disks would be truncated at a radius @xmath00.2 - 0.5@xmath11 the projected semimajor separation ( e.g. , * ? ? ?
if the 883@xmath1 m continuum depression noted here is caused by such clearing , we would expect the projected binary separation to be @xmath090 - 180au ( @xmath006 - 13 ) for a reasonable range of eccentricities , and the remnant individual circumstellar disks might be truncated at radii of @xmath020 - 90au .
this configuration is similar to the uy aur system , where a ring with @xmath020 diameter is detected in both scattered light and co around a roughly equal - mass binary with a similar projected separation @xcite .
the primary difference is in the masses involved ; the stellar mass in the iras 04158 + 2805 system is @xmath04@xmath11 less than for uy aur , but the circumstellar mass ( or rather 850@xmath1 m luminosity ) is @xmath04@xmath11 higher . while this possibility is certainly appealing in its reconciliation of both the information on the central source ( see 4.1 ) and
the structure of the gas and dust emission , testing the multiplicity of iras 04158 + 2805 will be a challenge given the obscuration of the central region by the dust structure .
decomposing high resolution spectra of scattered starlight may represent the best opportunity in this case .
the iras 04158 + 2805 sed , shown in figure [ sed ] , exhibits a strong infrared excess with a flat or slightly rising slope from 2 - 25@xmath1 m .
a variety of solid - state absorption features are noted in higher resolution infrared spectra , including h@xmath62o ice ( 3@xmath1 m ; not shown ) , silicates ( 10@xmath1 m ) , and co@xmath62 ice ( 15@xmath1 m ; * ? ? ?
* ; * ? ? ?
the bright and steep submillimeter spectrum has already been discussed above . while all of these sed properties are not necessarily inconsistent with a large , cold ,
highly inclined circumstellar disk ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , taken together they are more common for class i / flat - spectrum objects that harbor a remnant accretion envelope ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite demonstrate that the iras 04158 + 2805 sed features are strikingly similar to those for several other class i sources in taurus ( see their fig . 2 ) and
can be successfully reproduced in detail with an envelope model ( although the new observations of this source warrant a modification of their geometric assumptions ) .
so , all of the observational evidence presented here is also consistent with a flattened envelope structure around iras 04158 + 2805 .
such a flattened envelope geometry is predicted by molecular cloud core collapse models that incorporate rotation or magnetic fields @xcite , and have been clearly detected in other cases , albeit on significantly larger spatial scales than noted here ( e.g. , * ? ? ?
in this scenario , a partially evacuated outflow cavity at the center of the flattened envelope may explain the central depression in the submillimeter continuum data ( e.g. , * ? ? ? * their models 3 or 4 ) .
@xcite indicate that iras 04158 + 2805 drives a substantial h@xmath63 jet oriented in the north - south direction , perpendicular to the observed dust structure .
the velocity - integrated co @xmath2=3@xmath32 morphology does resemble the bases of some molecular outflows ( e.g. , * ? ? ?
* ) , but the absence of both a velocity gradient in the proposed flow direction ( north - south ) and larger scale molecular outflow signatures @xcite are difficult to reconcile with that interpretation .
nevertheless , some contribution to the co emission from an outflow and even the envelope ( i.e. , motions unrelated to keplerian rotation ) would complicate the observed spatio - kinematics .
the dynamical estimate of @xmath15 discussed above should be treated with caution with regards to this possibility .
of course , the envelope and binary ideas are not mutually exclusive , and dynamical clearing of the inner envelope and disks is still a viable alternative .
the images in figures [ images]c and [ chanmaps ] demonstrate that , unlike the submillimeter continuum , the co emission is centrally concentrated ; most of the line emission is from radii inside the toroidal structure traced in the continuum . while this may suggest that the bulk of the line emission is generated in the disk(s ) interior to that structure , it is difficult to provide a reliable origin without a more complete physical model for the iras 04158 + 2805 environment .
future studies of this source should focus on searching for definitive molecular outflow signatures , resolving dense gas tracers that are commonly associated with envelopes ( e.g. , * ? ? ?
* ) , observing co transitions at higher spectral resolution to provide a better tracer of the velocity field , and reconciling the optical / near - infrared scattered light observations with the submillimeter results in an effort to converge on a consistent model of the local gas and dust structure . considering the low stellar mass range implied for iras 04158 + 2805
, it is difficult to reconcile the large mass reservoir and size of this environment with any formation scenario that relies on it being dynamically ejected from a more massive system ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . rather , the observed gas and dust structure is more consistent with the standard scenario , where an accretion disk and perhaps a remnant envelope developed around a single ( or binary ) source during the gravitational collapse of a molecular cloud core fragment .
regardless of the precise value of @xmath15 , iras 04158 + 2805 is a remarkable example of the fact that the standard picture for star formation worked out for higher - mass stars is also applicable at the low end of the stellar mass function .
we have presented new submillimeter observations of the molecular gas and dust around iras 04158 + 2805 , a young low - mass object in the taurus star - forming region .
the data reveal a complex and surprising environment , from which we infer the following : \1 .
a high aspect ratio ( @xmath64 ) dust structure with a projected diameter of @xmath01120au and estimated ( gas+dust ) mass of @xmath00.03m@xmath5 is noted in optical absorption and submillimeter continuum emission images .
the resolved 883@xmath1 m data clearly demonstrate a central depression in the emission , suggesting a ring - like geometry with an inner radius of @xmath0250 - 300au . a variety of structures that may explain the observations are investigated , with a flattened envelope or circumbinary ring perhaps the most likely .
spatially resolved co @xmath2=3@xmath32 line emission shows a velocity gradient along the major axis of the dust structure . assuming the gas traced by this transition is in keplerian orbital motion around a central source , the spatio - kinematic properties of the emission are used to estimate a central mass , @xmath65m@xmath5 ( with @xmath050% uncertainty ) .
this estimate is roughly a factor of 2 higher than the mass range inferred from a reasonable effective temperature range and pre - main - sequence models , @xmath00.09 - 0.16m@xmath5 , but the significant uncertainties on both measurements do not formally rule out their mutual consistency .
however , a roughly equal - mass binary system represents an intriguing possibility in this case , as it would naturally account for this possible @xmath15 discrepancy , a very high luminosity , and the ring - like geometry of the dust structure .
the large spatial extent and mass of the complex gas and dust environment surrounding iras 04158 + 2805 would be difficult to retain in the dynamical ejection models used to explain the formation of such low - mass objects . instead , this case appears to illustrate that the standard gravitational collapse models applied to stars with masses up to @xmath02 - 3m@xmath5 can be extrapolated down to objects with @xmath66 - 0.3m@xmath5 .
more detailed models of the iras 04158 + 2805 gas and dust structure and specialized observations are recommended .
we thank f. mnard and g. duchne for encouraging our interest in this source , and elise furlan for kindly providing the mid - infrared spectrum .
we are grateful to an anonymous referee for a helpful , constructive review and for emphasizing the potential circumbinary nature of iras 04158 + 2805 .
the sma is a joint project between the smithsonian astrophysical observatory and the academia sinica institute of astronomy and astrophysics and is funded by the smithsonian institution and the academia sinica .
support for this work was provided by nasa through hubble fellowship grant # hf-01203.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 .
mcl and kna acknowledge support for this work from nsf grant ast-0407441 .
allers , k. n. , kessler - silacci , j. e. , cieza , l. a. , & jaffe , d. t. 2006 , , 644 , 364 andrews , s. m. , & williams , j. p. 2005
, , 635 , 1134 2007 , , 659 , 705 ( 2007a ) 2007 , , 671 , 1800 ( 2007b ) arce , h. g. , & sargent , a. i. 2006 , , 646 , 1070 armitage , p. j. , & clarke , c. j. 1997 , , 285 , 540 artymowicz , p. , & lubow , s. h. 1994 , , 421 , 651 baraffe , i. , chabrier , g. , allard , f. , & hauschildt , p. h. 1998 , , 337 , 403 , 2002 , , 382 , 563 bate , m. r. , bonnell , i. a. , & bromm , v. 2002 , , 332 , l65 beck , t. l. 2007 , , 133 , 1673 beckwith , s. v. w. , sargent , a. i. , chini , r. , & gsten , r. 1990 , , 99 , 924 beckwith , s. v. w. , & sargent , a. i. 1991 , , 381 , 250 beichman , c. a. , et al . 1986 , , 307 , 337 bontemps , s. , andr , p. , terebey , s. , & cabrit , s. 1996 , , 311 , 858 boogert , a. c. , et al . 2004 , , 154 , 359 bouy , h. , et al . 2008 , , in press ( astro - ph/0803.2051 ) brown , j. m. , blake , g. a. , qi , c. , dullemond , c. p. , & wilner , d. j. 2008 , , 675 , l109 burgasser , a. j. , reid , i. n. , siegler , n. , close , l. , allen , p. , lowrance , p. , & gizis , j. 2007 , in protostars & planets v , eds . b. reipurth , d. jewitt , & k. keil ( univ .
arizona press : tucson ) , 427 chiang , e. i. , & goldreich , p. 1999
, , 519 , 279 close , l. m. , et al .
1998 , , 499 , 883 cutri , r. m. , et al . 2003 , 2mass all - sky point source catalog ( pasadena : ipac ) dantona , f. , & mazzitelli , i. 1994 , , 90 , 467 1997 , mem .
soc . astron .
italiana , 68 , 807 doppmann , g. w. , greene , t. p. , covey , k. r. , & lada , c. j. 2005 , , 130 , 1145 duvert , g. , dutrey , a. , guilloteau , s. , mnard , f. , schuster , k. , prato , l. , & simon , m. 1998 , , 332 , 867 eisner , j. a. , hillenbrand , l. a. , carpenter , j. m. , & wolf , s. 2005 , , 635 , 396 elias , j. h. 1978 , , 224 , 857 fischer , d. , & marcy , g. 1992 , , 396 , 178 furlan , e. , et al .
2008 , , 176 , 184 galli , d. , & shu , f. h. 1993 , , 417 , 220 glauser , a. m. , mnard , f. , pinte , c. , duchne , g. , gdel , m. , monin , j .- l . , &
padgett , d. l. 2008 , , in press ( astro - ph/0804.3483 ) gomez , m. , whitney , b. a. , & kenyon , s. j. 1997 , , 114 , 1138 guilloteau , s. , dutrey , a. , & simon , m. 1999 , , 348 , 570 guilloteau , s. , dutrey , a. , pety , j. , & gueth , f. 2008 , , 478 , l31 hillenbrand , l. a. , & white , r. j. 2004 , , 604 , 741 ho , p. t. p. , moran , j. m. , & lo , k. y. 2004 , , 616 , l1 hughes , a. m. , wilner , d. j. , calvet , n. , dalessio , p. , claussen , m. j. , & hogerheijde , m. r. 2007 , , 664 , 536 hughes , a. m. , wilner , d. j. , qi , c. , & hogerheijde , m. r. 2008 , , 678 , 1119 ireland , m. j. , & kraus , a. l. 2008 , , 678 , l59 jensen , e. l. n. , mathieu , r. d. , & fuller , g. a. 1994 , , 429 , l29 jrgensen , j. k. , et al .
2007 , , 659 , 479 kenyon , s. j. , brown , d. i. , tout , c. a. , & berlind , p. 1998
, , 115 , 2491 klein , r. , apai , d. , pascucci , i. , henning , t. , & waters , l. b. f. m. 2003 , , 593 , l57 lay , o. p. , carlstrom , j. e. , & hills , r. e. 1997 , , 489 , 917 looney , l. w. , & mundy , l. g. 2000 , , 529 , 477 looney , l. w. , tobin , j. j. , & kwon , w. 2007 , , 670 , l131 luhman , k. l. 1999 , , 525 , 466 2006 , , 645 , 676 luhman , k. l. , stauffer , j. r. , muench , a. a. , rieke , g. h. , lada , e. a. , bouvier , j. , & lada , c. j. 2003 , , 593 , 1093 luhman , k. l. , et al .
2006 , , 647 , 1180 luhman , k. l. , joergens , v. , lada , c. , muzerolle , j. , pascucci , i. , & white , r. 2007 , in protostars & planets v , eds .
b. reipurth , d. jewitt , & k. keil ( univ .
arizona press : tucson ) , 443 luhman , k. l. , & rieke , g. h. 1998 , , 497 , 354 mathieu , r. d. 1994 , , 32 , 465 menshchikov , a. b. , & henning , t. 1997 , , 318 , 879 montalbn , j. , dantona , f. , kupka , f. , & heiter , u. 2004 , , 416 , 1081 motte , f. , & andr , p. 2001
, , 365 , 440 natta , a. , & testi , l. 2001 , , 376 , l22 pascucci , i. , apai , d. , henning , t. , & dullemond , c. p. 2003 , , 590 , l111 padoan , p. , & nordlund , .
2004 , , 617 , 559 pitu , v. , dutrey , a. , guilloteau , s. , chappilon , e. , & pety , j. 2006 , , 460 , l43 pitu , v. , dutrey , a. , & guilloteau , s. 2007 , , 467 , 163 pontoppidan , k. m. , et al .
2005 , , 622 , 463 2008 , , 678 , 1005 potter ,
d. e. , et al .
2000 , , 540 , 422 reid , i. n. , & gizis , j. 1997 , , 113 , 2246 reipurth , b. , & clarke , c. 2001 , , 122 , 432 scholz , a. , jayawardhana , r. , & wood , k. 2006 , , 645 , 1498 simon , m. , dutrey , a. , & guilloteau , s. 2000 , , 545 , 1034 stassun , k. g. , mathieu , r. d. , & valenti , j. a. 2006 , nature , 440 , 311 terebey , s. , shu , f. h. , & cassen , p. 1984
, , 286 , 529 umbreit , s. , burkert , a. , henning , t. , mikkola , s. , & spurzem , r. 2005 , , 623 , 940 watson , d. m. , et al .
2004 , , 154 , 391 watson , a. m. , stapelfeldt , k. r. , wood , k. , mnard , f. 2007 , in protostars & planets v , eds . b. reipurth , d. jewitt , & k. keil ( univ .
arizona press : tucson ) , 523 white , r. j. , & hillenbrand , l. a. 2004 , , 616 , 998 whitney , b. a. , wood , k. , bjorkman , j. e. , & wolff , m. j. 2003 , , 591 , 1049 wilner , d. j. , & welch , w. j. 1994 , , 427 , 898 wolf , s. , schegerer , a. , beuther , h. , padgett , d. l. , & stapelfeldt , k. r. 2008 , , 674 , l101 | we present high spatial resolution submillimeter array observations and supplementary single - dish photometry of the molecular gas and dust around iras 04158 + 2805 , a young source with spectral type m5-m6 in the taurus star - forming region . a bright , highly elongated dust structure that extends 8 ( @xmath01120au ) in diameter
is revealed in a 883@xmath1 m thermal continuum image .
this emission geometry is in good agreement with optical observations that show a similar structure in absorption , aligned perpendicular to bipolar scattered light nebulae .
however , the interferometric data also clearly demonstrate that the submillimeter continuum emission is not centrally concentrated , but rather appears to have a toroidal geometry with substantially lower intensities inside a radius of @xmath0250 - 300au .
spatially resolved emission from the co @xmath2=3@xmath32 transition exhibits a velocity gradient along the major axis of the dust structure . if this kinematic pattern is interpreted as the signature of rotation around a central object , a relatively low mass is inferred ( @xmath4m@xmath5 , with a @xmath050% uncertainty ) .
we discuss several possible explanations for the observed gas and dust environment around iras 04158 + 2805 , including a flattened envelope with an outflow cavity and a large circumbinary ring .
this source offers unique views of the gas and dust environment surrounding a young low - mass stellar system .
its properties are generally not commensurate with formation scenarios for such low - mass objects that rely on dynamical ejection , but rather confirms that a single mechanism @xmath3 molecular cloud core collapse and fragmentation @xmath3 can produce stars over a wide range of stellar masses ( at least an order of magnitude ) . |
for a few decades organic charge - transfer ( ct ) salts built of donor and acceptor molecular complexes have attracted a lot of attention due to the variety of ground states in their phase diagrams @xcite .
application of external or chemical pressure can lead to antiferromagnetic insulating , charge ordered , spin - density wave , spin liquid , or unconventional superconducting ground states .
tendencies in the dimensionality of the electronic transport are often determined by the choice of conducting molecules : compounds containing tmttf ( tetramethyltetrathiafulvalene ) molecules , for example , are typically one - dimensional @xcite , whereas several phases of bedt - ttf ( bisethylenedithio - tetrathiafulvalene ) based salts show two - dimensional behavior .
however , the arrangement of the ( donor ) molecules in these complexes is decisive . among the bedt - ttf family of ct salts , many different packings classified as @xmath9 , @xmath8 , @xmath10 , @xmath11 , @xmath12 , @xmath13 , @xmath3 , and @xmath14
have been experimentally realized and a wide range of different physical properties was found @xcite .
depending on the preparation conditions , different polymorphs of one structure can be synthesized ; for example , ( bedt - ttf)@xmath0i@xmath1 crystallizes in @xmath9 , @xmath10 , @xmath14 , and @xmath3 forms @xcite .
polymorphs provide an opportunity to explore the influence of the packing motif on the electronic properties .
effects originating from differences in the anion layer composition can be excluded in this case since this layer remains unaltered in the polymorph family .
here we consider the polymorph charge - transfer salt family ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) ( see fig .
[ fig : structure ] ) first synthesized by schlueter and collaborators @xcite .
tce stands for 1,1,2-trichloroethane and in the following we will make use of the common abbreviation et for bedt - ttf .
these systems show a metallic behavior at low temperatures and a @xmath5 of @xmath15 k to a superconducting state was measured for the single - layered compound ( fig .
[ fig : structure ] ( a ) ) where the et molecules form dimers arranged in a so - called @xmath3 pattern .
the term @xmath16 phase was coined for this structure , with the index @xmath17 referring to the low @xmath5 .
structural refinement of the other two multiphase polymorphs @xcite ( figs .
[ fig : structure ] ( b ) , ( c ) and fig . [ fig : structure_overview ] ) showed the presence of charge - ordered layers in @xmath8 packing between the @xmath3-type layers ; @xmath8 phases have also been characterized as mott - hubbard insulators @xcite .
even though as insulating layers they do not contribute directly to superconductivity , their existence seems to enhance the superconducting transition temperature in ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) .
while the dual - layered @xmath3-@xmath18 compound exhibits superconductivity at a critical temperature of @xmath6 k , which is approximately 3.5 times higher than the @xmath5 of the @xmath16 phase , the four - layered @xmath3-@xmath19 phase shows superconductivity at @xmath7 k and therefore belongs to the organic superconductors with the highest measured critical temperatures . in this work
we perform density functional theory ( dft ) calculations for the three polymorphs .
we especially focus on the effects of the @xmath8 layers on the electronic properties of the @xmath3 layers in the dual and four - layered systems and perform a comparative analysis of the three systems in terms of _ ab initio _ derived tight - binding hamiltonians using the projective wannier method .
while all three systems show apparently similar @xmath3 bands , the charge ordering in the @xmath8 layer in @xmath3-@xmath18 and @xmath3-@xmath19 influences significantly the magnitude of the hoppings in the conducting @xmath3 layer .
analysis of the degree of frustration within a minimal triangular lattice model hints to the different superconducting @xmath5 in these systems .
ag(cf@xmath1)@xmath2(tce ) . ] -@xmath18-(bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) together with cuts illustrating the two alternating et patterns . in ( b ) and
( c ) , colored outlines mark symmetry inequivalent et molecules . ]
-type packing in the low @xmath5 ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) compound .
( b ) @xmath3-type packing in the two high @xmath5 compounds .
@xmath18-type packing in the two - layer compound .
( c ) @xmath19-type packing in the four - layer compound .
symmetry inequivalent molecules are marked by numbers and colored differently .
r ( l ) indicate tilting of the molecules to the right ( left ) in the @xmath8 layers . ]
* @xmath3 phase.- * the single - layered phase crystallizes in the orthorhombic @xmath20 space group and its unit cell contains two donor layers which are separated by an insulating layer consisting of the anion [ ag(cf@xmath1)@xmath2]@xmath22 and the solvent tce @xcite
. however , the anion layer is disordered . in order to simplify the density functional theory calculations we choose one of the two symmetry - allowed orientations of the anion and lower the symmetry to @xmath23 ;
the corresponding simplified structure is shown in fig .
[ fig : structure ] ( a ) .
note that for better comparability with the @xmath3-@xmath18 and @xmath3-@xmath19 phases we denote with @xmath24 the short in - plane axis , with @xmath25 the long in - plane axis and with @xmath26 the stacking direction ( even though the original @xmath20 structure has a @xmath25 stacking direction ) .
in the two donor layers , the et molecules are in the @xmath3 packing motif ( see fig . [
fig : layerscheme ] ( a ) ) but with alternating tilt of the molecules with respect to the acceptor layer .
* @xmath3-@xmath18 phase.- * the triclinic crystal structure @xmath27 of the dual - layered compound is shown in fig .
[ fig : structure ] ( b ) . compared to the single - layered phase ,
the layer in the center of the unit cell is here replaced by an @xmath8 packed layer .
this packing motif is characterized by molecules that are lined up on a rectangular lattice ( see fig . [
fig : structure_overview ] and fig .
[ fig : layerscheme ] ( c ) ) .
the rows of molecules along @xmath25 alternate between a right and a left tilt with respect to the @xmath24 axis .
this leads to the characteristic cross pattern when viewed along @xmath24 ( fig .
[ fig : structure_overview ] and fig .
[ fig : layerscheme ] ( c ) ) .
in contrast to other @xmath8 packed structures @xcite , no dimerization and therefore also no shift along the long unit cell axis has been observed .
the low symmetry of the space group means that in the @xmath3 layer , the two et molecules in a dimer are still related by inversion symmetry but there are two symmetry inequivalent dimers , and in the @xmath8 layer there are even four inequivalent et molecules ( see fig .
[ fig : structure_overview ] and figs .
[ fig : layerscheme ] ( b ) and ( c ) ) . * @xmath3-@xmath19 phase.- * the polymorph with the highest superconducting critical temperature has the largest unit cell @xcite , where two @xmath3 layers alternate with two @xmath8 layers ( fig .
[ fig : structure ] ( c ) ) .
due to the monoclinic @xmath28 symmetry ( space group no . 14 ) , every second @xmath3(@xmath8 ) layer is shifted by half the lattice vector @xmath24 .
note that dft codes usually allow only one setting of space group no .
14 ; we therefore perform the calculations in a @xmath23 setting . as in the case of @xmath3-@xmath18
, the monoclinic space group leads to two symmetry inequivalent ( et)@xmath0 dimers in the @xmath3 layer ; the two et molecules in a dimer are related by inversion symmetry ( see fig .
[ fig : layerscheme ] ( b ) ) .
the symmetry also leads to a checkerboard pattern of symmetry related et molecules in the @xmath8 layer as displayed in fig .
[ fig : layerscheme ] ( d ) .
the blue bands are determined by the projective wannier function method and originate from the et homo orbitals . ]
-@xmath18 phase .
the wannier bands ( blue symbols for @xmath3 orbitals and red symbols for @xmath8 orbitals ) are in excellent agreement with the dft bands . ]
-layer , while the wiggly lines in red are the almost one - dimensional features resulting from the @xmath8-layer . ]
-@xmath19 phase .
the dft band structure is shown with black lines , while the wannier bands are displayed with blue ( @xmath3 ) and red ( @xmath8 ) symbols . ]
we determine the electronic structure of ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) using the all electron full potential local orbital basis as implemented in the fplo code @xcite and the generalized gradient approximation functional @xcite .
a @xmath29 @xmath30 mesh was employed to converge the energy and charge density .
* @xmath3 phase.- * the band structure and density of states ( dos ) of the @xmath3 phase are presented in fig .
[ fig : kappaband ] .
there are eight bands in the energy window @xmath31 $ ] deriving exclusively from the highest occupied molecular orbitals ( homos ) of the eight et molecules in the unit cell . as there is only a very small dispersion in the stacking direction ( @xmath32-@xmath33 path )
the bands originating from the two layers in the unit cell are almost degenerate .
the dimerization of the et molecule pairs is reflected in the bonding ( @xmath34 $ ] ) and antibonding ( @xmath35 $ ] ) character of the bands .
the bandstructure of @xmath3-(bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) is very similar to @xmath3-(bedt - ttf)@xmath36 compounds with other anions @xmath37 @xcite . integrating the partial densities of states for symmetry inequivalent molecules in the range @xmath38 $ ] ev results in a charge transfer of one hole per ( et)@xmath0 dimer to the anion layer . * @xmath3-@xmath18 phase.- * in fig .
[ fig : alpha1band ] we show the band structure and layer - resolved density of states of the dual - layered compound .
the anion layer has no weight at the fermi level and therefore does not contribute to the electronic transport .
the bands originating from the @xmath3 layer are similar in shape to the bands of the @xmath3 phase ( fig .
[ fig : kappaband ] ) while the @xmath8 bands are narrower and less dispersive than the @xmath3 bands . the corresponding fermi surface of the @xmath3-@xmath18 phase is shown in fig .
[ fig : alpha1fs ] . here , the elliptical shape corresponds to the @xmath3 layer and is typical for this packing motif @xcite .
in contrast , the @xmath8 fermi surface ( wiggly lines in fig . [
fig : alpha1fs ] ) is quasi - one - dimensional and has no dispersion in the @xmath39 direction .
recent de haas - van alphen experiments for the dual - layered compound observed only elliptic orbitals originating from the @xmath3 bands @xcite and , as expected , no @xmath8 bands were detected .
however , one should mention that on the one hand the @xmath8 packed systems have been reported in the past as mott - hubbard insulators @xcite and , on the other hand , dft underestimates correlation effects and can not reproduce the insulating behavior of a mott system
. a better treatment of correlations beyond dft in organic materials @xcite ( presently beyond the scope of this study ) could lead to an opening of a gap at the fermi level in the @xmath8 bands and to a fermi surface with only @xmath3 bands .
note , however , that direct hybridization between @xmath8 and @xmath3 layers is almost negligible ( maximum @xmath8 to @xmath3 hopping parameters in @xmath3-@xmath18 and @xmath3-@xmath19 are 0.2 - 0.3 mev ) so that it is justified to focus our investigation on the properties of the @xmath3 layers even in @xmath3-@xmath18 and @xmath3-@xmath19 as these layers will be responsible for the observed superconductivity .
the charge transfer in this system has more features than in the pure @xmath3 phase compound due to the low symmetry and the presence of the @xmath8 layer .
we find that the right tilted et molecules in the @xmath8 plane contribute a charge of the order of 1/3 electron ( 0.343 for et molecule 3 as denoted in fig .
[ fig : layerscheme ] ( c ) and 0.316 for molecule 4 ) .
the other two left tilted molecules show a larger charge transfer of the order of 2/3 of an electron ( 0.658 for molecule 1 and 0.705 for molecule 2 ) . for clarity ,
we depict in fig .
[ fig : scheme ] ( a ) this charge ordering .
a similar charge disproportionation has already been observed in other @xmath8 charge transfer salts ( e.g. in @xmath8-et@xmath0ag(cn)@xmath0 @xcite ) , where the homogeneous charge transfer of the et molecules at high temperature is redistributed to a 1/3 versus 2/3 order upon cooling . in the @xmath3 layer
, the symmetry of the crystal leads to two symmetry inequivalent ( et)@xmath0 dimers with distances between the et molecules in a dimer of @xmath40 and @xmath41 .
however , the charge transfer from these dimers - 0.493 electrons for et molecule a ( see fig . [
fig : layerscheme ] ( b ) ) and 0.494 for molecule b - to the anion layer is the same within the computational accuracy . dimers in the charge ordered environment as created by the @xmath8 layer in the @xmath3-@xmath18 and @xmath3-@xmath19 phase .
et molecules are shown as spheres ; the two symmetry inequivalent @xmath3 layer et molecules are labeled a and b. ( a ) the four inequivalent @xmath8 layer et molecules in the @xmath3-@xmath18 phase carry numbers 1 through 4 . ( b )
there are only two inequivalent @xmath8 layer molecules in the @xmath3-@xmath19 phase that are labeled 1 and 2 . ] * @xmath3-@xmath19 phase.- * fig .
[ fig : alpha2band ] shows the band structure and density of states of the four - layered compound .
as the unit cell consists of four donor layers , there are 16 bands corresponding to the homos of the et molecules .
however , as the dispersion along the stacking direction is extremely small , the bands originating from the two identically packed layers in the unit cell are nearly pairwise degenerate and the band structure is very similar to that of the @xmath3-@xmath18 compound ( see appendix b ) .
nevertheless , the subtle quantitative differences between the electronic structure of the two systems , in particular in the @xmath8 bands , have important consequences on the behavior of the materials as it is reflected , for instance , in the charge transfer .
the @xmath23 symmetry in @xmath3-@xmath19 poses stronger restrictions on the charge transfer in the @xmath8 layer compared to the @xmath27 @xmath3-@xmath18 system .
the symmetry equivalent et molecules form a checkerboard pattern on the rectangular lattice as shown in fig .
[ fig : layerscheme ] ( d ) .
the et molecules on one of the @xmath8 sublattices donate 0.367 electrons , while the other molecules transfer approximately 0.611 electrons .
thus , the average charge transfer from the @xmath8 layers to the anion layer is slightly less than half an electron per et molecule . as in the @xmath3-@xmath18 system , there are two symmetry inequivalent ( et)@xmath0 dimers in the @xmath3 layer .
the distances of the molecules within the dimers differ slightly , at @xmath42 and @xmath43 .
there is a small charge disproportionation between the dimers , with et molecules of dimer a transferring 0.520 electrons to the anion , while those of dimer b transfer 0.500 electrons ( compare fig .
[ fig : layerscheme ] ( b ) ) .
thus , the @xmath3 layer compensates for the slightly too low charge transfer of the @xmath8 layer .
an important difference between the @xmath3-@xmath18 and @xmath3-@xmath19 phases is that the constraint imposed by the higher symmetry in @xmath3-@xmath19 translates into a more symmetric charge order in the @xmath8 layer in @xmath3-@xmath19 than in @xmath3-@xmath18 and the corresponding @xmath8-@xmath3 stacking in @xmath3-@xmath19 shows the center of the ( et)@xmath0 dimers of the @xmath3 layers always aligned between the two differently charged et molecules in the @xmath8 layer ( fig .
[ fig : scheme ] ( b ) ) , contrary to what happens in @xmath3-@xmath18 .
note , that the further inclusion of correlations may change the degree of charge order while the symmetry constraints will keep the pattern the same .
this charge arrangement has important consequences on the hamiltonian description of these systems as we show below .
we use the projective wannier function method as implemented in fplo @xcite in order to obtain the tight - binding parameters for the conducting @xmath3 layers in ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) from wannier function overlaps . considering the central point of the inner c - c bond of the et molecules as sites
, the tight - binding ( tb ) hamiltonian can be written as : @xmath44 where @xmath45 creates ( destroys ) an electron on site @xmath46 . when the dimerization of the et molecule pairs is strong , the separation between bonding and antibonding bands is large and the analysis of the low - energy tight - binding hamiltonian can be reduced to the two partially occupied antibonding bands .
this case corresponds to a half - filled anisotropic triangular lattice where the sites denote the centers of the ( et)@xmath0 dimers ( centers of the two inner c - c bonds on neighboring et molecules ) : @xmath47,\sigma } t ' \left(c_{i\sigma}^\dagger c_{j\sigma } + c_{j\sigma}^\dagger c_{i \sigma}\right)\end{aligned}\ ] ] @xmath48 and @xmath49 correspond to nearest- and next - nearest - neighbor hopping contributions .
the hoppings between dimers ( eq . [ eq : ham ] ) can be connected to the hoppings between molecules ( eq . [ eq : ham_mol ] ) using geometrical as well as analytical considerations @xcite , @xmath50 here @xmath51 and @xmath52 are the hoppings between one molecule and the two closest molecules on the orthogonal oriented dimer .
@xmath53 belongs to the hopping between the closest et molecules on neighboring equally oriented dimers . in table
[ tab : allti ] we list the hopping parameters between et molecules ( eq . [ eq : ham_mol ] and fig . [
fig : tbdual ] ( a ) ) in the @xmath3 layer of the three polymorphs ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) . due to the presence of two inequivalent ( et)@xmath0 dimers in @xmath3-@xmath18 and @xmath3-@xmath19 , these phases have twice as many tb parameters compared to the @xmath3 phase . for the @xmath3-only system ,
the approximate symmetry of @xmath23 we employ means that @xmath3 molecules are all equivalent within the layer but inequivalent between neighboring layers ; however , as differences between tb parameter sets are below @xmath54 mev , we report only one of them in table [ tab : allti ] .
note , that the absolute value of the tb parameters depends on the strength of the overlap of the wannier orbitals and is very sensitive to variations in the distance and the orientation of the et molecules , while the sign originates from the phase factors of the tb hamiltonian . for completion
, we also show in table [ tab : alphadual ] and table [ tab : alphafour ] the hopping parameters between et molecules in the @xmath55 layer for @xmath3-@xmath18 and @xmath3-@xmath19 , respectively .
we observe that the different charge distribution among the et molecules in the @xmath8 layers is a manifestation of the different crystal field environment of the molecules as quantified by the onsite parameters ( @xmath56 ) in table [ tab : alphadual ] and table [ tab : alphafour ] . in table
[ tab : ttprime ] , the tight - binding parameters for the dimer model ( eq . [ eq : ham ] and fig .
[ fig : tbdual ] ( b ) ) are listed .
[ cols="^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] ratios of the three polymorphs which indicate the degree of frustration . ]
our microscopic analysis of the polymorph family ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) with density functional theory and projective wannier function derived tight - binding hamiltonians showed that , even though the @xmath8 layer in these materials is most probably insulating , its electronic structure decisively influences the behavior of the conducting @xmath3 layer in @xmath3-@xmath18 and @xmath3-@xmath19 .
in particular , the tb hopping parameters in @xmath3-@xmath18 reflect the stripe - like charge ordered pattern shown by the et molecules in the @xmath8 layer . in @xmath3-@xmath19
there is no such pattern since the ( et)@xmath0 dimers in @xmath3-@xmath19 are always aligned between the two distinctly charged et molecules in the @xmath8 layer .
this different charge arrangement has its origin on the symmetry constraint imposed by @xmath28 in @xmath3-@xmath19 , which leads to a different crystal field acting on the et molecules in the @xmath3 layer . our tight - binding model parameters for the molecule - based model ( eq . [ eq : ham_mol ] and table [ tab : allti ] )
provide an adequate and reliable starting point for a many - body description of these systems in terms of a hubbard - like hamiltonian where intramolecular and intermolecular coulomb interaction terms should be added to the tight - binding hamiltonian . in contrast , an analysis in terms of the simplified dimer model ( eq . [ eq : ham ] and table [ tab : ttprime ] ) shows that the half - filled triangular lattice is not a very good starting point for describing these materials .
this is due to the relatively weak dimerization of the ( et)@xmath0 dimers and therefore the contribution of the @xmath3 bonding bands should not be neglected when describing the electronic properties of these systems .
however , the information obtained from the tight - binding hopping parameters in the dimer model is still useful to roughly classify the degree of frustration in ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) and have first hints for understanding the different superconducting critical temperatures in these systems .
our comparison of the trends of the frustration parameters with earlier studies shows that the @xmath3-phase system lies in the range of frustration where other @xmath3 systems are antiferromagnetic insulators , while the multi - layered @xmath3-@xmath18 and @xmath3-@xmath19 show a slightly higher frustration degree as also observed in the superconductor @xmath3-(et)@xmath0cu(scn)@xmath0 .
nevertheless , a detailed understanding of the different critical temperatures requires a many - body analysis of the molecular hubbard - like model proposed here which is beyond the scope of the present work . summarizing , in this work we investigated the electronic properties of the charge - transfer compound ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) .
we demonstrated and quantified the importance of the @xmath8 layers with respect to the conducting @xmath3 layers and suggested a molecule - based model hamiltonian to describe these systems .
we hope that this work will motivate other groups to investigate these multi - layered materials which hold promise of increasing the superconducting critical temperatures in organic charge - transfer superconductors .
_ acknowledgements _ we thank d. guterding and t. kawamoto for helpful discussions .
we acknowledge support by the deutsche forschungsgemeinschaft through grant sfb / tr 49 as well as the center for scientific computing ( csc ) in frankfurt , germany .
[ fig : mol_dim ] shows a comparison of the dft calculated @xmath3 band structure in the @xmath3-@xmath18 phase to the molecular tb model based on the eight parameters listed in table [ tab : allti ] , and to the dimer tb model based on the four parameters of table [ tab : ttprime ] .
the parameters of the molecule model are calculated using projective wannier functions , while the dimer model parameters are derived from them via geometrical relations , eq . .
it is clear that the dimer model provides only a rough approximation to the two half - filled @xmath3 bands at the fermi level .
bands of the three polymorphs of ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) on an averaged path through the brillouin zone ( as the in - plane lattice parameters for the three compounds differ slightly ) . ]
j. a. schlueter , k. d. carlson , u. geiser , h. h. wang , j. m. williams , w .- k .
kwok , j. a. fendrich , u. welp , p. m. keane , j. d. dudek , a. s. komosa , d. naumann , t. roy , j. e. schirber , w. r. bayless , and b. dodrill , physica c * 233 * , 379 ( 1994 ) . | the charge - transfer compound ( bedt - ttf)@xmath0ag(cf@xmath1)@xmath2(tce ) crystallizes in three polymorphs with different alternating layers : while a phase with a @xmath3 packing motif has a low superconducting transition temperature of @xmath4 k , two phases with higher @xmath5 of @xmath6 and @xmath7 k are multi - layered structures consisting of @xmath8 and @xmath3 layers .
we investigate these three systems within density functional theory and find that the @xmath8 layer shows different degrees of charge order for the two @xmath3-@xmath8 systems and directly influences the electronic behavior of the conducting @xmath3 layer .
we discuss the origin of the distinct behavior of the three polymorphs and propose a minimal tight - binding hamiltonian for the description of these systems based on projective molecular wannier functions . |
exceptional lie groups @xmath4,}f_{4 } { \rm [ 2],}e_{6 } { \rm [ 3],}e_{7 } { \rm [ 4]\ ; and\ ; } e_{8 } { \rm [ 5]}$ ] have been proposed as models in high energy physics .
in particular , the largest exceptional group @xmath5 turned out to be the unique gauge symmetry @xmath6 of the heterotic superstring theory [ 6 ] .
it is not yet clear as to how any one of these groups will describe the symmetry of any natural phenomenon . a recent experiment by coldea et.al [ 7 ] on neutron scattering over @xmath7 ( cobalt niobate ) forming a one - dimensional quantum chain reveals evidence of the scalar particles describable by the @xmath8 [ 8 ] symmetry .
the coxeter - weyl groups associated with exceptional lie groups describe the symmetries of certain polytopes .
for example , the coxeter - weyl groups @xmath9 describe the symmetries of the gosset s polytopes [ 9 ] in six , seven and eight dimensions respectively .
the coxeter - weyl group @xmath1 is a unique group in the sense that it describes the symmetry of a uniform polytope 24-cell in 4-dimensions which has no correspondence in any other dimensions .
it is also interesting from the point of view that the automorphism group @xmath10 can be described by the use of quaternionic representation of the binary octahedral group [ 10 ] that will be described in the next section . in this paper
we study regular and semi - regular 4d polytopes whose symmetries and vertices are described by the quaternionic representations of the @xmath1 symmetry and the discrete quaternions respectively .
we also give the projections of the polytopes under the symmetries @xmath11 and @xmath12 .
the latter decomposition helps how to view 4d polytopes in three dimensions .
the dual polytopes of the group @xmath1 have not been constructed in any mathematical literature . by following
the method which we have developed for the catalan solids [ 11 ] , duals of the archimedean solids , we construct the duals of the regular and semi - regular @xmath13-polytopes .
the paper is organized as follows . in section
2 we construct the group @xmath10 in terms of the quaternionic elements of the binary octahedral group [ 12 ] .
a review of construction of the maximal subgroups @xmath11 and @xmath12 in terms of quaternions is presented [ 13 ] .
section 3 deals with the regular and semi - regular @xmath0-polytopes and their branchings under the maximal subgroups @xmath14 . in section 4
we construct the dual polytopes of the @xmath0-polytopes .
section 5 is devoted to discussion and results .
in this section we introduce the quaternionic root system of the coxeter diagram of @xmath0 and give a proof that the @xmath15 can be represented as the left - right action of the quaternionic representation of the binary octahedral group . the @xmath13 diagram with the quaternionic simple roots
is shown in figure 1 .
note that all simple roots of the coxeter diagrams have equal norms .
diagram with quaternionic simple roots , height=188 ] let @xmath16 , @xmath17 be a real unit quaternion with its conjugate defined by @xmath18 and the norm @xmath19 where the quaternionic imaginary units satisfy the relations @xmath20 here @xmath21 and @xmath22 are the kronecker and levi - civita symbols and summation over the repeated indices is implied .
we define the scalar product by the relation @xmath23 in general , a reflection generator @xmath24 of an arbitrary coxeter group with respect to a hyperplane represented by the vector @xmath25 is given by the action @xmath26 when @xmath27 and @xmath28 are represented by quaternions the equation reads @xmath29 where we define the product in symbolically as @xmath30^ { * } \lambda $ ] and drop the factor @xmath31
. then the quaternionic generators of the coxeter group @xmath1are given by @xmath32^ { * } , r_{2 } = [ -e_{3 } , e_{3 } ] ^ { * } , \nonumber \\
r_{3 } = [ -\frac{(e_{2 } -e_{3 } ) } { \sqrt{2 } } , \frac{(e_{2 } -e_{3 } ) } { \sqrt{2 } } ] ^ { * } , r_{4 } = [ -\frac{(e_{1 } -e_{2 } ) } { \sqrt{2 } } , \frac{(e_{1 } -e_{2 } ) } { \sqrt{2 } } ) ] ^ { * } .\end{aligned}\ ] ] the coxeter group @xmath1 generated by these generators can be compactly written as follows [ 14 ] @xmath33\oplus [ r , s]\oplus [ p , q]^ { * } \oplus [ r , s]^ { * } \ } ; { \rm \ ; \ ; \ ; \ ; } p , q\in t,{\rm \ ; \ ; } r , s\in t'.\ ] ] here the notation stands for @xmath34\lambda = p\lambda q$ ] where @xmath35 are the subsets of the binary octahedral group @xmath36 of order 48 .
they can further be written as the union of subsets @xmath37 where the subsets are given by the quaternions @xmath38 the set @xmath39 is the quaternion group and the set @xmath40 is the binary tetrahedral group .
the set @xmath39 is an invariant subgroup of the binary octahedral group @xmath41 .
the left or right coset decomposition of the binary octahedral group under the quaternion group can be written as @xmath42 with @xmath43 taking the values @xmath44
.multiplication table of the binary octahedral group .
[ cols="^,^,^,^,^,^,^",options="header " , ] the coset representatives in form a group isomorphic to the permutation group @xmath45 .
this is reflected in the multiplication table 1 .
+ of course , the binary octahedral group has a larger invariant subgroup @xmath40 which follows from the multiplication @xmath46 .
now we can write the equation in a symbolic form as @xmath47\oplus [ t',t']\oplus [ t , t]^ { * } \oplus [ t',t']^ { * } \}.\ ] ] it is a group of order @xmath48 from which one can identify many maximal subgroups of order 576 [ 14 ] . the diagram symmetry of @xmath49 can be generated by @xmath50 $ ] .
the extension of the coxeter group by the generator @xmath51 leads to the automorphism group @xmath10 of order 2304 and can be compactly written as [ 14 ] @xmath52\oplus [ o , o]^ { * } \ } .\ ] ] the cartan matrix of the coxeter group @xmath13 and its inverse are given respectively by the matrices @xmath53,\end{aligned}\ ] ] @xmath54\end{aligned}\ ] ] where the simple roots and the basis vectors in the dual space satisfy respectively the relations @xmath55 . here
the basis vectors of the dual space are defined through the relation @xmath56 .
the @xmath13-polytopes can be generated by the orbit @xmath57 where the vector @xmath58 is defined in the dual space which can also be represented as @xmath59 .
the components of the vector @xmath27 are real numbers @xmath60 , ( i=1,2,3,4 ) .
the maximal coxeter subgroups of the group @xmath61 that we deal with are @xmath2 and @xmath62 .
there are two subgroups @xmath63 and @xmath64 which are not conjugates in the group @xmath1 but they are conjugates in the larger group @xmath65 .
we use the notation @xmath66 for the coxeter group generated by reflection generators @xmath67 , @xmath68 .
the octahedral group @xmath69 plays an important role when an arbitrary polytope of the group @xmath1 is projected into 3d .
the quaternionic representations of the groups @xmath70 are given by [ grindeq__13 _ ] @xmath71\oplus [ t',\pm \bar{t}']\oplus [ t,\pm \bar{t}]^ { * } \oplus [ t',\pm \bar{t}']^ { * } \label{subeqn-1 } \\ w(b_{3l } ) \times c_{2 } & = \ { [ t,\pm \bar{q}\bar{t}q]\oplus [ t',\pm \bar{q}\bar{t}'q]\oplus [ t,\pm q\bar{t}q]^ { * } \oplus [ t',\pm q\bar{t}'q]^ { * } \ } , \nonumber \\ q=\frac{1+e_{1 } } { \sqrt{2 } } \label{subeqn-2}.\end{aligned}\ ] ] here @xmath72 $ ] or any other pair represents the group element @xmath73,{\rm \ ; } p\in t$ ] .
the coxeter group @xmath69 is isomorphic to the octahedral group @xmath74 of order 48 .
therefore the group @xmath75 of order 96 can be embedded in the group @xmath1 in 12 different ways , in each embedding , a quaternion with @xmath76sign left invariant . in equations ( 13a ) and ( 13b ) the groups leave the quaternions @xmath77 and @xmath78 invariant respectively .
it is clear from these arguments that while the conjugates of the group @xmath79 leaving the vectors of the set @xmath80 invariant the conjugates of the group @xmath81 leave invariant the vectors from @xmath40 .
diagram with quaternionic simple roots , height=226 ] the coxeter diagram of the group @xmath2 is shown in figure 2 . the cartan matrix and its inverse are respectively given by the matrices @xmath82,\end{aligned}\ ] ] @xmath83.\end{aligned}\ ] ]
if we chose the quaternionic simple roots as @xmath84 the coxeter group can be represented as [ 10 ] @xmath85\oplus [ v_{+ } , v_{- } ] \oplus [ v_{- } , v_{+ } ] \oplus \nonumber \\ & & [ v_{1 } , v_{1 } ] \oplus [ v_{2 } , v_{2 } ] \oplus [ v_{3 } , v_{3 } ] \nonumber \\ & & \oplus [ v_{0 } , v_{0 } ] ^ { * } \oplus [ v_{+ } , v_{- } ] ^ { * } \oplus [ v_{- } , v_{+ } ] ^ { * } \nonumber \\ & & \oplus [ v_{1 } , v_{1 } ] ^ { * } \oplus [ v_{2 } , v_{2 } ] ^ { * } \oplus [ v_{3 } , v_{3 } ] ^ { * } \}.\end{aligned}\ ] ] the validity of the above equation can be checked by using table 1 . the group @xmath2 can be embedded in the group @xmath86 triply symmetric way which leads to the coset decomposition @xmath87,g_{2}=[\omega _ { 0 } , 1],g_{3}=[\omega _ { 0 }
^{2 } , 1 ] , \nonumber \\ & & \omega _ { 0}=\frac{1}{2 } ( 1+e_{1 } + e_{2}+e_{3 } ) .\end{aligned}\ ] ] when the vector @xmath59 is expressed in terms of quaternions we can find the vectors @xmath88 and operate by @xmath2 on the left do determine the branching of the orbits of the group @xmath2 in the orbit of the group @xmath89 . if all components of the vector @xmath27 are different from zero then the orbit @xmath57 represents a polytope with 1152 vertices .
the branching of the orbit @xmath59 under the orbits of the group @xmath2 can be written as follows @xmath90 we use the notation @xmath91 for the orbit of the coxeter group where @xmath92 represents the coxeter diagram .
let us give a few examples .
a complete list of decomposition of the regular and semi - regular polytopes are given in the appendix 1 .
the polytope @xmath93 represents the set of 24 quaternions @xmath94 up to a scale factor @xmath95 .
they are the elements of the the binary tetrahedral group and represents the vertices of the regular polytope 24-cell .
this is a unique polytope which has no correspondence in any other dimensions .
this is perhaps due to the fact that @xmath1 is associated with the exceptional lie group @xmath0 .
the polytope 24-cell consists of 24 octahedral cells , every vertex of which , is shared by 6 octahedra . to illustrate this
let us consider the vertex represented by the quaternion 1 .
the following quaternions , e.g. , represent the vertices of an octahedron involving the quaternion 1 : @xmath96 the center of this octahedron is represented , up to a scale factor , by the quaternion @xmath97 .
if we introduce the orthogonal vectors @xmath98 then can be written as @xmath99 , @xmath100 which represent an octahedron shifted in the 4@xmath101 dimension by @xmath102 we tabulate in equation the vertices of the six octahedra and their centers sharing the vertex 1 .
@xmath103 when we discuss the dual polytope of the @xmath104 we will see that those centers of the six octahedra form another octahedron with the vertices @xmath105 where the vertices belong to the set @xmath106 which forms another 24-cell .
actually the diagram symmetry operator @xmath51 exchanges the two 24-cells so that the polytope 24-cell is said to be self - dual .
now we can use to check how 24-cell decomposes under the subgroup @xmath2 .
it is straightforward to see that + @xmath107 this is nothing but writing the set @xmath40 in the form @xmath108 .
the set @xmath109 represents the vertices of a hyperoctahedron ( octahedron in 4d or called 16 cell ) whose cells are tetrahedra , 8 of which , share the same vertex .
the set @xmath110 represents the hypercube ( 8-cell ) with 16 vertices .
the hypercube consists of 8-cubes , 4 of which , meet at one vertex .
the hyperoctahedron and the hypercube are the duals of each other ; the centers of the 16 tetrahedra are represented by the vertices of the hypercube and vice versa .
the dual polytope 24-cell @xmath111 decomposes under the subgroup @xmath2 as @xmath112 where the vertices , edges , faces and cells respectively decompose as @xmath113 , 96=96 , 96=32 + 64 , 24=16 + 8 .
as we mentioned earlier there are two non - conjugate octahedral groups in the coxeter group @xmath1 . here
we will give the projections of the @xmath1 polytopes under the octahedral group @xmath114 . decomposition with respect to the other group @xmath115
can be done using a similar technique .
the group @xmath114 up to a conjugation can be represented by the quaternions as follows + @xmath116\oplus [ t',\bar{t}']\oplus [ t,\bar{t}]^ { * } \oplus [ t',\bar{t}']^ { * } \}\ ] ] which leaves the unit quaternion 1 invariant .
the right coset decomposition of the coxeter group @xmath1 under the octahedral group @xmath114 can be given as @xmath117.$ ] note that @xmath118{\rm \ ; and\ ; } [ 1,t]$ ] are two invariant subgroups of the coxeter group @xmath1 . applying the group @xmath1 on the vector @xmath27 we obtain the orbit @xmath119 . defining the vectors
@xmath120 we obtain , in general , 24 different orbits of the group @xmath114 in the decomposition of the orbit @xmath121 .
let a vector represented in the dual space of the root system of the coxeter group @xmath114 be given by @xmath122 .
when the quaternionic simple roots of the group @xmath114 are given by @xmath123 then basis vectors in the dual space are given by @xmath124 applying the group @xmath114 on the quaternionic representations of the vectors @xmath125 and expressing them in the basis @xmath126 then we obtain the following decomposition : @xmath127 the term starting with the @xmath76 represents a @xmath128 vector . for a general @xmath1 orbit
the formula shows the sphere @xmath129 representing the vertices of the @xmath1 polytope sliced by 24 hyperplanes .
each crossection of the sphere @xmath129 with an hyperplane orthogonal to the vector represented by the group @xmath130 is an @xmath131 sphere representing a particular @xmath69 polyhedron .
let us discuss a few cases .
consider the 24-cell represented by the orbit @xmath132 which can be written as @xmath133 when expressed in terms of quaternionic vertices they correspond to the decomposition @xmath134 either from the analysis of the equation [ 15 ] or directly reading the equation the points @xmath135 corresponds to the two opposite poles of the sphere @xmath129 .
the set of vertices @xmath136represent an octahedron obtained as the intersection of the hyperplane through the center of the sphere @xmath129 .
the last two sets of quaternions in represent cubes oppositely placed with respect to the center of the sphere @xmath129 .
the dual 24-cell represented by the orbit @xmath137 decomposes as @xmath138 which can also be represented by the set of quaternions @xmath139 the first set represents two octahedra oppositely placed with respect to the origin along the fourth direction , the second set represents a cuboctahedron with 12 vertices around the center of the sphere @xmath140 .
the decompositions of quaternionic vertices of 24- cell indicate that it can have two different decompositions under the group @xmath114 .
if the subgroup @xmath141 were chosen for the projection of the 24-cell into 3d then the above decomposition of the sets represented by @xmath142 would be interchanged .
using equation , any interested reader can work out the projection of any @xmath1 polytope into 3d with the octahedral residual symmetry .
the catalan solids which are the duals of the archimedean solids can be derived from the coxeter diagrams @xmath143 with a simple technique [ 11 ] . following the same group theoretical technique
we have already constructed the dual polytopes of the @xmath144 polytopes [ 16 ] . here
we apply the same technique to determine the vertices and the symmetries of the dual polytopes of the uniform * @xmath1 * polytopes . to find the vertices of the dual polytope one should determine the vectors representing the centers of the cells at the vertex @xmath145 .
the relative magnitudes of the vectors of interest can be obtained by noting that the hyperplane formed by the vectors representing the centers of the cells are orthogonal to the vertex @xmath146 .
the dual polytopes are cell transitive while the polytopes are vertex transitive . in this section
we will construct the vertices of a dual polytope and determine its cell structure .
we shall also use the group theoretical technique to determine the numbers of vertices @xmath147 , edges @xmath148 , faces @xmath149 and the cells @xmath150 of an arbitrary polytope .
all 4d polytopes satisfy the euler s topological formula @xmath151 . * * the 24-cell has 24 vertices , 96 edges , 240 faces and 24-cells .
first we discuss how these numbers follow from the group theoretical calculations .
we note first that the vertex represented by the vector @xmath153 is left invariant under the subgroup @xmath154 of order 48 .
the number of vertices then is the number of left coset representatives under the decomposition @xmath155 which is equal to the index @xmath156 .
the number of edges can be determined as follows .
the difference of the vectors @xmath157 or @xmath158 represents an edge .
the set @xmath159is left invariant by the group generated by @xmath160 of order 12 so that its index in the group * @xmath1 * equals 96 .
similarly from the vector @xmath161 of the subgroup * @xmath162 * one can generate an equilateral triangle by the group * @xmath163 * of order 6 . therefore the group leaving the triangle invariant is the subgroup generated by * @xmath164 * of order 12 and its index in the group * @xmath165 * , 96
, is the number of triangular faces .
the octahedral cell is generated by the group @xmath166 whose index , 24 , is the number of octahedral cells hence the name of the polytope 24-cell .
the center of the octahedron @xmath167 can be represented by the vector @xmath168 up to a scale factor .
the number of octahedra at the vertex @xmath169 is given by the formula @xmath170 which is the index of the subgroup @xmath171 of order 8 fixing the vertex @xmath153 of the octahedron in the group @xmath172 which fixes the vertex @xmath169 .
the vertices of the cell representing the centers of the 6 octahedra can be determined by applying @xmath154 on the vector @xmath168 which leads to another octahedron .
we have discussed this case in detail in section 3 .
applying the technique discussed in section 5.1 we determine @xmath174 vertices , @xmath175 edges , @xmath176 faces and @xmath177 cells .
the polytope has 96 triangular , 144 square faces and 48 cells which consist of 24 cuboctahedra and 24 cubes whose symmetries are respectively are the conjugate groups of the groups @xmath178
. the number of cuboctahedral cells at the vertex @xmath179 is determined by the index @xmath180 .
the group in the numerator is a subgroup of the group * @xmath1 * fixing the vertex @xmath181 and the group in the denominator is the subgroup of the octahedral group @xmath141 fixing the same vertex .
the group @xmath141 which generates the cuboctahedra leaves the vector @xmath168 invariant which can be taken as the vector representing its center .
so the vectors representing the centers of the 3 cuboctahedra at the vertex @xmath182 can be taken as @xmath183 .
similarly the number of the cubes at the same vertex is 2 whose centers can be represented by the vectors @xmath184 where@xmath185 is a scale factor determined from the equation@xmath186 these five vertices form a cell which is a semi - regular bipyramid whose center is represented by the vector @xmath181 .
the bipyramid consists of six isosceles triangles of sides @xmath187 .
it has the symmetry @xmath188 of order 12 which fixes the vector @xmath181 .
the set of vertices of the dual polytope is the union of the orbits @xmath189 .
they define two concentric spheres @xmath129 with the fraction of radii @xmath190 .
let us define the quaternionic unit vectors @xmath191 .
note that all the cells of the dual polytope are bipyramids and their centers are the vertices of the polytope @xmath192 after expressing five vertices of the bipyramid in terms of the new quaternionic units and omitting the common 4@xmath101 component proportional to @xmath193 , a scaled set of the vectors representing the vertices of the bipyramid will read @xmath194 a plot of this bipyramid is shown in figure 3 .
this polytope is obtained from the polytope @xmath196 by exchanging @xmath197 this polytope has @xmath199 vertices , @xmath200 edges , @xmath201 faces and @xmath202 cells .
the faces are of two types : 96 hexagons and 144 squares .
the cells consist of 24 truncated octahedra and 24 cubes . to find the dual polytope we have to determine the vectors representing the centers of these cells .
the vectors representing the centers of the truncated octahedra and the cube at the vertex @xmath203 are given by @xmath204 the relative scale parameter is determined to be @xmath205 .
the cell is a pyramid consisting of an equilateral triangular base of side @xmath95 and three isosceles triangular faces of sides @xmath206 and @xmath207 defining the unit quaternion @xmath208 one obtains the vertices in 3d of the cell of the dual polytope as @xmath209 a plot of this cell is shown in figure 4 .
it is clear that this solid has a symmetry group generated by @xmath210 .
the orbit of the vertices of the dual cell consists of two concentric spheres @xmath129 with the ratio of the radii @xmath211 this polytope has @xmath213 vertices , @xmath214 edges , @xmath215 faces , and @xmath216 cells .
reading from left to right the faces consist of 96 triangles , 288 + 144 squares , and 192 triangles .
the cells consist of 24 small rhombicuboctahedra , 24 cubeoctahedra , and 96 triangular prisms .
the vectors representing the centers of the cells at the vertex @xmath217 are given by @xmath218 the relative scale parameters are determined to be @xmath219 .
the cell of the dual polytope consists of five vertices . defining the unit quaternion @xmath220
one obtains the vertices in 3d of the cell of the dual polytope as @xmath221 , \nonumber \\
\lambda r_{4 } \omega _ { 4 } & \approx & \frac{1 + 9\sqrt{2 } } { 7 } [ -p_{1}+(1+\sqrt{2 } ) p_{2 } + p_{3 } ] , \nonumber \\
\rho \omega_{1 } & \approx & \frac{2 - 5\sqrt{2 } } { 2 } [ p_{1 } + p_{2 } ] , \nonumber \\ \omega _ { 2 } & \approx & [ p_{1 } + ( 1-\sqrt{2})p_{2 } + ( 1+\sqrt{2 } ) p_{3 } ] , \nonumber \\ r_{2 } \omega _ { 2 } & \approx & [ ( 1-\sqrt{2 } ) p_{1 } + p_{2}-(1+\sqrt{2})p_{3}].\end{aligned}\ ] ] a plot of this cell is shown in figure 5 .
it is clear that this solid has a symmetry of the klein four - group generated by @xmath222 .
the dual polytope consists of the union of the orbits @xmath223 with 144 vertices , 288 cells , 864 faces and 720 edges .
the radii of the @xmath224 spheres are respectively @xmath225 2.536 , @xmath226 2.450 , @xmath227 2.774 .
the polytope has @xmath229 vertices , @xmath230 edges , @xmath231 faces and @xmath232 cells .
reading from left to right the faces consist of 192 triangles , 288 squares , and 192 triangles .
the cells consist of 24 + 24 octahedra , 96 + 96 triangular prisms .
the vectors representing the centers of the cells at the vertex @xmath233 are given by @xmath234 the scale factor is found to be @xmath235 this cell of the dual polytope is a solid called tetragonal trapezohedron with 10 vertices with the symmetry @xmath236 of order 16 .
the dihedral group @xmath237 is generated by @xmath238 is the group generated by the diagram symmetry . defining the new set of unit vectors by @xmath239
@xmath240 we obtain the vertices of the solid in 3d as + @xmath241 a plot of this tetragonal trapezoid is shown in figure 6 .
it has eight faces and each face is a kite with sides @xmath242 and the area 0.934 .
so the dual polytope has 144 cells of tetragonal trapezohedra shown in figure 6 .
the dual polytope is cell transitive and it has 240 vertices 672 edges and 576 faces of kites described above . the polytope is the union of the orbits @xmath243 .
the vertices of the dual polytope is on two concentric radii of the spheres @xmath224 with radii @xmath2442.414 and @xmath2452.449 .
the polytope has @xmath213 vertices , @xmath247 edges , @xmath248 faces and @xmath202 cells . reading from left to
right the faces consist of 96 triangles , 144 octagons , and 96 triangles .
the cells consist of 24 + 24 truncated cubes .
the vectors representing the centers of the cells at the vertex @xmath249 are given by @xmath250 .
it is a solid with four vertices which has a symmetry @xmath251 of order 8 .
the dihedral group is generated by the generators @xmath252 and the group @xmath253 is generated by the diagram symmetry given by @xmath254 $ ] .
the cell of the dual polytope consists of four faces of isosceles triangles of two sides @xmath255 and one side @xmath206 . defining the new set of unit vectors by @xmath256@xmath257 we obtain the vertices of the cell as @xmath258 a plot is depicted in figure 7 .
the dual polytope is the union of the vertices determined by the polytopes @xmath259 the 48 vertices of the dual polytope are on the same @xmath224 sphere with radius @xmath95 .
of course it is cell transitive just like any dual polytope is .
in addition it is also vertex transitive under the group @xmath260 , height=188 ] the polytope has @xmath262 vertices , @xmath263 edges , @xmath215 faces and @xmath216 cells . reading from left to
right the faces consist of 96 hexagons , 288 squares , 144 octagons and 192 triangles .
the cells consist of 24 great rhombicuboctahedra , 24 truncated cubes , and 96 triangular prisms .
the vectors representing the centers of the cells at the vertex @xmath264 are given by @xmath265 it is a solid with four vertices which has a symmetry generated by @xmath266 .
the parameters are determined to be @xmath267 the vertices of the dual cell is the union of the orbits @xmath268 .
the vertices lie on three @xmath129 spheres with the radii @xmath2691.31:1.21:1.41 . defining the new set of unit vectors by @xmath270
@xmath271 we obtain the vertices of the cell as @xmath272 , \nonumber \\ \omega _ { 4 } & \approx & ( 4+\sqrt{2 } ) p_{1 } -\sqrt{2 } p_{2 } -(2+\sqrt{2 } ) p_{3 } , \nonumber \\
r_{4 } \omega _ { 4 } & \approx & -(2+\sqrt{2 } ) p_{1 } + ( 4+\sqrt{2 } ) p_{2 } + \sqrt{2 } p_{3}.\end{aligned}\ ] ] a plot of this solid is depicted in figure 8 . , height=188 ] the polytope has @xmath262 vertices , @xmath274 edges , @xmath275 faces and @xmath276 cells . reading from left to right
the faces consist of 192 hexagons , 144 + 288 + 288 squares , and 192 triangles .
the cells consist of 24 truncated octahedra , 24 small rhombicuboctahedra , 96 hexagonal prisms , and 96 triangular prisms .
the vectors representing the centers of the cells at the vertex @xmath277 are given by @xmath278 it is a solid with five vertices which has a symmetry generated by @xmath279 .
the parameters are determined to be @xmath280 and @xmath281 the vertices of the dual cell is the union of the orbits @xmath282 defining the new set of unit vectors by @xmath283 @xmath284 we obtain the vertices of the cell as @xmath285,{\rm \ ; } } \\ { \eta \omega _ { 2 } \approx \frac{6(1 + 2\sqrt{2 } ) } { 4 + 9\sqrt{2 } } [ ( 1-\sqrt{2 } ) p_{1 } + p_{2 } + ( 1+\sqrt{2 } ) p_{3 } ] , } \\ { \lambda \omega _ { 4 } \approx \frac{12(1 + 2\sqrt{2 } ) } { 2 + 3\sqrt{2 } } p_{1 } -\frac{6(1 + 2\sqrt{2 } ) } { 2 + 3\sqrt{2 } } p_{3 } , } \\ { \omega _ { 3 } \approx ( 2-\sqrt{2 } ) p_{1 } + ( 4+\sqrt{2 } ) p_{2 } + ( -2+\sqrt{2 } ) p_{3 } , } \\ { r_{3 }
\omega _ { 3 } \approx ( 4-\sqrt{2 } ) p_{1 } -(2+\sqrt{2 } ) p_{2 } + ( 2+\sqrt{2 } ) p_{3}. } \end{array}\ ] ] a plot of this solid is depicted in figure 9 .
, height=188 ] the polytope has @xmath287 vertices,@xmath288 edges , @xmath289 faces and @xmath276 cells .
reading from left to right the faces consist of 192 hexagons , 288 + 288 + 288 squares , 144 octagons and 192 hexagons .
the cells consist of 24 + 24 great rhombicuboctahedra and 96 + 96 haxagonal prisms .
the vectors representing the centers of the cells at the vertex @xmath290 are given by @xmath291 , and @xmath292 with @xmath293 .
it is a solid with four vertices which has a diagram symmetry only .
the 240 vertices of the dual cell is the union of the orbits @xmath294 defining the new set of unit vectors by @xmath295 @xmath257 we obtain the vertices of the cell as @xmath296 , } \\
{ \rho \omega _ { 3 } \approx \frac{5 + 3\sqrt{2 } } { 9 + 6\sqrt{2 } } [ ( -\sqrt{2 } + 2)p_{1 } + ( \sqrt{2 } + 4)p_{2 } + ( \sqrt{2 } -2)p_{3 } ] . } \end{array}\ ] ] note that the coxeter diagram symmetry leads to the transformation @xmath297 a plot of this solid is depicted in figure 10 .
, height=188 ]
4d polytopes can be classified with respect to their symmetries represented by the coxeter groups @xmath298 , @xmath2 , @xmath299 and @xmath1 .
here we constructed the regular and semiregular polytopes of the coxeter group @xmath1 with their dual polytopes .
we have represented the group elements in terms of quaternions .
it is an interesting observation that the automorhism group of the coxeter diagram @xmath13 can be determined in terms of the binary octahedral group as @xmath300\oplus [ o , o]^ { * } \}$ ] .
the dual polytopes of the @xmath1 have been constructed for the first time in this work .
we have explicitly given the decomposition of a given polytope under the maximal subgroup @xmath2 .
the projection of an arbitrary @xmath1 polytope in 3d space has been made with the use of the octahedral group @xmath69 which allows us to view the 4d polytope in 3d space . for a detail decomposition
see appendix 2 .
this work can be related to the quantum information theory based on 2-qubits since we use a finite subgroup of the orthogonal group @xmath301 .
the decompositon of the @xmath302 wieghts under @xmath303 is given as follows : + @xmath304= @xmath305 @xmath306= @xmath307 @xmath308= @xmath309 @xmath310= @xmath311 .
@xmath312= @xmath313 .
@xmath314= @xmath315 .
@xmath316= @xmath317 .
@xmath318= @xmath319 .
@xmath320= @xmath321 .
@xmath322= @xmath323 .
@xmath324= @xmath325 .
@xmath326= @xmath327 .
@xmath328= @xmath329 .
@xmath330= @xmath331 .
@xmath332= @xmath333 .
the decompositon of the @xmath302 wieghts under @xmath334 is given as follows : + @xmath335= + @xmath336+@xmath337 .
+ + @xmath338= + @xmath339+@xmath340+@xmath341 .
+ + @xmath342= + @xmath343+@xmath344+@xmath345+@xmath346 .
+ + @xmath347= + @xmath348+@xmath349+@xmath350+@xmath351 .
+ + @xmath352= + @xmath353+@xmath354+@xmath355+@xmath356+@xmath357+@xmath358 .
+ + @xmath359= + @xmath360+@xmath361+@xmath362+@xmath363+@xmath364+@xmath365
. + @xmath366= + @xmath367+@xmath368 + @xmath369+@xmath370+@xmath371+@xmath372+@xmath373+@xmath374 .
+ + @xmath375= + @xmath376+@xmath377+@xmath378 .
+ + @xmath379= + @xmath380+@xmath381+@xmath382+@xmath383+@xmath384 .
+ + @xmath385= + @xmath386+@xmath387+@xmath388+@xmath389+@xmath390+@xmath391 .
+ + @xmath392= + @xmath393+@xmath394+@xmath395+@xmath396+@xmath397+@xmath398+@xmath399+@xmath400 .
+ + @xmath401= + @xmath402+@xmath403+@xmath404+@xmath405+@xmath406+@xmath407 .
+ + @xmath408= + @xmath409+@xmath410+@xmath411+@xmath412+@xmath413+@xmath414+@xmath415+@xmath416+@xmath417 .
+ + @xmath418= + @xmath419+@xmath420+@xmath421+@xmath422+@xmath423+@xmath424+@xmath425+@xmath426+@xmath427 .
+ + @xmath428= + @xmath429+@xmath430+@xmath431+@xmath432+@xmath433+@xmath434+@xmath435+@xmath436+@xmath437+@xmath438+@xmath439+@xmath440 .
+ e. catalan , memoire sur la theorie des polyedres__. _ _ j. lecole polytechnique ( paris ) * 41 * ( 1865 ) 1 ; m koca , n.o .
koca and r. ko , j. math . phys .
* 51 * ( 2010 ) * * 043501 ; m. koca , m. al - ajmi and s. shidhani , the african rev .
( * 2011 * ) * 6*:0007 53 .
du val , homographies , quaternions and rotations , ( oxford university press , 1964 ) ; h.s.m .
coxeter , regular complex polytopes ( cambridge univ . press , 1974 ) ; j. h. conway and d. a. smith , on quaternions and octonions : their geometry , arithmetic , and symmetry , a k peters , ltd.(2003 ) . | 4-dimensional @xmath0 polytopes and their dual polytopes have been constructed as the orbits of the coxeter - weyl group * @xmath1 * where the group elements and the vertices of the polytopes are represented by quaternions .
branchings of an arbitrary * @xmath1 * orbit under the coxeter groups * @xmath2 * and * @xmath3 * have been presented .
the role of group theoretical technique and the use of quaternions have been emphasized . |
major cellular activities are effectively governed by multiple protein - dna interactions @xcite .
the starting point of these interactions is a process of protein searching and recognizing for the specific binding sites on dna .
this is a critically important step because it allows a genetic information contained in dna to be effectively transferred by initiating various biological processes @xcite . in recent years
, the fundamental processes associated with the protein search for targets on dna have been studied extensively using a wide variety of experimental and theoretical methods @xcite .
although a significant progress in our understanding of the protein search phenomena has been achieved , the full description of the mechanisms remains a controversial and highly - debated research topic @xcite .
experimental investigations of the protein search phenomena revealed that many proteins find their targets on dna very fast , and the corresponding association rates might exceed the estimates from 3d diffusion limits @xcite .
these surprising phenomena are known as a _ facilitated diffusion _ in the protein search field .
more recent single - molecule experiments , which can directly visualize the dynamics of individual molecules , also suggest that during the search proteins move not only through the bulk solution via 3d diffusion but they also bind _ non - specifically _ to dna where they hop in 1d fashion @xcite . several theoretical approaches that incorporate the coupling between 3d diffusion and 1d sliding in the protein search have been proposed @xcite , but they had a variable success in explaining all experimental observations .
one of the most interesting problems related to the protein search on dna is the effect of multiple targets .
the question is how long will it take for the proteins to find _ any _ specific binding site from several targets present on dna .
naively , one could argue that in this case the search time should be accelerated proportionally to the number of targets , i.e. , the association reaction rate should be proportional to the concentration of specific binding sites .
however , this effectively mean - field view ignores several important observations .
first , it is clear that the search time for several targets lying very close to each other generally should not be the same as the search time for the same number of targets which are spatially dispersed .
second , the experimentally supported complex 3d+1d search mechanism suggests that varying spatial distributions of the specific binding sites should also affect the search dynamics .
thus , it seems that the simple mean - field arguments should not be valid for all conditions .
surprisingly , this very important problem was addressed only in one recent work @xcite .
hammar et al .
@xcite using high - quality single - molecule measurements in the living cells investigated the dynamics of finding the specific sites for _ lac _ repressor proteins on dna with two targets .
it was found that the association rates increase as a function of the distance between targets @xcite .
an approximate theoretical model for the protein search with two targets was proposed
. however , this theoretical approach has several problems .
it was presented for infinitely long dna chains using a continuum approximation . at the same time , it was shown recently that the continuum approach might lead to serious errors and artifacts in the description of protein search dynamics @xcite .
in addition , this theory predicted that the acceleration due to the presence of two targets in comparison with the case of only one target should disappear in the limit of very large sliding lengths .
this is clearly a nonphysical result . in this limit
, the protein spends most of the searching time on dna and it is faster to find any of two targets than one specific binding site . in this article ,
we present a comprehensive theoretical method of analyzing the role of multiple targets in the protein search on dna . our approach is based on a discrete - state stochastic framework that was recently developed by one of us for the search with one specific binding site @xcite .
it takes into account most relevant biochemical and biophysical processes , and it allows us to obtain fully analytical solutions for all dynamic properties at all conditions .
one of the main results of the discrete - state stochastic method was a construction of dynamic phase diagram @xcite .
three possible dynamic search regimes were identified .
when the protein sliding length was larger than the dna chain length , the search followed simple random - walk dynamics with a quadratic scaling of the search time on the dna length . for the sliding length smaller than the dna length but larger than the the size of the specific binding site , the search dynamic followed a linear scaling
when the sliding length was smaller than the target size , the search was dominated by nonspecific bindings and unbindings without the sliding along dna . in this paper , we extend this method to the case of several specific binding sites at arbitrary spatial positions .
it allows us to explicitly describe the role of multiple targets and their spatial distributions in the protein search .
our theoretical calculations agree with available experimental observations , and we also test them in monte carlo computer simulations .
the original discrete - state stochastic approach can be generalized for any number of the specific binding sites at arbitrary positions along the dna chain . but
to explain the main features of our theoretical method , we analyze specifically a simpler model with only two targets as shown in fig .
1 . a single dna molecule with @xmath0 binding sites and a single protein molecule are considered .
the analysis can be easily extended for any concentration of proteins and dna @xcite .
two of the bindings sites @xmath1 and @xmath2 ( @xmath3 and @xmath4 ) are targets for the protein search ( see fig .
the protein starts from the bulk solution that we label as a state @xmath5 . since 3d diffusion
is usually much faster than other processes in the system , we assume that the protein can access with equal probability any site on the dna chain ( with the corresponding total binding rate @xmath6 ) . while being on dna , the protein can move with a diffusion rate @xmath7 along the chain with equal probability in both directions .
the protein molecule can also dissociate from dna with a rate @xmath8 to the bulk solution ( fig .
the search process ends when the protein reaches for the first time _ any _ of two targets .
the main idea of our approach is to utilize first - passage processes to describe the complex dynamics of the protein search on dna @xcite .
one can introduce a function @xmath9 defined as a probability to reach any target at time @xmath10 for the first time if initially ( at @xmath11 ) the protein molecule starts at the state @xmath12 ( @xmath13 ) .
these first - passage probabilities evolve with time as described by a set of the backward master equations @xcite , @xmath14+k_{off } f_0(t)-(2u+k_{off } ) f_n(t),\ ] ] for @xmath15 . at dna ends ( @xmath16 and @xmath17 ) the dynamics is slightly different , @xmath18 and @xmath19 in addition , in the bulk solution we have @xmath20 furthermore , the initial conditions require that @xmath21 the physical meaning of this statement is that if we start at one of two targets the search process is finished immediately .
it is convenient to solve eqs .
( [ me1 ] ) , ( [ me2 ] ) , ( [ me3 ] ) and ( [ me4 ] ) by employing laplace transformations of the first - passage probability functions , @xmath22 @xcite . the details of calculations are given in the appendix .
it is important to note here that the explicit expressions for the first - passage probability distribution functions in the laplace form provide us with a full dynamic description of the protein search @xcite .
for example , the mean first - passage time to reach any of the target sites if the original position of the protein was in the solution ( @xmath23 ) , which we also associate with the search time , can be directly calculated from @xcite @xmath24 as shown in the appendix , the average search time is given by @xmath25}{k_{on } k_{off } s_{i}(0)},\ ] ] where @xmath26 is a new auxiliary function with a subscript specifying the number of targets ( @xmath27 for the system with two targets ) . for this function
we have @xmath28}{(1-y)(1+y^{2 m_1 -1})(1+y^{1 + 2(l - m_2)})(1+y^{m_2-m_1})},\ ] ] with @xmath29 it is important that for @xmath30 , as expected , our results reduce to expressions for the protein search on dna with only one target @xcite .
similar procedures can be used to estimate all other dynamic properties for the system with two targets .
we can extend this approach for any number of targets and for any spatial distribution of binding sites .
this is discussed in detail in the appendix .
surprisingly , the expression for the search times are the _ same _ in all cases but with different @xmath31 functions that depend on the number of specific binding sites and their spatial distributions
. analytical results for @xmath31 for the protein search on dna with three or four targets , as well as a general procedure for arbitrary number of specific binding sites , are also presented in the appendix .
because our theoretical method provides explicit formulas for all relevant quantities , it allows us to fully explore many aspects of the protein search mechanisms . the first problem that we can address
is related to the role of the spatial distribution of targets on the search dynamics . in other words ,
the question is how the search time is influenced by exact positions of all targets along the dna .
the results of our calculations for two specific binding sites are presented in fig .
the longest search times are found when two targets are at different ends of the molecule , and the distance between them along the dna curve , @xmath32 , is the largest possible and equal to @xmath33 .
the search is faster if targets are moved closer to each other and both distributed symmetrically with respect to the middle point of the dna molecule ( fig .
2 ) . moving the targets too close ( @xmath34 ) starts to increase the search time again : see fig .
2 . for short dna chains
, it can be shown that there is an optimal distance between two targets , @xmath35 , that yields the fastest search ( fig .
it corresponds to the most optimal positions of the specific sites to be at @xmath36 and @xmath37 .
the last result is slightly unexpected since simple symmetry arguments suggest that the fastest search would be observed for the uniform distribution of targets , i.e. when the distance between the specific sites and the distance between the ends and targets are the same , i.e. , for @xmath38 and @xmath39 .
this is not observed in fig .
2 . to explain this
, one can argue that the search on the dna molecule of length @xmath0 with @xmath12 targets can be mapped into the search on @xmath12 dna segments of variable lengths with only one target per each segment . in this case , positioning each target in the middle of the corresponding segment leads to the fastest search dynamics.@xcite this suggests that the most optimal distribution of @xmath12 symmetrically distributed targets is a uniform distribution with the distance between two neighboring targets equal to @xmath40 .
but then the first and the last targets will be separated from the corresponding ends by a shorter distance @xmath41 .
this is exactly what we see in fig . 2 for @xmath42 targets .
the reason that distances between the ends and the closest targets deviate from the distances between the targets is the reflecting boundary conditions at the ends that are assumed in our model : see eqs .
( [ me2 ] ) and ( [ me3 ] ) . the results presented in fig .
2 also illustrate another interesting observation .
increasing the length of dna effectively eliminates the minimum in the search time for specific symmetric locations of the targets .
essentially , for @xmath43 , which is much closer to realistic conditions in most cases , any two position of the targets inside the dna chain will be optimal and will have the same search time as long as they are not at the ends .
we will discuss the reason for this below .
one of the main advantages of our method is the ability to explicitly analyze the search dynamics for all ranges of relevant parameters .
this allows us to construct a comprehensive dynamic phase diagram that delineates different search regimes .
the results are presented in fig . 3 for the systems with different numbers of specific binding sites .
the important observation is that general features of the search behavior are independent of the number of targets . more specifically , there are three dynamic phases that depend on the relative values of the length of dna @xmath0 , the average scanning length @xmath44 and the size of the target ( taken to be equal to unity in our model ) . for @xmath45 the random - walk regime
is observed with the search time being quadraticaly proportional to the size of dna @xcite . in this case , the protein non - specifically binds dna and it does not dissociate until it finds one of the targets .
the quadratic scaling is a result of a simple random - walk unbiased diffusion of the protein molecule on dna during the search . for the intermediate sliding regime , @xmath46 ,
the protein binds to dna , scans it , unbinds and repeats this cycle at average @xmath47 times ( @xmath12 is number of the targets ) for symmetrically distributed specific sites . for more general distributions
the number of search cycles is also proportional to @xmath48 .
this leads to the linear scaling in the search times .
for @xmath49 we have the jumping regime where the protein can bind to any site on dna and dissociate from it , but it can not slide along the dna chain .
the search time is again proportional to @xmath0 because on average the protein must check @xmath0 sites .
these changes in the dynamic search behavior are illustrated in fig .
4 , in which the search times as a function of the dna lengths are presented for different scanning lengths . the slope variation indicates a change in the scaling behavior in the search times from @xmath50 to @xmath0 as the dna length increases for fixed @xmath51 .
it is also important to note here that the concept of the most optimal positions of targets is not working for the sliding regime ( @xmath52 ) because the protein during the search frequently unbinds from the dna , losing all memory about what it already scanned .
this concept also can not be defined in the jumping regime where the protein does not slide at all . from this point of view , any position of the targets are equivalent .
the only two positions that differ from others are the end sites in the sliding regime .
this is because they can be reached only via one neighboring site , while all other sites can be reached via two neighboring sites ( see fig .
1 ) . the most interesting question for this system is to analyze quantitatively the effect of multiple targets on search dynamics . to quantify this
we define a new function , @xmath53 , which we call an acceleration , @xmath54 this is a ratio of the search times for the case of one target and for the case of @xmath12 targets .
the parameter @xmath53 gives a numerical value of how the presence of multiple targets increases the rate of association to any specific binding site .
the results for acceleration are presented in figs .
first , we analyze the situation when targets in all cases are in the most optimal symmetric positions , which is shown in fig .
5 . for dna with the single target it is in the middle of the chain , while for dna with @xmath12 targets they are distributed uniformly , as we discussed above , with the distance @xmath40 between the internal targets and @xmath55 for boundary targets and dna ends .
the acceleration for these conditions depends on the dynamic search regimes , and it ranges from @xmath12 to @xmath56 : see fig .
5 . for the case of @xmath57 ( jumping and sliding regimes ) , on average
the number of search excursions to dna before finding the specific site is equal to @xmath40 , and this leads to a linear behavior in the acceleration ( @xmath58 ) . for @xmath45 ( random - walk regime ) , the search is one - dimensional and the protein must diffuse on average the distance @xmath40 before it can find any of the targets .
the quadratic scaling for the simple random walk naturally explains the acceleration in the search in this dynamic regime , @xmath59 .
however , the acceleration is also affected by the distance between the targets .
if we maintain the most optimal conditions for the dna with one target but vary the distance between multiple targets , while keeping the overall symmetry , the results are shown in fig .
6 . in this case
, putting targets too close to each other or moving them apart lowers the acceleration .
eventually , there will be no acceleration for these conditions ( @xmath60 ) .
but the results are much more interesting if we consider the non - symmetric distributions of targets .
surprisingly , the search time for the system with multiple targets can be even slower than for the single target system ! this is shown in fig .
7 where @xmath53 can be as low as 1/4 for the two - target system in the random - walk regime , or it can reach the value of 1/2 in the sliding and jumping regimes ( not shown ) . the single target in the most optimal position in the middle of the dna chain can be found much faster in comparison with the case of two targets seating near one of the ends .
these observations suggest that the degree of acceleration of the search process due to the presence multiple targets is not always a linear function of the number of specific binding sites .
it depends on the nature of the dynamic search phase , the distance between the targets and the spatial distribution of the targets .
varying these parameters can lead to larger accelerations as well as to unexpected decelerations .
it is a consequence of the complex mechanism of the protein search for targets on dna that combines 3d and 1d motions .
this is the main result of our paper .
recently , single - molecule experiments measured the facilitated search of _ lac _ repressor proteins on dna with two identical specific binding sites @xcite .
these experiments show that the association rate increases before reaching the saturation with the increase in the distance between the targets .
our theoretical model successfully describes these measurements , as shown in fig .
fitting these data , we estimate the 1d diffusion rate for the lac repressors as @xmath61 s@xmath62 , which is consistent with _ in vitro _ measured values @xcite .
our estimates for the sliding length , @xmath63 bp , and for the non - specific association to dna , @xmath64 s@xmath62 , also agree with experimental observations @xcite .
it is important to compare our results with predictions from the theoretical model presented in ref @xcite .
this continuum model was developed assuming that the length of dna is extremely long , @xmath65 .
it was shown that the acceleration for the search for the case of two targets can be simply written as @xcite @xmath66 where @xmath67 is the distance between targets .
the comparison between two theoretical approaches is given in figs .
one can see from fig .
9 that both models agree for very large dna lengths , @xmath43 , while for shorter dna chains there are significant deviations .
the continuum theory @xcite predicts that the acceleration is always a linear or sub - linear function of the number of targets , i.e. , @xmath68 .
our model shows that the acceleration can have a non - linear dependence on the number of the targets , @xmath59 .
more specifically , this can be seen in fig .
10 , where the acceleration is presented as a function of the scanning length @xmath51 .
the prediction of the continuum theory that for @xmath69 the acceleration always approaches the unity is unphysical .
clearly , if we consider , e.g. , the optimal distribution of targets , then the larger the number of specific binding sites , the shorter the search time .
the reason for the failure of the continuum model at this limit is its inability to properly account for all dynamic search regimes .
this analysis shows that the continuum model @xcite has a very limited application , while our theoretical approach is consistent with all experimental observations and provides a valid physical picture for all conditions .
we investigated theoretically the effect of the multiple targets in the protein search for specific binding sites on dna .
this was done by extending and generalizing the discrete - state stochastic method , originally developed for single targets , that explicitly takes into account the most important biochemical and biophysical processes . using the first - passage processes ,
all dynamic properties of the system can be directly evaluated .
it was found that the search dynamics is affected by the spatial distribution of the targets for not very long dna chains .
there are optimal positions for specific sites for which the search times are minimal .
we argued that this optimal distribution is almost uniform with a correction due to the dna chain ends .
we also constructed a dynamic phase diagram for the different search regimes .
it was shown that for any number of targets there are always three phases , which are determined by comparing the dna length , the scanning length and the size of the target .
furthermore , we investigated the quantitative acceleration in the search due to the presence of multiple targets for various sets of conditions .
it was found also that the acceleration is linearly proportional to the number of targets when the scanning length is less than the dna length . for larger scanning lengths ,
the acceleration becomes faster with the quadratic dependence on the number of targets .
however , changing the distances between the targets generally decreases the effect of acceleration .
unexpectedly , we found that varying also the spatial distributions can reverse the behavior : it might take longer to find the specific site in the system with multiple targets in comparison with properly positioned single target .
our model allows us to explain this complex behavior using simple physical - chemical arguments .
in addition , we applied our theoretical analysis for describing experimental data , and it is shown that the obtained dynamic parameters are consistent with measured experimental quantities . a comparison between our discrete - state theoretical method the continuum model is also presented .
we show that the continuum model has a limited range of applicability , and it produces the unphysical behavior at some limiting cases . at the same time , our approach is fully consistent at all sets of parameters .
our theoretical predictions were also fully validated with monte carlo computer simulations .
the presented theoretical model seems to be successful in explaining the complex protein search dynamics in the systems with multiple targets .
one of the main advantage of the method is the ability to have a fully analytical description for all dynamic properties in the system .
however , one should remember that this approach is still quite oversimplified , and it neglects many realistic features of the protein - dna interactions .
for example , dna molecule is assumed to be frozen , different protein conformations that are observed in experiments are not taken into account , and the possibility of correlations between 3d and 1d motions is also not considered .
it will be critically important to test the presented theoretical ideas in experiments as well as in more advanced theoretical methods .
the work was supported by the welch foundation ( grant c-1559 ) , by the nsf ( grant che-1360979 ) , and by the center for theoretical biological physics sponsored by the nsf ( grant phy-1427654 ) .
this appendix includes detailed derivations of the equations from the main text and explicit expressions for functions utilized in our calculations . to solve the backward master equations ( 1)-(5 ) for the system with two targets we use the laplace transformation which leads to @xmath70+k_{off }
\widetilde{f_0}(s);\ ] ] @xmath71 @xmath72 @xmath73 with the condition that @xmath74 we are looking for the solution of these equations in the form , @xmath75 , where @xmath76 and @xmath77 are unknown coefficients that will be determined after the substitution of the solution into eqs . ( [ l1 ] ) , ( [ l2 ] ) , ( [ l3 ] ) and ( [ l4 ] ) .
this gives the following expression , @xmath78+k_{off } \widetilde{f_0}(s).\ ] ] after rearranging , we obtain @xmath79 = ( s + k_{off } ) b - k_{off } \widetilde{f_0}(s).\ ] ] requiring that the right - hand - side of this expression to be equal to zero , yields @xmath80 since the parameter @xmath81 , we can find @xmath82 by solving @xmath83 or @xmath84 there are two roots of this quadratic equation , @xmath85 and @xmath86 with @xmath87 . the next step is to notice that two targets at the positions @xmath88 and @xmath89 divide the dna chain into 3 segments which can be analyzed separately .
then the general solution should have the form @xmath90 with the parameter @xmath77 is specified by eq .
( [ eqb ] ) and @xmath91 . using the corresponding boundary conditions
, it can be shown that for @xmath92 @xmath93 while for @xmath94 we have @xmath95 and for @xmath96 @xmath97 this leads to the following expression for @xmath98 : @xmath99 where the auxiliary function @xmath26 is introduced via the following relation @xmath100 note that eq.([eqf0 ] ) is identical to the corresponding equation for the single - target case @xcite , but with the different auxiliary function @xmath26 .
finally , we can obtain the explicit expressions for the search times as given in the main text in eq .
the explicit form of the search time depends on the auxiliary functions @xmath31 , which can be directly evaluated .
for example , for the two targets we have @xmath101 which after simplifications leads to eq .
( 8) in the main text .
similar analysis can be done for any number of targets with arbitrary positions along the chain .
the final expression for the search times is the same in all cases [ given by the eq.(7 ) ] , but with the different auxiliary functions @xmath26 .
when the protein molecule searches the dna with three targets ( @xmath102 ) , it can be shown that fig .
1 . a general view of the discrete - state stochastic model for the protein search on dna with two targets . there are @xmath106 nonspecific and @xmath107 specific binding sites on the dna chain .
a protein molecules can diffuse along the dna with the rate @xmath7 , or it might dissociate into the solution with the rate @xmath8 . from the solution , the protein can attach to any position on dna with the total rate @xmath6 .
the search process is considered to be completed when the protein binds for the first time to any of two targets at the position @xmath88 or @xmath89 .
normalized search times as a function of the normalized distance between two targets .
the targets are positioned symmetrically with respect to the center of the dna chain .
the parameters used in calculations are the following : @xmath108 s@xmath62 and @xmath109 s@xmath62 .
the scanning length @xmath51 is varied by changing @xmath8 .
solid curves are theoretical predictions , symbols are from monte carlo computer simulations .
dynamic phase diagram for the protein search with multiple targets .
search times as a function of the scanning length are shown for systems with one , two or three targets .
the parameters used in calculations are the following : @xmath110 bp ; and @xmath108 s@xmath62 .
the scanning length @xmath51 is varied by changing @xmath8 .
protein search times as a function of dna length for different scanning lengths for the system with two targets .
the parameters used in calculations are the following : @xmath108 s@xmath62 .
solid curves are theoretical predictions , symbols are from monte carlo computer simulations .
the scanning length @xmath51 is varied by changing @xmath8 .
5 . acceleration in the search times as a function of the scanning length for the systems with two and three targets .
the parameters used in calculations are the following : @xmath108 s@xmath62 .
the scanning length @xmath51 is varied by changing @xmath8 .
acceleration in the search times as a function of the normalized distance between the targets for the systems with two and three targets .
the single target is in the middle of the dna chain .
other targets systems are symmetric but not optimal .
the parameters used in calculations are the following : @xmath111 s@xmath62 ; @xmath112 s@xmath62 and @xmath113 bp .
acceleration in the search times as a function of the normalized distance between the targets for the systems with targets .
the single target is in the middle of the dna chain . in the two - target system one of the specific binding sites
is fixed at the end and the position of the second one is varied .
the parameters used in calculations are the following : @xmath108 s@xmath62 ; @xmath112 s@xmath62 and @xmath113 .
9 . comparison of theoretical predictions for the acceleration as a function of the distance between the specific binding sites for the system with two targets for different dna lengths .
targets are distributed symmetrically with respect to the middle of the dna chain .
solid curves are discrete - state predictions , dashed curves are from the continuum model from ref .
the parameters used in calculations are the following : @xmath108 s@xmath62 ; and @xmath109 s@xmath62 .
comparison of theoretical predictions for the acceleration as a function of the scanning length for the system with two targets for different dna lengths .
targets are in the most optimal symmetric positions .
solid curves are discrete - state predictions , dashed curves are from the continuum model from ref .
the parameters used in calculations are the following : @xmath108 s@xmath62 .
the scanning length @xmath51 is varied by changing @xmath8 . | protein - dna interactions are crucial for all biological processes
. one of the most important fundamental aspects of these interactions is the process of protein searching and recognizing specific binding sites on dna .
a large number of experimental and theoretical investigations have been devoted to uncovering the molecular description of these phenomena , but many aspects of the mechanisms of protein search for the targets on dna remain not well understood .
one of the most intriguing problems is the role of multiple targets in protein search dynamics . using a recently developed theoretical framework we analyze this question in detail .
our method is based on a discrete - state stochastic approach that takes into account most relevant physical - chemical processes and leads to fully analytical description of all dynamic properties .
specifically , systems with two and three targets have been explicitly investigated .
it is found that multiple targets in most cases accelerate the search in comparison with a single target situation .
however , the acceleration is not always proportional to the number of targets .
surprisingly , there are even situations when it takes longer to find one of the multiple targets in comparison with the single target .
it depends on the spatial position of the targets , distances between them , average scanning lengths of protein molecules on dna , and the total dna lengths .
physical - chemical explanations of observed results are presented .
our predictions are compared with experimental observations as well as with results from a continuum theory for the protein search .
extensive monte carlo computer simulations fully support our theoretical calculations . |
the fundamental existence result @xcite of sacks - uhlenbeck asserts that there exists a set of free homotopy classes of maps from @xmath2 to a compact manifold @xmath3 so that the elements in these classes form a generating set for @xmath4 acted on by @xmath5 and each class contains a minimizing harmonic map @xmath6 .
this can be used to show that there are only finitely many distinct free homotopy classes in @xmath4 which are representable by maps in @xmath7 with energies stay below a given positive constant @xcite . for homotopy classes
, one needs to count contribution of @xmath5 by fixing a base point , and the set @xmath8 where @xmath9 is a positive constant and @xmath10 is the dirichlet energy of @xmath11 may contain infinitely many homotopy classes .
an example describing such behavior is first constructed in @xcite and then modified in @xcite to the compact setting by mapping a portion of the domain surface to curves whose lengths go to infinity while the total energies remain uniformly bounded . for harmonic maps ,
if the domain is a single @xmath2 and the energies are uniformly bounded , it follows from parker s bubble tree convergence theorem @xcite that the limiting map ( of a subsequence ) is from a stratified sphere , i.e. a chain of 2-spheres obtained by pinching finitely many loops on @xmath2 to points ( see definition in section 2 ) , and consequently there are only finitely many homotopy classes for harmonic maps from @xmath2 under a fixed energy level .
there are interesting geometric situations , e.g. harmonic map heat flow and mean curvature flow , where we need to consider a sequence of harmonic maps from stratified spheres , possibly different , and study the homotopy of these maps .
in this paper we prove : [ str ] let @xmath3 be a compact riemannian manifold without boundary . then for any @xmath12 , the set of harmonic maps from stratified 2-spheres to @xmath3 with energies @xmath13 contains only finitely many homotopy class . to keep track of the homotopy classes when a sequence of harmonic maps from ( possibly different ) stratified spheres converge as a bubble tree , we reformulate the construction of bubble trees @xcite , @xcite and introduce _ a normalization _ of a stratified sphere ( see definition [ homo ] ) , which is , roughly , a surjective continuous map from @xmath2 to a stratified sphere @xmath14 and collapses finitely many multiply connected regions arising from the blowup points to the singular points of @xmath14 .
this procedure allows us to use the homotopy class of a normalization to capture the homotopy of a map from a stratified sphere .
+ there are interesting applications of theorem [ str ] to geometric flows . in this regard , it is useful to find manifolds which have infinitely many homotopy classes representable by mappings from @xmath2 with uniformly bounded energies . in theorem
[ 3-manifold ] , we show that if @xmath15 are closed 3-manifolds and @xmath16 and @xmath17 is the connected sum of @xmath15 then there are infinitely many smooth maps from @xmath2 to @xmath18 with uniformly bounded energies and are mutually nonhomotopic .
the key topological result is proposition [ h.sequence ] derived in appendix . as applications of theorem [ str ] and theorem [ 3-manifold ] ,
we prove finite time blowup for the harmonic map flow and the mean curvature flow of @xmath2 in the 3-manifold @xmath18 above .
the formation of finite time singularities is due to the topology of @xmath18 . for the harmonic map flow ,
these new examples and their construction are completely different from the one in @xcite by chang - ding - ye .
[ flow ] let @xmath19 and @xmath20 be closed 3-dimensional riemannian manifolds and @xmath17 .
then \(a ) there exist infinitely many smooth maps @xmath21 such that @xmath22 are not homotopic for any @xmath23 , @xmath24 and the harmonic map flow that begins at @xmath25 develops a finite time singularity .
\(b ) there exist infinitely many embeddings @xmath26 such that @xmath27 are nonhomotopic for any @xmath23 with uniformly bounded area , and the mean curvature flow of @xmath2 initiated at @xmath28 develops a singularity in a finite time .
to apply theorem [ str ] in the proof of theorem [ flow ] , for each of the two flows , we need a convergence result for a subsequence along the flow with the chosen initial data .
it is shown by qing - tian in @xcite that for the bubbling of a palais - smale sequence for the dirichlet energy functional with a uniform @xmath29-bound on the tension fields , there are no necks between bubbles in the limit .
when the domain surface is @xmath2 , the limit is a harmonic map from a stratified 2-dimensional sphere .
hence , absence of harmonic maps from any stratified spheres with energies less than a given bound @xmath9 in certain homotopy classes implies that the harmonic map flow starting from an initial map with energy below @xmath9 in such homotopy classes must develop finite time singularities , as pointed out already in introduction of @xcite . for the mean curvature flow ,
we need to invoke the compactness theorem in @xcite to extract a limiting harmonic stratified sphere if the flow exists for all time .
in this section , we introduce a notion , called normalization , that provides a systematic way to associate a mapping @xmath30 , for any given stratified sphere @xmath14 .
the normalization allows us to study homotopy classes of mappings from ( possibly different ) stratified spheres to a manifold @xmath3 by using mappings from a single @xmath2 to @xmath3 .
let @xmath31 be a connected compact metric space .
as in @xcite , we say @xmath18 is a _ stratified riemann surface with singular set @xmath32 _
if @xmath33 is finite set such that 1 .
@xmath34 is a smooth riemann surface without boundary ( possibly disconnected ) and @xmath35 is a smooth metric @xmath36 ; 2 . for each @xmath37
, there is @xmath38 , such that @xmath39 and @xmath40 , where @xmath41 , and each @xmath42 is topologically a disk with its center deleted .
moreover , @xmath43 can be extended on each @xmath42 to be a smooth metric on the disk . a connected component of @xmath44 is called a _ component of @xmath18_.
the genus of @xmath18 is @xmath45 when @xmath46 , @xmath18 is called a _
stratified sphere_. a stratified sphere can be obtained by shrink finitely many disjoint embedded loops in @xmath2 to points . a continuous map @xmath47 from a stratified surface @xmath18 to @xmath3 _ harmonic _ if @xmath47 is harmonic on each component of @xmath18 .
the map @xmath47 may be trivial on some components .
a domain in a smooth sphere is a _ multiply connected collar _
if it is topologically the sphere minus @xmath48 open disks @xmath49 , @xmath50 , @xmath51 for some positive integer @xmath48 , where @xmath52 for any @xmath23 .
we will use the multiply connected collars to describe the neck regions in blowup analysis . _
_ let @xmath53 be a stratified sphere with @xmath54 singular points @xmath55 .
a continuous surjective map @xmath56 is called _ a normalization of @xmath53 _ , if there exist multiply connected collars @xmath57 , @xmath50 , @xmath58 , such that 1 .
@xmath59 , @xmath60 ; 2 .
@xmath61 is injective on @xmath62 and @xmath63 ; 3 .
@xmath64 .
we denote a normalization by @xmath65 . when @xmath66 , i.e. @xmath67 , @xmath53 is a smooth @xmath2 and any homeomorphism from @xmath2 to @xmath53 is a normalization .
is a multiply connected collar , @xmath68 . ]
the following result is useful to construct a normalization for a given stratified sphere . [ nor ] let @xmath57 , @xmath50 , @xmath69 be multiply connected collars in @xmath2 with @xmath70 for @xmath23 .
let @xmath11 be a continuous map from @xmath2 to a stratified sphere @xmath53 with singular point set @xmath32 .
if @xmath71 for each @xmath72 and @xmath11 is bijective from @xmath73 to @xmath74 , then @xmath32 has @xmath54 elements and @xmath75 is a normalization .
we argue by induction on the number of components of @xmath53 . when @xmath53 has only one component , @xmath32 is empty and the lemma obviously holds .
suppose the result is valid for any @xmath53 with @xmath76 components .
now assume @xmath77 has @xmath48 components .
since @xmath53 is a stratified sphere , it has a component @xmath78 whose topological closure @xmath79 intersects @xmath32 only at one point @xmath80 . by assumption , @xmath81 is bijective from the topological disk @xmath82 to its image @xmath83 , so @xmath84 is a topological disk . without loss of generality , we assume @xmath85 . since @xmath11 is bijective from @xmath86 to @xmath74 , we have @xmath87 and @xmath88 . set @xmath89 .
clearly , @xmath90 is a stratified sphere with @xmath91 components and let @xmath92 be its singular set .
if @xmath80 is not in @xmath92 , then @xmath69 is an annulus , and @xmath93 is a disk . since @xmath94 does not intersect @xmath42 for any @xmath95 , we can take a disk type neighbourhood @xmath96 of @xmath94 in @xmath2 , such that @xmath96 does not intersect @xmath42 for any @xmath95
. then we can extend @xmath97 to a map @xmath98 such that @xmath99 and @xmath100 is bijective from @xmath96 onto the disk @xmath101 .
then @xmath100 is bijective from @xmath102 to @xmath103 . by the induction hypothesis
, @xmath90 has exact @xmath91 singular points @xmath104 , @xmath50 , @xmath105 , and we may assume @xmath106 for @xmath95 .
hence @xmath107 is a normalization of @xmath53 .
if @xmath80 is a singular point of @xmath90 , then @xmath108 with @xmath109 , where @xmath110 are disks in @xmath2 and @xmath52 if @xmath111 ( equivalently , @xmath69 is @xmath2 with at least 3 non - intersecting disks removed ) .
note @xmath84 is one of the @xmath112 s .
define @xmath113 and @xmath114 then @xmath100 is bijective from @xmath115 to @xmath103 . by the induction hypothesis , @xmath90 has exact @xmath91 singular points @xmath104 , @xmath50 , @xmath105 except @xmath80 , and we may assume @xmath106 . hence @xmath116 is a normalization of @xmath53 .
now the induction is complete . for a given stratified sphere ,
its normalization is unique up to homeomorphisms : [ lemma ] let @xmath53 be a stratified sphere with @xmath117 and let @xmath118 be two normalizations of @xmath53 . then there is a homeomorphism @xmath119 such that @xmath120 .
we will argue by induction on the number @xmath54 .
the lemma obviously holds when @xmath66 as both @xmath121 are homeomorphisms from @xmath2 to @xmath2 , and assume it also holds for all @xmath122 . for @xmath123 ,
take @xmath37 .
assume @xmath124 has @xmath125 connected components @xmath126 , @xmath50 , @xmath127 ( each @xmath128 can be a union of 2-spheres ) . since @xmath53 has genus 0 , there is no closed chain of @xmath2 s in @xmath53 , therefore @xmath129 is a stratified sphere with no more than @xmath76 points in its singular set @xmath130 , by noting that @xmath80 is not a singular point of @xmath131 , for each @xmath132 .
then @xmath133 is a multiply connected collar with @xmath125 boundary circles , i.e. , @xmath134 has @xmath125 connected components @xmath135 which are all topological disks . note that @xmath136 and @xmath137 .
let @xmath138 be a continuous map from @xmath139 to @xmath2 such that @xmath140 and @xmath138 is bijective from @xmath112 to @xmath141 , where @xmath3 is the north pole of @xmath2 .
define @xmath142 by @xmath143 then it is easy to check that @xmath144 is a normalization of @xmath131 corresponding to the multiply connected collars @xmath145 subject to @xmath146 . by the same argument
, @xmath147 is a multiply connected collar with @xmath125 boundary curves , and we define a normalization @xmath148 of @xmath149 as above .
then , by the induction hypothesis , we can find a homeomorphism @xmath150 such that @xmath151 , for each @xmath152 . since both @xmath134 and @xmath153 are unions of @xmath125 disjoint topological disks @xmath135 and @xmath154 , respectively , we can find a homeomorphism @xmath155 , such that ( 1 ) @xmath156 and ( 2 ) @xmath157 and @xmath158 .
we now show @xmath159 : \(a ) for any @xmath160 , @xmath161 and @xmath162 by ( 1 ) , hence @xmath163 .
\(b ) for any @xmath164 , we have @xmath165 . if @xmath166 then by ( 2 ) @xmath167 so @xmath162 and in turn @xmath168 by ( 1 ) .
this is contradiction .
therefore @xmath169 provided @xmath170 . but
@xmath171 would imply @xmath172 hence @xmath160 which is impossible . in conclusion
, we have @xmath173 .
we generalize the standard notion of homotopy equivalence of two maps from @xmath2 to @xmath3 to two maps from stratified spheres , possibly different , to @xmath3 .
[ homo]_let @xmath174 and @xmath175 be two stratified spheres .
let @xmath176 be continuous maps from @xmath131 to @xmath3 , @xmath177 .
we say @xmath178 is _ homotopic _ to @xmath179 if there exist normalizations @xmath180 of @xmath181 , respectively , such that @xmath182 is homotopic to @xmath183 as maps from @xmath2 into @xmath3 .
we write @xmath184 if @xmath185 are homotopic .
_ two maps being homotopic is independent of the choice of normalizations : suppose @xmath184
. then there exist normalizations @xmath144 of @xmath131 so that @xmath186 .
let @xmath187 be two arbitrary normalizations of @xmath188 , respectively .
by lemma [ lemma ] , there exist homeomorphisms @xmath150 such that @xmath189 for @xmath177 .
then @xmath190 .
since any homeomorphism of @xmath2 to itself is homotopic to the identity map , it follows @xmath191 .
[ homotopy ] suppose @xmath53 is a stratified sphere with components @xmath78 , @xmath50 , @xmath192 and each @xmath149 is a 2-sphere @xmath2 .
let @xmath193 be a stratified sphere , where each @xmath194 is a stratified sphere and @xmath195 consists of singular points of @xmath196 for any @xmath60 .
suppose that @xmath197 is a continuous map such that @xmath198 is a normalization of @xmath194 and @xmath11 maps each singular point of @xmath53 to a singular point of @xmath196 .
then @xmath199 for any @xmath200 .
it suffices to show that there is a normalization @xmath201 such that @xmath202 is a normalization of @xmath196 , because @xmath203 .
we argue by the induction on @xmath54 . when @xmath204 , @xmath205 and @xmath206 is a normalization of @xmath196 , we may take @xmath207 to be the identity map of @xmath2 .
assume the claim is true for @xmath123 . for @xmath208 , without loss of generality , we assume @xmath78 is a component of @xmath53 which contains only one singular point @xmath80 of @xmath53 . set @xmath209 and @xmath210 .
let @xmath211 .
then @xmath212 satisfies the conditions in the lemma .
thus , by the induction hypothesis , we can find a normalization @xmath213 of @xmath90 such that @xmath214 is a normalization .
next , we construct @xmath207 on @xmath53 .
note that @xmath215 is a normalization of @xmath216 by assumption .
set @xmath217 we have four cases : case 1 : @xmath80 is not a singular point of @xmath90 and @xmath218 is not a singular point of @xmath219 . in this case
, there exists a unique @xmath220 with @xmath221 .
pick @xmath222 such that @xmath213 is bijective on @xmath223 .
let @xmath224 be a continuous map such that 1 .
@xmath225 , 2 .
@xmath226 is bijective from @xmath227 to @xmath228 , 3 .
@xmath226 is the identity map on @xmath229 .
let @xmath230 be a continuous map such that @xmath231 and @xmath61 is bijective from @xmath232 onto @xmath233 .
set @xmath234 obviously , @xmath207 is a normalization of @xmath53 .
it follows from lemma [ nor ] that @xmath235 is a normalization of @xmath196 .
+ case 2 : @xmath80 is not a singular point of @xmath90 but @xmath218 is a singular point of @xmath219 .
in this case , there exists a unique @xmath236 with @xmath221 .
however , we have @xmath237 .
we choose @xmath238 , @xmath239 , @xmath226 , @xmath61 and @xmath207 as before .
we may assume @xmath240 and modify @xmath241 such that @xmath242 .
set @xmath243 .
note @xmath244 .
then @xmath245 is a multiply connected collar homeomorphic to @xmath126 .
clearly , @xmath207 is a normalization of @xmath53 and by lemma [ nor ] , @xmath246 is a normalization of @xmath196 .
+ case 3 : @xmath80 is a singular point of @xmath90 and @xmath218 is not a singular point of @xmath219 . in this case
, we may assume @xmath247 and @xmath248 . let @xmath249 .
it is easy to see @xmath250 : @xmath251 , @xmath252 , then @xmath253 , so @xmath254 .
choose @xmath239 and @xmath238 , such that @xmath255 .
let @xmath61 be a continuous map from @xmath256 into @xmath78 such that @xmath231 and @xmath61 is bijective from @xmath232 to @xmath233 .
set @xmath257 then @xmath207 is a normalization of @xmath53 , and @xmath258 is a normalization of @xmath196 .
+ case 4 : @xmath80 is a singular point of @xmath90 and @xmath218 is a singular point of @xmath219 . without loss of generality
, we may assume @xmath259 , @xmath260 . let @xmath261 .
choose @xmath239 and @xmath238 , such that @xmath255 . modify @xmath241 such that @xmath242 and set @xmath262 .
then @xmath263 is a multiply connected collar .
define @xmath207 as in case 3 .
then @xmath264 is a normalization of @xmath196 .
this section is divided into three parts .
firstly , we assign an ordering for the blowup sequences according to their blowup rates and describe the adjacent ones .
secondly , we construct the neck region and show it is a union of disjoint multiply connected collars .
this construction is convenient for using the normalizations to keep track the homotopy classes .
lastly , we collect known results which are needed for the bubble tree convergence .
we begin by recalling the following important results in @xcite .
[ epsilon ] ( @xmath265-regularity ) there exists an @xmath266 , such that for any harmonic map @xmath47 from the unit disk @xmath267 in @xmath268 into @xmath3 , if @xmath269 , then @xmath270 [ gap](gap constant ) there exists @xmath271 which only depends on @xmath272 , such that there is no non - trivial harmonic map from @xmath2 into @xmath3 , whose energy is in @xmath273 .
let @xmath25 be a sequence of harmonic maps from a compact surface @xmath14 with a riemannian metric @xmath274 to a compact riemannian manifold @xmath3 with @xmath275 .
we may assume @xmath276 in @xmath277 .
the blowup set of the sequence @xmath278 is @xmath279 then we can find a subsequence , still denote by @xmath278 , such that for any @xmath280 , there is @xmath281 with @xmath282 thus , we may assume that @xmath283 for any @xmath284 , and that @xmath25 ( possibly a subsequence ) converges smoothly to @xmath285 in @xmath286 , as @xmath3 is compact .
the blowup set @xmath287 consists of at most finitely many points which are called the ( energy ) concentration points of @xmath278 . near a point @xmath288 , we take an isothermal coordinate system @xmath289 centered at @xmath80 which is the only concentration point in @xmath267 .
the maps @xmath25 can be regarded as harmonic maps from @xmath267 into @xmath3 .
_ a sequence @xmath290 is a _ blowup sequence of @xmath278 at @xmath291 _ if @xmath292 and @xmath293 , @xmath294 , and @xmath295 for some finite set @xmath296 , and @xmath297 .
_ by removability of singularity , @xmath298 extends to a harmonic map from @xmath2 into @xmath3 .
we call the non - constant harmonic map @xmath299 a _
bubble_. [ essen ] _ two blowup sequences @xmath300 and @xmath301 of @xmath278 at @xmath291 are said to be _ essentially different _
if one of the following happens @xmath302 otherwise , they are called _ essentially same_. _ in the sequel , we will write @xmath303 for a blowup sequence for simplicity . [ same ] if two blowup sequences @xmath303 and @xmath304 of @xmath278 at @xmath80 are essentially same , then the bubbles @xmath305 are the same in the sense @xmath306 , where @xmath307 is a linear transformation . since ( [ bubble1 ] ) does not hold , after passing to a subsequence @xmath308 if necessary , we may assume @xmath309 by the definition of blowup sequences , we have @xmath310 and @xmath311 .
observe @xmath312 letting @xmath313 we then have @xmath314 we can then take @xmath315 .
[ different ] let @xmath278 be a sequence of harmonic maps @xmath267 into @xmath3 . if @xmath316 are mutually essentially different blowup sequences of @xmath278 at @xmath317 , where @xmath318 ,
then @xmath319 where @xmath320 is the gap constant in proposition [ gap ] .
let @xmath321 .
we assume @xmath322 converges to @xmath323 in @xmath324 , where @xmath325 is the set of concentration points of @xmath326 , which is finite .
set @xmath327 as @xmath323 is nontrivial , for any @xmath265 , we can find @xmath239 , such that @xmath328 then @xmath329 by definition [ essen ] , we may assume one of the following alternatives holds : 1 .
@xmath330 , which is equivalent to @xmath331 and @xmath332 , 2 .
@xmath333 and @xmath334 ( @xmath308 is a sequence of positive integers , @xmath335 ) . when ( 1 ) holds , we have @xmath336 when @xmath48 is sufficiently large .
since @xmath337 and @xmath338 , we conclude @xmath339 for large @xmath48 .
when ( 2 ) holds , for any @xmath281 , there exists @xmath340 , for all @xmath341 we have @xmath342 then , by noting that @xmath343 and writing @xmath344 , we have @xmath345 thus , @xmath346 and in turn @xmath347 .
then @xmath348 for large @xmath308 , hence @xmath349 for large @xmath308 .
now it follows from @xmath350 that @xmath351 for large @xmath308 .
therefore , we have established : for any @xmath352 , there exists an increasing sequence @xmath353 of positive integers such that @xmath354 applying ( [ differentbubble ] ) and the diagonal sequence procedure , we can find a sequence @xmath355 such that @xmath356 thus , @xmath357 because @xmath358 is arbitrary , we are done .
_ we say the sequence _
@xmath278 has @xmath54 bubbles _
if @xmath278 has @xmath54 essentially different blowup sequences and no subsequence of @xmath278 has more than @xmath54 essentially different blowup sequences . _ although arising from essentially different blowup sequences , some of the @xmath54 bubbles may be essentially same .
a consequence of lemma [ different ] and lemma [ same ] is that @xmath278 has only finitely many bubbles , provided @xmath359 .
the number of essentially different blowup sequences only depends on the upper bound of the energies .
now , we assume @xmath278 has @xmath54 essentially different blowup sequences @xmath360 , @xmath50 , @xmath361 at @xmath80 , where @xmath54 depends on @xmath362 .
note that any subsequence @xmath363 of @xmath278 has the same blowup set @xmath287 and @xmath364 are essentially different blowup sequences of @xmath363 .
therefore , to describe the bubble tree convergent it is convenience for us to make the following assumption on the sequence of harmonic maps : for any @xmath352 , whenever @xmath333 for any @xmath352 , we always assume , by selecting a subsequence if needed , that holds and @xmath365 a diagonal sequence procedure ensures the above holds simultaneously for all @xmath366 and we will still use the index @xmath48 for the subsequence .
we now build a hierarchy for the bubbles . put @xmath367 [ less]_for two essentially different blowup sequences @xmath368 , we say @xmath369 , if @xmath333 and @xmath370 converges as @xmath371 . _ under the assumption ( [ bubble2 ] ) , it holds @xmath372 assume @xmath369 .
take @xmath373 for any @xmath374 , @xmath375 thus @xmath376 .
the other direction holds true as @xmath377 is uniformly bounded hence it converges as @xmath371 by ( [ bubble2 ] ) .
[ still seen ] if @xmath369 , then the bubble @xmath323 of @xmath278 at 0 is also a bubble of @xmath378 at @xmath379 .
we have @xmath380 since @xmath369 , we have @xmath381 and @xmath382 converges as the left hand side @xmath383 converges .
hence @xmath384 is a blowup sequence of @xmath378 at @xmath385 , and @xmath323 is a bubble of @xmath378 .
intuitively , lemma [ still seen ] says that a bubble arises from the lower " side in the relation @xmath386 is still captured from the upper " side , so it sits on an upper " level in the bubble tree .
if the sequence @xmath322 has no concentration points , then @xmath323 must be at the top of the bubble tree .
we now make it more precise .
[ ontop]_a blowup sequence @xmath316 is said to be _
right on top of _ another blowup sequence @xmath387 , if @xmath369 and there is no blowup sequence @xmath388 that is essentially different from @xmath389 and @xmath390 , such that @xmath391 _ the bubble @xmath392 is rooted at a point on the bubble @xmath393 if @xmath394 is right on top of @xmath395 .
is right on top of @xmath396 . ]
[ empty ] if @xmath316 and @xmath387 are essentially different and are both right on top of @xmath388 , then for any fixed @xmath397 , there holds @xmath398 and sufficiently large @xmath48 .
assume the result were not true for some @xmath399 . then there would be a subsequence @xmath400 wiere @xmath401 since @xmath316 and @xmath387 are essentially different blowup sequence , from ( [ essen ] ) we
must then have @xmath402 or @xmath403 . without losing generality
, we assume @xmath404 . by assumption ( [ bubble2 ] ) , @xmath405 converges , hence @xmath406 by definition .
thus @xmath407 which contradicts the fact that @xmath316 is right on top of @xmath388 .
is not right on top of @xmath396 and vice versa . ] to describe the bubble tree convergence of a harmonic map sequence , we need some notations . for convenience , we call @xmath408 to be the 0th blow up sequence , and @xmath285 be the 0th bubble
. then we may say @xmath409 for any @xmath410 . for a blowup sequence @xmath316 of @xmath278 at the center of @xmath267 ,
let @xmath411 be the finite set of blowup points of @xmath412 , here we recall @xmath413 .
since there are only finitely many essentially different blowup sequences , there exists @xmath414 such that for @xmath415 @xmath416 is a multiply connected collar and @xmath322 converges smoothly on @xmath417 when @xmath239 is fixed . for large @xmath48 ,
the following region is in @xmath267 : @xmath418 for a blowup sequence @xmath389 , let @xmath419 , be all of the blowup sequences of @xmath278 which are right on top of @xmath389 with @xmath420 lemma [ still seen ] asserts @xmath421 .
lemma [ empty ] and ( [ equiv ] ) imply @xmath422 is a multiply connected collar . for each @xmath48 , we call @xmath423 the _ neck region for the blowup sequence @xmath424 at @xmath425 _ , where @xmath80 which satisfies .
similarly , we define the neck region for a blowup sequence which is right on top of @xmath389 , and so on .
note that , according to our convention , the neck region for the 0th blowup sequence @xmath426 at a blowup point @xmath427 is @xmath428 , where the union is taken over all essentially different blowup sequences @xmath429 of @xmath278 at @xmath317 that are right on top of @xmath426 .
we will show below that @xmath430 is a finite union of disjoint multiply connected collars for small @xmath239 and large @xmath48 .
for simplicity of notations , write @xmath325 for @xmath431 .
we choose @xmath414 to satisfy : @xmath432 and @xmath433 then for any fixed @xmath415 , it holds that for any @xmath434 and @xmath435 , @xmath436 since @xmath437 implies @xmath438 [ nointersection ] for sufficiently large @xmath48 and @xmath439 , it holds 1 .
@xmath440 for any @xmath441 .
2 . @xmath442 for any @xmath352 .
\(a ) by definition , @xmath443 , and @xmath444 . thus @xmath440 .
\(b ) for any @xmath352 , since @xmath445 are essentially different and we have assumed ( [ bubble2 ] ) , it holds either 1 . @xmath446 and @xmath447 for some positive constant @xmath448 ; or 2 .
@xmath449 and @xmath450 converges . if @xmath451 for all large @xmath48 , then ( b ) holds since @xmath452 and @xmath453 .
if there are @xmath454 for infinitely many @xmath48 , then @xmath455 and @xmath456 , here we still use index @xmath48 for simplicity . therefore ( 1 ) can not happen as @xmath457 if ( 2 ) happens , then @xmath458 and @xmath370 converges to some blowup point @xmath459 of the sequence @xmath460 .
since @xmath462 , it follows @xmath463 given @xmath464 , we can write @xmath465 for some @xmath466 . then @xmath467 recalling that @xmath468 and @xmath333 , we have @xmath469 for a fixed @xmath239 and sufficiently large @xmath48 .
thus , @xmath470 which can be combined with to yield @xmath471 by definition , @xmath472 now it follows from that @xmath473 case 2 .
@xmath474 and @xmath316 is right on top of @xmath387 . in this case
, the definition of @xmath475 directly leads to @xmath476 .
note @xmath477 by .
it then follows @xmath478 .
@xmath474 and there is @xmath388 with @xmath479 . as there are only finitely many blowup sequences , we may assume that @xmath480 is right on top of @xmath390 .
then we have @xmath481 then we have @xmath482 for any @xmath483 , we have for @xmath48 large @xmath484 since @xmath485 , then we get @xmath486 the definition of @xmath487 implies @xmath488 . by , @xmath489
, so we conclude @xmath478 .
this completes the proof of ( b ) . by results in @xcite , a sequence of harmonic maps @xmath278 with uniformly bounded energy , or more generally
, a palais - smale sequences for the energy functional , with uniformly @xmath29-bounded tension fields , converges in the sense of the bubble tree as follows @xmath490 where @xmath285 is a harmonic map from @xmath267 and each @xmath491 is a bubble @xcite , with the following properties : 1 .
@xmath276 in @xmath492 .
2 . there exist essentially different blowup sequences , i.e. satisfy , @xmath316 , @xmath493 , such that @xmath494 3 .
when @xmath333 , holds .
4 . for any fixed @xmath239
, @xmath495 converges smoothly to @xmath323 on @xmath496 , where @xmath417 is defined by .
energy identity : @xmath497
no neck : for any @xmath498 , we have @xmath499 + the bubble tree limit @xmath500 can be viewed as a continuous map from a bubble tree formed by @xmath267 and @xmath54 copies of @xmath2 .
in this section , we first construct normalizations for harmonic maps from a single @xmath2 , and use them to show that there are only finitely many homotopy classes below a given energy level .
this of course follows from parker s bubble tree convergence theorem @xcite , without using the normalizations .
next , we construct normalizations of stratified spheres from the ones for a single @xmath2 and prove theorem [ str ] .
suppose that @xmath278 is a sequence of harmonic maps from @xmath2 into @xmath3 with uniform bounded energies which converges to @xmath501 ; @xmath392 , @xmath50 , @xmath502 in the sense of bubble tree , where @xmath503 are all non - trivial harmonic maps from @xmath2 to @xmath3 .
we will construct a stratified sphere @xmath504 and put @xmath285 , @xmath392 , @xmath50 , @xmath505 together to obtain a harmonic map from @xmath504 to @xmath3 .
we follow the following steps to construct @xmath504 : step 1 . around each blowup point of @xmath278 , call it @xmath506 , take an isothermal coordinate system @xmath289 around @xmath506 such that @xmath507 .
let @xmath392 , @xmath393 , @xmath50 , @xmath508 be all of the bubbles which are right on top of @xmath285 , i.e. the corresponding blowup sequences are right on top of that of @xmath285 , which , in this initial step , is @xmath426 in our convention . each @xmath323 is a harmonic map from a standard sphere @xmath149 to @xmath3 .
we identify the south pole of @xmath149 with @xmath509 .
repeat step 1 at @xmath510 each concentration point of @xmath511 where @xmath512 , @xmath513 each concentration point of @xmath514 where @xmath515 @xmath516 similar to the above two cases , at all higher levels in the bubble tree . in the end
, we get a stratified sphere @xmath504 and a harmonic map @xmath517
. next we construct a normalization @xmath518 of @xmath504 such that @xmath519 is very close to @xmath25 .
let @xmath520 be the blowup points of @xmath278 .
now , on an isothermal coordinate system around each concentration point @xmath521 ( we may assume that @xmath522 is the only blowup point of @xmath278 in the local chart ) , we define a continuous map @xmath523 as follows . at first level : outside the disks @xmath524 s : 1 .
@xmath518 is the identity map on @xmath525 ; 2 .
@xmath518 is bijective from @xmath526 onto @xmath527 , for each @xmath528 ; 3 .
@xmath529 ; at the second level : inside @xmath524 s : we first recall from proposition [ nointersection ] for all large @xmath48 and @xmath439 that @xmath530 s are disjoint and lie inside @xmath524 , and @xmath531 .
select those @xmath498 and @xmath80 such that the corresponding disks @xmath532 are disjoint .
define @xmath518 bijectively from @xmath533 onto @xmath534 and maps @xmath535 to @xmath536 .
further , define @xmath518 on @xmath537 by sending every point to @xmath538 .
third and higher levels : on the disks @xmath539 which are inside @xmath540 .
repeat the second step .
finally , on the remaining disks which correspond to the bubbles at the highest level , i.e. there are no blowup sequences right on top of those in the previous steps , note the boundary of each of these disks is mapped to a point in the previous step , now define @xmath518 be a diffeomorphism from the disk to @xmath2 with the point deleted .
this completes the definition of @xmath518 . as there is no neck
, we have @xmath541 thus , @xmath542 .
it is easy to check that @xmath518 is a normalization of @xmath504 . then by definition @xmath543 .
hence , let @xmath25 be a sequence of harmonic maps from the sphere @xmath2 to @xmath3 with uniformly bounded energies . then after passing to a subsequence , all @xmath544 are in the same homotopy class .
let @xmath545 be a sequence in @xmath2 .
we say @xmath546 is the _
@xmath61-limit of @xmath545 _ , if @xmath547 by the no - neck result , we have @xmath548 it is easy to check that for any sequence @xmath549 , we can find a subsequence which has a @xmath61-limit .
+ the bubble tree convergence of @xmath278 can also be described via @xmath518 .
let @xmath32 be the set of singular points of @xmath504 .
let @xmath550 .
clearly , for any @xmath281 , @xmath551 is well defined on @xmath552 when @xmath239 is sufficiently small .
moreover @xmath553 is harmonic and converges smoothly .
if @xmath554 is a connected component of @xmath555 , then for any @xmath37 @xmath556 let @xmath557 and @xmath558 be the set of singular points of @xmath559 , where @xmath149 s are smooth 2-spheres .
note that @xmath560 may happen .
let @xmath25 be a harmonic map from @xmath559 into @xmath3 , i.e. @xmath25 is continuous on @xmath559 and harmonic on each component of @xmath559 .
we may assume @xmath561 converges to @xmath562 in the sense of bubble tree , where @xmath563 is not trivial .
then we get a stratified sphere @xmath564 , and a harmonic map @xmath565
. now , let @xmath566 be the singular point which connects @xmath78 and @xmath567 . by passing to subsequences
if needed , we may assume @xmath568 is the @xmath61-limit of @xmath569 on @xmath570 and @xmath571 is the @xmath61-limit of @xmath569 on @xmath572 . by the no - neck result , @xmath573 .
we then identify @xmath568 and @xmath571 to get a stratified sphere @xmath574 , and a harmonic map from @xmath574 into @xmath3 . repeating this construction for @xmath575 whenever @xmath576 are joint at a singular point
, we get a stratified sphere @xmath504 and a harmonic map @xmath577 .
we may view @xmath504 as the union of @xmath578 . by the gap theorem for harmonic maps from @xmath2 @xcite , the number of components of @xmath559 is uniformly bounded in @xmath48 .
therefore there are only finitely many homeomorphic types .
it suffices for us to assume @xmath559 are of the same homeomorphic type .
let @xmath579 be harmonic maps with the uniformly bounded energies .
we assume @xmath25 converges in the bubble tree sense to a harmonic map @xmath517 as we described in the previous subsection .
moreover , we may set @xmath580 , @xmath581 , @xmath582 , such that @xmath583 converges in the sense of bubble tree to @xmath584 , where @xmath585 .
let @xmath586 be the normalization of @xmath578 , which is defined in subsection [ b.t . ] .
to use lemma [ homotopy ] , we need to combine @xmath586 over @xmath587 to produce a continuous map from @xmath559 to @xmath504 . to achieve this
, we need to modify @xmath586 on neighborhoods of the singular points .
let @xmath588 , such that @xmath545 has a @xmath61-limit @xmath589 .
we have two cases : case 1 .
@xmath238 is a singular point of @xmath578 . in this case
, we can suitably choose @xmath239 , so that @xmath590 .
set @xmath591 of @xmath564 .
@xmath238 is not a singular point of @xmath578 .
choose a disk @xmath592 , such that @xmath593 for large @xmath48 and @xmath594 contains no singular points of @xmath564 .
let @xmath595 be a homeomorphism of @xmath578 onto itself such that @xmath596 and @xmath597 is the identity map .
thus , @xmath598 is a normalization of @xmath578 with @xmath599 .
thus , @xmath600 , defined by @xmath601 on each @xmath149 , is a continuous map and @xmath602 is a normalization of @xmath578 . by lemma [ homotopy ] , @xmath603 .
furthermore , in case 2 , we may suitably choose @xmath595 , as @xmath238 is a @xmath61-limit of @xmath545 , such that @xmath595 converges to the identity map as @xmath604 ( this is always true for case 1 ) .
note that @xmath605 and recall that @xmath606 converges smoothly on @xmath594 .
then for any @xmath607 we can suitably choose @xmath239 , such that @xmath608 by , we may assume @xmath609 then @xmath610 thus , @xmath611 and it follows @xmath612 since we have shown @xmath603 .
in this section , we will construct a family of smooth maps from @xmath2 into some 3-dimensional manifold which are not homotopic to each other while their energies are uniformly bounded above .
the elements @xmath613 in the fundamental group @xmath614 of @xmath18 which yield distinct homotopy classes are produced in proposition [ h.sequence ] in appendix .
[ 3-manifold ] let @xmath17 , where @xmath19 and @xmath615 or @xmath616 .
then there is a sequence of smooth mappings @xmath617 , such that @xmath618 and @xmath176 is not homotopic to @xmath619 for any @xmath23 .
let @xmath620 be the universal cover of @xmath18 and let @xmath621 , @xmath613 be the ones in proposition [ h.sequence ] .
let @xmath32 and @xmath3 be the south pole and the north pole of @xmath2 respectively .
let @xmath622 be a map from @xmath2 to @xmath623 which is homotopic to @xmath621 , such that @xmath622 is a constant in a neighborhood of @xmath32 and a neighborhood of @xmath3 respectively .
let @xmath624 , where @xmath625 is the stereographic projection from @xmath32 .
we assume @xmath626 .
let @xmath627\rightarrow \hat{m}$ ] be an embedded curve with @xmath628 and @xmath629 .
we define @xmath630 as follows : @xmath631 where @xmath38 be sufficiently small and @xmath397 be sufficiently large , such that @xmath632 is constants on @xmath633 and @xmath634 .
we have @xmath635 @xmath636 and @xmath637
so , we can find suitable @xmath265 , such that @xmath638 since @xmath639 on @xmath634 , we can view @xmath640 as a map from @xmath2 into @xmath623 . we need to prove that @xmath641 is not homotopic to @xmath642 for any @xmath23 . by the hurewicz theorem
, we only need to check that @xmath643_{h_2}\neq [ v_j]_{h_2}$ ] .
obviously , @xmath644_{h_2}=[\tilde{u}']_{h_2}+[\gamma_i\circ\tilde{u}]_{h_2},\s [ v_j]_{h_2 } = [ \tilde{u}']_{h_2}+[\sigma_j\circ\tilde{u}]_{h_2}.\ ] ] since @xmath645_{h_2}\neq [ \gamma_j\circ\tilde{u}]_{h_2}$ ] , we have @xmath643_{h_2}\neq [ v_j]_{h_2}$ ] .
now , we let @xmath646 .
then @xmath176 is not homotopic to @xmath619 for any @xmath23 .
moreover , it is easy to check @xmath647 .
theorem [ str ] can be used to produce harmonic map flows of @xmath2 in closed 3-manifolds which develop singularities in a finite time .
[ harmonicblowup ] let @xmath19 and @xmath615 or @xmath616 be closed 3-dimensional riemannian manifolds and @xmath17 . then there exist infinitely many smooth maps @xmath21 such that @xmath22 are not homotopic for any @xmath23
, @xmath648 and the harmonic map flow that begins at @xmath25 develops a finite time singularity .
theorem [ 3-manifold ] guarantees existence of a sequence of mutually nonhomotopic smooth maps @xmath649 with uniformly bounded energies , and only finitely many ( may be none ) of them contain harmonic maps from some stratified spheres to @xmath18 in their homotopy classes by theorem [ str ] .
select the @xmath640 s which do not have harmonic maps from stratified sphere to @xmath18 in their homotopy classes and denote them by @xmath25 .
if the harmonic map flow @xmath650 admits a longtime solution , then by results in @xcite , as @xmath651 , @xmath652 will converge to a harmonic map from a stratified sphere to @xmath18 which is homotopic to @xmath25 .
this contradicts the choice of @xmath25 .
it is well - known that a minimal sphere is also a harmonic sphere . in this section ,
we combine theorem [ str ] , theorem [ 3-manifold ] and the compactness theorem in @xcite to produce mean curvature flows that blowup in a finite time in closed 3-manifolds .
[ mcf ] let @xmath19 and @xmath615 or @xmath616 be closed 3-dimensional riemannian manifolds and @xmath17 . then there exist infinitely many embeddings @xmath26 such that @xmath27 are nonhomotopic for any @xmath23 and the areas @xmath653 are uniformly bounded , and the mean curvature flow of @xmath2 initiated at @xmath28 develops a singularity in a finite time .
first , select @xmath25 as in the proof of theorem [ harmonicblowup ] .
next , we replace the line @xmath654 in @xmath25 by a thin cylinder to get an embedding @xmath640 from @xmath2 with uniformly bounded area and @xmath655 . this can be done as follows .
recall , in the proof of theorem [ 3-manifold ] , @xmath25 maps the annulus @xmath656 to the curve @xmath654 and @xmath632 maps the domain @xmath657 to a point and the disk @xmath658 to a point for some large @xmath399 ( these two domains correspond to the two small caps in @xmath2 that are mapped to points by @xmath622 ) , and @xmath632 is diffeomorphic on the complement of these two domains
. extend @xmath654 a little at its end points in @xmath18 , and take the thin embedded cylinder in the tubular neighborhood of the extended @xmath654 such that it is the circle of radius @xmath239 in the normal plane at each point of the extended @xmath659 .
we can assume that one boundary circle of the cylinder is @xmath660 pointwise , and the other boundary circle of the cylinder is @xmath661 twisted by a local diffeomporphism .
define @xmath640 by sending each ray @xmath662 in the annuls @xmath663 to the corresponding curve in the twisted cylinder in @xmath18 , and outside the annulus let @xmath640 be @xmath25 .
we can assume @xmath640 is smooth by slight modification in a neighborhood of the boundary circles .
note that @xmath640 is conformal , in fact isometric , with respect to the pull back metric @xmath664 on @xmath2 from @xmath18 .
since there is only one conformal structure on @xmath2 , there exists a diffeomorphism @xmath665 such that @xmath666 , where @xmath667 is the round metric on @xmath2 .
due to conformality , area of @xmath668 , and conformal invariance of energy implies @xmath669 .
set @xmath670 .
as any diffeomorphim of @xmath2 to @xmath2 is homotopic to the identity map of @xmath2 , we have @xmath671 . in conclusion ,
@xmath28 s are conformal embeddings from @xmath2 to @xmath18 with uniformly bounded energies in the round metric , and @xmath672 is not homotopic to @xmath673 for @xmath60 .
assume the mean curvature flow @xmath674 begins at @xmath28 exists for longtime .
write @xmath675 for the image of @xmath2 at time @xmath676 .
since @xmath677 for all @xmath678 , there exists a sequence @xmath679 such that @xmath680 each surface @xmath681 is immersed , and @xmath682 is conformal with respect to the pull - back metric from @xmath18 . as there is only one conformal class on @xmath2 ,
@xmath683 is conformal for some diffeomorphism @xmath684 . along the mean curvature flow ,
the area of the conformal maps @xmath685 is uniformly bounded @xmath686 and so is the willmore energy @xmath687 we now recall a special case ( namely , domain is @xmath2 ) of the compactness theorem in @xcite : suppose that @xmath688 is a sequence of @xmath689 branched conformal immersions of @xmath690 in a compact manifold @xmath18 .
if @xmath691 then either @xmath688 converges to a point , or there is a stratified sphere @xmath692 and a @xmath689 branched conformal immersion @xmath693 , such that a subsequence of @xmath694 converges to @xmath695 in the hausdorff topology , and the area and the willmore energy satisfy @xmath696 by this compactness theorem , there is a stratified sphere @xmath697 , where each @xmath698 is a 2-sphere @xmath2 , and a @xmath689 branched conformal immersion @xmath699 with @xmath681 converging to @xmath700 in the sense of bubble tree , therefore @xmath701 is homotopic to @xmath28 .
moreover , @xmath702 we see that @xmath703 is a branched minimal immersion from @xmath704 , hence @xmath701 is harmonic and conformal , and its dirichlet energy can be estimated as @xmath705 this contradicts the fact that there is no harmonic map from a stratified sphere to @xmath18 which is homotopic to @xmath28 but with energy not exceeding @xmath706 , by the choice of @xmath28 .
in this section , we set @xmath707 , where @xmath708 and @xmath709 are 3-dimensional closed manifold and @xmath18 is the connected sum of @xmath708 and @xmath709 . that is to say , we delete a small ball @xmath710 inside @xmath708 and a small ball @xmath711 inside @xmath709 and glue together their boundary spheres .
we assume both @xmath708 and @xmath709 are not @xmath712 .
we will use the following notations in this secton : * @xmath713 , @xmath714 , and @xmath715 . * @xmath623 : the universal cover of @xmath18 with covering map @xmath620 .
* @xmath47 : a diffeomorphism from @xmath2 to @xmath77 .
* @xmath621 : a lift of @xmath47 .
* @xmath716 : the deck transformation composed with @xmath621 for @xmath717 . * for @xmath718 , @xmath719_{h_2}$ ] denotes the element in @xmath720 represented by @xmath298 .
assume @xmath47 is trivial in @xmath721 . by proposition 3.10 in @xcite
, @xmath722 bounds a compact contractible submanifold @xmath723 in @xmath18 , and by the poincar conjecture , @xmath723 is a ball . since @xmath724 , @xmath77 separates @xmath18 and @xmath723 is connected , we have @xmath725 or @xmath726 .
we assume @xmath725 .
that @xmath727 implies @xmath728 is open in @xmath729 .
since @xmath730 and @xmath731 is connected , we conclude @xmath732 .
thus @xmath733 , which is a contradiction to our assumption .
assume there is a point @xmath735 .
let @xmath736\rightarrow \widetilde{m}$ ] be a lift of @xmath737 with @xmath738 .
let @xmath739 and @xmath740\rightarrow \gamma\circ\tilde{u}(s^2)$ ] be a continuous curve with @xmath741 and @xmath742 .
then the product path @xmath743 is a loop in @xmath623 . since @xmath744
, @xmath745 is trivial in @xmath614 .
the loop @xmath746 is trivial in @xmath614 as it is contained in @xmath722 .
so @xmath747 in @xmath614 , and it contradicts @xmath748 .
[ component ] @xmath749 has exactly two connected components , and the closure of each of them is noncompact in @xmath623 .
moreover , for any nontrivial @xmath717 , @xmath750 has exactly 3 connected components , and only one of them has boundary @xmath751 . since @xmath621 is an embedding
, @xmath749 is either connected or has two components .
assume it is connected .
let @xmath736\rightarrow \widetilde{m}$ ] be a smooth curve in a neighborhood of @xmath752 such that @xmath434 intersects @xmath621 at @xmath753 transversally and @xmath753 is the only intersection of @xmath621 and @xmath434 .
since @xmath749 is connected , we can find a curve @xmath754\rightarrow \widetilde{m}\backslash \tilde{u}(s^2)$ ] with @xmath755 and @xmath756 .
let @xmath757 for @xmath758 $ ] and @xmath759 for @xmath760 $ ] .
then @xmath761 is a loop which intersects @xmath752 transversally at only one point . thus @xmath761 is not trivial in @xmath762 , which contradicts the fact that @xmath763 .
so @xmath764 has two components .
let @xmath765 and @xmath766 be two components of @xmath767 . since @xmath768 by lemma [ no.intersection ] , we have @xmath769 or @xmath770 .
assume @xmath770 . using similar arguments
, we can prove @xmath771 also has exact two components ( @xmath766 is simply connected by the van kampen theorem ) .
obviously , there is only one component whose boundary consists of @xmath752 and @xmath772 .
if @xmath773 is compact in @xmath623 , it is a smooth compact manifold with boundary @xmath752 , which implies @xmath774_{h_2(\widetilde{m})}=0 $ ] .
hence @xmath752 is trivial in @xmath775 by the hurewicz theorem .
however , it is well - known that @xmath776 . lemma [ nontrivial ] yields a contradiction .
similarly , @xmath777 is noncompact in @xmath623 .
first , we prove @xmath785 restricts to a covering map from @xmath778 to @xmath731 .
it suffices to prove it is surjective . for any @xmath786
, we can find a curve @xmath736 \rightarrow m_1'$ ] with @xmath787 and @xmath788 .
let @xmath789 be the lift of @xmath434 in @xmath623 with @xmath790 .
since @xmath791 , @xmath789 has no intersection with @xmath792 . thus @xmath793 as @xmath778 is a connected component of @xmath779 .
hence @xmath794 .
next , we show @xmath778 is simple connected .
assume there is a close curve @xmath795 which is not trivial in @xmath796 .
then @xmath797 is a closed curve in @xmath731 . if @xmath798 is nontrivial in @xmath799 , it is nontrivial in @xmath800 as @xmath801 . since @xmath802 , then @xmath798 is also nontrivial in @xmath614 .
note that @xmath38 is the lift of @xmath798 , so @xmath38 can not be a closed curve in @xmath778 . we get a contradiction .
if @xmath798 is trivial in @xmath799 , since @xmath803 is a covering map , @xmath38 must be trivial in @xmath778 .
again , we have a contradiction . [ h2 ]
assume @xmath804 and @xmath778 is a connected component of @xmath779 , which is a universal cover of @xmath780 .
if @xmath805 , such that @xmath806 and @xmath807 are in the boundary of @xmath778 , then @xmath808_{h_2}$ ] and @xmath809_{h_2}$ ] are linearly independent in @xmath720 . since @xmath810 , we have @xmath811 and @xmath812 .
thus @xmath813 contains at least 3 elements
. then we can find @xmath814 which is not @xmath815 , such that the closure of @xmath816 contains at least 3 boundary components @xmath817 , @xmath807 and @xmath818 .
assume @xmath819_{h_2}$ ] and @xmath809_{h_2}$ ] are linearly dependent .
equip @xmath623 with the metric pulled back from @xmath18 via @xmath785 .
then fixing a point @xmath820 , we can find a sufficiently large @xmath397 , such that the metric ball @xmath821 in @xmath623 contains @xmath822 for @xmath823 , and @xmath808_{h_2(b_r(p)\cap \widetilde{m})}$ ] and @xmath809_{h^2 ( b_r(p)\cap \widetilde{m})}$ ] are linearly dependent in @xmath824 .
we can choose @xmath397 via sard s theorem and by lemma [ component ] , such that the closure of @xmath825 is a smooth manifold with nonempty smooth boundary .
take a smooth curve @xmath736\rightarrow { \hat{m}}\cup\partial\hat{m}$ ] so that @xmath434 intersects @xmath817 and @xmath830 at @xmath831 and @xmath832 transversally , respectively , and demand @xmath833 , @xmath834 .
by lemma [ component ] , @xmath765 is unbounded , so @xmath835 .
we can extend @xmath434 in @xmath765 to a curve @xmath836\rightarrow \widetilde{m}$ ] such that @xmath837 .
note that @xmath838 , @xmath839)\cap \gamma_2\circ\tilde{u}(s^2)=\emptyset$ ] .
similarly , we can extend @xmath434 to a curve @xmath840\rightarrow \widetilde{m}$ ] , such that @xmath841 and @xmath842)\cap \gamma_2\circ\tilde{u}(s^2 ) = \emptyset$ ] .
let @xmath843 be the manifold obtained by identifying two copies of @xmath844 along the boundary .
then @xmath808_{h_2(m^*)}$ ] and @xmath809_{h_2 ( m^*)}$ ] are linearly dependent in @xmath845 .
the two copies of @xmath434 , however , yields a closed curve in @xmath843 which intersects @xmath817 transversally only at one point and has no intersection with @xmath807 .
this is impossible if @xmath846_{h_2}$ ] and @xmath847_{h_2}$ ] are linearly dependent . by lemma [ no.intersection ]
, @xmath854 has a component @xmath851 , whose boundary consists of @xmath752 and @xmath772 .
let @xmath855 and @xmath856 be the other two components with @xmath857 and @xmath858 .
we now prove the closure of @xmath859 is compact .
let @xmath860\rightarrow \widetilde{m}$ ] be a smooth curve which intersects @xmath752 at @xmath831 transversally with @xmath861 and @xmath862 . if the closure of @xmath859 is not compact , then fixing a point @xmath783 , for any large @xmath397 , @xmath863 . extend @xmath864}$ ] to a curve from
@xmath865 $ ] to @xmath866 so that @xmath867 and @xmath868 . by lemma [ component ] ,
@xmath855 is unbounded , then @xmath869 .
similarly we extend @xmath870}$ ] in @xmath855 to a curve @xmath871\rightarrow\hat{m}_1 $ ] such that @xmath872 , which implies that @xmath873 . then similar to the proof of lemma [ h2 ]
, we can construct a manifold @xmath843 , such that @xmath774_{h_2(m^*)}=[\gamma\circ\tilde{u}(s^2)]_{h_2(m^*)}$ ] , and there exists a closed curve @xmath798 , such that @xmath798 intersect @xmath752 in @xmath843 transversally at only one point and has no intersection with @xmath874 , which yields a contradiction .
next , we show @xmath851 is simple connected .
let @xmath875 be the manifold obtained by gluing a 3-ball to @xmath876 along the boundary and let @xmath877 be the manifold obtained by gluing a 3-ball to @xmath851 along each boundary . since @xmath878 , @xmath879 it follows @xmath880 . by the poincar conjecture ,
the closed simply connected @xmath877 is @xmath712 and in turn @xmath881 .
we claim that there is no @xmath889 that satisfiest @xmath890 or @xmath737 and @xmath891 .
assume there is such a @xmath892 .
let @xmath778 be the component of @xmath779 which is a universal cover of @xmath780 and @xmath893 belongs to @xmath894 .
lemma [ h2 ] asserts that @xmath895_{h_2}$ ] and @xmath896_{h_2}$ ] are linearly independent in @xmath897 .
but it follows from lemma [ cylinder ] that @xmath898 . a contradiction .
note that @xmath802 contains infinitely many elements since @xmath903 . by lemma [ not.homotopy ] ,
when @xmath904 , for any sequence @xmath905 with @xmath906 for any @xmath23 , we have @xmath902_{h_2}\neq [ \gamma_j\circ u]_{h_2}$ ] .
when @xmath885 , choose a sequence such that @xmath907 for any @xmath23 , where @xmath887 is the nontrivial element of @xmath888 .
by lemma [ not.homotopy ] again , @xmath645_{h_2}\neq [ \gamma_j\circ\tilde{u}]$ ] for any @xmath23 . | we show that the set of harmonic maps from the 2-dimensional stratified spheres with uniformly bounded energies contains only finitely many homotopy classes .
we apply this result to construct infinitely many harmonic map flows and mean curvature flows of 2-sphere in the connected sum of two closed 3-dimensional manifolds @xmath0 and @xmath1 , which must develop finite time singularity . |
ads backgrounds of string theory are often fruitful arenas for studying holographic dualities and for constructing sigma models with r - r fluxes , among other things . ads backgrounds of type 0a string theory should be no exception . in particular , the recent discovery of a matrix quantum mechanics dual to two - dimensional type 0a string theory @xcite suggests a promising direction for understanding ads@xmath0/cft@xmath1 @xcite .
also , two - dimensional ads presents one of the simplest backgrounds in which to study sigma models in r - r flux . as a first step in these pursuits ,
we present here a two - parameter family of ads@xmath0 solutions to the two - dimensional type 0a effective action .
one of the parameters is the tachyon vev @xmath2 ( equivalently , the ratio of dualized r - r field strengths , @xmath3 ) .
the other parameter is the string coupling @xmath4 ( equivalently , the magnitude of the field strength @xmath5 ) . in these solutions ,
string loops can be suppressed by reducing the string coupling , but the high curvature of the spaces makes higher order @xmath6 corrections important . in section 2 ,
we briefly review the 2d 0a string theory . in section 3
, we present our family of ads@xmath0 solutions . in section 4 ,
we discuss possible corrections to our solutions from higher order terms in the effective action .
the ten - dimensional type 0a string theory is given by the same worldsheet action as the type iia string , but with a gso projection onto the closed string sectors @xmath7 where @xmath8 and @xmath9 denote the eigenvalue of the worldsheet fermion number operator @xmath10 . in ten dimensions ,
each of these sectors contains a tower of states corresponding to the possible transverse oscillations . in two dimensions ,
however , there is no room for transverse oscillations , so the situation is much simpler .
we have the graviton @xmath11 and dilaton @xmath12 in the ( ns@xmath8,ns@xmath8 ) sector , the tachyon @xmath2 in the ( ns@xmath9,ns@xmath9 ) sector , and two gauge fields @xmath13 from the r - r sectors that give rise to two field strengths , @xmath14 . an allowed background for this theory
is the two - dimensional linear dilaton vacuum ( @xmath15 and @xmath16 ) with an exponential tachyon wall ( @xmath17 ) and zero field strengths ( @xmath18 ) . the worldsheet action for this string theory is the action of @xmath19 super liouville theory plus the action for a free scalar superfield .
the action for @xmath20 super liouville theory can be written in superfield formalism for both the liouville superfield and the background dilaton . ] as @xmath21 where @xmath12 is the scalar superfield @xmath22 and the covariant derivatives are given by @xmath23 in the case @xmath24 , this yields a theory with central charge @xmath25 . when combined with the @xmath26 theory of a free scalar superfield @xmath27 with action @xmath28 we get a critical scft with central charge @xmath29 .
this is the worldsheet action for two - dimensional type 0a in the linear dilaton vacuum .
the effective spacetime action was calculated in @xcite and was found to be @xmath30\ , .\end{gathered}\ ] ] the first few terms in a taylor expansion of the @xmath31 functions are @xmath32 there is evidence that the exact expression for @xmath33 is @xmath34 @xcite , and we will use this form for @xmath35 in our calculations .
to simplify the action ( [ spacetimeaction ] ) , we will dualize the r - r field strengths : @xmath36 in the second line , we have introduced an auxiliary field @xmath37 .
the equation of motion for @xmath37 is @xmath38 which , when substituted in , gives the original action . in the third line , we have integrated out @xmath39 which constrains @xmath37 to be a constant . therefore
, in the third line , the fields @xmath39 and @xmath37 are no longer functionally integrated . the full action can now be written as @xmath40 \ , .\end{gathered}\ ] ] varying with respect to the metric @xmath11 , dilaton @xmath12 , and tachyon @xmath2 gives the equations of motion @xmath41 \\
+ \frac{e^{-2\phi}}{2\kappa^2 } \left(-2 \nabla^\mu \nabla^\nu
\phi + f_1(t ) \nabla^\mu t \nabla^\nu t \right ) = 0\ , , \end{gathered}\ ] ] @xmath42 and @xmath43 \\ - \frac{1}{4\pi \alpha ' } f_3^\prime ( -t ) q_+^2 - \frac{1}{4\pi \alpha ' } f_3^\prime ( t ) q_-^2 = 0\ , , \end{gathered}\ ] ] where primes denote differentiation with respect to @xmath2 . setting @xmath12 and
@xmath2 constant , we find @xmath44 @xmath45 and @xmath46 with the @xmath47 metric @xmath48 the ricci scalar is @xmath49 the solution with @xmath51 and @xmath50 satisfies the equations of motion with ads radius given by @xmath52 and dilaton given by @xmath53 a notable feature of this solution is that the curvature radius is fixed at a value of order the string length .
this implies that higher order @xmath6 terms in the effective action will be important .
this will be addressed in section 4 .
also , note that we are free to tune the string coupling to zero by ramping up the strength of the r - r flux . in this case
, the `` tachyon '' is massive for all values of @xmath54 .
this can be seen as follows .
the @xmath55 equation of motion , to first order in @xmath2 , gives us @xmath56 the @xmath57 equation of motion , to zero order in @xmath2 , tells us that @xmath58 and , when substituted into the linearized @xmath55 equation , gives us @xmath59 finally , substituting our background expressions for @xmath12 and @xmath60 , we get @xmath61 so that the tachyon mass is @xmath62 .
the authors of @xcite noted that , in ten dimensions , r - r flux could stabilize the tachyon potential . in our two - dimensional case
, we see that the r - r flux makes the otherwise - massless tachyon massive .
solutions to the wave equation in ads@xmath0 are most readily attained in poincare coordinates , in which @xmath63 in these coordinates , the wave equation takes the form @xmath64 using separation of variables , we can write the general time - dependent , positive - frequency solution as @xmath65 . the normalizable solution is readily obtained in terms of a bessel function as @xmath66 where @xmath67 .
the static solutions are obtained by noting that the wave equation @xmath68 implies that @xmath69 where @xmath70 .
therefore , the general static solution is @xmath71 note that , although these static solutions are non - normalizable , they may make an appearance as approximate solutions in regions of spacetime that are ads - like .
solutions to the wave equation in global coordinates are a little more difficult to come by , but they have been worked out in @xcite . in
the global coordinates @xmath72 the normalized positive - frequency solutions are @xmath73 where @xmath74 , @xmath75 is the gegenbauer polynomial , and @xmath76 is once again related to @xmath77 by @xmath78 . note that , unlike in poincare coordinates , the spectrum in global coordinates is discrete .
the solution given in the previous section can be deformed by moving the constant value of @xmath2 away from zero .
the solution is given by @xmath80 @xmath81 and @xmath82 again , it is clear that , for all solutions in this family , we can send the string coupling to zero while holding fixed both the tachyon vev @xmath2 and the ads radius @xmath83 .
this is accomplished by sending @xmath84 and @xmath85 to infinity while holding the ratio @xmath86 fixed .
it is not evident from these equations whether or not there exists an ads@xmath0 solution with one of the @xmath54 s , say @xmath87 , set to zero .
setting @xmath88 would require a @xmath2 of order 1 , but to understand such large values of @xmath2 would require a more complete knowledge of @xmath89 .
specifically , @xmath88 would require that @xmath90 and it is not known whether this equation has solutions .
it should be noted that the ads spaces presented here are solutions to the first few terms in the effective action . since the ads radius is of order the string length ,
we expect higher order terms in @xmath6 to change some of the quantitative features of the solutions , such as the exact value of the ads radius or the true mass of the tachyon . however , as we shall discuss here , the qualitative features of the ads solutions are rather generic and are not expected to be changed by the higher order @xmath6 terms
. we can ask what other terms might make contributions to the equations of motion , and , therefore , might change features of the ads solution . for simplicity ,
let us concentrate on the @xmath50 solution found in section 3.3 .
since we seek a solution with @xmath12 constant and @xmath2 zero , any terms linear or higher order in @xmath91 or @xmath2 will make no contribution to the @xmath92 or @xmath57 equation of motion . therefore , the only higher order ns - ns terms that will contribute are contractions of the riemann tensor .
since we are in two dimensions , the riemann tensor is related to the ricci scalar as @xmath93 as a result , any scalar combination of @xmath94 riemann tensors can be rewritten as @xmath95 for some constant @xmath96 .
therefore , without loss of generality , we may consider replacing @xmath60 in the effective action with @xmath97 . in our @xmath50 solution
, these terms will change equation ( [ radiusequation ] ) to @xmath98 so long as there are real roots to this equation , then there will exist an ads solution with the corresponding radius .
the equation ( [ dilatonequation ] ) for the string coupling is modified to @xmath99 once again , it is clear that the string coupling may be taken to zero by making @xmath100 large .
the @xmath55 equation of motion is , of course , unaffected by these @xmath101 terms . on the other hand
, the @xmath55 equation will be affected by terms in the action that are linear in @xmath2 . however , there are no ns - ns terms involving odd powers of @xmath2 @xcite
. terms with odd powers of @xmath2 are possible when they also involve r - r fields , for example , the @xmath102 terms
. however , the proposed symmetry @xcite of the theory under @xmath103 and @xmath104 would imply that setting @xmath105 makes such terms zero
. we may also ask how the tachyon mass would be altered by the terms higher order in @xmath6 . for example , ns - ns terms of the form @xmath106 would change the tachyon mass . in fact , we fully expect the value of the tachyon mass to be corrected by higher order terms .
it is clear that the form of the ads solutions are rather generic and are not likely to be dramatically changed by terms higher order in @xmath6 .
this fact motivates a search for the corresponding worldsheet sigma model describing type 0a strings propagating in these ads@xmath0 spaces . because of the existence of nonzero r - r fluxes
, the correct sigma model will most likely not be found using the nsr formalism .
fortunately , several other worldsheet formalisms have been developed that have allowed for quantization of the string in r - r backgrounds .
for example , the hybrid formalism has been used to study superstring quantization in ads@xmath107 @xmath108 s@xmath109 @xcite , ads@xmath0 @xmath108 s@xmath110 backgrounds @xcite , and curved 2d backgrounds @xcite .
this work was supported in part by doe grant de - fg02 - 91er40654 .
i am grateful to a. simons and a. strominger for helpful conversations .
i would also like to thank my friends at eastern electronics for their last - minute technical support . | we present a two - parameter family of ads solutions to the two - dimensional type 0a effective action . hep - th/0312156 2.5 cm 1.0 cm david mattoon thompson 0.5 cm _ jefferson physical laboratory + harvard university + cambridge , ma 02138 _ 0.8 cm december 2003 1.5 cm |
since 1985 , when the presence of non - statistical fluctuations has been evidenced in the dependence of the cross section of deep inelastic processes on the incident energy @xcite , the excitation function statistical analysis became a method to obtain information on the di - nuclear system ( dns ) lifetime .
naturally it was addressed the question if the lifetime extracted from the ef analysis is identifiable with the value extracted previously from the angular distributions of the final fragments .
the authors of ref .
@xcite pointed out that the dns lifetime extracted by these two methods are close to each other .
they reached this conclusion based on a reaction model in which the partial coherence mechanism already addressed in @xcite has been considered .
indeed the model leads to an angular distribution pattern having a degree of focusing given by the number of interfering partial waves and to fluctuating excitation functions .
the analysis within this model has been done for the @xmath5si + @xmath6ni and @xmath7c + @xmath8 mg systems for which experimental information was available by that time .
afterwards the excitations functions in dissipative heavy ion collisions ( dhic ) have been measured and analyzed for other light and medium mass systems : @xmath9f + @xmath10y , @xmath5si + @xmath11ti , @xmath9f + @xmath12cu , @xmath5si + @xmath5si , @xmath9f + @xmath13v @xcite .
a new model , partial overlapping molecular level model ( pomlm ) , has been also developed in order to explain the presence of the fluctuations in the ef for dhic introducing , besides the partial coherence , the hypothesis that the dns is excited in the low density region of molecular levels @xcite.the large amplitude of the fluctuations persists due to the lower density of the states in the vicinity of the yrast line .
an unified explanation for fluctuation phenomenom in the excitation functions of elastic , inelastic and deep inelastic collision has been obtained .
the aim of the excitation functions for @xmath9f + @xmath0al and @xmath0al + @xmath0al collisions measured by us was to see to what extent the fluctuation phenomenom is present in the case of these light systems where none of the participants to the reaction is an @xmath14-conjugate nucleus . for the system
@xmath9f + @xmath0al the excitation functions have been obtained for the final fragments with z = 6 - 12 for a total kinetic energy loss ( tkel ) window of 5 mev width centered at 20 and 30 mev in order to study the dependence of this phenomenom on tkel .
such a study has not been addressed before .
non - statistical fluctuations have been evidenced in the cross section efs of final fragments and for the corresponding fluctuation correlation widths , @xmath15 , no dependence on z , tkel and fragment emission angle , @xmath16 , has been observed within error limit @xcite .
an average value of @xmath15 of ( 170@xmath265 ) kev has been obtained .
the corresponding dns lifetime ( 3.9 @xmath2 1.1)@xmath310@xmath4 sec is in agreement with the values extracted from angular distribution studies @xcite .
the present paper presents results obtained for the @xmath0al + @xmath0al collision . after a short review of the experimental procedures and the criteria used to select deep inelastic events in section [ expr ] , results obtained from cross section efs analysis for the @xmath0al + @xmath0al system
are reported in subsection [ sa ] of the paper .
the subsection [ evcor ] is dedicated to the presentation of a study of the influence of the evaporation corrections on the fluctuating structure of ef .
a pattern of the energy autocorrelation function with secondary damped oscillations besides the lorentzian structure from @xmath17 = 0 , where @xmath17 is the energy increment in the center of mass system ( cms ) , was obtained .
the observed oscillation periodicity could be explained if hyperdeformed rotational states are supposed to be excited in dns during the collision @xcite .
this aspect is discussed in subsection [ deform ] .
conclusions are presented in section [ conclu ] .
the experiment was performed at the smp tandem accelerator from lns -catania using @xmath0al ions with incident energies e@xmath18 = ( 122 - 132 ) mev .
the energy loss in self supported targets of @xmath0al ( @xmath19 75 kev ) is less than the energy increment of 250 kev ( 125 kev in cms ) between two successive experimental points .
the beam current was measured with a tantalum plated faraday cup provided with an electron suppressing guard ring .
the reaction products were detected and identified using the experimental device dracula @xcite from which only the large area position sensitive ionization chambers ( ics ) and the corresponding parallel plate avalanche counters in front of them have been operated .
the ics were filled with ar(90% ) + ch@xmath20(10% ) at 106.8 torr .
the polar and azimuthal angles spanned by the ic were @xmath21 = 24@xmath22 , @xmath23 = 4@xmath22 , respectively .
the energy resolution at the elastic peak was 2.5% , the angular resolution 0.5@xmath22 and the charge resolution better than 0.3 charge units .
the ics being centered at 24@xmath22 in the laboratory system , an angular range 12@xmath24 was covered continuously in a single measurement for each incident energy . the grazing angle at e@xmath18 = 132 mev is of @xmath19 15 @xmath22 , so the weight of deep inelastic processes was increased within the measured angular interval . nevertheless , a quasielastic component is present in the tke spectra of fragments with z = 11 , 12 as could be seen in fig .
[ fig1 ] for e@xmath18 = 128 mev .
the arrows on the figure represent the energies corresponding to complete energy damping @xcite .
the excitation functions for deep inelastic processes are obtained taking events with tke larger than these values and well bellow the quasielastic component . the tke window
used for e@xmath18 = 128 mev is represented by the hatched area on fig .
[ fig1 ] and corresponds to the total kinetic energy loss window from 20 to 32 mev , kept the same for all incident energies .
the obtained excitation functions present fluctuations with quite large amplitude as could be observed in figs .
[ fig2 ] , where the efs corresponding to z = 8 - 12 , 14 are given .
the full circles represent the cross section of the primary products and the empty triangles of the secondary ones .
the influence of the evaporation corrections on the efs , obtained from comparing the results for these two sets of ef , will be discussed in subsection [ evcor ] . in the following we refer only to the analysis of the ef for primary products .
fluctuations with amplitude larger than the statistical errors , which are of symbol size , could be observed .
the statistical analysis of the efs has been done following the recipe from ref .
fluctuations present in the efs corresponding to different final channels are strongly correlated , evidenced by large channel cross correlation coefficients , @xmath25 , as can be seen in table [ tab1 ] .
this shows that the observed fluctuations are not of compound nucleus origin .
taking into account this correlation , in order to improve the statistics , it was obtained an ef by summing the efs for reaction products having higher yields ( z = 8 , 11 and 12 ) , which is given in the bottom panel of fig .
[ fig2 ] . .[tab1]channel cross correlation coefficients [ cols="^,^,^,^,^,^ " , ] the correlation width obtained by us is very close to the value of @xmath15 = 135 kev obtained from the analysis of the ef for dissipative collisions in the @xmath0al + @xmath0al system on a lower incident energy interval e@xmath18 = 114 - 127 mev in ref .
@xcite .
the lower value of the correlation width obtained for the @xmath0al + @xmath0al system compared to the case of @xmath9f + @xmath0al system could be explained by the lower excitation energy per nucleon available in heavier system .
the position of the entry region ( horizontal lines correspond to upper and lower e@xmath26 energy for which measurements have been done ) for the two systems is given in fig .
the entry regions for both systems lie in the vicinity of the yrast line ( thick continuous line ) given by the orbital cluster model ( ocm ) energies : @xmath27 where @xmath28 is the coulomb barrier , @xmath29 the angular momentum and @xmath30 the total moment of inertia of di - nuclear system . the thin continuous line on the same figure represents @xmath31 as given by the calculation based on the transport model ( dona code ) @xcite for the higher limit of the measured energy interval .
one can see that the description of the @xmath32 dependence on @xmath33 within the transport model extensively used for heavy systems works quite well for light systems too as it was already noticed in our first study of a light system @xcite .
the extended theory of ericson fluctuations and the other models developed in order to explain the presence of non - statistical fluctuations in dhic @xcite are elaborated for primary reaction products .
experimentally are measured the observables of secondary products and for comparison with models , evaporation corrections have to be done .
the energy spectra were corrected for nucleon evaporation using the iterative procedure described in ref .
a parameterization of neutron and proton separation energies as a function of n / z ratio elaborated for the mass region up to 50 has been used @xcite . in order to estimate the influence of these corrections on the efs of the reaction products resulted from the @xmath0al + @xmath0al interaction , we decided to construct also a set of efs before the evaporation corrections
these are represented with empty triangles in fig .
[ fig2 ] . at the first sight
no sizable differences between the two sets of efs could be seen . for a more quantitative evaluation
, we calculated eafs for efs obtained without evaporation corrections .
these are represented with empty triangles in fig .
the extracted @xmath34 values are given in the second column of the table [ tab2 ] .
one can conclude that within error limits a modification of oscillation structure in the efs the by evaporation corrections procedure is not observed , for the present data .
it is not clear that the obtained result could be extended for higher excitation energy .
experimental eaf of dhic shows , besides the lorentzian structure at @xmath17 = 0 , secondary oscillations @xcite .
it was shown that these oscillations appear when the lifetime of dns is equal or greater than its rotation period as a result of interference between different revolutions @xcite .
the period of the secondary structures is given by @xmath35 , where @xmath36 is the angular velocity of the di - nucleus .
the experimental value of @xmath17 in connection with the most probable kinetic energy @xmath37tke@xmath38 could be used to obtain an evaluation of the separation distance at the break - up moment of the composite ( di - nuclear ) system and , in this way , information on the deformation of the system during the interaction .
the energy autocorrelation functions from fig .
[ fig3 ] present secondary structures with a period of @xmath19 0.85 mev .
such a period of oscillation implies for the distance of separation a value of @xmath39 @xmath19 10.9 fm , taking for the angular momentum the sticking configuration value given by @xmath40 , where @xmath41 and @xmath42 are the intrinsic and total momenta of inertia , respectively .
for @xmath43 we used the value of 47.8@xmath44 calculated with the dona code for @xmath45 = 128 mev , and @xmath46 = 43.7@xmath44 was calculated using the formula given in ref .
the value of the separation distance is compatibile with that obtained by us for the @xmath9f + @xmath0al system @xcite . for this value of the separation distance one
obtains , for fragmentation @xmath47na + @xmath48p , a value of @xmath37tke@xmath38 equal to 41.6 mev .
it was supposed that @xmath37tke@xmath38 equals the final channel barrier given by the sum of the coulomb and centrifugal energies @xcite .
the centrifugal energy is calculated for @xmath49 = @xmath50 @xmath19 32.7 .
this value is in a good agreement with the experimental most probable tke of @xmath19 42 mev corresponding to z = 11 ( see corresponding panel of fig .
[ fig1 ] ) .
one has to remark also that the used value of @xmath50 is within the @xmath49-window ( @xmath49 = 27 - 33 ) corresponding to the integration tkel window ( tke = 32 - 44 mev ) given by the transport model calculation ( thin continuous line in fig .
[ fig4](b ) ) .
the separation distance values otained for both studied systems mean a quite large deformation for the reaction partners of the dns .
for relative momentum of inertia at break - up , the following relation can be written : @xmath51 where @xmath52 = @xmath53 is the usual interaction radius .
it was also necessary to consider larger momenta of inertia in order to explain the period of secondary oscillations observed in the eaf of other systems ( @xmath7c + @xmath8 mg , @xmath8 mg + @xmath8 mg and @xmath5si + @xmath5si ) in this mass region @xcite , while the period of secondary structures of eaf for heavier systems ( @xmath9f + @xmath10y , @xmath54ni + @xmath54ni , @xmath54ni + @xmath55ni , @xmath54ni + @xmath56ti ) from refs .
@xcite could be explained using the moment of inertia calculated at @xmath52 .
this shows the possibility of the existence of super- and hyper- deformation in this mass region , as it was predicted thirty years ago @xcite .
the authors of ref .
@xcite concluded their paper : the outstanding problem is to devise methods that would identify the presence of such superdeformed nuclei which in some existing experiments have been produced without having been detected .
meanwhile experimental methods to evidence the existence of such nuclei have been developed .
one should notice results of recent experiments providing information on nuclear deformation in the mass region close to the mass of the di - nuclear systems studied in the present paper , which appears to be a very interesting one from this point of view .
for example light charge particle spectra emitted by @xmath57ni and @xmath58ti compound nuclei could be described by using in formula ( 1 ) of the yrast line an effective moment of inertia @xmath59 = @xmath60(1+@xmath61 ) , where @xmath62 , @xmath63 are the deformability parameters and @xmath60 is the rigid body moment of inertia @xcite . for deformability parameter values deduced from the fit of experimental spectra , one obtaines for the effective moment of inertia @xmath59 = 1.4@xmath60 and @xmath59 = 1.5@xmath60 for the @xmath57ni and @xmath58ti cn , respectively .
( [ two ] ) obtained for the di - nucleus @xmath64fe is similar to these relations .
the relative momentum of inertia at break - up obtained by us can be also seen as an effective momentum introduced in order to reproduce the experimental periodicity of eaf . the @xmath65-ray spectrum from the decay of gdr built in hot @xmath56ti was recently measured and its shape suggests the presence of an elongated 3-axial equilibrium shape for this nucleus at j=30@xmath44 @xcite .
the excitation functions for the system
@xmath9f + @xmath0al ( di - nucleus @xmath56ti ) have been constructed for tkel windows centered at 20 and 30 mev .
one can see from fig .
[ fig4](a ) that for these tkel windows ( tke = 50 - 55 mev and 40 - 45 mev , respectively ) correspond angular momentum values near the region where very elongated shapes could appear .
thus , one can consider the analysis of the cross section efs as a robust experimental method to evidence and extract information on such elongated objects .
the results obtained in the present paper and in the previous ones @xcite , using this method , evidence a large deformation of composite systems formed in the first stage of elastic , inelastic scatering and deep inelastic processes in the mass region 36 - 56 .
the efs analysis for deep inelastic processes in the @xmath0al + @xmath0al collision shows that rotational states with large deformation are excited in the first stage of the reaction .
the correlation width of the fluctuations extracted from cross section ef is equal to ( 115@xmath229 ) kev to which coresponds a dns lifetime of ( 5.7 @xmath2 2.2)@xmath310@xmath4sec .
the large channel cross correlation coefficients show the non - statistical origin of the fluctuations . they support a reaction mechanism where special states of rotational ( molecular ) nature play the role of doorway configurations towards a regime characterized by stochastic exchange of nucleons between interacting nuclei as the main mechanism behind the dissipative phneomena in light heavy ion collisions .
we are grateful to the operating crew of lns tandem accelerator for the quality of delivered beam . for the high quality targets we owe special acknowledgement to mr .
c. marchetta .
we acknowledge prof .
e. migneco and prof .
d. vinciguerra for their permanent support .
a. glaesner , w. dnnweber , w. hering , d. konnerth , r. ritzka ; r. singh and w. trombik , phys . lett . *
b169 * , 153 ( 1986 ) ; t. suomijrvi , b. berthier , r. lucas , m. c. mermaz , j. p. coffin , g. guillaume , b. heusch , f. jundt and f. rami , phys . rev . *
c36 * , 181 ( 1987 ) ; a. de rosa , g. inglima , m. romano , v. russo , m. sandolini , g. cardella , g. pappalardo , f. rizzo , g. fortuna , a. m. stefanini , s. beghini , g. montagnoli and c. signorini , phys . rev . *
c37 * , 1042 ( 1988 ) ; g. cardella , m. papa , g. pappalardo , f. rizzo , q. wang , a. de rosa , e. fioretto , g. inglima , m. romoli , m. sandoli , r. setola , l. corradi , g. montagnoli , a. m. stefanini , z. phys .
* a336 * , 387 ( 1990 ) ; f. rizzo , g. cardella , a. de rosa , a. di pietro , a. donofrio , e. fioretto , g. inglima , m. papa , g. pappalardo , m. romano , m. romoli , f. terrasi , m. sandoli and g. s. wang , z. phys . *
a349 * , 169 ( 1994 ) ; m. papa , g. cardella , a. di pietro , s. l. li , a. musumarra , g. pappalardo , f. rizzo , a. de rosa , g. inglima , m. la comarra , d. pierroutsakou and m. romoli , z. phys . *
a353 * , 205 ( 1995 ) ; wang qi , lu jun , xu hushan , li songlin , zhu yongtai , fan enjie , yin xu , zhang yuhu , li zhichang , zhao kui , lu xiuqin , hu xiaoqing , phys . lett .
* b388 * , 462 ( 1996 ) i. berceanu , , a. andronic , m. duma , d. mois , m. petrovici , a. pop , v. simion , s. aiello , g.imm@xmath67 , g.lanzan@xmath68 , a. pagano , g. raciti , g. derasmo , a. pantaleo , g. guarino , h.saccetti _ phys.rev . _
* c57 * , 2359 ( 1998 ) .
a.pop , a.andronic , i. berceanu , m. duma , m. mois , m. petrovici , v. simion , a. bonasera , g. imm@xmath67 , g.lanzan@xmath68 , a. pagano , g. raciti , n. colonna , g. derasmo , a. pantaleo , h. feldmeier , j. schnack _ nucl .
phys . _ * a679 * , 793 ( 2001 ) m. petrovici , v. simion , i. berceanu , a. demian , m. duma , m. ivacu , d. mois , e. osvath , g. pascovici , r. bock , a. gobbi , k. d. hildenbrand , u. lynen , h. sann , g. augustinski , h. beeskov , h. w. daues , ipne preprint np-55 - 1986 v. simion , rom .
. phys . * 38 * , 513 ( 1993 ) g. pappalardo , phys . lett . * 13 * , 320 ( 1964 ) wang qi , li song - lin , tian wen - dong , hu peng - yun , lu jun , f. amorini , m. cabibbo , g. cardella , p. figuere , a. musumarra , m. papa , g. pappalardo , f. rizzo , s. romano , s. tudisco , b. heusch , chin .
* 16 * , 876 ( 1999 ) m. petrovici , a. andronic , i. berceanu , a. bu , m. duma , d. mois , a. pop , v. simion , a. bonasera , g. imm ' e , g. lanzan ` o , a. pagano , g. raciti , n. colonna , g. derasmo and a. pantaleo , z. phys . * a354 * , 11 ( 1996 ) c. k. gelbke , p. braun - munzinger , j. barette , b. zeidman , m. j. levine , a. gamp , h. l. harney and th .
walcher , nucl . phys . *
a269 * , 460 ( 1976 ) j. b. natowitz , m. n. namboodiri , r. eggers , p. gonthier , k. geofferoy , r. hanus , c. towsley and k. das , nucl . phys .
* a277 * , 477 ( 1977 ) s. yu .
kun , u. abbodanno , m. bruno , n. cindro , m. d ' agostino , p. m. milazzo , r. a. ricci , t. ritz , b. a. robson , w. scheid , a. v. vagov , g. vannini , l. vannucci , z. phys . * a359 * , 145 ( 1997 ) s. cohen , f. plasil , w. j. swiatecki , ann
82 * , 557 - 596 ( 1974 ) c. bhattacharya , m. rousseau , c. beck , v. rauch , r. m. friedman , d. mahboub , r. nouicer , p. papka , o. stezowski , a. hachem , e. martin , a. k. dummer , s. j. sanders , a. szanto de toledo , phys . rev . *
c65 * 014611 ( 2001 ) p. papka , c. beck , f. haas , v. rauch , m. rousseau , p. bednarczyk , s. courtin , o. dorvaux , k. eddahbi , j. robin , a. sanchez , i. zafra , o. stezowski , a. p@xmath67vost , acta phys . pol .
* b34 * , 2343 ( 2003 ) a. maj , m. kmiecik , m. brekiesz , j. grebosz , w. meczyy@xmath69ski , j. stycze@xmath69 , m. ziebli@xmath69ski , k. zuber , a. bracco , f. camera , g. benzoni , b. million , n. blasi , s. brambilla , s. leoni , m. pignanelli , o. wieland , b. herskind , p. bednarczyk , d. curien , j. p. vivien , e. farnea , g. de angelis , d. r. napoli , j. nyberg , m. kici@xmath69ska - habior , c. m. petrache , j. dudek , k. pomorski , eur .
j. * a20 * , 135 ( 2004 ) , ( nucl - ex/0302004 ) a. maj , m. kmiecik , a. bracco , f. camera , p. bednarczyk , b. herskind , s. brambilla , g. benzoni , m. brekiesz , d. curien , g. de angelis , e. farnea , j. grebosz , m. kici@xmath69ska - habior , s. leoni , w. meczyy@xmath69ski , b. million , d. r. napoli , j. nyberg , c. m. petrache , j. stycze@xmath69 , o. wieland , m. ziebli@xmath69ski , k. zuber , n. dubray , j. dudek , k. pomorski , nucl . phys . *
a731 * , 319 ( 2004 ) | the excitation functions ( efs ) for different fragments produced in the @xmath0al + @xmath0al dissipative collisions have been measured in steps of 250 kev in the incident energy range 122 -132 mev .
deep inelastic processes have been selected by integrating events over a total kinetic energy loss ( tkel ) window of 12 mev between 20 and 32 mev .
large fluctuations have been observed in all the studied efs .
non - statistical origin of these fluctuations is confirmed by large channel cross correlation coefficients
. energy autocorrelation function ( eaf ) of the cross section ef presents damped oscillation structure as predicted for the case when a di - nuclear system with a lifetime ( @xmath1 = ( 5.7 @xmath2 2.2)@xmath310@xmath4sec ) similar with its revolution period ( t = 4.9@xmath310@xmath4sec ) is formed .
the periodicity of the eaf oscillations is used to obtain information on the deformation of the @xmath0al + @xmath0al di - nucleus . |
sagittarius a * ( sgr a * ) is the radiative manisfestation of our galactic supermassive black hole
. discovered almost four decades ago ( balik & brown 1974 ) , this intriguing radio source has been well studied at all wavelengths , generating interest among both theorists and observers ( see melia 2007 for a comprehensive review ) . a significant development with sgr a * occurred in 2001 , with the discovery of bright x - ray flares emanating from its vicinity ( baganoff et al . 2001 ) .
lasting anywhere from 40 minutes to over 3 hours , these events are associated with enhanced x - ray emission , 30200 times greater than that of its quiescent state . in the intervening years
, many more x - ray flares have been observed by both _
chandra _ and xmm-_newton _ , and analyzed extensively ( see , e.g. , goldwurm et al . 2003 , porquet et al .
2003 , blanger et al 2005 , porquet et al .
2008 ) . among the many intriguing features characterizing these events
are temporal substructures as short as 200 s , indicating that the physical processes responsible for the outbursts occur within a mere 1020 schwarzschild radii of a @xmath0 black hole ( schdel et al .
2002 , ghez et al .
2003 ) .
flaring events in sgr a * have also been detected in the nir ( genzel et al .
2003 ) and even at mm and sub - mm wavelengths ( see , e.g. , zhao et al .
2004 , trap et al .
2011 ) . at least in the infrared ,
the flare spectra appear to be power laws with variable linear polarization , suggesting optically thin synchrotron emission as the likely mechanism producing the excess flux .
but though the observational profile of these bursts continues to be refined , in tandem with significant theoretical discourse and modeling , we still do nt completely understand where or why these flares originate , nor exactly how the x - rays are produced , or even if the variable mm and sub - mm emission is really indicative of the flares spectral extension into the radio domain .
one thing is certain , however that no significant progress is likely to occur without multiwavelength observations and a broadband approach to the data analysis and modeling .
for this reason , several groups have systematically organized large simultaneous observational campaigns , utilizing both ground - based and orbital platforms ( see , e.g. , blanger et al .
2004 , 2006 , yusef - zadeh et al .
2006 , marrone et al .
2008 , yusef - zadeh et al .
2009 , dodds - eden et al .
2009 , kunneriath et al .
2010 , trap et al .
2011 ) .
the key results emerging from these ( and other similar ) studies are as follows ( see dodds - eden et al .
2009 for a more comprehensive compilation ) : ( 1 ) ir / nir flares occur more frequently than x - ray flares , typically @xmath1 times per day , compared with roughly 1 per day for the latter ; ( 2 ) consequently , every x - ray flare appears to be associated with an nir flare , but not every nir event is correlated with an x - ray flare ; additionally , ( 3 ) though x - ray and nir flares occur simultaneously , with no significant delay , the former are typically shorter in duration ; ( 4 ) polarimetric measurements in the nir show that the source is significantly polarized ( by as much as @xmath2 ; see eckart et al . 2006 , nishiyama et al . 2009 ; ( 5 ) the brightest flares have a constant spectral index @xmath3 between 3.8 and 1.6 @xmath4 m , where the flux is given as @xmath5 ( see , e.g. , hornstein et al .
2007 ) . at lower intensities
, there may also be a possible trend of spectral index with flux , though this is still in dispute .
additional results emerge when we include observations at longer wavelengths , but our focus in this _ letter _ will be on the nir and x - ray emission , so we will defer the more complete discussion to a subsequent paper .
as we shall see , three of the critical questions that must be resolved for a complete understanding of the flare phenomenon in sgr a * ( at least in the nir / x - ray ) are ( 1 ) how are the x - rays produced ? ( 2 ) how do we understand sgr a * s polarized nir flare emission ? and
( 3 ) can we predict the associated polarization fraction and position angle for the variable x - ray component , in anticipation of future polarimetric measurements at @xmath6 kev energies ? based on preliminary work ( some of it by our group )
, we now suspect that the nir / x - ray flares originate close to sgr a * s event horizon . as such
, it will not be possible to carry out a meaningful simulation of the emission and polarization profiles without a general relativistic ( gr ) approach .
addressing these ( and related ) questions within the context of gr is therefore the principal goal of this _
sgr a * s transient x - ray emission has variously been attributed to thermal processes or invese compton scattering of either seed mm / sub - mm photons ( by the relativistic particles producing the ir / nir synchrotron emission ; see , e.g. , yusef - zadeh et al .
2006 ) or of the ir / nir photons themselves as part of a synchrotron self - compton ( ssc ) approach ( see , e.g. , markoff et al .
2001 , liu & melia 2001 , eckart et al .
it has also been suggested that both the nir and the x - ray components may be due to the same synchrotron process ( see , e.g. , yuan et al .
2004 ) .
but the analysis of several bright flares detected over the past few years has all but ruled out these earlier proposals ( see dodds - eden et al .
2009 , trap et al .
single component synchrotron models ( with a particle distribution @xmath7 ) are problematic because the high - energy electrons required to generate x - ray synchrotron emission have very short cooling times ( much shorter than a typical x - ray flare duration ) .
thus , even a continuous injection to replenish the energetic population can not sustain the same power - law index @xmath8 at both low and high energies . if instead the x - rays are submm photons inverse - compton scattered by the electrons producing the nir emission , the largest permissible size of the quiescent radio - producing region is about @xmath9 ( dodds - eden et al .
2009 , trap et al .
2011 ) , far smaller than the measured fwhm size ( @xmath10 ; doeleman et al .
2008 ) of sgr a * at 1.3 mm .
this uncomfortably tight restriction ( see also liu & melia 2001 ) is compounded by several other difficulties , but even the size issue on its own already rules out the external " compton scattering scenario .
attempts at fitting an ssc spectrum to the combined nir / x - ray data are equally problematic because such a model requires low electron energies ( with @xmath11 1015 ) and unrealistically strong magnetic fields ( @xmath12 g ) and very large particle densities ( much larger than those inferred for the quiescent emission ) .
we are thus left with the following rather tightly constrained indicators . during a typical flare
, there is little if any detectable emission at @xmath13 m , implying that the flare emission spectrum ( characterized by the power density @xmath14 ) must rise from the mir towards the nir .
this is consistent with the spectral index @xmath3 described above .
it is clear , therefore , that the electron population producing the @xmath15-band flare has a different distribution of energies than that associated with the submm bump ( see , e.g. , melia 2007 ) , so a nir flare can not simply be a small change in the overall properties of the steady radio - submm emitting region .
however , it is unrealistic to expect a power - law particle distribution such as this to maintain the same power - law index @xmath8 at all energies .
synchrotron energy losses , not to mention the escape time from the acceleration region , both depend on the particle energy ( see , e.g. , liu et al .
it is well known ( see , e.g. , pacholczyk 1970 ) that while energetic electrons are injected continuously into the system by the acceleration process , the emitted steady - state photon spectrum has an index @xmath16 ( with @xmath8 the particle index ) up to a cooling break " frequency @xmath17 , steepening to an index @xmath18 ( corresponding to a particle index @xmath19 ) above it .
the cooling break corresponds to the electron energy ( or lorentz factor @xmath20 ) at which the escape time @xmath21 from the system is equal to the ( synchrotron ) cooling time @xmath22 .
the cooling timescale is given as ( pacholczyk 1970 ) @xmath23 but the escape timescale depends on several factors , including @xmath24 and the ambient particle density @xmath25 . nonetheless , detailed calculations of the acceleration process ( see nayakshin & melia 1998 , liu et al .
2004 , 2006 ) indicate that for a broad range of conditions , @xmath21 corresponds roughly to 3 times the light transit time in the system .
( for an analogous , though much smaller system than sgr a * , see also misra & melia 1993 . )
thus , if the emission region is near the marginally stable orbit ( see below ) , we would expect @xmath26 for a black hole mass @xmath27 . in the overall spectrum
, the cooling break is therefore expected to occur at a frequency @xmath28 i.e. , somewhere above the nir component and well below the x - ray , as long as the magnetic field intensity is of order 30 g , essentially the value required to produce sgr a * s quiscent emission . for the simulations reported in this _
, we will therefore assume that the relativistic particle distribution producing the nir / x - ray flare has the following form : @xmath29 where the choice of @xmath8 , @xmath20 ( or , equivalently , @xmath24 in equation 3 ) , and the normalization constant @xmath30 , are all based on earlier fits to the data ( see , e.g. , dodds - eden et al . 2009 ,
trap et al .
2011 ) .
an additional constraint on the flaring region is provided by the temporal substructure seen in typical bursts , particularly in the nir .
for example , one sees @xmath15-band flux variations up to @xmath31 of the peak value on a timescale of only @xmath32 minutes . taken at face value
, these fluctuations could be telling us that most of the burst activity is occurring near the marginally stable orbit surrounding a @xmath0 black hole ( see , e.g. , melia et al .
however , light - travel arguments constrain the size of the emitting region even further when one uses the shortest time - scale variations seen , e.g. , in the 4 april , 2007 burst reported by dodds - eden et al .
( 2009 ) . there
, in the @xmath15 lightcurve , very rapid changes in flux ( by factors of 120170 @xmath33 , with a statistical significance @xmath34 ) were observed within a timescale @xmath35 seconds .
thus , a large fraction of the emitting plasma must be confined to a region @xmath36 which , for a @xmath0 black hole , corresponds to about @xmath37 , where @xmath38 is the schwarzschild radius .
we may be witnessing the emergence of a very confined region , such as a magnetic flux tube breaking out above the disk , or perhaps a ring - like hot " region near the marginally stable orbit .
the latter scenario , in particular , would be expected on the basis of an instability in the disk , such as the rossby - wave instability studied in detail by tagger & melia ( 2006 ) and falanga et al .
( 2007 ) . our goal in this _
letter _ is to ascertain whether these observed nir / x - ray flare characteristics can be explained in terms of the geometry and the particle distributions described above , taking into account all of the essential general relativistic effects , such as light bending and area amplification .
unfortunately , what is not known precisely is the fraction of transient x - ray emission hidden below sgr a * s quiescent high - energy luminosity ( which presumably originates from a different location , as discussed earlier ) . as such , our calculation of the x - ray to nir intensity ratio necessarily provides only an upper limit to the observed value , but because we here work principally with a single particle population ( equation 4 ) , even this upper limit is very probative .
( as our understanding of the conditions in the ambient medium surrounding sgr a * improves , the uncertainty in this fraction will not doubt diminish ; see , e.g. , crocker et al .
2010 . )
this is because in addition to the total flux densities themselves , we will also carefully calculate the polarization fractions expected in this scenario for both the nir and x - ray components . the discovery of significant linear polarization in the nir ( see , e.g. , eckart et al .
2006 ) was an important milestone in the study of these flares because it unambiguously confirmed the signature of nonthermal synchrotron radiation .
but we still do nt know if this measured _ degree _ of polarization is consistent with the geometry and particle distribution suggested by the other flare characteristics .
our simulations here should help us resolve this question .
and an equally important question concerns the polarized fraction of x - rays .
of course , we do not yet have data relevant to this issue . however , it is not unreasonable to expect the x - ray polarization fraction to differ from that of the nir ( since the polarized emissivities are energy dependent ; see below ) , and to differ considerably between two possible origins , ( i ) synchrotron radiation due to a cooled power - law particle distribution , or ( ii ) synchrotron - self - comptonization .
thus , that portion of our simulation involving x - rays is motivated not only by our desire to provide a firm prediction of the x - ray polarization fraction for comparison with future observations , but also to demonstrate how the degree of x - ray polarization relative to that in the nir differs depending on the mechanism producing the x - rays .
in so doing , we hope to establish the viability of an important diagnostic for examining whether or not the x - rays are indeed produced by a cooled power - law population of electrons , and whether they originate from a small region near the marginally stable orbit .
our group has been developing methods to produce synthetic images of sgr a * using general - relativistic ray - tracing codes since the 1990 s ( hollywood & melia 1997 , falcke et al .
2000 , bromley et al .
2001 , falanga et al .
other investigators have by now successfully produced their own algorithms for carrying out similar ray - traced simulations , producing a rich ensemble of complementary results ( see , e.g. , broderick & loeb 2005 , zamaninasab et al .
2011 ) . in a broader context
, the transfer of polarized radiation through black - hole spacetimes has also been explored by dovciak et al .
( 2004 ) , noble et al .
( 2007 ) , dolence et al .
( 2009 ) , dexter et al .
( 2009 ) , davis et al .
( 2009 ) , schnittman & krolik ( 2010 ) , and shchenbakov & huang ( 2011 ) . spurred by the evident need to carry out a comprehensive broadband investigation of sgr a * s flaring state , in which the radiation we observe at earth may be polarized , but which may also include a comptonized component ( possibly the x - rays ) originating from locations other than those where the primary ( or seed ) photons are emitted , we have recently developed a more comprehensive code called pollux ( polarized lux ) .
this algorithm ( more fully described in falanga et al .
2011 ) incorporates several indispensible physical attributes associated with the emission and propagation of a multi - component spectrum through the schwarzschild spacetime surrounding a non - spinning black hole .
often , the presence of a relativistic jet is taken to be the signature of a high spin rate .
the fact that no such structure has ever been detected in sgr a * suggests that this object may be spinning slowly , if at all ( but see also melia et al .
2002 , liu & melia 2002 ) .
thus , in order to keep the simulations as straightforward as possible , we have chosen to work with the schwarzschild metric in this instance . at the inner edge of the disk , where the nir and x - ray photons are presumably being emitted ,
the gas is circularized and settled into keplerian motion , where a magnetohydrodynamic dynamo produces a predominantly azimuthal magnetic field @xmath39@xmath40@xmath41 ( hawley , gammie & balbus 1996 ) .
overcoming the rate of field destruction in the differentially rotating portion of the inflow , the field reaches a saturated intensity since the dynamo timescale is shorter than the dissipation timescale in this region ( melia et al .
however , it is well - known by now that the disk surrounding sgr a * can not be a large standard disk , which would produce a detectable quiescent - state ir emission ( see , e.g. , falcke & melia 1997 ) .
thus , all of these processes , even in the quiescent state , must be restricted to a compact region no bigger than 1020 schwarzschild radii from the black hole . for the incipient emission at nir wavelengths , we track both the extraordinary ( perpendicular to * b * ) and ordinary ( parallel to * b * ) waves ,
whose emissivities may be written ( see , e.g. , rybicki & lightman 1985 ) : @xmath42\omega^{(1-p)/2}\;,\ ] ] and @xmath43\omega^{(1-p)/2}\;,\ ] ] where @xmath44 in these expressions , @xmath45 is the angular frequency in the comoving frame , @xmath46 is the gamma function , and @xmath47 is the angle between the local magnetic field vector and the outwardly pointing ray ( which is also very nearly the pitch angle since for relativistic particles most of the emission is projected into a very tiny cone about the particle s velocity ) . when the x - rays are produced predominantly by the cooled electrons in the broken power law ( equation 4 ) , as opposed to inverse compton scattering , which we also consider in this _
letter _ , their extraordinary and ordinary emissivities are given by expressions analogous to these , except with @xmath48 .
the overall specific intensity @xmath49 observed at infinity is an integration of the emissivity @xmath50 over the path length along geodesics , though most of the contribution arises in the geometrically thin disk , close to the marginally stable orbit .
however , a novel feature of our code is that we also allow the incipient radiation to scatter with the ambient plasma , either thermal or nonthermal , as the case may be , as it propagates along the geodesics towards the observer . throughout this process , we preserve ( or update as needed ) the photon energy , the overall intensity , and the polarization fraction and position angle .
as we shall see , this allows us to form polarimetric images at infinity of both the _ direct _ emission from the transiently energized inner region of the disk and the light _ scattered _ above or beyond this location .
it is straightforward to relate the specific intensity of each polarization separately in the emitter and detector frames from the relativistic invariant @xmath51 , and the degree of polarization is itself an invariant , though the position vector may rotate along geodesics .
likewise , the frequency @xmath52 at radius @xmath53 from the central object , is known precisely in terms of the frequency @xmath54 in the frame comoving with the emitter , and the lapse function , @xmath55 , which includes both the effects of gravitational time dilation ( between the emitter s radius @xmath56 and the pertinent radius @xmath53 ) and the doppler shift associated with motion of the emitting particles at @xmath56 : @xmath57 where @xmath58 in terms of the emitting particle s velocity @xmath59 , and @xmath47 is the aforementioned angle between the local magnetic field vector and the outwardly pointing ray , since * b * and * v * are parallel in the assumed keplerian geometry . at radius @xmath53 , we have @xmath60 , and the intensity then follows from the relativistic invariant . for the image of the direct emission viewed at infinity
, we simply set @xmath61 , while for the scattered light , @xmath53 becomes the radius @xmath62 at which the scattering has occurred , and an additional scattered " lapse function @xmath63 is introduced to incorporate the effects of gravitational time dilation between @xmath62 and infinity .
( note that we are not including an additional doppler shift factor into the expression for @xmath64 because the frequency shift associated with the scattering itself is already included in the evaluation of the comptonized photon s energy . )
the position angle of polarized light may be calculated from a relativistic invariant related to the parallel transport of a polarization vector along the null rays ( see , e.g. , connors & stark 1977 ) .
we perform the parallel transport operation quite directly by defining a reference vector at the detector and numerically propagating it along with the null ray itself .
this allows us to consistently map a position angle from one frame to the next , an essential feature for calculating radiative transfer for the ordinary and extraordinary waves as they propagate around the black hole .
scattering introduces a significant complication to this process because it not only redirects the photon s trajectory , but also creates its own polarization .
we follow the method described in connors et al .
( 1980 ) , based on the use of normalized stokes parameters @xmath65 where @xmath66 is the intensity and @xmath67 , @xmath68 are the stokes parameters determining the linear polarization . the stokes parameters @xmath69 and @xmath70 of the scattered beam can be determined from @xmath71 and @xmath72 , once the scattered photon s new direction is known and a new set of reference axes have been defined from the wavevectors before and after scattering .
we refer the reader to this well - written paper for further details .
the observer is located at infinity with viewing angle @xmath73 relative to the @xmath74-axis in the non - rotating frame ( see figure 1 ) .
the deflection angle of a photon emitted by plasma in the transiently energized inner region of the disk is @xmath75 , where @xmath76 , @xmath77 being the azimuthal angle of the emission point relative to the reference @xmath78-axis .
these emitted photons are deflected by the black hole and intersect the observer s detector plane at infinity .
( note that this same geometry is valid for the deflection of scattered photons as well . )
the distance between the line - of - sight and the point at which the photon reaches the detector is defined as the impact parameter @xmath79 . using this geometry , the deflection angle of the photon s trajectory
may be obtained with the light - bending relation between @xmath80 ( the angle between the emission direction of the photon and the direction from the center of the black hole to the location of the emitter ) and @xmath75 , from the geodesic equation ( see beloborodov 2002 ) @xmath81 this procedure yields the impact parameter @xmath82 of the photons in terms of the emitting radius @xmath56 .
a detailed description of this geometry is provided in , e.g. , luminet ( 1979 ) , and a more complete accounting of pollux , together with several other sample applications , will appear in falanga et al .
given the above considerations
some from prior fitting to the data ( e.g. , dodds - eden et al .
2009 , trap et al .
2011 ) , others from earlier theoretical work ( e.g. , melia 2007)we have selected the following physical environment to model with pollux , the results of which are summarized in the next two subsections .
we assume a radiating ring ( energized by an as yet unidentified instability ) , @xmath83 wide and inner radius at @xmath84 .
the non - thermal particle distribution is described by the synchrotron - cooled power law in equation ( 4 ) , with an ambient azimuthal magnetic field @xmath85 .
the break frequency ( equation 3 ) is then @xmath86 hz . aside from the clear observational motivation for this geometry , there are also compelling theoretical reasons for expecting this type of disk instability , as discussed more extensively in tagger & melia ( 2006 ) and falanga et al .
further , to examine how the polarimetric image of this instability compares with one in which scattering is important , we also assume ( and report in subsection 4.2 below the impact of ) a straightforward scattering halo , with uniform electron density @xmath87 @xmath88 and radius @xmath89 , centered on the black hole . again , for specificity , we take these particles to be thermal , with temperature @xmath90 k. figure 2 gives a sample of the results from this simulation . in all of these images ,
the vertical ( or @xmath91 ) coordinate is parallel to the disk s symmetry axis .
the top panel shows the intensity @xmath92 of light polarized in the horizontal ( or @xmath78 ) direction on a linear color scale .
the corresponding intensity @xmath93 of light polarized in the @xmath91 direction appears in the middle panel .
the bottom panel is an image ( using the same spatial scale as the first two ) of the polarization fraction , defined as @xmath94 where @xmath95 and @xmath96 are the fluxes of polarized light calculated in the @xmath78 and @xmath91 directions , respectively , by integrating @xmath92 and @xmath93 over their respective solid angles , with all the usual gravitational effects e.g . ,
area amplification taken into account .
the interpretation of these results is rather straightforward ( see also bromley et al .
2001 ) . as we discuss in 3 above , the dominant magnetic field component in this region
is expected to be azimuthal .
the extraordinary emissivity dominates over the ordinary , so most of the @xmath91-polarized light comes from the front and rear of the disk , whereas most of the @xmath78-polarized light comes from the sides .
in addition , the disk is assumed to rotate counter - clockwise in this geometry , so the left is blue - shifted , whereas the right is red - shifted .
all the general relativistic and emissivity effects together conspire to produce a shift to the left of @xmath92 relative to @xmath93 .
however , the polarized fraction is not affected by redshift , so the bottom image is symmetric about the vertical axis , and more clearly demonstrates the spatial dependence of the extraordinary versus ordinary emissivities . in principle
, the polarization fraction varies over a broad range in values , from @xmath97 ( the negative sign meaning that @xmath93 is dominant over @xmath92 ) to @xmath98 ( when @xmath92 dominates over @xmath93 ) . what we see
, however , is an integration of the intensity over the whole image , yielding a single polarized flux , so the net polarization fraction is never quite this high ( see figure 4 below ) .
the polarimetric images at other energies , say @xmath99 kev , are similar to these , and will be reported as part of a more complete survey of results in falanga et al .
but we will continue our discussion of these characteristics in conjunction with figure 4 below . before doing so , however , let us examine the impact on our results of a scattering halo .
let us assume now that , in addition to the synchrotron emitting ring described above , we also have a scattering halo surrounding the black hole and inner disk region .
we are not necessarily espousing the view that this is a viable model for the nir / x - ray flare characteristics observed in recent years . if comptonization is important in producing the x - rays , the incipient emission need not be that due to a synchrotron - cooled particle distribution .
our main goal here is to demonstrate how the images and polarization fraction ( and possibly the position angle ) are altered by scattering from those corresponding to the _ direct _ images .
scattering clearly broadens the image , by a degree dependent on the density @xmath100 of scattering particles . in this simulation , we have chosen a value @xmath87 @xmath88 , corresponding to a photon mean free path roughly equal to the size of the halo .
thus , virtually all of the photons emitted by the inner ring scatter at least once on their way out .
but scattering does more than this .
it is well known that radiation passing through a scattering medium acquires a linear polarization fraction dependent on the inclination angle relative to the surface of the scattering medium ( see , e.g. , conors et al . 1980 ) .
thus , light originally polarized at the point of emission , may become more or less polarized , depending on its trajectory through to the observer .
we see the tell - tale signature of this effect in the lower panel of figure 3 .
notice , e.g. , that portions of the image polarized in the vertical direction ( in blue ) now extend much farther above and below the disk . and
we shall see in figure 4 below that an important consequence of scattering is a general dilution of otherwise more strongly polarized x - ray emission .
some of the most important results of our simulations are summarized in figure 4 . with these
, we can now begin to answer some of the questions posed in the introduction .
first of all , a broken power - law distribution of particles emitting synchrotron radiation from an energized ring at the marginally stable orbit produces a polarization fraction at infinity of up to @xmath101 ( even more for inclination angles @xmath102 ) when all the relevant general relativistic effects are taken into account .
polarimetric measurements in the nir reveal polarization fractions as high as @xmath103 .
the geometry we have adopted here can therefore account for these observations without the need to induce linear polarization using more complicated procedures .
detailed fits to the observations led to the idea that the x - rays may be due primarily to the same population of electrons producing the nir emission , though synchrotron - cooled at higher energies .
the results of our simulations show that this remains a viable model even when all of the relativistic effects are taken into account .
null geodesics ( and parallel transport ) are not energy dependent , so the polarization properties of the x - ray component echo those of the nir .
therefore , a prediction of this model is that both the polarization fraction and position angle @xmath104 of the x - rays should be similar to those of the infrared .
panel ( c ) in figure 4 shows that @xmath104 , defined by the expression @xmath105 lies somewhere between @xmath106 and @xmath107 for most inclination angles , except above @xmath108 where it can reach @xmath109 or more .
these values of position angle are consistent with the fact that @xmath96 dominates over @xmath95 , as one would expect when the extraordinary component is much greater than the ordinary .
the increase in @xmath104 with @xmath73 is due to the increasing influence of redshift seen , e.g. , in figure 2 , which moves the core of the observed emission towards the left in these images .
notice , however , that a different picture emerges for the x - ray characteristics when scattering is important .
panel ( e ) in figure 4 shows that the polarization fraction is smaller ( typically smaller than @xmath110 ) when x - rays emerge from scattering . for many inclination angles
, @xmath111 is actually closer to @xmath112 .
in addition , the position angle differs from that in panel ( c ) , by anywhere from @xmath113 to as much as @xmath114 , depending on the inclination angle . of course , the capability to measure @xmath111 and @xmath104 for x - rays may be many years away , but that technology is being developed .
for example , several x - ray polarimetry missions have already been studied and proposed to open up this very promising field of research over the next few years .
the most advanced project today the gravity and extreme magnetism ( gem ) mission ( swank et al . 2010 ) , is planned for launch in 2014 , with the primary goal of studying the polarimetric properties of the soft x - ray emission from bright black holes and neutron stars .
unfortunately , the gem instruments will not be sensitive enough to study a relatively weak source , such as sgr a*. indeed , for fluxes less than 1 millicrab ( the intensity reached by sgr a * at the peak of the brightest flares ) , gem would need a net observing time of more than 100 ks to detect a polarization fraction of 10% or lower .
but sgr a * s flares never last longer than 3 hr , so it would not be feasible to obtain such a deep exposure .
moreover gem , and even other future polarimetry projects , such as polarix ( costa et al . 2010 ) , could not provide angular resolutions better than @xmath115 , making it very difficult to identify the origin of a polarized signal from a crowded region , such as we have at the galactic center . nonetheless , this field is only in its infancy .
more promising missions , with better sensitivity and spatial resolution , will no doubt follow in the years to come . and
our results demonstrate the power of future simultaneous polarimetric observations in both the nir and x - ray portions of the spectrum .
this research was supported by nasa grant nnx08ax34 g at the university of arizona .
partial support was also provided by onr grant n00014 - 09-c-0032 . in addition , fm is grateful to amherst college for its support through a john woodruff simpson lectureship .
we also acknowledge the international space science institute ( issi ) in bern , where a large portion of this work was carried out . | the galaxy s supermassive black hole , sgr a * , produces an outburst of infrared radiation about once every 6 hours , sometimes accompanied by an even more energetic flurry of x - rays .
it is rather clear now that the nir photons are produced by nonthermal synchrotron processes , but we still do nt completely understand where or why these flares originate , nor exactly how the x - rays are emitted .
circumstantial evidence suggests that the power - law electrons radiating the infrared light may be partially cooled , allowing for the possibility that their distribution should be more accurately described by a broken power law with a ( cooling break " ) transition frequency .
in addition , the emission region ( energized by an as yet unidentified instability ) appears to be rather compact , possibly restricted to the inner edge of the accretion disk . in that case , the x - ray outburst may itself be due to synchrotron processes by the most energetic particles in this population . in this paper
, we examine several key features of this proposal , producing relativistically correct polarimetric images of sgr a * s nir and x - ray flare emission , in order to determine ( 1 ) whether the measured nir polarization fraction is consistent with this geometry , and ( 2 ) whether the predicted x - ray to nir peak fluxes are confirmed by the currently available multi - wavelength observations .
we also calculate the x - ray polarization fraction and position angle ( relative to that of the nir photons ) in anticipation of such measurements in the coming years .
we show that whereas the polarization fraction and position angle of the x - rays are similar to those of the nir component for synchrotron - cooled emission , these quantities are measurably different when the x - rays emerge from a scattering medium .
it is clear , therefore , that the development of x - ray polarimetry will represent a major new tool for studying the spacetime near supermassive black holes .
[ firstpage ] acceleration of particles ; black hole physics ; galaxy : centre ; gravitation ; magnetic fields ; polarization ; relativistic processes ; scattering |
europium chalcogenides are prospective candidates to serve as spin - filters in prototype spintronic devices @xcite and novel magneto - optical devices ( tunable lasers , detectors and modulators ) @xcite , where the principal source of the interesting physical properties is the _ d - f _ exchange interaction between band - edge electrons and localized eu@xmath4 spins .
the optical properties of europium chalcogenides have already been investigated some 20 - 30 years ago , and it was established that the band - edge absorption spectrum is associated with a @xmath7 electronic transition ( surveys are given in refs .
the experiments were performed on powders @xcite , polycrystalline thin layers @xcite and bulk crystals @xcite , whereby the @xmath0 absorption spectrum is characterized by a very broad band - full width at half maximum of almost @xmath81 ev .
this absorption band was analyzed by some researchers @xcite in terms of an excitonic transition extending mainly in the nearest neighbor eu sites , with an exciton binding energy dependence on external fields and temperature that is governed by the _
d - f _ exchange interaction between the electron and the eu@xmath4 spins in the lattice ( the magnetic exciton model ) .
the large breadth of the absorption line , however , has remained unexplained @xcite , and it has lead some researches to propose , instead , that the absorption spectrum is associated with a transition to extended conduction band states @xcite .
the photoluminescence of europium chalcogenides was also studied , and it is described by a broad line ( full width at half maximum fwhm=150 mev in eute ) with a large stokes shift ( photon energy @xmath9 ev in eute ) @xcite . recently , molecular beam epitaxy ( mbe ) was used to produce eute , which has lead to the observation of a much narrower photoluminescence line ( full width at half maximum fwhm=10 mev ) , and at a much photon higher energy ( @xmath10 ev ) @xcite than in previous investigations .
the photoluminescence line displays a giant red shift ( 34 mev / t ) when a magnetic field is applied .
this sensitivity was interpreted in the framework of the magnetic exciton model @xcite .
the theory relies on such basic parameters as the _ d - f _ exchange interaction energy and effective bohr radius , whose accepted values vary within a large interval . for example , the _ d - f _ exchange interaction energy , @xmath11 in eute , has been estimated to be ( a ) 0.35 ev , based on measurements on the temperature dependence of the optical absorption edge @xcite , ( b ) 0.11 ev from the red shift of the faraday rotation spectrum , and also of the optical band - edge , when a high magnetic field is applied @xcite , ( c ) 0.17 ev from theoretical calculations @xcite . in this work we studied the @xmath0 optical absorption of epitaxial layers of eute grown by mbe , using circularly polarized light and high magnetic fields . at zero magnetic field ,
the optical absorption spectrum is described by a sharp threshold around 2.25 ev .
when the intensity of the applied magnetic field is increased above 6 t , a very sharp symmetric absorption line ( full width at half maximum of 41 mev ) appears at the low energy side of the absorption onset in one of the photon polarizations ( @xmath3 ) . with increasing magnetic field
this absorption line displays a huge red shift , which saturates when the eu@xmath4 spins in the crystal arrange ferromagnetically .
the peak position dependence on the magnetic field fits very well into the model of a localized excitation whose energy is tunable through the _ d - f _ exchange interaction . from the absorption line shift with magnetic field we obtained an estimate of the _ d - f _ exchange interaction energy , @xmath12 ev , with an uncertainty less than 10% .
the optical absorption dependence on magnetical field was also studied for a pb@xmath1eu@xmath2te sample .
the optcial absorption of pb@xmath13eu@xmath14te for @xmath15 is eute - like , i.e. it is associated with an @xmath7 electronic transition , whose energy is pinned at 2.3 ev for @xmath16@xcite . in the pb@xmath1eu@xmath2te sample ,
the same absorption line seen in eute was detected in high magnetic fields ( above 6 t ) , however , the maximum red shift of the line was smaller , due to a smaller concentration on eu@xmath4 in the lattice .
in addition , the width of the absorption line ( 43 mev ) was larger than in eute ( 41 mev ) . assuming that the increase in the breadth of the line is due to alloy disorder
, the radius of localization of the excitation is estimated to be @xmath810 .
recent investigations of the photoluminescence spectrum of eute grown by mbe have discovered almost the same red shift of the optical line ( 34 mev / t ) when a magnetic field is applied , as in this work .
it should be pointed out , however , that the physical mechanisms involved in the photoluminescence could be much more complex than in the optical absorption . according to the franck - condon principle @xcite , whereas absorption occurs when the crystal lattice is undistorted , emission occurs when the atoms surrounding the excitation move to their new equilibrium positions , and in eute a large distortion occurs due to the a large charge transfer from to excited eu@xmath4 to the surrounding atoms @xcite ; in addition to the mechanical distortion , an alignment of the eu@xmath4 spins inside the excitation radius ( forming a magnetic exciton ) , would also contribute to reduce the energy of the excitation .
the amount by which energy is reduced could be dependent on the intensity of the magnetic field applied , due to a changing size of the excitation , making the magnetic field induced red shift seen in the pl much more complex than the red shift of the absorption spectrum reported in this work , in which the positions of the atoms in the lattice , and their spin orientations , are fixed at the unperturbed crystal state .
eute and pb@xmath1eu@xmath2te epitaxial layers ( @xmath17 m thick ) were grown by molecular beam epitaxy ( riber 32p mbe ) onto freshly cleaved ( 111 ) baf@xmath18 substrates . in the case of eute
, the constituents ( eu and te ) were sublimated from elemental sources , while a pbte source was used for the pb@xmath1eu@xmath2te growth .
pb@xmath13eu@xmath14te crystallizes in the rock salt structure , and its lattice mismatch to the baf@xmath18 substrate ( fluoride structure ) varies from 4 to 6% , with increasing eu content in the ternary compound . in addition , the thermal expansion coefficient of baf@xmath18 is well matched to those of pbte and eute , thus preventing the build up of high thermal strains when cooling the samples to cryogenic temperatures .
the eu content values in the pb@xmath1eu@xmath2te sample was determined by magnetization squid measurements , with an estimated uncertainty of less than one percent .
the crystalline quality was investigated by measuring the x - ray rocking curve of the ( 222 ) bragg reflection of the samples .
the peak position corresponded to the bulk lattice parameter of 6.600 , and the full width at half maximum of the eute diffraction peak was about 400 arcsec , which is indicative of a good structural quality ( for pb@xmath1eu@xmath2te the fwhm of the peak was larger than in eute by about 100 arcsec , due to alloy disorder ) .
pl measurements at t=2k were made , using the 488 nm line of an ar laser , and the sharp luminescence ( fwhm@xmath810 mev ) at 1.922 ev was detected , demonstrating the high optical quality of the samples @xcite .
the optical absorption spectra of eute samples and pb@xmath1eu@xmath2te were measured at 1.8k , using left and right circularly polarized light , in magnetic fields of intensity up to 17 teslas . the sample ( square shaped , of side length 2.5 mm ) , was immersed in superfluid helium , and light was conveyed to the sample , and collected from the sample , in the faraday configuration , using optical fibers coupled to _ in situ _ miniature focusing optics and circular polarizers .
the absorption spectrum taken at zero field shows the characteristic absorption onset at @xmath82.3 ev ( see fig.[fig : fig1 ] ) , which is associated with the @xmath0 transition @xcite . in the transparency region ,
the absorption coefficient shows weak oscillations , due to the effect of fabri - perot interference in the thin epitaxial film .
when a magnetic field is applied , the absorption edges taken in both polarizations show remarkably different behaviours . in the @xmath19 polarization
there is simply a red shift of the absorption edge , which saturates when the applied field , @xmath20 , reaches 8.3 t. in the @xmath3 polarization , however , the absorption spectrum does not red shift as a whole , but develops a new sharp absorption peak that emerges from the main absorption edge . with increasing magnetic field the new absorption peak gains intensity , and it red shifts , until saturation is reached at @xmath20=8.3 t. the same absorption spectra were measured for a number of eute samples of various thickness in the range of 1 - 3@xmath21 m , some of which contained a baf@xmath18 protective layer , some of which did not .
the position and width of the peak was the same in all samples , and the optical density of the sharp absorption peak was proportional to the thickness of the eute epitaxial layer , demonstrating that the absorption line is originated in the bulk of eute .
for our sample geometry ( a thin layer of eute subject to a perpendicular magnetic field ) , the internal field will be given by @xmath22 , where @xmath23 is the magnetization of the sample , and @xmath24 varies linearly with @xmath20 and saturates to @xmath25 t at @xmath26 t @xcite , meaning that @xmath20=8.3 t corresponds to an internal field equal to the well known critical field value of 7.2 t for eute .
we have employed fields as large as 17 t , but no further changes were seen in the absorption spectra .
the absorption spectra at @xmath20=9.6 t in both polarizations are shown in figure 1 . to show the evolution of the absorption spectra taken in the @xmath19 and @xmath3 polarizations when
the applied magnetic field is increased from zero , a contour plot of the absorption spectra is shown in figure 2 .
this figure shows that in the @xmath19 polarization the absorption edge red shifts almost linearly with the field intensity , until saturation is reached .
in contrast , the evolution of the absorption spectrum in the @xmath3 polarization does not show a linear red shift of the absorption edge , but rather the changes in the spectrum are described by the emergence of a sharp absorption line at the low energy side of the spectrum .
when @xmath27 is less than the saturation value , the strength of the sharp absorption line increases with @xmath27 , and it red shifts , leading to a completely resolved line at the highest fields .
the red shift of the sharp peak is described by a huge value of 35 mev / t , which is about the same value seen in the shift of excitonic pl @xcite .
notice also that the main absorption edge remains nearly in the same place as at @xmath28 .
the peak position and full width at half maximum of the sharp absorption peak seen in the @xmath3 polarization at fields above 8.3 t were obtained from a fit of the absorption spectrum with a linear combination of a gaussian curve and a straight line ( see fig.[fig : fig3 ] ) . for eute ,
the maximum occurs at 2.169 ev ( the peak is shifted by @xmath29 ev from the main @xmath0 absorption ) , and the full width at half maximum ( fwhm ) equals to 41 mev .
the pb@xmath1eu@xmath2te sample showed the same qualitative behavior as seen in eute , but the low energy side peak was broader than for eute , and centered at a higher energy , as shown in the right panel of fig.[fig : fig1 ] .
the absorption maximum is seen at 2.181 ev , i.e. an energy 12 mev greater than for eute , and the fwhm=43 mev , greater by 2 mev compared to the same value for eute ( see right panel in fig.[fig : fig3 ] ) .
the magnetic field dependence of the newly discovered absorption line will be discussed in the framework of the @xmath30 exchange interaction .
the @xmath30 exchange interaction energy operator is given by @xcite @xmath31 where @xmath32 is the @xmath30 exchange interaction constant , @xmath33 is the creation operator of a wannier function centered at the @xmath34th eu@xmath4 with spin @xmath21 , @xmath35 is a pauli spin operator , and @xmath36 the spin of the @xmath34th eu@xmath4 .
we treat @xmath37 as a perturbation over a photoexcited electron in a @xmath38band with definite spin , whose wave function is given by @xmath39 .
the spin of the photoexcited electron is subject to the constraint that it must be parallel to the eu@xmath4 spin at the site where the excitation occurred @xcite .
we assume that the wave function describing the excited state encloses a large number of eu@xmath4 sites , hence at zero magnetic field ( t=0k ) , when the arrangement of the eu ions is antiferromagnetic , ( the eu spins lie in the ( 111 ) plane @xcite ) , the @xmath30 interaction energy of the photoexcited electron averages out to zero .
as the magnetic field applied normal to the ( 111 ) increases , the eu@xmath4 spins ca nt in the magnetic field direction , the _ d - f _ exchange interaction energy increases and at arbitrary field it will be given by @xmath40 where @xmath41 is the angle between the sublattice spin directions , determined by @xcite @xmath42 where @xmath43 is the saturation field for eute @xcite .
fig.[fig : fig4 ] shows the measured position of the side peak as a function of applied magnetic field both for eute and pb@xmath1eu@xmath2te . the same peak position dependence on magnetic field as for the eute sample shown in figure [ fig : fig4 ]
was obtained for a second eute sample , whose epitaxial layer thickness was 1.75@xmath21 m . the solid lines in fig.[fig : fig4 ]
were obtained by a simultaneous fit of all of the experimental values shown to the equation @xmath44 where @xmath45 is the position of the excitonic line at @xmath28 , @xmath46 is the molar fraction of eu in the pb@xmath13eu@xmath14te sample , and @xmath47 is given by eq.([eq : deltae ] ) . equation ( [ eq : deltaex ] ) should be valid in the range @xmath16 , where the pb@xmath13eu@xmath14te gap is eute - like and nearly constant @xcite . in the fitting procedure ,
the spin of the eu@xmath4 was fixed at @xmath48 , and the @xmath30 exchange interaction constant , @xmath32 , and @xmath45 were the fitting parameters , leading to the best fit values of @xmath45=2.321 ev and @xmath5 ev . for pb@xmath1eu@xmath2te , the energy position of the side peak is greater than in eute , which is due to a lower density of spins in the lattice .
then , assuming that in going from eute to pb@xmath1eu@xmath2te , the increase in the full width at half maximum ( fwhm ) of the side peak is due to inhomogeneous broadening , caused by random fluctuations in the local eu concentration , the number of eu@xmath4 spins inside the exciton radius can be estimated through @xmath49 where @xmath50 and @xmath51 are the shift in the position of the maximum of the excitonic line and the change in eu molar fraction on going from eute to pb@xmath13eu@xmath14te , respectively . using the results obtained for eute and pb@xmath1eu@xmath2te
, we find @xmath52 , which justifies the assumption made above that the exciton is large , and means that the exciton sphere encloses about 10 conventional unit cells of the fcc structure .
this estimate puts the exciton diameter at approximately 18 . in conclusion ,
we have studied the optical band - edge absorption in eute and pb@xmath1eu@xmath2te grown by mbe at low temperatures and high magnetic fields . under @xmath3 polarization
, we detect a well - resolved sharp peak ( full width 0.041 ev ) .
the sharp absorption line displays a huge red shift of about 35 mev / t when the magnetic field increases , and its separation from the main absorption band reaches @xmath8 0.2 ev at the saturation field . in pb@xmath1eu@xmath2te , the same absorption line was also detected , but it is broader , which is an indication of a localized excitation .
the displacement of the absorption line with magnetic field fits very well into a model of a localized excitation whose energy is tuned by the _
d - f _ exchange interaction . from a fit of the experimental points to the theory we obtain a direct estimate of the _ d - f _ exchange interaction energy , @[email protected] ev .
we are greatful to prof .
f. ikawa for help with the photoluminescence measurements , done at unicamp . l.k.h . acknowledges support from fapesp ( grant no .
02/00720 - 9 ) .
this work was supported by grants cnpq-306335/88 - 3 and fapesp-99/10359 - 7 .
99 d. p. divincenzo , j. app . phys . * 85 * , 4785 ( 1999 ) .
w. heiss , g. prechtl , and g. springholtz , appl .
. lett . * 78 * , 3484 ( 2001 ) ; t. kasuya and a. yanase , rev .
phys . * 40 * , 684 ( 1968 ) .
p. wachter , in _
handbook on the physics and chemistry of rare earths _ , edited by k. a. gschneidner and l. eyring ( north holland , amsterdam , 1979 ) , chap .
a. mauger and c. godart , phys
. rep . * 141 * , 51 ( 1986 ) .
j. schoenes , in _ ferromagnetic materials _ , edited by k. h. buschow and e. p. wohlfarth ( elsevier , new york , 1990 ) , vol . 5 , p.165 . g. busch , p. streit , and p. wachter , solid state commun . * 8 * , 1759 ( 1970 ) .
p. wachter , critical rev .
solid state sci .
* 3 * , 131 ( 1972 ) .
j. feinleib and c. r. pidgeon , phys .
* 23 * , 1391 ( 1969 ) .
r. akimoto , m. kobayashi , and t. suzuki , j. phys .
japan * 63 * , 4616 ( 1994 ) .
t. kasuya , j. magn .
magn . mater . *
195 * , 141 ( 1999 ) .
w. heiss , g. prechtl , and g. springholtz , phys .
b * 63 * , 165323 ( 2001 ) ; z. h. chen , w. heiss , i. souma , g. springholz , a. murayama , and y. oka , j. supercond . *
16 * , 403 ( 2003 ) .
m. umehara , phys .
b * 52 * , 8140 ( 1995 ) .
j. schoenes , j. magn .
magn . mater . * 11 * , 102 ( 1979 ) .
s. j. cho , phys .
b * 1 * , 4589 ( 1970 ) .
m. umehara , phys .
205208 ( 2002 ) . h. krenn , w. herbst , h. pascher , y. ueta , g. springholz , and g. bauer , phys .
b * 60 * , 8117 ( 1999 ) .
e. u. condon , phys .
rev . * 32 * , 858 ( 1928 ) .
n. f. oliveira , jr .
, s. foner , and y. shapira , phys .
b * 5 * , 2634 ( 1972 ) .
* fig . 1 * absorption spectra of eute for b=0 t ( dotted line ) and b=9.6 t ( full curves ) , in the latter case the polarization corresponding to each curve ( @xmath19 or @xmath3 ) is indicated . + + * fig . 2 * contour plot for the absorption spectrum dependence on the magnetic field intensity ( a ) @xmath3 ; ( b ) @xmath19 . + + * fig .
3 * dots represent the @xmath3 absorption spectrum of eute ( left ) and pb@xmath1eu@xmath2te ( right ) measured at b=9.6 t .
full lines are fitting curves to the measured spectra .
+ + * fig . 4 * magnetic field dependence of the energy position of the maximum in the @xmath0 excitonic absorption measured in @xmath3 polarization . | the optical absorption spectra in the region of the @xmath0 transition energies of epitaxial layers of of eute and pb@xmath1eu@xmath2te , grown by molecular beam epitaxy , were studied using circularly polarized light , in the faraday configuration . under @xmath3
polarization a sharp symmetric absorption line ( full width at half - maximum 0.041 ev ) emerges at the low energy side of the band - edge absorption , for magnetic fields intensities greater than 6 t. the absorption line shows a huge red shift ( 35 mev / t ) with increasing magnetic fields .
the peak position of the absorption line as a function of magnetic field is dominated by the _
d - f _ exchange interaction of the excited electron and the eu@xmath4 spins in the lattice .
the _ d - f _ exchange interaction energy was estimated to be @xmath5 ev . in pb@xmath1eu@xmath2te
the same absorption line is detected , but it is broader , due to alloy disorder , indicating that the excitation is localized within a finite radius . from a comparison of the absorption spectra in eute and
pb@xmath1eu@xmath2te the characteristic radius of the excitation is estimated to be @xmath6 . |
quantum mechanics , after more than 70 years , still poses fundamental problems of understanding .
many physicists believe these problems are ` only ' problems of ` physical interpretation ' of the mathematically well defined ` standard formalism ' . in this paper
we will show that this is not necessarily so .
we will show that the problem of quantum mechanics connected to the existence of nonproduct states in the situation of the description of the joint entity of two quantum entities may well indicate that a change of the standard formalism is necessary .
if we mention the ` standard formalism ' of quantum mechanics we mean the formalism as exposed , for example , in the book of john von neumann ( 1932 ) , and we will refer to it by sqm .
often the name ` pure state ' is used to indicate a state represented by a ray of the hilbert space .
we will however use it in the physical sense : a ` pure state ' of an entity represents the reality of this entity . as a consequence it is natural to demand that an entity ` exists ' if and only if at any moment it is in one and only one ` pure state ' .
let us formulate this as a principle , since it will play a major role in our analysis .
[ physprinc01 ] if we consider an entity , then this entity ` exists ' at a certain moment iff it ` is ' in one and only one pure state at that moment .
we denote pure states of an entity @xmath0 by means of symbols @xmath1 and the set of all pure states by @xmath2 .
we mention that in ( aerts , 1984a , 1999a ) , where aspects of the problem that we investigate in the present paper are analyzed , a ` pure state ' is called a ` state ' . a state represented by a ray of the hilbert space
will be called a ` ray state ' .
we denote rays by symbols @xmath3 , where @xmath4 and we denote @xmath5 the ` ray state ' represented by the ray @xmath6 . with each ray @xmath7 , @xmath8 ,
corresponds one and only one ray state @xmath5 .
one of the principles of standard quantum mechanics is that ` pure states ' are ` ray states ' .
[ qmprinc01 ] consider an entity @xmath0 described by sqm in a hilbert space @xmath9 .
each ray state @xmath5 , @xmath8 is a pure state of @xmath0 , and each pure state of @xmath0 is of this form .
the problem that we want to consider in this paper appears in the sqm description by of the joint entity @xmath0 of two quantum entities @xmath10 and @xmath11 .
[ qmprinc02 ] if we consider two quantum entities @xmath10 and @xmath11 described by sqm in hilbert spaces @xmath12 and @xmath13 , then the joint quantum entity @xmath0 consisting of these two quantum entities is described by sqm in the tensor product hilbert space @xmath14 .
the subentities @xmath10 and @xmath11 are in ray states @xmath15 and @xmath16 , with @xmath17 and @xmath18 , iff the joint entity @xmath0 is in a ray state @xmath19 . in relation to the situation of a joint entity consisting of two subentities
there is another physical principle we generally imagine to be satisfied .
[ physprinc02 ] if an entity is the joint entity of two subentities then the entity exists at a certain moment iff the subentities exist at that moment .
let us introduce the concept of ` nonproduct vectors ' of the tensor product .
for @xmath20 we say that @xmath21 is a nonproduct vector iff @xmath22 .
we are now ready to formulate the theorem that points out the paradox we want to bring forward . physical principle [ physprinc01 ] , physical principle [ physprinc02 ] , sqm principle [ qmprinc01 ] and sqm principle [ qmprinc02 ] can not be satisfied together .
[ theor01 ] proof : suppose the four principles are satisfied .
this leads to a contradiction . consider the joint entity @xmath0 of two subentities @xmath10 and @xmath11 , described by sqm . from sqm principle [ qmprinc02 ]
follows that if @xmath10 and @xmath11 are described in hilbert spaces @xmath12 and @xmath23 then @xmath0 is described in the hilbert space @xmath24 .
let us consider a nonproduct vector @xmath25 and the ray state @xmath26 .
from sqm principle [ qmprinc01 ] follows that @xmath26 represents a pure state of @xmath0 .
consider a moment where @xmath0 is in state @xmath26 .
physical principle [ physprinc01 ] implies that @xmath0 exists at that moment and from physical principle [ physprinc02 ] we infer that @xmath10 and @xmath11 also exist at that moment .
physical principle [ physprinc01 ] then implies that @xmath10 and @xmath11 are respectively in pure states @xmath27 and @xmath28 . from sqm principle [ qmprinc01 ]
it follows that there are two rays @xmath29 and @xmath30 , @xmath31 and @xmath32 such that @xmath33 and @xmath34 . from sqm principle [ qmprinc02 ] then follows that @xmath0 is in the state @xmath35 which is not @xmath26 since the ray generated by @xmath36 is different from @xmath37 .
since both @xmath26 and @xmath35 are pure states , this contradicts physical principle [ physprinc01 ] .
the fundamental problems of the sqm description of the joint entity of two subentities had already been remarked a long time ago . the einstein podolsky rosen paradox and later research on the bell inequalities are related to this difficulty ( einstein et al . , 1935 ;
bell , 1964 ) .
it are indeed states like @xmath26 , with @xmath21 a nonproduct vector , that give rise to the violation of the bell inequalities and that generate the typical epr correlations between the subentities .
most of the attention in this earlier analysis went to the ` non local ' character of these epr correlations .
the states of type @xmath26 are now generally called ` entangled ' states .
the problem ( paradox ) related to entangled states that we have outlined in section [ sec01 ] has often been overlooked , although noticed and partly mentioned in some texts ( e.g. van fraassen , 1991 , section 7.3 and references therein ) .
the problem of the description of a joint entity has also been studied within axiomatic approaches to sqm .
there it was shown that some of the axioms that are needed for sqm are not satisfied for certain well defined situations of a joint entity consisting of two subentities ( aerts , 1982 , 1984a ; pulmannov , 1983 , 1985 ; randall and foulis , 1981 ) .
more specifically , it has been shown in aerts ( 1982 ) that the joint entity of two separated entities can not be described by sqm because of two axioms : weak modularity and the covering law .
this shortcoming of sqm is proven to be at the origin of the epr paradox ( aerts , 1984b , 1985a , b ) .
it has also been shown that different formulations of the product within the mathematical categories employed in the axiomatic structures do not coincide with the tensor product of hilbert spaces ( aerts , 1984a ; pulmannov , 1983 , 1985 ; randall and foulis , 1981 ) . again
certain axioms , orthocomplementation , covering law and atomicity , are causing problems .
all these findings indicate that we are confronted with a deep problem that has several complicated and subtle aspects .
a very extreme attitude was to consider entangled states as artifacts of the mathematical formalism and hence not really existing in nature .
meanwhile , entangled states are readily prepared in the laboratory and the corresponding epr correlations have been detected in a convincing way .
this means that it is very plausible to acknowledge the existence of entangled states as ` pure states ' of the joint entity in the sense of physical principle [ physprinc01 ] . as a result of earlier research
we have always been inclined to believe that we should drop physical principle [ physprinc02 ] to resolve the paradox ( aerts , 1984a , 1985a , b ; see also aerts , broekaert and smets , 1999 ) .
this comes to considering the subentities @xmath10 and @xmath11 of the joint entity @xmath0 as ` not existing ' if the joint entity is in an entangled state .
we still believe that this is a possible solution to the paradox and refer for example to valckenborgh ( 1999 ) and aerts and valckenborgh ( 1999 ) for further structural elaboration in this direction . in the present paper we would like to bring forward an alternative solution . to make it explicit
we introduce the concept of ` density state ' , which is a state represented by a density operator of the hilbert space .
we denote density operators by symbols @xmath38 and the corresponding density states by @xmath39 . with each density operator
@xmath40 on @xmath9 corresponds one and only one density state @xmath41 . within sqm density states
are supposed to represent mixtures , i.e. situations of lack of knowledge about the pure state .
the way out of the paradox we propose in the present paper consists of considering the density states as pure states .
hence , in this sense , sqm principle 1 would be false and substituted by a new principle .
[ cqmprinc ] consider an entity @xmath0 described in a hilbert space @xmath9 .
each density state @xmath41 , where @xmath40 is a density operator of @xmath9 , is a pure state of @xmath0 , and each pure state of @xmath0 is of this form
. we call the quantum mechanics that retains all the old principles except sqm principle [ qmprinc01 ] , and that follows our new principle cqm principle [ cqmprinc ] , ` completed quantum mechanics ' and refer to it by cqm .
the first argument for our proposal of this solution comes from earlier work in relation with the violation of bell inequalities by means of macroscopic entities ( aerts , 1991a ) .
there we introduce a macroscopic material entity that entails epr correlations .
let us briefly describe this entity again to state our point .
first we represent the spin of a spin @xmath42 quantum entity by means of the elastic sphere model that we have introduced on several occasions ( aerts , 1986 , 1987 , 1991a , b , 1993 , 1995 , 1999a , b ) , and that we have called the ` quantum machine ' .
it is well known that the states , ray states as well as density states , of the spin of a spin @xmath43 entity can be represented by the points of a sphere @xmath44 in three dimensional euclidean space with radius @xmath45 and center @xmath46 .
let us denote the state corresponding to the point @xmath47 by @xmath48 . to make the representation explicit we remark that each vector @xmath49 can uniquely be written as a convex linear combination of two vectors @xmath50 and @xmath51 on the surface of the sphere ( fig .
2 ) i.e. @xmath52 with @xmath53 $ ] and @xmath54 . in this way
we make correspond with @xmath55 the density operator @xmath56 : @xmath57 each density operator can be written in this form and hence the inverse correspondence is also made explicit .
we remark that the ray states , namely the density operators that are projections , correspond to the points on the surface of @xmath44 .
it is much less known that the experiments on the spin of a spin @xmath42 quantum entity can be represented within the same picture .
let us denote the direction in which the spin will be measured by the diametrically opposed vectors @xmath58 and @xmath59 of the surface of @xmath44 ( fig . 1 ) , and
let us consider @xmath58 as the @xmath21 direction of the standard spin representation ( this does not restrict the generality of our calculation ) . in this case , in sqm , the spin measurement along @xmath58 , which we denote @xmath60 ,
is represented by the self adjoint operator @xmath61 with @xmath62 being the spectral projections .
the sqm transition probabilities , @xmath63 , the probability for spin up outcome if the state is @xmath48 , and @xmath64 , the probability for spin down outcome if the state is @xmath48 , are then : @xmath65 let us now show that , using the sphere picture , we can propose a realizable mechanistic procedure that gives rise to the same probabilities and can therefore represent the spin measurement .
0.5 cm 2 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 7pt fig . 1 : a representation of the quantum machine . in ( a )
the particle is in state @xmath66 , and the elastic corresponding to the experiment @xmath67 is installed between the two diametrically opposed points @xmath68 and @xmath69 . in ( b )
the particle falls orthogonally onto the elastic and sticks to it . in ( c )
the elastic breaks and the particle is pulled towards the point @xmath68 , such that ( d ) it arrives at the point @xmath68 , and the experiment @xmath67 gets the outcome @xmath70 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ our mechanistic procedure starts by installing an elastic strip ( e.g. a rubber band ) of 2 units of length , such that it is fixed with one of its end - points in @xmath58 and the other end - point in @xmath59 ( fig .
once the elastic is installed , the particle falls from its original place @xmath55 orthogonally onto the elastic , and sticks to it ( fig .
then , the elastic breaks in some arbitrary point . consequently the particle , attached to one of the two pieces of the elastic ( fig . 1,c ) , is pulled to one of the two end - points @xmath58 or @xmath59 ( fig .
now , depending on whether the particle arrives in @xmath58 ( as in fig .
1 ) or in @xmath59 , we give the outcome _ ` up ' _ or _ ` down ' _ to this experiment @xmath60 .
let us prove that the transition probabilities are the same as those calculated by sqm .
the probability @xmath63 , that the particle ends up in point @xmath58 ( experiment @xmath60 gives outcome _ ` up ' _ ) is given by the length of the piece of elastic @xmath71 divided by the total length of the elastic .
the probability , @xmath64 , that the particle ends up in point @xmath59 ( experiment @xmath60 gives outcome _ ` down ' _ ) is given by the length of the piece of elastic @xmath72 divided by the total length of the elastic .
this means that we have : @xmath73 comparing ( [ probquant ] ) and ( [ probform ] ) we see that our mechanistic procedure represents the quantum mechanical measurement of the spin .
0.5 cm 3.5 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ = 7pt fig . 2 : a representation of the experimental process .
the elastic of length 2 , corresponding to the experiment @xmath74 , is installed between @xmath68 and @xmath69 .
the probability , @xmath75 , that the particle ends in point @xmath68 under influence of the experiment @xmath74 is given by the length of the piece of elastic @xmath76 divided by the total length of the elastic .
the probability , @xmath77 , that the particle ends in point @xmath69 is given by the length of the piece of elastic @xmath78 divided by the total length of the elastic .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ to realize the macroscopic model with epr correlations we consider two such ` quantum machine ' spin models , where the point particles are connected by a rigid rod , which introduces the correlation .
we will only describe the situation where we realize a state that is equivalent to the singlet spin state @xmath79 , where @xmath80 , and refer to aerts ( 1991a ) for a more detailed analysis .
suppose that the particles are in states @xmath81 and @xmath81 where @xmath82 and @xmath83 are respectively the centers of the spheres @xmath84 and @xmath85 ( fig .
3 ) connected by a rigid rod .
we call this state ( the presence of the rod included ) @xmath48 .
the experiment @xmath86 consists of performing @xmath87 in @xmath84 and @xmath88 in @xmath85 and collecting the outcomes _ ( up , up ) , ( up , down ) , ( down , up ) _ or _ ( down , down)_. in fig .
3 we show the different phases of the experiment .
we make the hypothesis that one of the elastics breaks first and pulls one of the particles _ up _ or _
down_. then we also make the hypothesis that once that one of the particles has reached one of the outcomes , the rigid connection breaks down .
the experiment continues then without connection in the sphere where the elastic is not yet broken .
the joint probabilities can now easily be calculated : @xmath89 where @xmath90 is the angle between @xmath91 and @xmath92 .
these are exactly the quantum probabilities when @xmath48 represents the singlet spin state . as a consequence
our model is a representation of the singlet spin state .
this means that we can put @xmath93 . 0.5 cm 0.5 cm _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 0.7 cm = 7pt fig . 3 : a macroscopic mechanical entity with epr correlations .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ why does this example inspire us to put forward the hypothesis that density states are pure states ?
well , if we consider the singlet spin state , then this is obviously a nonproduct state , and hence the states of the subentities are density states .
in fact they are the density states @xmath94 and @xmath95 where @xmath96 however , the state of the joint entity is clearly not given by the density state corresponding to the density operator @xmath97 because this state does not entail correlations .
it is due the presence of the epr correlations that the state of the joint entity is represented by a ray state . in our macroscopic mechanistic model
however all the states ( also the states of the subentities ) are ` pure states ' and not mixtures ( remark that we use the concept ` pure state ' as defined in section [ sec01 ] ) . if our proposal were true , namely if density states as well as ray states in principle represented pure states , we could also understand why , although the state of the joint entity uniquely determines the states of the subentities , and hence physical principle [ physprinc02 ] is satisfied , the inverse is not true : the states of the subentities do not determine the state of the joint entity . indeed , a state of one subentity can not contain the information about the presence of an eventual correlation between the subentities . this way , it is natural that different types of correlations can give rise to different states of the joint entity , the subentities being in the same states .
this possibility expresses the philosophical principle that the whole is greater than the sum of its parts and as our model shows it is also true in the macroscopic world .
let us now say some words about the generality of the construction that inspired us for the proposed solution . it has been shown in coecke ( 1995a ) that a quantum machine like model can be realized for higher dimensional quantum entities .
coecke ( 1995b , 1996 ) also showed that all the states of the tensor product can be realized by introducing correlations on the different component states .
this means that we can recover all the nonproduct ray states of the tensor product hilbert space by identifying them with a product state plus a specific correlation for a general quantum entity and hence that our solution of the paradox is a possible solution for a general quantum entity .
if we carefully analyze the calculations that show the equivalence of our model to the quantum model , we can understand why the distinction between ` interpreting density states as mixtures ' and ` interpreting density states as pure states ' can not be made experimentally .
indeed , because of the linearity of the trace , used to calculate the quantum transition probabilities , and because the inner points of the sphere can be written as convex linear combinations of the surface points , an ontological situation of mixtures must give the same experimental results as an ontological situation of pure states . if we would be able to realize experimentally a nonlinear evolution of one of the subentities that has been brought into an entangled state with the other subentity as subentity of a joint entity , it would be possible to test our hypothesis and to detect experimentally whether density states are pure states or mixtures . indeed
, suppose that we a nonlinear evolution of one of the entangled subentities could be realized .
then , we can distinguish the two possibilities in the following way .
if the density state @xmath94 of the entangled subentity is a mixture , then this state evolves while staying a convex linear combination of the ray states @xmath98 and @xmath99 ( referring to the situation of fig . 3 ) .
the nonlinear evolution makes evolve the ray states @xmath98 and @xmath99 and this determines the evolution of the density state @xmath94 , but the correspondence between @xmath94 and @xmath98 and @xmath99 remains linear .
if the density state @xmath94 of the entangled subentity is a pure state , then the nonlinear evolution will make it evolve independent of the way in which the ray states @xmath98 and @xmath99 evolve .
this means that in general the relation between @xmath94 and @xmath98 and @xmath99 will not remain that of a convex linear combination .
so we can conclude that for a nonlinear evolution the change of the density state of an entangled subentity under this evolution will be different depending on whether it is a mixture or a pure state .
this difference can experimentally be detected by a proper experimental setup .
we believe that such an experiment would be of great importance for the problem that we have outlined here .
the quantum axiomatic approaches make use of piron s representation theorem where the set of pure states is represented as the set of rays of a generalized hilbert space ( piron , 1964 , 1976 ) . this theorem has meanwhile been elaborated and the recent result of solr has made it possible to formulate an axiomatics that characterizes sqm for real , complex or quaternionic hilbert spaces ( sler 1995 , aerts and van steirteghem 1999 ) .
this standard axiomatic approach aims to represent pure states by rays of the hilbert space .
if our proposal is true , an axiomatic system should be constructed that aims at representing pure states by means of density operators of the hilbert space . within the generalization of the geneva - brussels approach that we have formulated recently , and where the mathematical category is that of state property systems and their morphisms , such an axiomatic can be developed ( aerts , 1999a ; aerts et al .
1999 ; van steirteghem , 1999 ; van der voorde 1999 ) . in aerts ( 1999c )
we made a small step in the direction of developing such an axiomatic system by introducing the concept of ` atomic pure states ' and treating them as earlier the pure states were treated , aiming to represent these atomic pure states by the rays of a hilbert space .
we proved that in this case the covering law remains a problematic axiom in relation to the description of the joint entity of two subentities ( theorem 18 of aerts 1999c ) .
we are convinced that we would gain more understanding in the joint entity problem if a new axiomatic would be worked out , aiming to represent pure states by density operators of a hilbert space , and we are planning to engage in such a project in the coming years .
diederik aerts is senior research associate of the belgian fund for scientific research .
this research was carried out within the project bil 96 - 03 of the flemish governement .
aerts , d. , ( 1982 ) .
description of many physical entities without the paradoxes encountered in quantum mechanics , _ found .
phys . _ , * 12 * , 1131 - 1170
. aerts , d. , ( 1985b ) .
the physical origin of the epr paradox and how to violate bell inequalities by macroscopic systems , in _ on the foundations of modern physics _ ,
p. lathi and p. mittelstaedt , world scientific , singapore , 305 - 320 .
aerts , d. , ( 1987 ) .
the origin of the nonclassical character of the quantum probability model , in _ information , complexity , and control in quantum physics _ , eds .
a. blanquiere , s. diner , and g. lochak , springer - verlag , wien - new york , 77 - 100 .
aerts , d. , ( 1991a ) . a mechanistic classical laboratory situation violating the bell inequalities with @xmath100 , exactly in the same way as its violations by the epr experiments , _ helv .
acta _ , * 64 * , 1 - 24 .
aerts , d. , ( 1991b ) . a macroscopic classical laboratory situation with only macroscopic classical entities giving rise to a quantum mechanical probability model , in _ quantum probability and related topics , volume vi _ , ed .
l. accardi , world scientific publishing , singapore , 75 - 85 .
aerts , d. , ( 1999b ) .
the stuff the world is made of : physics and reality , to appear in _
einstein meets magritte : an interdisciplinary reflection _ , eds .
d. aerts , j. broekaert and e. mathijs , kluwer academic , dordrecht , boston , london .
aerts , d. , coecke , b. , dhooghe , b. , durt , t. and valckenborgh , f. , ( 1996 ) . a model with varying fluctuations in the measurement context , in _ fundamental problems in quantum physics ii _ , eds .
ferrero , m. and van der merwe , a. , plenum , new york , 7 - 9 .
randall , c. and foulis , d. , ( 1981 ) . operational statistics and tensor products , in _ interpretations and foundations of quantum theory _ , h. neumann , ed .
wissenschaftsverslag , bibliographisches institut , mannheim , 21 . | we formulate a paradox in relation to the description of a joint entity consisting of two subentities by standard quantum mechanics .
we put forward a proposal for a possible solution , entailing the interpretation of ` density states ' as ` pure states ' .
we explain where the inspiration for this proposal comes from and how its validity can be tested experimentally .
we discuss the consequences on quantum axiomatics of the proposal . |
ions or molecules with degenerate electronic orbitals coupled to local vibrations can display a rich variety of phenomena.@xcite the most notable of these phenomena is the so - called static jahn - teller ( jt ) effect , i.e. , the system gains energy by lowering the symmetry through a static distortion , thereby removing the electronic degeneracy . for finite systems , like a molecule , quantum mechanics can play an important role in that , if the frequency of the relevant vibronic modes is high enough , it can dynamically restore the original symmetry through a quantum superposition of `` degenerate '' distorted configurations ( dynamical jt effect ) .
many systems containing jt ions or molecules are found to be insulators with quite a rich variety of spin and orbital ordering effects .
such a richness of exchange effects is due to the interplay between the spin and orbital degrees of freedom.@xcite the simplest cases can be modeled by the so - called @xmath1 jt - coupling , i.e. , at each site @xmath2 a two - fold degenerate electronic orbital is coupled to a two - fold degenerate local vibron.@xcite although conceived as the simplest workable model , such a situation is not totally unrealistic , as higher molecular degeneracies can often be reduced , by the crystal field of the solid state environment , to residual doublets .
( if the crystal field symmetry is very low , the degeneracy will be completely removed ) .
if exactly one electron sits on each jt site , and if the coulomb repulsion is very large , the system will have an insulating gap , due to mott - hubbard correlations , and the low - energy spectrum will consist of spin and orbital excitations . for the @xmath1 case , which we are going to consider , the physics of the insulating state is governed by an exchange hamiltonian involving , besides the actual spin @xmath3 , also a pseudo - spin-1/2 variable @xmath4 , representing the orbital degrees of freedom . unlike the case of a non - degenerate orbital , where antiferromagnetism is favored , one finds here a rich phase diagram , depending on the energy of the two - particle intermediate states available for virtual hopping.@xcite if hund s rule coupling is the dominant effect , the lowest doubly occupied site is a spin triplet , and spin ferromagnetism turns out to be favored.@xcite many experimental systems,@xcite including possibly the case of tdae - c@xmath5,@xcite may fit into such a scenario . in the present paper we are going to consider a different regime , i.e. , when the coupling to vibrons is strong enough that an inter - orbital singlet state becomes the lowest - energy doubly - occupied virtual state .
we are going to argue that new interesting phases can show up in this case .
the exchange hamiltonian for this case can be shown to be related to the following spin - model:@xcite @xmath6 we will show that , in this case , valence bond ( vb ) states tend to be favored , at least in low enough spatial dimensions .
the paper is organized as follows .
section [ intro_model : sec ] introduces the hopping model that we will consider . in section [ spin_ham : sec ] we will discuss in detail the hamiltonian @xmath7 in eq .
( [ hst_intro : eqn ] ) . in section [ tst : sec ] we study the behavior of the model upon introduction of holes , in one dimension . finally , section [ conclusion : sec ] contains our conclusions , and some open issues .
suppose that at each site @xmath2 of the lattice we have two degenerate electronic orbitals and two degenerate phonon modes , of frequency @xmath8 , both labeled by an orbital quantum number @xmath9 .
let @xmath10 be the creation operator for an electron with spin @xmath11 and orbital quantum number @xmath12 at site @xmath2 , and @xmath13 the phonon operator . in a standard @xmath1 jahn - teller problem , the on site
linear electron - phonon interaction term can be written as@xcite @xmath14 + ( g\omega_0/2 ) \ { t^+ \ , [ b^{\dagger}_{- } + b_{+ } ] + \rm{h.c.}\ } \;,\ ] ] where @xmath15 is the orbital ( pseudo ) spin operator , @xmath16 are pauli matrices and we have omitted the obvious site - label .
we will also consider electron - electron terms , which include a hubbard intra - orbital repulsion @xmath17 , an inter - orbital repulsion @xmath18 , and an exchange ( hund s ) coupling @xmath19 , @xmath20 where @xmath21 is the spin operator for the @xmath12orbital .
assume now that there are ( small ) electronic hopping processes between the different sites described by a standard tight - binding hamiltonian .
the total hamiltonian is therefore written as : @xmath22 \,+\ , \sum_{r } h_{\rm site}(r ) \;,\ ] ] where we have neglected overlap integrals between orbitals with different orbital quantum numbers , and @xmath23 we will be interested in the limit in which double occupancy of a site is strongly inhibited by the repulsive on - site interactions .
the low - energy physics is then described by an effective hamiltonian in which the double occupancy is only treated as a virtual process to second - order in the hopping matrix element @xmath24 . for a two - fold degenerate
orbital , there are six possible two - particle states to be considered as intermediate states : an inter - orbital singlet @xmath25 , two intra - orbital singlets @xmath26 , and an inter - orbital triplet @xmath27 . for the case of one electron per site , @xmath28 ,
the resulting effective hamiltonian involves two spin-1/2 operators at each site , the spin @xmath29 and the orbital pseudo - spin @xmath30 .
the derivation is standard,@xcite yielding @xmath31 \nonumber \\ & & \hspace{5 mm } \,+\ , \sum_{<rr ' > } [ 4(j_s+j_t ) ( { \bf s}_r \cdot { \bf s}_{r ' } ) ( { \bf t}_r \cdot { \bf t}_{r ' } ) + 8(j_{\pm}-j_s ) ( { \bf s}_r \cdot { \bf s}_{r ' } ) ( t^z_r t^z_{r ' } ) ] \;.\end{aligned}\ ] ] the couplings @xmath32 , @xmath33 , and @xmath34 , generalization of the famous antiferromagnetic coupling @xmath35 of the non degenerate hubbard model , are essentially given by @xmath36 where @xmath37 is the energy difference between a doubly occupied site of type @xmath38 and two singly occupied sites , and @xmath39 is an effective hopping matrix element reduced by electron - phonon interactions ( ham s factor).@xcite in absence of electron - phonon coupling ( @xmath40 ) , @xmath41 and @xmath42 , @xmath43 , @xmath44.@xcite the triplet state is therefore the lowest energy virtual state ( under the realistic assumption @xmath45 ) , and @xmath33 the largest coupling in the problem .
spin ferromagnetism is the natural outcome of such a scenario.@xcite the jahn - teller coupling can change this picture considerably .
first of all , standard polaronic effects lead to a decrease of the energy of the singlet states , primarily the inter - orbital singlet @xmath46 , with respect to the triplet states @xmath47.@xcite second , the effective hopping matrix element @xmath39 turns out to be larger for the singlet @xmath46 than for the other states .
@xcite gives the results for the @xmath48 s obtained from perturbation theory in @xmath49 , accurate to order @xmath50 : @xmath51 in the opposite strong electron - phonon coupling case , @xmath52 , it was shown in ref .
@xcite that , again , the most important doubly - occupied state is the inter - orbital singlet @xmath46 , and the model can be mapped into a single - band hamiltonian with a spin-1 variable attached to each site.@xcite this also leads to a situation in which , effectively , @xmath32 is the only coupling left in the problem.@xcite the mean field phase diagram of the model ( [ general_h : eqn ] ) has been studied in ref .
@xcite , together with a few special points where exact solutions can be obtained in @xmath53 . as it turns out , there is a region of the phase diagram , not considered so far , where , as we will argue below , valence bonds ( vb ) are favored .
the physical regime in which this region is of relevance is precisely the jt limit in which @xmath32 is the largest coupling in the problem , i.e. , the two - particle inter - orbital singlet has lower energy than the corresponding triplet .
more in detail , let us assume that @xmath54 and define @xmath55 . in a bipartite lattice the effective hamiltonian , after performing the canonical transformation @xmath56 , can be written as : @xmath57 where @xmath58 \;,\ ] ] and @xmath59 \nonumber \\ & & \,+ 4j_{\pm } \sum_{<rr ' > } ( 2 { \bf s}_r \cdot { \bf s}_{r ' } -1/2 ) ( t^z_r t^z_{r ' } + 1/4 ) \;.\end{aligned}\ ] ] in the next section we will concentrate on @xmath60 , showing first of all that it certainly has a vb - solid ground state in @xmath53 , and suggesting that the same might be true in @xmath61 . the presence of a gap in the excitation spectrum implies the robustness of such a phase with respect to a small enough perturbation @xmath62 .
next , we will consider the question of the behavior of a valence bond state upon doping .
the resulting generalization of the @xmath63 model will be discussed , for the @xmath53 case , in section [ tst : sec ] .
as previously discussed , when there is exactly one electron per site ( @xmath28 ) and one assumes the inter - orbital singlet @xmath46 to be the only important intermediate two - particle state , the exchange hamiltonian governing the spin and orbital degrees of freedom can be written as @xmath64 where @xmath65 is a permutator between the spins at sites @xmath2 and @xmath66 , and similarly @xmath67 for the pseudo - spins .
this form will be particularly useful in the following .
this model shows an obvious spin and pseudospin rotation invariance described by the @xmath68 group .
however , the full symmetry group displayed by ( [ st_2:eqn ] ) is much larger and includes @xmath69 .
indeed , one can show that , in a bipartite lattice , @xmath7 commutes with the 15 operators @xmath70 , @xmath71 , and @xmath72 which generate the lie algebra @xmath73 @xcite . due to the staggering in the @xmath74 operators , these generators _
do not _ commute with the spatial symmetries which interchange the two sublattices .
therefore , the full symmetry group is even larger than the aforementioned @xmath69 .
finally , we notice that this hamiltonian differs from the exactly soluble @xmath69 sutherland model @xmath75 obtained by a natural generalization of the heisenberg hamiltonian .
rather , it is possible to show that the @xmath76-invariant model introduced by affleck @xcite @xmath77 is unitarily equivalent to @xmath78 for @xmath79 . here
@xmath80 and @xmath81 generate , respectively , the fundamental and anti - fundamental representations of @xmath76.@xcite in @xmath53 , affleck has shown that @xmath82 , in the limit @xmath83 , has two simple vb - solid ground states , illustrated in fig . [
vbgs : fig ] .
moreover , the @xmath84 case has also been shown to have a dimerized ground state , although with a large spin - spin correlation length.@xcite as the tendency to dimerization is likely to increase with @xmath0,@xcite we expect that the @xmath79 case , of interest to us , will also have a vb - solid type of ground state .
this expectation will be explicitly shown to be correct in @xmath53 .
let us first show that singlet bonds between @xmath85 and @xmath86 variables are very natural objects to introduce in our problem , for any @xmath87 .
if we denote by @xmath88 a singlet bond formed between the spins at @xmath2 and @xmath66 , it is easy to show that @xcite @xmath89 similar results apply to @xmath90 when acting on valence bonds formed from the pseudo - spin variables , @xmath91 . if we denote by @xmath92 the product of singlet bonds between @xmath2 and @xmath66 for both @xmath85 and @xmath86 , and by @xmath93 [ p^{(t)}_{rr'}-1]$ ] , we find that @xmath94 and @xmath95 first observation : the subspace of valence bond states in which bonds are simultaneously s- and t - singlets is left invariant by the hamiltonian .
( in all figures , we will denote by a line such simultaneous s- and t - singlets . )
such a subspace is , however , still overcomplete.@xcite in @xmath53 , nevertheless , if we apply @xmath7 any number of times to the valence bond solid in fig . [ vbgs : fig ] , we generate a subset of valence bond configurations which are linearly independent ( although not orthogonal ) .
these are the valence bond configurations associated to non - crossing lewis diagrams.@xcite it is known that such a basis can be conveniently used to work out efficient exact diagonalization algorithms in small chains.@xcite the ground state wave function for a chain of @xmath96 sites , for instance , can be easily obtained by solving a simple @xmath97 problem .
[ 8siti : fig ] shows the result obtained for the ground state wave function , and the lowest excited state of momentum @xmath98 , in terms of vb configurations .
second observation : as is apparent from eq .
[ qaction : eqn ] , a nearest - neighbor bond contributes a diagonal energy of @xmath99 , whereas the off - diagonal matrix elements connecting different vb configurations are smaller by a factor @xmath100 , i.e. , @xmath101 . for the @xmath76-model of affleck , the off - diagonal matrix element are smaller by a factor @xmath102 , which implies that only diagonal energies are retained in the limit @xmath83 .
the ground states , for @xmath103 , are therefore the vb configurations with the maximum number ( @xmath104 ) of nearest - neighbor bonds connected by vbs . in @xmath53 , there are only two such states , i.e. , the vb - solids in fig .
[ vbgs : fig ] . to check
if the vb scenario remains correct for our @xmath79 case , we resort to exact lanczos diagonalizations of chains up to @xmath105 sites , and to green function monte carlo for longer chains . fig .
[ gap1d : fig ] shows the finite - size value of the gap between the ground state and the first excited state both in the @xmath106 ( @xmath107 ) and in the @xmath108 sectors ( @xmath109 ) .
the full symbols refer to @xmath7 , whereas the open symbols refer to two decoupled heisenberg chains ( no quartic term in eq . [ st_1:eqn ] ) .
all these excited states have momentum @xmath98 relative to the ground state . for the heisenberg case ,
the excited state is a triplet and the model has gapless excitations with gaps decreasing as @xmath110 for @xmath111.@xcite for @xmath7 , the situation is different : the lowest excited state is a singlet , with a finite size gap @xmath112 decreasing faster than @xmath110 for @xmath111 .
the triplet , instead , lies above with a gap @xmath113 extrapolating to a finite value as @xmath111 .
this is a clear signal of a _ degenerate _ infinite volume ground state with a _
spontaneous breaking of translational symmetry _ and a gap to all excitations , consistent with a vb phase
. further support to this interpretation can be gained by direct inspection of the spin - spin correlations , @xmath114 , the dimer - dimer correlations along a chain @xmath115 , and on different chains @xmath116 .
[ sq1d : fig ] shows a log - log plot of the size scaling of the peak of the different structure factors , located at momentum @xmath98 . while the dimer - dimer structure factor peaks , @xmath117 and @xmath118 , tend to diverge linearly with the length @xmath119 of the chain , the usual spin - spin structure factor tends to a finite value in the thermodynamic limit .
finally , simple variational arguments also point to a vb ground state in @xmath53 , while being , we believe , not conclusive in @xmath61 . a vb phase benefits
highly from the large values of @xmath120 on the singlet bonds , due to the presence of the `` square '' of the singlet contribution in the quartic term of @xmath7 , which enters with a factor @xmath100 .
let us compare , for instance , the energy of the following two states : ( i ) the product of two independent heisenberg ground states , @xmath121 , and ( ii ) the product of two vb states , @xmath122 where each vb state has the simple form of a product of dimers on adjacent sites . in @xmath53 we take @xmath123 where @xmath124 denotes a singlet between sites @xmath125 and @xmath126 .
in @xmath61 we consider the simplest short range vb state : a columnar dimer state , or , equivalently , any state with nearest - neighbor pairs coupled into singlets in an arbitrary manner ( there is a huge degeneracy ) .
the energy per site of the two states are : @xmath127 here @xmath128 ( @xmath2 and @xmath66 nearest neighbors ) and @xmath129 are the energy per site of the heisenberg ground state and of the vb state , respectively . for the heisenberg model we have @xmath130 ( in @xmath53 ) and @xmath131 ( in @xmath61 ) , whereas for the crude vb state considered here , the expression for @xmath132 turns out to be independent of @xmath87 .
the factor @xmath133 in the expression for @xmath132 is due to the fact that only half of the bonds in the vb state enjoy the large singlet - singlet pairing @xmath134 . in spite of this reduction ,
the coefficient of the quartic term is substantially larger for the vb state than for the heisenberg state in both @xmath53 and @xmath61 ( @xmath135 for the vb state , to be compared to @xmath136 and @xmath137 for the @xmath53 and @xmath61 heisenberg , respectively ) . in @xmath53
the vb state wins over the heisenberg state . generally speaking , an increase in
the coordination number tends to stabilize a nel - like antiferromagnet with respect to vb states . in @xmath61 ,
the previous crude variational estimate would indeed give the heisenberg state as favored .
the vb state considered here is , however , very poor : for instance , its energy per site , neglecting the quartic term , is only @xmath138 , whereas it is well known that short - range rvb states close in energy to the heisenberg ground state can be constructed.@xcite as a consequence , we believe the present variational estimate to be not conclusive in @xmath61 . in case the vb picture
should be energetically favored , we face the problem of determining the possible orderings of the dimers in the lattice . in principle , either a vb crystal with broken translational symmetry or a homogeneous spin liquid may occur .
following affleck s @xmath53 approach , @xcite a possible way of tackling this problem , in @xmath61 , is from the @xmath83 limit . for @xmath103 , once again , all the states made up of the maximum possible number ( @xmath104 ) of nearest - neighbor s- and t - vbs we refer to them as nnvb configurations
are degenerate , linearly independent ( although not orthogonal ) ground states .
contrary to the @xmath53 case , where only two such states exist ( see fig .
[ vbgs : fig ] ) , in @xmath61 this implies a huge degeneracy . for @xmath83
, however , there are off - diagonal matrix elements connecting different nnvb states ( smaller by a factor @xmath102 , as previously discussed ) , which lift this degeneracy to first order .
it is straightforward to verify , using eq .
[ qaction : eqn ] , that the effective hamiltonian within the subspace of nnvb states ( to first order in @xmath102 ) is just the quantum dimer model ( qdm ) first introduced in ref .
@xcite @xmath139 for the particular case of @xmath140 .
( here @xmath141 and @xmath142 schematically indicate adjacent s- and t - vbs in the y and x direction . )
the qdm is known to have a _
columnar solid _ phase in a region of parameter space close to @xmath140.@xcite recently , an exact diagonalization study of the qdm has revealed that such a columnar solid phase has , more precisely , the features of a plaquette rvb , i.e. translational invariance in one direction is spontaneously broken , but the spin - peierls `` columns '' are disordered as a consequence of a resonance of parallel bonds in each plaquette.@xcite this picture would yield a fourfold degenerate ground state in two dimensions .
we conclude that , for @xmath83 , our model has a columnar vb phase in @xmath61 .
given the fact that our crude variational calculation appears to favor a nel state in @xmath61 , we believe that the issue of the actual behavior of our @xmath79-case , @xmath7 , in @xmath61 is still open .
this issue will be addressed in a future numerical study.@xcite
the motion of holes in a vb state has been the subject of considerable research starting from the early works on high temperature superconductivity @xcite .
now , we are able to investigate this problem on the basis of a consistent microscopic model which does have a vb solid as a ground state . this study may shed light on the issue whether holons and spinons decouple in vb states and whether the vb spin background favors effective hole - hole attraction .
therefore , we analyze in some detail the case of filling @xmath143 in one dimension . for large repulsive interactions
( small @xmath48 ) the effective hamiltonian describing the system is simply the analog of the @xmath63 model @xmath144 where @xmath145 is the hopping part of the original hamiltonian ( eq . [ model : eqn ] ) , @xmath146 is the projector onto states without double occupancy , and @xmath147 is the occupation number operator . in @xmath53 ,
analogously to the hubbard case , @xcite the ( real - space ) wavefunction _ factorizes _ , for @xmath148 , into a spinless fermion part times the wavefunction of a suitable spin hamiltonian @xmath149 on the _ squeezed chain _
@xcite @xmath150 the position of the electrons is determined by the spinless fermion wavefunction @xmath151 , while the spin and orbital ordering is governed by a spin hamiltonian which is obtained by taking the average of @xmath152 over the spinless fermion state .
the result is simply related , by an overall change of energy scale and a canonical transformation , to the previous spin model @xmath7 ( eq . [ st_2:eqn ] ) : @xmath153 \sum_{<ij > } ( { \bf s}_i \cdot { \bf s}_{j } - \frac{1}{4 } ) ( -{\bf t}_i \cdot { \bf t}_{j } + 2 t^z_i t^z_j - \frac{1}{4 } ) \;.\ ] ] the physical origin of this factorization relies on a general feature of one dimensional models with nearest neighbor hopping : in the @xmath154 limit , the spatial ordering of the particles is conserved by the dynamical evolution , irrespectively of the spin configuration .
therefore , the hamiltonian can be diagonalized in each subspace defined by a given spin ordering and the result does not depend on the chosen spin configuration .
the effective hamiltonian governing the charge degrees of freedom is that of a free spinless fermion gas ( with suitable boundary conditions which do depend on the spin wavefunction @xcite ) whose eigenfunctions are @xmath155 .
the degeneracy among different spin configurations is lifted to @xmath156 by the magnetic term in @xmath152 ( [ htst : eqn ] ) , leading to the complete classification ( [ factor ] ) of the eigenstates of @xmath152 to lowest order in @xmath48 .
the spin hamiltonian in eq .
[ hspin : eqn ] , we argued before , has a spin gap .
therefore , we conclude that , while the charge sector is gapless ( metallic ) for large positive @xmath157 and @xmath158 , _ the spin sector is gapped at all densities _ and has a doubly degenerate ground state in the thermodynamic limit .
however , such a spin gap is not related to the formation of bound electron pairs .
in fact , the charge degrees of freedom behave as free particles with no tendency towards pairing : the superconductive correlations are not enhanced in the ground state and decay as @xmath159 .
this state can be rather interpreted as the one dimensional realization of an `` itinerant '' valence bond state where spinons bind , leaving holons in a luttinger liquid state .
( the situation is quite similar to the strong coupling phase of the @xmath160 hubbard model in @xmath53 , for @xmath161.@xcite ) the very presence of such a phase is then a remarkable consequence of one dimensional spin charge decoupling .
the factorization of the wavefunction at @xmath148 can be used to calculate various physical quantities of this spin - gapped metallic phase , following analogous calculations in the hubbard case.@xcite in particular , density correlation functions just coincide with those of a spinless fermion gas , showing power law decay at large distances ( @xmath159 ) while the spin - spin ( and pseudospin ) correlation still decays exponentially , similarly to the half filled case .
another interesting property which can be evaluated is the one - hole green s function providing a quantitative description of the motion of one hole in a short range vb state .
the picture of a static vb crystal would yield an exponentially decaying green s function in real space , but 1d fluctuations severely affect the asymptotic form of the green s function .
the calculation can be carried out exactly in the @xmath148 limit , where the factorization property of all eigenfunctions holds .
in fact , in this limit , the energy levels of a single hole in a vb state coincide with those of a free particle in vacuum plus @xmath156 corrections depending on the energy of the spin background .
moreover , the energy , to lowest order , just depends on the _ holon momentum _ , i.e. on the momentum @xmath162 of the spinless fermion part of the wavefunction @xmath155 : @xmath163 . on the other hand ,
the holon momentum is related to the total momentum @xmath164 of the state by the conservation law : @xmath165 . here
@xmath166 is the momentum of the spin part of the wavefunction @xmath167 and is quantized in units of @xmath168 , the spin model being defined on the squeezed chain of @xmath169 sites .
therefore , for @xmath148 , the one - hole energy levels of the model , at a given total momentum @xmath164 , can be written in terms of the spinon momentum : @xmath170 , i.e. , of the momentum of the spin part of the factorized wavefunction .
the corresponding density of states coincides with that of a single free fermion : @xmath171 with singularities at the band edges @xmath172 .
these singularities are due to the quadratic dependence of the energy spectrum on the holon momentum which will be probably cut - off by the presence of spin excitations to @xmath156 .
the calculation of the single hole green s function @xmath173 is conveniently carried out by use of the lehmann representation of the corresponding spectral function : @xmath174 where , in the last equation , use has been made of the special form of the energy spectrum as @xmath148 .
the usefulness of this representation rests on the knowledge of the weight function @xmath175 which , following ref .
@xcite , can be written in terms of a non - local correlation function involving only spin ( and pseudospin ) variables : @xmath176 such a correlation function can be evaluated on the ground state of the @xmath7 hamiltonian by lanczos diagonalization and , similarly to the heisenberg model case , shows negligible size dependence .
calculations have been carried out up to @xmath177 sites , and the results are shown in fig . [ z : fig ] . a clear singularity in the quasiparticle weight @xmath175 is visible at the spinon fermi momentum @xmath178 .
the analysis of lanczos data suggests a power law divergence as @xmath179 , in close similarity to the analogous behavior encountered in the _ gapless _ heisenberg model .
a quantitative determination of the critical exponent is however precluded by the severe limitation in system size associated to the lanczos method . in the thermodynamic limit , this singularity gives rise to branch cuts in the analytic structure of the spectral function ( [ spectral ] ) : @xmath180 n(\omega)\ ] ] along the lines defined by @xmath181 , where @xmath182 i.e. at @xmath183 .
the form of this dispersion relation suggests nearest neighbor hopping processes of the hole across the lattice .
the overall picture of hole motion in a valence bond state is therefore quite similar to the case of holes in one dimensional quantum antiferromagnets , being characterized by power law tails in the single particle green s function at the dispersion energy @xmath184 .
this behavior can be traced back to the presence of branches of _ gapless _ collective modes above the spin gap which allow the decay of the spin @xmath185 excitation created by the hole motion .
these features of hole propagation should be contrasted to the properties of charges in ising antiferromagnets , where the excitation spectrum is discrete and the band is dispersionless .
in this paper we have discussed in some detail a hopping model of strongly repelling electrons in the presence of local orbital degeneracies not removed by crystal field effects . for the insulating case ( one electron per site and large coulomb repulsions ) we have studied a particular exchange hamiltonian , eq .
( [ hst_intro : eqn ] ) .
it describes the low - energy physics of this system in the regime in which the inter - orbital singlet is the lowest - energy intermediate state available to hopping .
this regime tends to be further favored by a relatively strong electron - phonon coupling . at variance with the standard hund s rule case ( the high spin two - particle state has lowest energy ) , which would instead favor spin ferromagnetism
, we find in our case a strong tendency to the formation of a statically ordered valence bond ( vb ) phase . in @xmath53 ,
this has been firmly established by using variational arguments , a large-@xmath0 limit , and exact diagonalization results .
given the presence of a sizeable spin - gap , weakly coupled chains should also have a vb ground state .
more uncertain is , instead , the outcome in the isotropic @xmath61 case , where a large-@xmath0 approach would predict a columnar vb phase , whereas a crude variational calculation rather points to a nel - like antiferromagnet .
a numerical study , presently under way , will clarify , we hope , this matter .
finally , we have considered the behavior of the system upon hole doping in the @xmath53 case .
the factorization of the wavefunctions , valid in the strong repulsive limit and peculiar to the @xmath53 case , is the crucial ingredient used to tackle the problem .
the outcome is a kind of `` itinerant '' vb state , where charge degrees of freedom are gapless , but the spin and orbital degrees of freedom are gapped and vb - like .
the motivation for our study has been primarily theoretical .
originally inspired by dynamical jahn - teller molecular lattices , such as @xmath186 compounds , the ingredients used are , admittedly , extremely idealized , and not easily applicable to an existing realistic situation . in a real jt lattice , both the crystal field effect , and the intermolecular elastic coupling can not normally be neglected , and may favor a cooperative static jt distortion of all molecules.@xcite second , intermolecular electron hopping tends to be comparable , or even larger in comparison with relevant vibron energies . apart from this , realistic hopping matrix elements have a strong orientational dependence in more than one dimension , and , in general , there is no reason for neglecting overlap integrals between orbitals with different orbital quantum numbers ( see eq .
( [ model : eqn ] ) ) .
third , a detailed description of the structure of the doubly occupied states at each site brings in correction terms to the exchange hamiltonian which turns out to be , in general , strongly asymmetric in orbital space ( see eq .
( [ delta_h : eqn ] ) ) . on the opposite side ,
the exchange hamiltonian we have studied is elegant , parameter - free and prototypical of strong correlations in an orbitally degenerate lattice .
the physics uncovered has a definite robustness against `` small '' perturbations , due to the gap present in the excitation spectrum .
the vb gapped state has a clear resemblance to a spin - peierls state in @xmath53 .
the spin - gapped metal found upon doping is interesting .
the basic ingredients needed for a realistic system to be a candidate vb described in this work are : i ) orbital degeneracy not trivially removed by cooperative jt and/or crystal field effects ; ii ) relatively large molecules with a strong electron - phonon coupling , so as to make the inter - orbital singlet favored as compared to the hund s rule triplet ; iii ) narrow bands with relatively large on - site hubbard @xmath17 , so as to stabilize a mott - hubbard insulator at exactly one electron per site ; iv ) reduced dimensionality .
@xmath186 charge transfer compounds , based on @xmath187 ions,@xcite are potential candidates to the realization of such a scenario .
@xmath186 has a triply degenerate @xmath188 molecular orbital , coupled to several intramolecular vibrations , resulting in an important dynamical jt effect.@xcite the electron - phonon coupling leads to a substantial pairing energy @xmath189 which is , however , overwon by a substantially larger hubbard @xmath190,@xcite resulting in a mott insulating behavior .
the lattices tend to have anisotropic lattice constants , with a pronounced quasi - one - dimensional character.@xcite as for the virtual intermediate states of @xmath191 , singlets and triplet are close in energy , at least in solutions.@xcite if the triplet prevails , as is possibly the case in tdae-@xmath186,@xcite there can be spin ferromagnetism .
we believe , however , there is room in some future system for the alternative possibility of a spin gap , and the formation of the spin - orbital vb state described here .
acknowledgments it is a pleasure to thank s. sorella and m. fabrizio for many stimulating discussions .
ap gratefully acknowledges hospitality at sissa .
work at sissa and ictp was co - sponsored by infm project htsc , and by eu contract erbchrxct940348 .
the case of tdae - c@xmath5 is still quite controversial .
first , the weak ferromagnetism observed reveals a strong anisotropy ; suggestions have been put forward that the actual exchange mechanism might be of antiferromagnetic nature ( dzyaloshinsky - moriya type ) with an overall canting , responsible for the observed ferromagnetism ( see r. blinc _ et al . _ , , 523 ( 1996 ) ) .
second , the c@xmath5 molecules might well be jt distorted , which would make the assumption of unbroken electronic degeneracy invalid .
third , there are recent reports of diamagnetism in this compound ( see m. ricc _
et al . _ ,
( to appear ) ) . | we discuss a hopping model of electrons between idealized molecular sites with local orbital degeneracy and dynamical jahn - teller effect , for crystal field environments of sufficiently high symmetry . for the mott - insulating case ( one electron per site and large coulomb repulsions ) , in the simplest two - fold degenerate situation
, we are led to consider a particular exchange hamiltonian , describing two isotropic spin-1/2 heisenberg problems coupled by a quartic term on equivalent bonds .
this twin - exchange hamiltonian applies to a physical regime in which the inter - orbital singlet is the lowest - energy intermediate state available for hopping .
this regime is favored by a relatively strong electron - phonon coupling .
using variational arguments , a large-@xmath0 limit , and exact diagonalization data , we find that the ground state , in the one dimensional case , is a solid valence bond state . the situation in the two dimensional case is less clear .
finally , the behavior of the system upon hole doping is studied in one dimension . |
jets are one of the fundamental components of radio - loud active galactic nuclei ( agn ) .
although different in detail , they all share the property of being relativistic on the parsec scale and of carrying magnetic fields that may be dynamically important . observations of their non - thermal synchrotron radiation , in fact , require the presence of a magnetic field and the high degree of polarization observed indicates the presence of some large scale structure in the field .
theoretical investigations require , therefore , a relativistic mhd description of such objects . in this effort , recent numerical simulations
@xcite have turned out as efficient investigation tools in understanding many aspects of the jet evolution .
this , in turn , justifies recent and new efforts to extend existing gas - dynamical godunov - type codes to the relativistic regime , see , , , and the extensive review by .
in this perspective , we extend the recently derived relativistic hydrodynamical hllc riemann solver @xcite to the magnetic case , with vanishing normal component of the magnetic field .
the novel numerical scheme is then applied to the propagation of relativistic jets in presence of toroidal magnetic fields .
the paper is organized as follows : in [ sec : equations ] we give the relevant relativistic mhd equations , in [ sec : hllc ] we briefly review the approximate riemann solver . finally , in
[ sec : tests ] we discuss the astrophysical jet applications .
the motion of a magnetized , ideal relativistic fluid is governed by the system of conservation laws @xcite @xmath0 where @xmath1 and @xmath2 is the vector of conservative variables with components given , respectively , by the laboratory density @xmath3 , the three components of momentum @xmath4 and magnetic field @xmath5 ( @xmath6 ) and the total energy density @xmath7 .
@xmath8 are the fluxes along the @xmath9 direction ; for @xmath10 one has @xmath11\ , , \quad\ ] ] where @xmath12 is the total pressure , @xmath13 is the thermal ( gas ) pressure and @xmath14 is the fluid three - velocity .
the lorentz factor is denoted with @xmath15 .
expressions for @xmath16 and @xmath17 follow by cyclic permutations of the indexes . introducing the primitive flow variables @xmath18 , one has @xmath19 where @xmath20 and @xmath21 are , respectively , the rest - mass density and specific enthalpy of the fluid .
an additional equation of state is necessary to close the system ( [ eq : rmhd_eq ] ) . throughout the following
we will assume a constant @xmath22-law , with specific enthalpy given by @xmath23 where @xmath22 is the constant specific heat ratio .
godunov - type codes @xcite are based on a conservative formulation where laboratory density , momentum , energy and magnetic fields are evolved in time . on the other hand , primitive variables , @xmath18 , are required for the computation of the fluxes ( eq . [ eq : fluxes ] ) and more convenient for interpolation purposes .
this task requires the solution of the nonlinear equation @xmath24 where @xmath25 , @xmath26 .
equation ( [ eq : press_fun ] ) follows directly from equation ( [ eq : cons_var_e ] ) together with ( [ eq : cons_var_m ] ) .
since , at the beginning of a time step @xmath7 , @xmath27 and @xmath28 are known quantities , both @xmath15 and @xmath13 may be expressed in terms of @xmath29 alone from @xmath30 * and from the equation of state ( [ eq : eos ] ) using @xmath31*. equation ( [ eq : press_fun ] ) can be solved by any standard root finding algorithm .
once @xmath29 has been found , the lorentz factor is easily found from ( [ eq : c2p_m ] ) , the thermal pressure from ( [ eq : eos ] ) and velocities are found by inverting equation ( [ eq : cons_var_m ] ) : @xmath32 finally , equation ( [ eq : cons_var_d ] ) is used to determine the proper density @xmath20 .
in the traditional godunov approach @xcite , the conservation laws ( [ eq : rmhd_eq ] ) are advanced in time by solving , at each zone interface @xmath33 , a riemann problem with initial condition axis as the normal direction . ]
@xmath34 where @xmath35 and @xmath36 are the left and right edge values at zone interfaces .
the exact solution to the riemann problem for the relativistic mhd equations has been recently derived by . as in the classical case ,
a seven - wave pattern emerges from the decay of an initial discontinuity separating two constant states .
the resulting structure can be found by iterative techniques and it can be quite involved . for the present purposes , however , we will be concerned with the special case where the component of magnetic field normal to a zone interface vanishes . under this condition ,
a degeneracy occurs where the tangential , alfvn and slow waves all propagate at the speed of the fluid and the solution simplifies to a three - wave pattern . in this special situation ,
the relativistic hllc approximate riemann solver originally derived for the gas - dynamical equations by and recently extended to the relativistic regime by ( paper i henceforth ) , can still be applied with minor modifications .
this statement stems from the fact that both state and flux vectors on either side of the tangential discontinuity ( denoted with the @xmath37 superscript ) retain the same form as in the non - magnetized case : @xmath38 @xmath39 where we can still set @xmath40 .
total pressure and normal velocity are continuous across the middle wave , i.e. @xmath41 , @xmath42 .
the method of solution is thus entirely analogous to the strategy shown in paper i and we will not repeat it here .
the two additional components related to the presence of transverse magnetic field are decoupled from the rest and can be solved without additional complications .
once @xmath43 has been found , in fact , we use the jump conditions to get @xmath44 from the pre - shock states .
finally , we need an estimate for the outermost right and left - going fast wave speeds , @xmath45 and @xmath46 .
these two signal velocities may be found by exploiting the characteristic decomposition of the rmhd equation , extensively analyzed by and . in the general case ,
the fast and slow magneto - sonic waves satisfy a quartic equation solvable by means of numerical and analytical methods , e.g. . for vanishing normal component of the magnetic field , however ,
the two roots corresponding to the degenerate slow waves have the trivial solution @xmath47 , whereas the fast speeds satisfy the following quadratic : @xmath48 with @xmath49 where @xmath50 is the speed of sound . the solutions of ( [ eq : eigenvalues_eq ] ) are used to provide suitable guesses for the outermost signal velocities in our hllc riemann solver . as in paper
i , we set @xmath51 where @xmath52 ( @xmath53 ) is the smallest ( biggest ) fast wave speed .
spatial and temporal discretization of equations ( [ eq : rmhd_eq ] ) is based on the second - order hancock predictor step already described in paper i. for the sake of conciseness we will not repeat it here . as an illustrative test , we consider an initial discontinuity separating two constant left and right states given by @xmath54 zones .
the final integration time is @xmath55 ; the courant number is @xmath56 .
the solution is comprised of a left - going fast rarefaction , a right - going tangential and shock waves moving close to each other.,width=434 ] the domain is the interval @xmath57 $ ] covered with @xmath58 uniform computational cells .
the ideal equation of state ( [ eq : eos ] ) is used with @xmath59 .
figure ( [ fig : sod1 ] ) compares the result computed with the hllc solver with the analytical solution * ( available from ) * , at the final time @xmath55 .
the breakup of the discontinuity results in a left - going fast rarefaction wave and a thin shell of high density material bounded by a tangential discontinuity and a gasdynamical shock propagating to the right .
the total pressure and magnetic field are continuous through the contact and the shock wave , respectively .
all discontinuities are correctly captured , with satisfactory resolutions : the tangential discontinuity and the shock smear out over @xmath60 and @xmath61 zones , respectively .
small overshoots appear at the tail of the rarefaction fan . a similar ,
perhaps more pronounced behavior is shown in the results by who used a linearized , roe - like riemann solver .
the relative error in the thin shell is @xmath62 for both pressure and density .
we now turn our attention to the propagation of a relativistic magnetized jet in cylindrical coordinates @xmath63 . for the sake of comparison ,
we have implemented three models already discussed in literature .
the first setup , denoted with @xmath64 , is model @xmath65 of ( k99 henceforth ) , the second and third setups , @xmath66 and @xmath67 , are taken from models @xmath68 and @xmath69 of ( l05 henceforth ) , respectively .
.relevant parameters for models @xmath64 , @xmath66 and @xmath67 described in the text . from left to right : jet density , ambient medium density , beam velocity , magnetization radius , average magnetization , specific heat ratio and magnetic to rest mass energy ratio .
[ cols="^,^,^,^,^,^,^,^",options="header " , ] in all models , the static external medium has uniform density @xmath70 and zero magnetic field . at the inlet , @xmath71 and @xmath72 ,
a supersonic beam with uniform velocity @xmath73 ( lorentz factor @xmath74 ) and density @xmath75 is injected with a toroidal magnetic field obeying @xmath76 where @xmath77 is the magnetization radius of the beam .
the ideal equation of state ( [ eq : eos ] ) is used in all calculations . in model @xmath64 , we set @xmath78 and the thermal pressure inside the beam , @xmath79 , is prescribed from the condition of hydromagnetic equilibrium , @xmath80 & \quad \textrm{if } \quad & r < r_m \ ; , \\
\noalign{\medskip } \alpha p_e & \quad \textrm{if } \quad & r_m < r < 1 \;,\\ \noalign{\medskip } \end{array}\right.\ ] ] where @xmath81 , @xmath82 and @xmath83 is the thermal pressure of the ambient medium . in models
@xmath66 and @xmath67 @xmath84 is computed from the average magnetization parameter @xmath85 : @xmath86 and the thermal pressure follows directly from the definition of the mach number , i.e. @xmath87 with @xmath88 .
the numerical values of the relevant parameters are given in table [ tab : ic ] .
the simulations are performed on the physical domain @xmath89 $ ] , @xmath90 $ ] using @xmath91 points per jet radius .
the total grid size is @xmath92 . for each model ,
we have carried two set of simulations : the first one employing the hllc approximate riemann solver described in
[ sec : hllc ] and the second one adopting the hll scheme , also used in l05 . .
the top and bottom panels show , respectively , the density logarithms and magnetization parameter @xmath93 for the hllc ( upper half of each panel ) and hll ( lower half of each panel ) runs .
notice that the two panels use opposite color gradients .
the grid resolution is @xmath91 zones per jet beam , giving a total grid size of @xmath94 ( in @xmath95 ) and @xmath96 zones ( in @xmath97 ) .
integration has been carried with cfl = 0.4 and the second - order hancock scheme outlined in paper i.,width=434 ] .
the top panel shows the density logarithm in the hllc ( upper half ) and hll ( lower half ) runs .
the bottom panel shows the magnetization parameter in the hllc ( upper half ) and hll ( lower half ) runs .
the final integration time is the same one as in l05.,width=434 ] models @xmath64 and @xmath66 are characterized by a high poynting flux @xmath98 and tend to develop prominent nose cones due to the strong magnetic pinching .
this is evident from fig .
( [ fig : s1jet ] ) and ( [ fig : s2jet ] ) , where the upper ( lower ) portion of each gray scale image refers to the hllc ( hll ) approximate riemann solver at @xmath99 and @xmath100 , respectively .
the nose cone appears as the extended high pressure region bounded by the terminal conical shock ( mach disk ) and the bow shock . at the mach disk ,
the beam is strongly decelerated and magnetic tension wards off sideways deflection of shocked material into the cocoon .
the magnetization parameter , defined as @xmath101 , is particularly strong in the cone above the symmetry axis .
as pointed out by k99 , morphological properties are prevalently dictated by the ratio of the kinetic energy flux to the poynting flux , @xmath102 , which determines the magnetization of shocked gas behind the terminal shock .
an alternative parameter ( l05 ) is the ratio of the magnetic energy density to the rest mass , @xmath103 : extended nose cones tend to form for low @xmath104 ( high @xmath105 ) , whereas familiar cocoon structures appear for high values of @xmath104 ( low @xmath105 ) , regardless of the jet magnetization parameter @xmath93 .
values of @xmath105 are reported in table [ tab : ic ] .
computed for three different resolution runs : low ( @xmath106 ) , medium ( @xmath107 ) and high ( @xmath108 ) .
the hllc runs are shown using , respectively , dotted , dashed and solid lines ; for the hll we use plus signs , triangles and squares.,width=434 ] the positions of the reverse shocks for the hll and hllc runs are , respectively , @xmath109 and @xmath110 in model @xmath64 , @xmath111 and @xmath112 in model @xmath66 . however , the most striking difference between the hll and hllc schemes is the jet propagation speed obtained from model @xmath64 .
indeed , at @xmath99 the head position in the hllc run is @xmath113 to be compared with a value of @xmath114 obtained from the hll integration .
note that the final integration time is not the same as in k99 , where the simulation ends at a later time , @xmath115 , yielding a head position value of @xmath116 , apparently more consistent with the hll run .
however , in order to further investigate these discrepancies , we carried out an additional set of simulations at half and twice the resolution .
figure ( [ fig : ko_pos ] ) plots the jet head position as function of time for the three resolutions in consideration : low ( @xmath117 ) , medium ( @xmath107 ) and high ( @xmath118 ) .
the hll scheme achieves the same convergence of the hllc scheme only at high resolution , whereas the propagation speeds in the hllc runs show to be by far less sensitive to the grid size . .
the final integration time is the same one used in l05 .
the meaning of the gray scale images in the same as in fig .
( [ fig : s1jet ] ) and ( [ fig : s2jet]).,width=434 ] this disagreement is remarkably reduced in the second model @xmath66 , where the hll runs shows a slightly higher propagation speed , but again more evident in model @xmath67 , although with an opposite trend .
once more we ascribe this behavior to resolution effects , based on the fact that the same model in l05 with twice the resolution ( i.e. @xmath119 zones per beam radius instead of @xmath91 ) shows stronger affinities with the hllc integration presented here . when the hllc solver is employed , the backflow turbulent patterns in the cocoon are better defined than with the hll scheme , where the larger numerical viscosity inhibits the formation of small scale structure at lower resolution .
this is corroborated by a simple qualitative comparison with the results of l05 , who used twice the resolution employed here and the hll solver .
the small scale structural details provided by the hllc approach appear to be almost equivalent even at half the resolution .
this consideration together with the convergence test previously shown confirms , indeed , that the ability to describe tangential and shear waves largely improves the quality of solution , particularly in the multidimensional case .
finally , we remind that the computational cost of the hllc scheme over the traditional hll riemann solver is less than @xmath120 and we believe that this largely justifies its use in godunov - type codes .
numerical simulations of relativistic magnetized jets with toroidal magnetic fields have been presented .
the three models considered show formation of conspicuous nose cones and jet confinement with increasing magnetization . the morphological properties favorably compare with the results of previous investigators .
simulations have been conducted by using the recent relativistic hydrodynamics hllc solver by mignone & bodo , opportunely extended to the special case of vanishing normal component of magnetic field . in order to assess the validity of the new algorithm ,
simulations have been directly compared with the traditional hll riemann solver @xcite .
it is found that the hllc approach largely improves the quality of solution over the hll scheme by considerably reducing the amount of numerical viscosity .
this is manifested in a sharper resolution of fine scale structures in proximity of shear and tangential flows .
furthermore , convergence tests at different resolutions have shown that , with the hllc scheme , the jet head position has little dependence on the grid size .
this is a remarkable property for a shock - capturing scheme and has to be consolidated through further numerical testing .
generalization to the full relativistic mhd case is underway .
leismann , t. , ant ' on , l. , aloy , m. a. , m " uller , e. , mart , j. m. , miralles , j. a. , and j. m. ib ' a~ nez : 2005 ` relativistic mhd simulations of extragalactic jets ' , _ astronomy & astrophysics _ * 436 * , pp . | this paper presents an application of the recent relativistic hllc approximate riemann solver by mignone & bodo to magnetized flows with vanishing normal component of the magnetic field .
the numerical scheme is validated in two dimensions by investigating the propagation of axisymmetric jets with toroidal magnetic fields .
the selected jet models show that the hllc solver yields sharper resolution of contact and shear waves and better convergence properties over the traditional hll approach . |
three - body systems and processes are of fundamental interest in physics @xcite .
one of these , with which a number of us have been concerned , is the recombination of three - particles to a dimer plus a free particle , in a many body system forming a bose - einstein condensate @xcite .
the condensate is not the lowest state of the system , but a metastable state .
the 3-body recombination is the dominant mechanism for cooling and lowering the overall energy of the system .
experimental and theoretical studies have shown that this recombination rate depends mainly on the two - body scattering length @xmath5 @xcite , as the collision energy is low and the interaction is weak - owing to large interparticle distances , and on the bound state energies .
this would suggest that zero - range potentials ( zrp ) , defined in terms of the scattering length @xcite , @xmath6=-1/a.\label{i0.1}\ ] ] can be applied to model the interaction between the particles of the condensate .
it has been shown by nielsen and macek using the hidden crossing technique that the zrp describes properly the recombination transition in a system of three @xmath3he atoms @xcite .
also , gasaneo and macek showed that the zrp gives a quite good representation for the adiabatic potential of the same system @xcite . a closed form solution for a system of three - particles interacting via
a zrp has been recently presented by gasaneo et al @xcite .
the fragmentation process @xmath3he@xmath7he@xmath8he@xmath9he@xmath9he was studied and relatively good agreement was found when compared with the hidden crossing calculations . in this paper , we seek to apply our techniques to a famous model : 3 particles in one dimension , subject to pair delta - function interactions . for this model , introduced by mcguire @xcite
, one can obtain exact solutions for the wave functions , the scattering matrix and the binding energies , in the case of particles of identical masses and equally weighted interactions .
as such it has been extended to a larger number of particles @xcite , using bethe s ansatz @xcite , and also found to be exceedingly useful when used as a test - bed for the development of a number of different methods ( pertubative , faddeev , hyperspherical adiabatic , etc . ) @xcite . here
, we note that using zrp and ( [ i0.1 ] ) , in 3-dimensions , leads to the thomas effect and the collapse of the 3-body ground state @xcite .
however , in one dimension , an equation similar to ( [ i0.1 ] ) - with @xmath5 not the scattering length - provides boundary conditions which correctly characterizes the wave functions and replace the use of the @xmath10-function interactions , and should therefore again give us exact results .
one of these , though , is that the recombination rate , for this model , is exactly zero . in section
ii we propose a solution , written in integral form , for the free particle schrdinger equation , written in hyperspherical coordinates . a linear combination of free particle solutions can then be found to satisfy the boundary condition that we alluded to earlier , and thus provide us with the solution of the problem with interaction .
the requirement that the wave function satisfy the boundary conditions leads us to one of the important results of this paper , namely that the weight of the free particle solutions , in the integral form , satisfies a recurrence relation similar to that obtained in the refs . @xcite and @xcite . in section
iii the method is applied to a particular case in which two of the particles are bound .
it is shown that the recurrence relation , defining the coefficient of the free - particle expansion , can be solved in closed form and , thus , the scattering matrix is also obtained in a closed form .
this allows us to have a detailed test of our method . in this section
it is also shown that the wave function obtained is equivalent to the mcguire plane wave solution , and that our expression for the @xmath11 matrix is the matrix obtained by mcguire , in the particular case discussed in this paper . in section
iv the relation between the hyperspherical adiabatic approach and the present one is discussed . in appendix
a , the pseudo - sturmian functions are derived . in appendix b ,
the wave function is written as the symmetric wave plane in cartesian coordinates .
to begin the study of the three identical - particle system ( therefore with equal masses ) , consider the center of mass and jacobi coordinates , @xmath12 [ fig1 ] the @xmath13 give us the locations of the 3 particles along the line , see fig .
1 . using polar coordinates , the 2 jacobi variables allow us to define , in turn , a hyper radius @xmath14 and an angle @xmath15 as @xmath16 where @xmath17and@xmath18 . in terms of these coordinates the schrdinger equation for the ` relative ' system can be written as @xmath19 where @xmath20 the function @xmath21 is defined by @xmath22 where the coefficient @xmath23 equals @xmath24 , @xmath25 being the strength of the interactions .
this @xmath23 is negative for attractive interactions and positive for repulsive ones .
the angles @xmath26 equal @xmath27 .
the lines @xmath28 divide the @xmath29 plane in six regions . in each region
the order of particles is fixed , so that between @xmath30 and @xmath31 , @xmath32 , etc .
a different permutation of particles is associated to each region .
> from now on , we will choose the units such that @xmath33 and @xmath34 . in each sector
, we now seek a free particle solution that satisfies the boundary condition that will replace the effect of the potential , i.e. @xmath35=-\frac{1}{a},\label{wfbc}\ ] ] where @xmath36 . in eq.([wfbc ] ) @xmath37 does not depend on @xmath38 because all the strengths of the interactions and all the masses are equal .
writing a solution for the free particle system as the product @xmath39 , where @xmath40 , leads to the set of free particle equations @xmath41r^{-1/2}z_{\nu}\left(\mathrm{k}r\right)\right.\nonumber \\ & & \left.=\frac{-1}{\theta(\nu,\theta)}\left(\frac{\partial^{2}}{}{\partial\theta^{2}}+\frac{1}{4}\right)\theta(\nu,\theta)=\nu^{2}-\frac{1}{4}\right.,\end{aligned}\ ] ] where @xmath42 is a bessel function and @xmath43 a separation constant .
if for the @xmath44 functions we choose the pseudo - sturmian functions @xmath45 , defined , for fixed @xmath43 , as the solutions of @xmath46s(\nu,\theta)=\left(\nu^{2}-\frac{1}{4}\right)s(\nu,\theta),\label{seq}\ ] ] then the functions @xmath47 are solutions of the schrdinger equation eq.([sr ] ) for values of @xmath48 note that eq.([seq ] ) may be replaced by the relation @xmath49s(\nu,\theta)=0\label{afree}\ ] ] subject to the boundary conditions @xmath50=-\frac{1}{a},\ ; j=0,1, ... ,5.\label{sbc}\ ] ] in the last equation , we assumed that @xmath45 is symmetric about each line @xmath28 .
we now propose to write the general wave function of the system as a kontorovich - lebedev transform , in terms of the base functions just discussed , that is , as @xmath51 provided that its derivative satisfies the boundary conditions , i.e. eq.([wfbc ] ) .
the contour of integration must be chosen so that the wave function has the correct asymptotic behaviour . following the reasoning of gasaneo et al @xcite , we will now show that the boundary conditions , eq.([wfbc ] ) , @xmath52can be transformed into a recurrence relation for @xmath53 .
first , we substitute eq.([klr ] ) in eq.([wfbc ] ) , and then interchange the order in which the integral and the derivative are taken , to obtain for each @xmath38@xmath54=0.\right.\end{aligned}\ ] ] second , we use eq.([sbc ] ) and the identity @xmath55 @xmath56 to transform the equation to @xmath57s(\nu,\theta)\right.\nonumber \\ & & \times\left[z_{\nu+1}(kr)-z_{\nu-1}(kr)\right]\nonumber \\ & & \left.\left.+\frac{2}{ka}\int_{\varsigma}d\nu a(\nu)s(\nu,\theta)z_{\nu}(kr)\right\ } = 0.\right.\end{aligned}\ ] ] we assumed in the previous equation that @xmath58 and @xmath59 , because we are mainly interested in negative energies . by selecting the appropriate contours we can now transform the last equation to @xmath60z_{\nu}(kr)=0\right.\end{aligned}\ ] ] since the set of bessel functions forms a complete set of basis functions , the function within the square brackets should be zero at the limit .
we arrive , finally , at the recurrence relation that we are looking for , @xmath61 where @xmath62 . in the following section
, we will apply this approach to a particular case of this three body system and show that we can obtain the wave function and the @xmath63-matrix .
consider now the case where two of the particles are bound .
the wave function @xmath64 can still be written in terms of the kontorovich - lebedev representation , eq.([klr ] ) .
the unnormalized angle pseudo - sturmian function @xmath45 , a six - fold symmetric function , is defined by the eqs.([afree ] ) and ( [ sbc ] ) .
as can be seen in appendix a , the function @xmath45 may written as @xmath65\qquad\left|\theta - j\frac{\pi}{3}\right|<\frac{\pi}{6},\;\label{st}\ ] ] with @xmath66 , where @xmath67 satisfies the relation @xmath68 > from the previous section , we can immediately conclude that @xmath69 satisfies the recurrence relation @xmath70-a(\nu-1)\sin\left[\left(\nu-1\right)\frac{\pi}{6}\right]\nonumber \\ & & = -\frac{\pi c}{3k}a(\nu)\cos\left[\nu\frac{\pi}{6}\right],\label{rec1}\end{aligned}\ ] ] .
the recurrence relation , displayed in eq.([rec1 ] ) , can be written as @xmath71+e^{-i\left(\pi/6\right)\nu}\left[-a\left(\nu+1\right)e^{-i\left(\pi/6\right)}\right.\nonumber \\ & & \left.\left.+a(\nu-1)e^{i\left(\pi/6\right)}+\frac{i\pi c}{3k}a(\nu)\right]=0.\right.\label{rec2}\end{aligned}\ ] ] an inspection , of the solution of the recurrence relation - eq.(25 ) in ref .
@xcite , leads us to propose a coefficient in the form of the series @xmath72\label{anu}\end{aligned}\ ] ] substituting this expression in eq.([rec2 ] ) , and equating to zero the coefficients of exponentials , with different arguments that depend on @xmath43 , we obtain the following values for the parameters : @xmath73 @xmath74 consequently , the solution for the coefficient can be written as @xmath75 or @xmath76\ ] ] where @xmath77 and @xmath78 in the next section , we demonstrate that @xmath63 represents the scattering matrix and , accordingly , @xmath10 the phase shift .
we should stress the remarkable fact that the s - matrix appears explicitly in the solution of the recurrence relation . in the next subsection
it is shown that the expression obtained for @xmath63 in this work is equivalent to the formula for the exact symmetric @xmath63-matrix for the 2 + 1 process , given in ref .
@xcite . to be specific we will restrict the following discussion to the case of total negative energies , and will write @xmath79 , in which @xmath80 is the two - body bound energy and @xmath81 is the effective energy .
next , we will show that the imaginary axis is the appropriate contour to obtain the correct asymptotic behaviour of the wave function . substituting the coefficients @xmath69 defined in eqs.([par ] ) and ( [ a ] ) , the pseudo - sturmian functions given in eq.([st ] ) and the modified bessel functions @xmath82 , into eq.([klr ] ) , as well as choosing the imaginary axis as the contour of integration , we find @xmath83-\sinh\left[\left(i\pi/3+\beta\right)\nu\right]\right)\nonumber \\ & & \times\cos\left[\left(\theta - j\frac{\pi}{3}\right)\nu\right]k_{\nu}(kr)\nonumber \\ & & + \mathcal{s}\int_{\varsigma}d\nu\left(\cosh\left[\left(i\pi/3-\beta\right)\nu\right]+\sinh\left[\left(i\pi/3-\beta\right)\nu\right]\right)\nonumber \\ & & \times\cos\left[\left(\theta - j\frac{\pi}{3}\right)\nu\right]k_{\nu}(kr).\nonumber \\ & & + \mathcal{s}_{3}\int_{\varsigma}d\nu\left(\cosh\left[\beta\nu\right]-\sinh\left[\beta\nu\right]\right)\nonumber \\ & & \times\cos\left[\left(\theta - j\frac{\pi}{3}\right)\nu\right]k_{\nu}(kr).\end{aligned}\ ] ] note that in the above expression the exponentials in the coefficients have been written in terms of hyperbolic functions .
the integral over the odd terms vanishes , leaving only the even terms in the integrand .
after a trigonometric identity , this leads to @xmath84\frac{\pi}{3}\right)\right]\nu\right)k_{\nu}(kr)\right.\nonumber \\ & & + \int_{\varsigma}d\nu\cosh\left(\left[-\beta+i\left(\theta-\left[j+1\right]\frac{\pi}{3}\right)\right]\nu\right)k_{\nu}(kr)\nonumber \\ & & + \mathcal{s}\left[\int_{\varsigma}d\nu\cosh\left(\left[-\beta+i\left(\theta-\left[j-1\right]\frac{\pi}{3}\right)\right]\nu\right)k_{\nu}(kr)\right.\nonumber \\ & & + \;\left.\int_{\varsigma}d\nu\cosh\left(\left[\beta+i\left(\theta-\left[j+1\right]\frac{\pi}{3}\right)\right]\nu\right)k_{\nu}(kr)\right]\nonumber \\ & & + \mathcal{s}_{3}\left[\int_{\varsigma}d\nu\cosh\left(\left[-\beta+i\left(\theta - j\frac{\pi}{3}\right)\right]\nu\right)k_{\nu}(kr)\right.\nonumber \\ & & + \left.\int_{\varsigma}d\nu\cosh\left(\left[\beta+i\left(\theta - j\frac{\pi}{3}\right)\right]\nu\right)k_{\nu}(kr)\right]\end{aligned}\ ] ] using the kontorovich - levedev transforms @xcite , we obtain @xmath85\right)\right\ } \right.\nonumber \\ & & + \exp\left\ { -kr\cosh\left(-\beta+i\left[\theta-\left(j+1\right)\frac{\pi}{3}\right]\right)\right\ } \nonumber \\ & & + \mathcal{s\,}\left[\exp\left\ { -kr\cosh\left(\beta+i\left[\theta-\left(j+1\right)\frac{\pi}{3}\right]\right)\right\ } \right.\nonumber \\ & & \left.+\exp\left\ { -kr\cosh\left(-\beta+i\left[\theta-\left(j-1\right)\frac{\pi}{3}\right]\right)\right\ } \right]\nonumber \\ & & + \mathcal{s}_{3}\left[\exp\left\ { -kr\cosh\left(-\beta+i\left[\theta - j\frac{\pi}{3}\right]\right)\right\ } \right.\nonumber \\ & & \left.+\exp\left\ { -kr\cosh\left(\beta+i\left[\theta - j\frac{\pi}{3}\right]\right)\right\ } \right]\end{aligned}\ ] ] introducing @xmath86 from eq.([beta ] ) into this expression , yields @xmath87\right.\right.\nonumber \\ & & \left .- ikr\sin\left[\theta-\left(j-1\right)\frac{\pi}{3}\right]\right\ } \nonumber \\ & & + \exp\left\ { \frac{\pi c}{6}r\cos\left[\theta-\left(j+1\right)\frac{\pi}{3}\right]\right.\nonumber \\ & & \left.ikr\sin\left[\theta-\left(j+1\right)\frac{\pi}{3}\right]\right\ } \nonumber \\ & + \mathcal{s\ , } & \left[\exp\left\ { \frac{\pi c}{6}r\cos\left[\theta-\left(j+1\right)\frac{\pi}{3}\right]\right.\right.\nonumber \\ & & \left .- ikr\sin\left[\theta-\left(j+1\right)\frac{\pi}{3}\right]\right\ } \nonumber \\ & & + \exp\left\ { \frac{\pi c}{6}r\cos\left[\theta-\left(j-1\right)\frac{\pi}{3}\right]\right.\nonumber \\ & & \left.\left.ikr\sin\left[\theta-\left(j-1\right)\frac{\pi}{3}\right]\right\ } \right]\nonumber \\ & + \mathcal{s}_{3 } & \left[\exp\left\ { \frac{\pi c}{6}r\cos\left[\theta - j\frac{\pi}{3}\right]\right.\right.\nonumber \\ & & \left.ikr\sin\left[\theta - j\frac{\pi}{3}\right]\right\ } \nonumber \\ & & + \exp\left\ { \frac{\pi c}{6}r\cos\left[\theta - j\frac{\pi}{3}\right]\right.\nonumber \\ & & \left.\left.\left .- ikr\sin\left[\theta - j\frac{\pi}{3}\right]\right\ } \right]\right),\nonumber \\ & & j=0,1, ... ,5\label{symf}\end{aligned}\ ] ] from eq.([st ] ) it is easily seen that this wave function is fully symmetric under the interchange of particles .
the function is invariant under the addition of @xmath88 to @xmath15 , together with the addition of one unit to @xmath89 which is what should be done to move from one region in the @xmath90plane to its next counterclockwise neighbour .
remember that for each region there is a specific order of the particles .
we can also see that the real part of each of the exponential arguments is negative , except when @xmath91 that is on the lines @xmath92 where its value is zero .
thus , when @xmath14 is large , the wave function is negligible except near the lines @xmath93 note that only the first 4 terms give a significant contribution in the asymptotic region .
we can be conclude that the form of the wave function is that of products of bound state functions , associated with two particles , with oscillatory functions , which describe the location of the third particle with respect to the 2 bound ones .
evaluating , then , the wave function for large values of @xmath14 , its asymptotic form can be written as @xmath94 where @xmath95 .
this asymptotic expression consists of a wave representing a two particles bound state multiplied by an incoming wave , together with an outgoing wave multiplied by @xmath11 .
> from the expression in eq.([par ] ) the matrix @xmath63 can be written as @xmath96 in terms of @xmath97 , @xmath98 therefore @xmath99 this is , precisely , the scattering matrix @xmath100\left[3+i\left(6\sqrt{3}/\pi c\right)k\right]}{\left[3+i\left(6\sqrt{3}/\pi c\right)k\right]\left[-1+i\left(6\sqrt{3}/\pi c\right)k\right]},\label{s}\ ] ] given as eq.(61 ) in ref .
it corresponds to the symmetric s matrix calculated for the specific process 2 + 1 .
the matrix @xmath101 , see again ( [ par ] ) , has the following form as a function of k @xmath102
@xmath101 multiplies the shorter ranged part of the exact wave function , that goes to zero when @xmath14 goes to @xmath103 . additional insight can be gained , by following the reasoning of mcguire @xcite .
the scattering of 3 asymptotically free particles , to 3 also asymptotically free particles , requires 3 ( successive ) collisions , and yields the part of the wave function associated with the calculation of the s - matrix .
intermediate stages , associated with fewer collisions , give rise to the shorter ranged part of the wave functions .
a similar reasoning holds for the @xmath104 processes . in conclusion
, we have shown that this integration contour , and the choice of bessel functions , have imparted the correct asymptotic behaviour .
furthermore , we can deduce from the asymptotic expression that the coefficient @xmath63 represents the s - matrix .
the eigenfunctions of the following eigenvalue equation @xcite form a complete set of orthogonal hyperspherical adiabatic basis functions .
changing , a bit , the usual notation : @xmath105b_{\kappa}(\theta;r')=0,\label{adi}\ ] ] where @xmath106 , a real parameter in this equation , is held fixed ; @xmath107 and @xmath108 as the interaction is turned off .
the unnormalized eigenfunctions @xmath109,\ ] ] where @xmath38 is an integer such that @xmath110 and @xmath111 satisfies @xmath112 in the adiabatic approach the parameter @xmath106 is identified with the hyper radius @xmath14 . for @xmath113
there is only one open channel , labeled by @xmath114 . for large @xmath14
, the channel function is concentrated along the lines defined by @xmath28 .
accordingly it can represent the two - body bound state . for @xmath115
, there is an infinite number of open channels , labeled by the successive numbers @xmath116 , equal and greater than zero .
they describe , asymptotically , three free particles or a two - body bound state , together with a free particle .
note that the function @xmath67 , eq.([rnu ] ) , is a real function if , and only if , @xmath43 takes on values along the imaginary or the real axis , see fig .
it can be seen that the pseudo - sturmian function defined in eq.([seq ] ) coincides , apart from normalization constants , with the lowest adiabatic function @xmath117 when @xmath118 is an imaginary number and @xmath119 . also , if @xmath120 are in the real intervals @xmath121,9+\left[\kappa-6\right]\right)$ ] with @xmath122 , then @xmath123 and the pseudo - sturmian functions become equal , except for the normalization constants , to the adiabatic eigenfunctions @xmath124 . thus , in the case of the example considered in this paper , that is @xmath113 and the @xmath104 system , the integral eq.([klr ] ) , along the imaginary axis in the complex @xmath43 plane , can be written in terms of the lowest adiabatic function as @xmath125 where @xmath43 runs from @xmath126 to @xmath127 .
the most important contribution of the adiabatic functions to the integral , at large @xmath14 , comes from the lines @xmath28 , where two of the particles are joined .
when these adiabatic functions are multiplied by the appropriate bessels functions , their linear combination ( eq . ( [ intad ] ) ) should have the correct asymptotic behaviour , and will represent a two - body bound system in the colliding with a third particle . . in ( a )
we plot @xmath67 as a function of @xmath128 . in ( b )
the plot of ( a ) is rotated and flipped to give @xmath128 as a function of @xmath129 for @xmath130 positive , @xmath131.,width=264 ] [ fig2 ]
we have shown that the integral representation approach within the hyperspherical context , when applied to mcguire s model , offers a reliable tool to study the collisional dynamics of the 3-body system .
we have obtained several interesting results , namely : -an exact solution to the corresponding schrdinger equation .
-a closed form for the angular basis for this system , the pseudo - sturmian functions .
-a recurrence relation for the coefficients , in the expansion of the wave function in terms of the free - particle basis .
-the s - matrix , obtained directly from the solution of the recurrence relation .
-the relation of the present approach to the traditional adiabatic approach .
-the relation of the present solution to the known plane wave exact solution .
the simplicity of the approach as compared with the adiabatic one , promises to be very useful in extending it to more complicated situations , like the system with different masses and systems with more particles , currently under research , or systems in three dimensions modeled by zrp potentials . in the last case ,
the method can be applied to a wide kind of systems to obtain asymptotic solutions which can be matched to solutions obtained with methods like the r - matrix one , simplifying substantially the calculations .
oak ridge national laboratory , is managed by ut - battelle , llc under contract number de - ac05 - 00or22725 .
support by the national science foundation under grant number phy997206 is gratefully acknowledged .
aat thanks to conacyt , project 4877-e for partial support and gratefully acknowledges the financial support from the national science foundation under grant number phy997206 together with the hospitality of the university of tennessee .
syl also thanks the university for making possible a visit .
one of us , g.g .
thanks the support under the pict nro .
0306249 of the anpcyt . with @xmath132 , can be written as the free angular wave solution @xmath133,\ ; j=0,1, ... ,5,$ ] provided that it be continuous through the boundary lines @xmath28 , that is , @xmath134=\cos\left[\nu\left(\theta_{j}-\gamma_{j}\right)\right],\ ] ] and satisfies the boundary conditions @xmath135\varphi(\theta)\right.\nonumber \\ & & \left.\left.-\left[\rho\left(\nu\right)\frac{\pi c}{3}\sum_{l=0}^{5}\delta\left(\theta-\theta_{l}\right)\right]\varphi(\theta)\right\ } = 0,\right.\;\;\label{co1}\end{aligned}\ ] ] with @xmath136 .
@xmath137 , which does not depend on @xmath38 for the symmetric solution , determined by normalizing the wave function@xcite .
the requirement of continuity leads to the conditions @xmath138\right)\pi/3 $ ] or to @xmath139
. the second condition does not satisfy ( [ co1 ] ) so we shall use the first one , which can be written as @xmath140 . now to focus on eq .
( [ co1 ] ) .
continuity implies that the integral of the second term gives zero . for each @xmath38 , the first and the third terms give @xmath141\right)_{\theta=\theta_{j}+\varsigma}\right.\nonumber \\ & & -\left.\frac{\partial}{\partial\theta}\left(\cos\nu\left[\theta-\gamma_{j}\right]\right)_{\theta=\theta_{j}-\varsigma}\right)\nonumber \\ & & -\rho\left(\nu\right)\frac{\pi c}{3}\cos\nu\left[\theta_{j}-\gamma_{j}\right]=0.\label{b4}\end{aligned}\ ] ] @xmath148\right.\right.\nonumber \\ & & + \left.\frac{x_{1}+x_{2}-2x_{3}}{\sqrt{6}}\sin\left[\left(j-1\right)\frac{\pi}{3}\right]\right)\nonumber \\ & & -ik\left(\frac{x_{1}+x_{2}-2x_{3}}{\sqrt{6}}\cos\left[\left(j-1\right)\frac{\pi}{3}\right]\right.\nonumber \\ & & -\left.\left.\frac{x_{1}-x_{2}}{\sqrt{2}}\sin\left[-\left(j-1\right)\frac{\pi}{3}\right]\right)\right\ } \nonumber \\ & & + \exp\left\ { -\frac{\pi c}{6}\left(\frac{x_{1}-x_{2}}{\sqrt{2}}\cos\left[\left(j+1\right)\frac{\pi}{3}\right]\right.\right.\nonumber \\ & & + \left.\frac{x_{1}+x_{2}-2x_{3}}{\sqrt{6}}\sin\left[\left(j+1\right)\frac{\pi}{3}\right]\right)\nonumber \\ & & -ik\left(\frac{x_{1}+x_{2}-2x_{3}}{\sqrt{6}}\cos\left[\left(j+1\right)\frac{\pi}{3}\right]\right.\nonumber \\ & & -\left.\left.\frac{x_{1}-x_{2}}{\sqrt{2}}\sin\left[-\left(j+1\right)\frac{\pi}{3}\right]\right)\right\ } \end{aligned}\ ] ] the incoming wave takes the form @xmath152 the outgoing wave for @xmath149
can be obtained from the incoming one by substituting @xmath153 by @xmath154 , which in turns means interchanging @xmath155 and inverting the sign of the whole argument within all exponentials , that is , the above results correspond to the sector @xmath149 in the @xmath29 plane , in which the order of particles is given by @xmath158 .
the waves in different sectors can be obtained by the appropriate permutation of the set of coordinates @xmath159 . the completely symmetric wave plane
may then be written as @xmath160_{j = j_{p}}\right.\right.\nonumber \\ & + \left.\exp\left[i\left(k_{2}x_{1}+k_{3}x_{2}+k_{1}x_{3}\right)\right]_{j = j_{p}}\right\ } \nonumber \\
+ \mathcal{s}(p ) & \left\ { \exp\left[-i\left(k_{1}x_{1}+k_{3}x_{2}+k_{2}x_{3}\right)\right]_{j = j_{p}}\right.\nonumber \\ & \left.+\exp\left[i\left(k_{3}x_{1}+k_{1}x_{2}+k_{2}x_{3}\right)\right]_{j = j_{p}}\right\ } \nonumber \\ + \mathcal{s}_{3}(p ) & \left\ { \exp\left[-i\left(k_{1}x_{1}+k_{2}x_{2}+k_{3}x_{3}\right)\right]_{j = j_{p}}\right.\nonumber \\ & + \left.\left.\exp\left[i\left(k_{2}x_{1}+k_{1}x_{2}+k_{3}x_{3}\right)\right]_{j = j_{p}}\right\ } \right]\end{aligned}\ ] ] where the sum runs over all permutations of the set @xmath159 . 10 _ xviith european conference on few - body problems in physics_. conference handbook .
ed . : a. stadler et al .
evora 2000 .
e. nielsen and j. h. macek , phys .
lett * 83 * , 1566 ( 1999 ) .
s. inouye , m. r. andrews , k stenger , h.j .
miesner , d. m. stamper - kurn and w. ketterle , nature ( london ) * 392 * , 151 ( 1998 ) .
a. j. moerdijk , h. m. j. m. boesten and b. j. verhaar , phys .
a 53 , r916 ( 1996 ) .
b. d. esry , c. h .. greene , y. zhou and c. d. lin , j. phys .
b 29 , l51 ( 1996 ) .
p. o. fedichev , m. w. reynolds and g. v. shlyapnikov , phys .
77 , 2921 ( 1996 ) .
n. demkov and v. n. ostrovsky , _ zero - range potentials and their applications in atomic physics_. plenum press , ny , 1988 .
g. gasaneo and j. h macek , jour .
b * 35 * 2239 - 2250 ( 2002 ) .
g. gasaneo , s. yu .
ovchinnikov and j. h. macek , phys .
a submitted ( 2002 ) . j. b. mcguire , j. math .
phys . * 5 * , ( 1964 ) 622 .
c. n. yang , phys .
* 19 * , 1312 ( 1967 ) ; c. n. yang , phys . rev .
* 168 * , 1920 ( 1968 ) .
h. a. bethe , z. phys .
* 71 * , ( 1931 ) 205 .
l. r. dodd , j. math .
phys . * 11 * , ( 1970 ) 207 ; h. b. thacker , phys .
d * 11 * , ( 1975 ) 838 ; a. amaya - tapia , s.y .
larsen and j. popiel , few - body sys .
* 23 * , ( 1997 ) 87 ; s. i. vinitsky , s. y. larsen , d. v. pavlov , d. v. proskurin . phys . at
. nucl . * 64 * , 27 ( 2001 ) ; o. chuluunbaatar , a. a. gusev , s. y. larsen , s. i. vinitsky . j. phys . *
a35 * , l513 ( 2002 ) l. h. thomas , phys . rev . * 47 * , 903 ( 1935 ) .
g. gasaneo , s. yu .
ovchinnikov and j. h. macek , jour . of phys .
* a34 * , ( 2001 ) 8941 . i. s. gradshteyn and i. m. ryzhik , _ table of integrals , series and products .
_ academic press , ny , 1965 , p.773 .
a. amaya - tapia , s.y . larsen and j. popiel , few - body sys . * 23 * , ( 1997 ) 87 ; w. gibson , s. y. larsen and j. j. popiel , phys .
a * 35 * ( 1987 ) 4919 .
larsen , in _ few body methods : principles and applications_. t. k. lim , c. g. bao , d. p. hou and h. s. huber , eds .
singapore , world - scientific 1986 . | the schrdinger equation , in hyperspherical coordinates , is solved in closed form for a system of three particles on a line , interacting via pair delta functions .
this is for the case of equal masses and potential strengths .
the interactions are replaced by appropriate boundary conditions .
this leads then to requiring the solution of a free - particle schrdinger equation subject to these boundary conditions .
a generalized kontorovich - lebedev transformation is used to write this solution as an integral involving a product of bessel functions and pseudo - sturmian functions .
the coefficient of the product is obtained from a three - term recurrence relation , derived from the boundary condition .
the contours of the kontorovich - lebedev representation are fixed by the asymptotic conditions .
the scattering matrix is then derived from the exact solution of the recurrence relation .
the wavefunctions that are obtained are shown to be equivalent to those derived by mcguire .
the method can clearly be applied to a larger number of particles and hopefully might be useful for unequal masses and potentials .
@xmath0 centro de ciencias fsicas , unam , ap 48 - 3,cuernavaca , mor .
62251 , mxico .
@xmath1 department of physics and astronomy , university of tennessee , knoxville , tn 37996 - 1501 , usa .
@xmath2 departamento de fsica , universidad nacional del sur , av .
alem 1253 , ( 8000 ) baha blanca , buenos aires , argentina .
@xmath3 oak ridge national laboratory , po box 2008 , oak ridge , tennessee 37831 , usa @xmath4 100 forest place , apt .
1305 , oak park , il 60301 , usa . |
the numbers of embedded clusters observed in giant molecular clouds and the inferred birthrates of their stellar populations suggest strongly that stars in the galaxy form predominantly within clustered environments and not in isolation ( lada & lada 2003 ) . at present
, however , no paradigm exists describing how stellar formation proceeds within clusters inside turbulent cloud cores ( meyer et al .
such cores can contain starless objects , i.e. , local maxima of arbitrarily high column density where no evidence for star formation has been detected , as well as extremely young protostellar objects , i.e. , locations where evidence for star formation has been detected . observations of these objects in clusters should significantly improve our understanding of how they form .
the gas and dust comprising or surrounding these objects can reveal the motions and physical conditions associated with the earliest stages of stellar formation , providing the necessary constraints to new models .
the ophiuchus star - forming molecular cloud is particularly well suited for an intensive study of the dense gas associated with star formation within clusters .
only 125 pc distant ( de geus , de zeeuw , & lub 1989 , de geus 1992 ) , ophiuchus contains the nearest embedded cluster to the sun ( see allen et al .
2002 , and references therein ) , residing within high column densities of gas and dust ( e.g. , see loren , wootten , & wilking 1990 ; lww90 ) . therefore
, appropriate observations of ophiuchus can reveal starless objects and extremely young protostellar objects within a clustered environment at the best possible sensitivities and the finest possible linear resolutions . over the last decade ,
observations of the cores associated with l1688 in ophiuchus , i.e. , oph a to oph f ( see lww90 ) , have revealed numerous potential starless and extremely young protostellar objects . in particular
, the oph a core contains a protostellar cluster including the prototypical class 0 " object , vla 1623 , in addition to starless objects to its north and east named sm1n , sm1 , and sm2 ( see andr , ward - thompson , & barsony 1993 ; awb93 ) .
wide - field 1.3 mm continuum maps by motte , andr , & neri ( 1998 ; man98 ) revealed oph a is the brightest of the oph cores , and a multiresolution wavelet analysis of its emission suggested the presence of eight other starless objects in oph a. the objects a - mm6 , a - mm7 , and a - mm8 , together with sm1n , sm1 , and sm2 , comprise an arc within the filamentary structure of oph a centered on the nearby b star s1 ( see figure 2a of man98 ) .
man98 also found the starless objects in the oph cores have a mass spectrum reminiscent of the stellar imf , suggesting a link exists between the formation of these objects and actual stars in clusters . in this paper , we examine emission from dense gas associated with the oph a core , using combined single - dish and interferometer data to attain a linear resolution of @xmath21000 au . we observed the 10 transition of the n@xmath0h@xmath1 molecular ion
since it is a tracer of dense gas ( e.g. , its critical density for excitation is @xmath7 = 2 @xmath8 10@xmath9 @xmath10 ) and n@xmath0h@xmath1 may deplete significantly only at densities @xmath1110@xmath9 @xmath10 ( see bergin & langer 1997 , bergin et al .
2002 ) . only two
of the five maxima of integrated line intensity found are coincident with peaks of millimeter continuum emission from dust . in addition , several peaks of dust emission are not coincident with maxima of integrated line intensity .
this situation is different from that seen in isolated cores where coincidence between dust and n@xmath0h@xmath1 emission is common .
the difference is likely due to variations in n@xmath0h@xmath1 abundance within oph a arising from differences in the relative density and dust temperature within each object .
focusing on the line emission alone , we find that its widths vary by a factor of @xmath25 in oph a , with several locations having fwhms @xmath4 0.3 km s@xmath5 .
the centroid velocities of the line emission vary little in oph a , however , having an rms of only @xmath20.17 km s@xmath5 .
small - scale centroid velocity gradients are found near the sm1 , sm1n , and sm2 objects .
the quiescent and slow motions of objects in oph a objects contrast sharply with the turblent and dynamic motions expected in some models of cluster formation .
one integrated line intensity maximum , oph a - n6 , has line widths as low as 0.193 km s@xmath5 fwhm , or just larger than the line width expected from thermal motions alone , indicating localized reduction of turbulent motions within oph a. details of our observations are described in 2 .
we present the results in 3 , including derivations of n@xmath0h@xmath1 column densities and h@xmath0 column densities , and descriptions of the widths , centroids , and profiles of the n@xmath0h@xmath1 10 line across oph a. a discussion of the implications of our data is found in 4 , and we summarize our conclusions in 5 .
observations of the oph a core at the 93.17 ghz frequency of n@xmath12h@xmath13 10 , consisting of 7 hyperfine components over 4.6 mhz ( see caselli , myers , & thaddeus 1995 ; cmt95 ) , were carried out with the iram 30 m telescope on pico veleta , spain .
data were obtained during three clear nights between 1998 june 25 and 1998 june 28 , in association with a comprehensive molecular line study of the cores and condensations of the ophiuchus protostellar cluster ( belloche et al .
2001 ; andr et al .
2004 , in preparation ) data were taken in the `` on - the - fly '' mapping mode ( ungerechts et al .
2000 ) , with an individual dump time of 2 seconds , a scanning speed of 1 s@xmath5 , and a spacing of 5 between consecutive scans .
the reference position was taken at the location of the b3 star s1 ( see figure 1 ) which had previously been seen to be free of emission in any dense gas tracer such as n@xmath0h@xmath1 10 .
the half - power beam width of the telescope was @xmath226 .
an sis heterodyne receiver was used with an autocorrelator as backend to give a spectral resolution of 19.5 khz over 17.5 mhz .
the corresponding spectral resolution was 0.06284 km s@xmath5 .
the observations were done in single sideband mode , with a sideband rejection of 0.003 and an associated calibration uncertainty of @xmath210% .
the forward and beam efficiencies of the telescope used to convert antenna temperatures @xmath14 into main beam temperatures @xmath15 were 92% and 73% , respectively .
the system temperature ranged from @xmath2170 k to @xmath2200 k on the @xmath14 scale .
the telescope pointing was checked every @xmath22 hours on the strong radio sources nrao 530 or 1514 - 241 and found to be accurate to better than @xmath25 ( i.e , the maximum deviation between two consecutive pointings ) .
the telescope focus , optimized on the same sources every @xmath23 - 4 hours , was very stable .
the data cube was produced and reduced with the class software package ( buisson et al .
2002 ) .
the oph a core was observed with the bima millimeter array located at hat creek , california , u.s.a .
, over 2 tracks in its c - configuration ( 1999 may 04 , 23 ) and 2 tracks in its b - configuration ( 2000 march 14 , 15 ) .
a total of 3 pointings was observed along the bright continuum filament associated with the a core detected by man98 .
the pointings were spaced by @xmath210 , slightly less than required for nyquist sampling , for a constant rms noise level along the filament .
each track consisted of cycled visits to the oph a pointings interspersed with observations of radio sources 1625 - 254 ( in 1999 ) or 1733 - 130 ( in 2000 ) approximately every 20 minutes for phase calibration .
flux calibration was obtained from observations of 3c 273 and neptune at the beginning and the end of each track respectively .
data reduction was performed using standard tasks in the miriad software package ( see sault , teuben & wright 1995 ) .
quasar data were self - calibrated using fluxes bootstrapped from neptune data , and the resulting amplitude and phase solutions were applied to the target pointings .
all 3 pointings were inverted simultaneously , and the resulting mosaics were deconvolved with a cleaning algorithm based on that presented by steer , dewdney , & ito ( 1984 ) .
the n@xmath0h@xmath1 10 data were convolved with a gaussian of 50 @xmath8 05 fwhm ( p.a . = 90 ) to improve the beam shape and increase brightness sensitivity .
the final synthesized beam size was 99 @xmath8 61 fwhm ( and p.a .
= -13.18 ) , corresponding to a linear fwhm resolution of 0.0060 pc @xmath8 0.0037 pc at the 125 pc distance to oph a. the bima and iram data of n@xmath0h@xmath1 10 were combined to recover missing flux from extended structure resolved out in the interferometer map .
the iram data were regridded in velocity from 0.06284 km s@xmath5 channels to 0.07854 km s@xmath5 channels to match the velocity resolution of the bima data .
the miriad task _ immerge _ was used for the combination , where the sum of the fourier transforms of the clean bima and iram images is itself transformed into the image plane as the combined image . to ensure a common flux scale for the independent datasets prior to combination ,
visibilities of both were compared over 29 k@xmath16 , the maximum range of spatial frequencies in common , and a derived scaling factor of 1.11 was applied to the iram data . comparing the total flux of the combined dataset convolved to a @xmath227 fwhm beam with that of the iram data , we estimate this method recovers 85% of the flux in the latter map over the region of interest .
figure 1a shows the 1.3 mm continuum emission from oph a detected by man98 using the iram 30 m telescope at 13 fwhm resolution .
figure 1b shows the integrated intensity of n@xmath0h@xmath1 10 line emission detected by us over the same region , using combined data from the iram 30 m telescope and the bima millimeter interferometer at @xmath210 @xmath8 @xmath26 fhwm resolution . the line emission from oph
a shows a filamentary structure similar to that shown by the continuum emission , although the width of the the integrated line emission structure is narrower than that of the continuum emission .
figure 1a shows positions of several objects in oph a identified by man98 from their multiresolution wavelet analysis .
figure 1b shows these positions again as well as the positions of the various maxima of integrated intensity of n@xmath0h@xmath1 10 , which we have named here oph a - n1 through n6 and whose positions are listed in table 1 .
some , but not complete , coincidence is seen between continuum objects and the maxima of integrated line intensity in oph a. for example , sm1 and sm1n are coincident ( within a beam fwhm ) of n4 and n5 respectively ( and possibly a - mm5 with n2 also ) . some oph
a continuum objects , however , have little - to - no detected n@xmath0h@xmath1 10 emission associated with them .
for example , relatively little line emission is found at the positions of vla 1623 , sm2 , a - mm6 , a - mm7 , and a - mm8 .
furthermore , some maxima of the n@xmath0h@xmath1 10 integrated intensity in oph a , although coincident with extended continuum emission , are not associated with continuum objects . for example , n1 , n3 , and n6 are not associated with objects specifically identified by man98 .
the lack of coincidence between peaks of dust emission and maxima of n@xmath0h@xmath1 emission in oph a is very different from the high level of coincidence between the two tracers seen in numerous isolated cores , albeit at lower resolution ( e.g. , tafalla et al .
2002 ) .
all 7 hyperfine components of the n@xmath0h@xmath1 10 line from the combined data were fit simultaneously using the hfs ( hyperfine structure ) fitting routine of the class data analysis package ( see buisson et al . ) for each 075 @xmath8 075 pixel of the data cube , these fits provided estimates of : 1 ) @xmath3 , the line fwhm , 2 ) @xmath6 , the centroid velocity , 3 ) @xmath17 , the sum of peak optical depths for all 7 hyperfine components , and 4 ) ( @xmath18@xmath19)@xmath17 , the difference between the ( rayleigh - jeans equivalent ) excitation and background temperatures times @xmath17 .
( in the latter , @xmath18 = ( @xmath20)/(exp(@xmath21)-1 ) , @xmath19 = ( @xmath20)/(exp(@xmath22)-1 ) , and @xmath23 and @xmath24 are the actual excitation and background temperatures respectively . ) the brightness temperature of the 10 line , @xmath25 , can be estimated from @xmath26 table 2 lists the mean , rms , minimum and maximum values of @xmath3 , @xmath6 , @xmath25 and @xmath17 from the combined data after averaging them into 105 @xmath8 675 pixels , i.e. , pixels with areas slightly larger than those of the synthesized beam .
most notably , @xmath27 in oph a varies widely but @xmath6 does not .
for example , @xmath3 varies by a factor of @xmath25 in oph a but the rms of @xmath6 is only 0.17 km s@xmath5 . in the following ,
we describe the n@xmath0h@xmath1 data in more detail , using the results obtained by hfs fitting to examine the n@xmath0h@xmath1 column densities and abundances as well as the widths and centroid velocities of the n@xmath0h@xmath1 10 line .
column densities of h@xmath0 , i.e. , @xmath28(h@xmath0 ) , can be determined in oph a with the 1.3 mm continuum data of man98 ( e.g. , see figure 1a ) from @xmath29 where @xmath30 is the 1.3 mm flux density , @xmath31 is the main - beam solid angle , @xmath32 is the mean molecular weight , @xmath33 is the mass of atomic hydrogen , @xmath34 is the dust opacity per unit mass at 1.3 mm ( assumed to be 0.005 @xmath35 g@xmath5 for starless objects by man98 ; the local gas - to - dust ratio is assumed to be 100 ) , and b@xmath36(@xmath37 ) is the planck function at t = @xmath37 . figure 2a shows @xmath28(h@xmath0 ) in oph
a obtained by assuming @xmath37 = 20 k at all locations ( following man98 ) . to facilitate comparision with the n@xmath0h@xmath1 data ,
the area shown is restricted to that where the peak brightness temperature of the brightest hyperfine component , @xmath38 @xmath39 3 @xmath40 in the n@xmath0h@xmath1 10 line data .
furthermore , the man98 continuum data were binned into pixels 105 @xmath8 675 in size using standard image interpolation methods in miriad to obtain pixels with areas approximately that of the beam of the n@xmath0h@xmath1 observations .
( note that each pixel in figure 2a is still not spatially independent of its neighbors . )
figure 2a shows the @xmath28(h@xmath0 ) distribution across oph a is similar to that of the 1.3 mm continuum emission shown in figure 1a , due of course to the linear proportionality between the two quantities in equation 2 if @xmath37 is constant .
table 3 lists the mean , rms , minimum , and maximum of @xmath28(h@xmath0 ) shown in figure 2a .
column densities of n@xmath0h@xmath1 , i.e. , @xmath28(n@xmath0h@xmath1 ) , can be determined along the oph a filament from the line characteristics obtained from hfs fitting ( see 3 ) .
assuming lte , i.e. , all level populations are described by a single excitation temperature , @xmath23 , @xmath28(n@xmath0h@xmath1 ) can be found following womack , ziurys , & wyckoff ( 1992 ) from @xmath41 where @xmath3 and @xmath17 are found from hfs fits to data averaged to 105 @xmath8 675 pixels .
in addition , @xmath42 is the lower rotational level number ( i.e. , 0 ) , @xmath43 is the dipole moment of the n@xmath0h@xmath1 molecular ion ( i.e. , 3.40 debye ; green , montgomery & thaddeus 1974 ) , @xmath44 is the rotational constant of n@xmath0h@xmath1 ( i.e. , 46.586702 ghz ; see cmt95 ) and @xmath45 is the energy above ground of the lower rotational level ( i.e. , 0 ) .
figure 2b shows @xmath28(n@xmath0h@xmath1 ) in oph a obtained using the results of our hfs fits and equation 3.80% of beam - sized pixels had derived values of @xmath23 between 6 k and 24 k with a peak at @xmath215 k , but the remaining @xmath220% had values extending up to 260 k. these latter values were found typically if @xmath17 @xmath4 0.2 ; high values of @xmath23 are obtained in such cases because only the product of @xmath18 and @xmath17 is constrained . in these instances ,
the derived @xmath23 values were deemed spurious , and @xmath28(n@xmath0h@xmath1 ) values were calculated using equation 3 after substituting @xmath25 for the factor ( @xmath23 + @xmath46)@xmath17 and assuming in the exponential factors @xmath23 = 14.9 k , the mean @xmath23 of the other @xmath280% of pixels . ] figure 2b shows the @xmath28(n@xmath0h@xmath1 ) distribution across oph a is not similar to the @xmath28(h@xmath0 ) distribution , but rather is similar to that of the n@xmath0h@xmath1 integrated intensities shown in figure 1b .
table 3 lists the mean , rms , minimum , and maximum of @xmath28(n@xmath0h@xmath1 ) shown in figure 2b .
we note that assuming lte in oph a can result in overestimates of @xmath28(n@xmath0h@xmath1 ) from n@xmath0h@xmath1 10 emission toward locations with densities @xmath4 10@xmath47 @xmath10 ( e.g. , by factors of @xmath22 where densities are @xmath210@xmath48 @xmath10 ; e. bergin , private communication . ) as discussed in 4.1 , however , mean densities are @xmath4910@xmath47 @xmath10 towards positions of bright n@xmath0h@xmath1 10 emission in oph a. figure 2c shows the n@xmath0h@xmath1 fractional abundance across oph a , determined using values from figures 3a and 3b ( i.e. , @xmath50(n@xmath0h@xmath1 ) = @xmath28(n@xmath0h@xmath1)/@xmath28(h@xmath0 ) ) .
the mean @xmath50(n@xmath0h@xmath1 ) derived , 1.9 @xmath8 10@xmath51 , is similar to values derived for other dark clouds , e.g. , @xmath50(n@xmath0h@xmath1 ) @xmath52 10@xmath51 has been found in various cores by caselli et al .
( 2002a ) , tafalla et al .
( 2002 ) , benson , caselli , & myers ( 1998 ) , womack et al .
( 1992 ) , and ohishi , irvine , & kaifu ( 1992 ) . within oph a itself , figure 2c
shows that @xmath50(n@xmath0h@xmath1 ) varies by a factor of @xmath240 , with lower values at the edges of the filament and near the continuum objects like sm2 , and higher values at positions at the extreme northern and southern ends of the filament .
table 3 lists the mean , rms , minimum , and maximum of @xmath50(n@xmath0h@xmath1 ) shown in figure 2c .
assuming lte , our accuracy in determining @xmath50(n@xmath0h@xmath1 ) is limited by our knowledge of @xmath37 within oph a , may vary in oph a by factors of @xmath22 or more due to grain evolution ( see man98 and references therein ) . in this paper , we assume constant @xmath34 since we are considering similar objects where differences in dust opacities are likely minimal . ] .
indeed , isothermal dust is arguably not expected in oph a given the number of nearby heating sources ( i.e. , embedded cluster members ) and the wide range of column densities ( i.e. , extinctions ) present . for the continuum objects ,
dust temperatures determined independently can be used to determine more refined , specific values of @xmath50(n@xmath0h@xmath1 ) .
table 4 lists values of @xmath37 for continuum objects in oph a by awb93 , determined by modeling their submillimeter spectral energy distributions .
( awb93 found dust emissivities of @xmath34 @xmath53 @xmath54 , i.e. , @xmath55 @xmath52 0.01 @xmath35 g@xmath5 , best fit their data . ) awb93 speculated these differences may be due to radiative heating by nearby sources to the northeast of sm1 , e.g. , the nearby b3 star s1 .
table 4 also lists values of @xmath50(n@xmath0h@xmath1 ) , @xmath28(n@xmath0h@xmath1 ) , and @xmath28(h@xmath0 ) for vla 1623 , sm1n , sm1 , and sm2 obtained by assuming the @xmath37 values of awb93 in equations 2 and 3 and averaging the resulting values over 3 - 4 beam - sized pixels coincident with each object .
( for vla 1623 , upper limits for @xmath28(n@xmath0h@xmath1 ) and @xmath50(n@xmath0h@xmath1 ) are given in table 4 since it is located where @xmath38 @xmath4 3 @xmath40 . ) the difference in @xmath50(n@xmath0h@xmath1 ) from one source to the next is smaller when constant @xmath37 is used .
n@xmath0h@xmath1 abundances are similar for sm1n and sm1 while those of sm2 and vla 1623 are lower by factors @xmath494 .
( oph a - n6 , also in table 4 , will be discussed in 3.2 . ) for locations in oph a beyond the continuum objects , specific values of @xmath37 have not been defined , e.g. , awb93 did not derive @xmath37 values there
. maps of submillimeter continuum spectral indicies by awb93 , however , indicate that a @xmath37 gradient exists across oph a , with lower @xmath37 ( i.e. , @xmath56 15 k ) to the south or southwest ( near vla 1623 and sm2 ) and higher @xmath37 ( i.e. , @xmath39 25 k ) to the north or northeast ( near sm1 and sm1n ) .
unfortunately , the awb93 submillimeter ( e.g. , 350 ) continuum maps do not cover well the extreme northern and southern ends of the filament where we find the highest values of @xmath50(n@xmath0h@xmath1 ) , i.e. , assuming @xmath37 = 20 k. without specific estimates for @xmath37 , it is difficult to refine our estimates of @xmath50(n@xmath0h@xmath1 ) at the ends of the filament .
we note , however , that @xmath50(n@xmath0h@xmath1 ) at these locations would be quite similar to those of sm1 and sm1n if @xmath37 @xmath52 1012 k , quite significantly reducing the abundance contrast shown by figure 2c . at the very least , our observations suggest differences in @xmath50(n@xmath0h@xmath1 ) for the continuum objects in oph a by factors @xmath394 . put simply , variations of both @xmath50(n@xmath0h@xmath1 ) and @xmath37 appear necessary to reconcile the 1.3 mm continuum and n@xmath0h@xmath1 line data of oph a. figure 3a shows a map of @xmath3 obtained from the hfs fits of the combined n@xmath0h@xmath1 data of oph a. the data shown are only those where @xmath38 @xmath39 3 @xmath40 .
the large variation of @xmath27 in oph a by a factor of @xmath25 differs from the relatively little variation of line widths ( i.e. , coherence " ) found over similar @xmath20.1 pc length scales in isolated , quiescent cores by goodman et al .
table 4 lists the @xmath3 of sm1n , sm1 , and sm2 , obtained by averaging over 4 beam - sized pixels surrounding each object .
these continuum objects are coincident with locations in oph a with relatively broad n@xmath0h@xmath1 lines , i.e. , @xmath57 0.50.6 km s@xmath5 .
we note also that the largest line widths are found adjacent to ( but not coincident with ) these continuum objects .
figure 3a shows fourteen locations in oph a with @xmath3 @xmath56 0.30 km s@xmath5 , which we arbitrarily define as very narrow . "
table 5 lists these positions , named here as locations a `` through ' ' n " , and gives their mean line width within the 0.3 km s@xmath5 contour .
no very narrow line width location is coincident with an identified continuum object , although location c " , near a - mm5 , may be an exception .
eleven very narrow line width locations are at the edges of the detected filament , where line peaks are relatively weak and near our detection limit ( i.e. , @xmath58 @xmath40 ) and where hfs fits are relatively poor ( e.g. , typical @xmath3 uncertainties are @xmath20.02 - 0.05 km s@xmath5 . )
conversely , only the b " , f , and l " locations are well - defined local minima of @xmath3 ; each is surrounded by larger @xmath3 and has line peaks well above our detection limit ( e.g. , typical @xmath3 uncertainties are @xmath20.005 - 0.01 km s@xmath5 ) .
the extents of most very narrow line locations are therefore not well defined spatially since narrow line emission below our detection limit may or may not reside adjacent to each .
finally , most narrow line locations have detected extents smaller than the beam .
only the c " , d " , l " , and
m " locations have detected extents larger than the beam . the very narrow line location
l " is especially notable in oph a given its resolved extent , the relative brightness and the narrowness of its n@xmath0h@xmath1 10 line emission , and its association with a maximum of integrated intensity ( i.e. , oph a - n6 ) .
the b " and f " locations are like l " in that they are also associated with bright emission ( i.e. , oph a - n1 ) , but unlike l " they are each spatially unresolved and their lines are slightly broader than those of
the c " , d " , and
m " locations are like l " in that they are each spatially resolved and their lines are similarly as narrow than those of
l " ( or less ) , but unlike l " they are not spatially well - defined and have weaker emission ( i.e. , no associations with integrated intensity maxima ) .
these limitations make the b " , c " , d " , f " , and m " locations more difficult to characterize than l " , so we concentrate on l " in the following discussion .
( the conclusions we reach on l " may also apply to these other very narrow line positions . )
since l " includes oph a - n6 we henceforth refer to this object " or location within oph a simply as oph a - n6 or n6 for short .
table 4 lists values of @xmath50(n@xmath0h@xmath1 ) , etc . , for n6 found by averaging values over 5 beam - sized pixels coincident with positions where @xmath3 @xmath56 0.30 km s@xmath5 , and assuming @xmath37 = 15 k at that location ( see 3.1 ) .
although the position of minimum line width in n6 is @xmath219 from its position of maximum integrated intensity , line emission is quite narrow throughout n6 ; when averaged over 5 adjacent beam - sized pixels , the @xmath3 of n6 is 0.25 km s@xmath5 ( see table 4 ) .
the minimum line width in oph a - n6 is remarkably small , only 0.193 @xmath59 0.008 km s@xmath5 .
figure 4 shows the spectrum of n@xmath0h@xmath1 10 at the location of minimum line width in oph a - n6 , along with its hfs fit .
all 7 hyperfine components of n@xmath0h@xmath1 10 are unblended and clearly discernable at this position .
given the finite 0.07854 km s@xmath5 velocity resolution , the actual line widths are slightly less than the values shown in figure 3a , depending on the line width .
for example , the minimum @xmath27 in oph a - n6 reduces to only 0.176 km s@xmath5 when the velocity resolution is subtracted in quadrature .
such narrow line widths are actually comparable to the thermal fwhms expected for n@xmath0h@xmath1 given expected gas temperatures , @xmath60 , within oph a. for example , wootten et al .
( 2004 , in preparation ) using vla nh@xmath61 data have determined gas temperatures for the oph a continuum objects that are similar to the @xmath62 = 20 k assumed by man98 , i.e. , 16 k for sm1 , 21 k for sm1n , and 18 k for sm2 .
other estimates of @xmath60 in oph a can be made using the values of @xmath23 found from hfs fits of the n@xmath0h@xmath1 10 line at positions where @xmath17 @xmath63 1 , assuming the transition is thermalized , i.e. , the local density is well above its 2 @xmath8 10@xmath9 @xmath10 critical density .
( below we show that this is reasonable within oph a - n6 , given its size and mass . ) for the minimum @xmath3 spectrum shown in figure 4 , values of @xmath23 = 18.1 k @xmath59 3.4 k and @xmath17 = 12.0 @xmath59 2.2 were found , suggesting @xmath60 @xmath52 18 k. within oph a - n6 where @xmath17 @xmath39 11.57 ( i.e. , where the optical depth of the brightest hyperfine component , 123 - 012 , is at least 3 ) , the mean @xmath23 is 17.4 k. if @xmath60 = 18.1 k at the position of the spectrum shown in figure 4 , the expected thermal fhwm is 0.169 km s@xmath5 .
the fwhm of the nonthermal component to @xmath3 at this position , @xmath64 , obtained by subtracting in quadrature the thermal fwhm , @xmath65 , from the resolution - corrected fwhm , is only 0.0500 km s@xmath5 .
although the spectrum shown in figure 4 is the narrowest in oph a - n6 , the gas velocities there are all remarkably dominated by thermal motions , with relatively little nonthermal contributions .
the occurrence of near - thermal line widths in oph a is significant because the ophiuchus star - forming clouds are turbulent on larger spatial scales and at lower densities ( see loren 1989a ) .
near - thermal line widths have been observed previously towards isolated cores ; e.g. , see jijina , myers , & adams ( 1999 ) for a detailed list based on nh@xmath61 observations and benson et al .
( 1998 ) and caselli et al .
( 2002a ) for n@xmath0h@xmath1 observations .
oph a - n6 is similar to the quiescent regions in serpens nw identified using n@xmath0h@xmath1 10 by williams & myers ( 2000 ) .
for example , the average @xmath66 for oph a - n6 , assuming @xmath60 = 18 k , is 0.17 km s@xmath5 , similar to objects at the lower end of the @xmath66 range reported by williams & myers .
there are , however , some important differences between oph a - n6 and the quiescent serpens objects . for example , the minimum @xmath64 of oph a - n6 is smaller than those reported for the serpens objects by a factor of @xmath23 .
furthermore , oph a - n6 contains an obvious maximum of n@xmath0h@xmath1 10 integrated intensity , whereas the serpens objects are not .
other instances of n@xmath0h@xmath1 10 as narrow as 0.2 km s@xmath5 were detected within ngc 1333 iras 4 by di francesco et al .
2001 ( see also walsh et al .
2004 , in preparation , for other locations in ngc 1333 . )
in contrast to oph a - n6 , @xmath66 = 0.46 - 0.57 km s@xmath5 for sm1n , sm1 , and sm2 , given their respective values of @xmath3 listed in table 4 and the @xmath60 values found by wootten et al . within ophiuchus , extremely narrow lines such as those described here have been uncommonly observed , possibly due to the relatively large beam sizes of earlier studies . for example , loren ( 1989b ) found @xmath3 = 1.7 km s@xmath5 for @xmath67co 10 in the general oph a region ( r22 ) when observed with a 2.4 beam ( although the line was considered saturated . ) in addition , tachihara , mizuno , & fukui ( 2000 ) found @xmath3 = 1.5 km s@xmath5 for the c@xmath68o 10 core in the neighbourhood of oph a ( @xmath69 oph 2 ) .
furthermore , lww90 found @xmath3 = 0.8 km s@xmath5 and 0.7 km s@xmath5 for dco@xmath1 32 at the core center " of oph a ( defined then as @xmath210 west of sm1n ) when observed with a 12 fwhm beam and a 05 fwhm beam respectively .
belloche et al .
( 2001 ) , however , detected relatively narrow n@xmath0h@xmath1 10 lines at various locations within oph , e.g. , @xmath3 = 0.26 km s@xmath5 toward oph e - mm2 , in iram 30 m telescope data alone ( obtained with a 045 fwhm beam . ) our iram 30 m telescope data is a subset of the belloche et al .
data and across oph a - n6 they show no evidence of the narrower lines seen in the combined data , i.e , @xmath3 = 0.35 km s@xmath5 .
our high angular resolution data have revealed locations of extremely narrow line emission in oph a. figure 3b shows a map of @xmath6 , the central velocity of the n@xmath0h@xmath1 10 line , in oph a obtained from hfs fits to the combined data . immediately striking in this figure is the narrow range of @xmath6 values within oph a ; the rms of @xmath6 values is only 0.17 km s@xmath5 ( see table 1 ) .
figure 3b reveals no large - scale gradient of @xmath6 along oph a but small - scale gradients are possibly found near the continuum objects sm1n , sm1 , and sm2 , roughly aligned east - west or west - east .
other possible gradients include one north of vla 1623 aligned north - south , and one northwest of sm1n ( and oph a - n1 ) aligned northwest - southeast .
in contrast , no @xmath6 gradient is found toward oph a - n6 .
( as noted earlier , weak emission in its periphery are at slightly lower @xmath6 . )
figure 5 shows a spectral grid of n@xmath0h@xmath1 101012 , the isolated component " of n@xmath0h@xmath1 across oph a , where each panel shows the line averaged over a ( beam - sized ) 105 @xmath8 675 area .
the variation of @xmath3 and lack of variation of @xmath6 across oph a are easily apparent in this figure , but the line shapes are more evident here than possible from maps of hfs fit parameters .
figure 5 reveals that the n@xmath0h@xmath1 line profiles in oph a are typically symmetric in most locations , with little evidence for self - reversal or asymmetry .
this appearance is consistent with the relatively low optical depths of the lines here as obtained from hfs fits ( see figure 2b ) . only at a position just west of sm1 is a single self - reversal found , and perhaps not coincidentally this location is also that of maximum @xmath28(n@xmath0h@xmath1 ) in oph a as derived from the hfs fits .
the line shapes reveal little evidence for the asymmetrically - blue line profiles that suggest infall motions , such as those seen in n@xmath0h@xmath1 10 towards l1544 by williams et al .
( 1998 ) , likely also due to the relatively low optical depths of the line in oph a. only one possible instance of blue asymmetry is found in oph a , immediately southwest of sm1 .
this spectrum , however , is adjacent to an instance of red asymmetry , and both asymmetries may be consistent with having been made from noise across the line .
the line shapes also show no evidence for broad line wings expected from high velocity outflows .
surprisingly , the line velocities of n@xmath0h@xmath1 10 appear correlated with its line width , but only for the brightest lines in oph a , i.e. towards the highest n@xmath0h@xmath1 column densities .
figure 6a shows that higher velocity ( redder ) lines are systematically wider in lines where @xmath38 @xmath39 10 @xmath40 .
a linear fit to the data shown in figure 6a reveals a fairly high correlation parameter @xmath70 of 0.82 ( see press et al .
figure 6b shows , however , that a similar correlation between @xmath6 and @xmath3 is less likely for lines with @xmath38 @xmath4 10 @xmath40 ; the @xmath70 obtained from a linear fit to this subsample is only 0.31 .
figure 7 shows evidence that the line widths are increasing because only their red sides are moving to higher velocities and their blue sides are not shifting in velocity .
three pairs of 2 spectra of the n@xmath0h@xmath1 101 - 012 line are shown , taken from different positions in oph a , where each pair consists of an average over 2 - 4 neighboring pixels with bright narrow emission in oph a and an average over 4 - 6 surrounding pixels with fainter , broader emission .
this figure suggests that the trend shown in figure 6a is not simply a matter of the widths of bright lines increasing as the entire line shifts to higher velocities . a similar trend
is suggested by n@xmath0h@xmath1 10 line data of the quiescent region q8 of serpens ( williams & myers ) and ngc 1333 iras 4b in perseus ( di francesco et al . )
our combined single - dish and interferometer data have revealed interesting variations of n@xmath0h@xmath1 abundance and the width and centroid velocity of the n@xmath0h@xmath1 10 line within the oph a filament .
together , these variations provide a qualitative picture of the protostellar evolution of gas and dust in a relatively dense star - forming core , assuming that the differences are due solely to evolution . in this picture , the object " oph a - n6 represents a relatively early , quiescent stage marked by high density , low @xmath37 , high @xmath50(n@xmath0h@xmath1 ) , low @xmath3 , and little local variation of @xmath6 .
sm2 , however , represents a more - evolved protostellar stage with higher density , low @xmath37 , low @xmath50(n@xmath0h@xmath1 ) , medium @xmath3 , and some local variation of @xmath6 .
sm1n and sm1 may be at a similar protostellar stage as sm2 given their similar @xmath3 and variation of @xmath6 , but their higher @xmath50(n@xmath0h@xmath1 ) may reflect their higher @xmath37 . finally , vla 1623 , which has just formed a protostar ( awb93 ) , is representative of the next stage of evolution , namely the beginning of the protostellar phase .
it has high density , low @xmath37 , and very low @xmath50(n@xmath0h@xmath1 ) . in the following ,
we describe in more detail how the differences in @xmath50(n@xmath0h@xmath1 ) , @xmath3 , and @xmath6 between the oph a objects result in this qualitative picture .
variations of n@xmath0h@xmath1 abundance in oph a point to subtle differences in the evolutionary stages of its starless objects due to the expected behavior of @xmath50(n@xmath0h@xmath1 ) within cold , dense environments .
although @xmath50(n@xmath0h@xmath1 ) can be reduced locally as a result of outflows , e.g. , when h@xmath0o and co are returned to the gas phase by shocks and quickly deplete n@xmath0h@xmath1 via ion - molecule reactions ( see bergin , neufeld , & melnick 1998 ; van dishoeck & hogerheijde 1999 ) , these objects do not contain or are not impacted by any outflowing gas ( kamazaki et al .
2001 , 2003 ) .
lee et al .
( 2003 ) have suggested molecular abundance differences between starless objects can be related also to the relative ratio of their dynamical and chemical time scales ( i.e. , their respective environments ) , but the oph a objects should have this factor in common since they are situated within the same core throughout oph a. ]
. gas - grain chemistry at high densities and low temperatures is a very plausible cause of these variations .
examples of the effect of gas - grain chemistry on @xmath50(n@xmath0h@xmath1 ) in cold , dense environments were recently provided by aikawa , ohashi , & herbst ( 2003 ; aoh03 ) .
aoh03 found from isothermal ( i.e. , @xmath71 = 10 k ) models of larson - penston collapsing starless objects that central @xmath50(n@xmath0h@xmath1 ) increases when central densities increase up to @xmath23 @xmath8 10@xmath47 @xmath10 ( e.g. , that expected for l1544 ) , as the principal destroyers of n@xmath0h@xmath1 like co become strongly depleted onto dust grains ( see also bergin & langer , shematovich et al .
at densities @xmath52 3 @xmath8 10@xmath72 @xmath10 , however , aoh03 found that @xmath50(n@xmath0h@xmath1 ) _ decreases _ by a factor of @xmath22 from its maximum within a @xmath21000 au radius as n@xmath0 , from which n@xmath0h@xmath1 forms , is adsorbed onto dust grains .
although not explored explicitly by aoh03 , further n@xmath0h@xmath1 depletion may be expected at still higher central densities .
observational evidence for central n@xmath0h@xmath1 depletions have been reported recently in isolated starless objects with central densities of @xmath210@xmath73 @xmath10 , e.g. , b68 ( bergin et al . ) , l1544 ( caselli et al . 2003 ) , and iram 04191 + 1522 ( belloche & andre 2004 , in preparation ) .
the models of aoh03 imply that depletion of n@xmath0h@xmath1 can be hard to observe directly with large beams .
for example , no central minimum in @xmath28(n@xmath0h@xmath1 ) can be seen when their model core is observed " with a beam of @xmath24000 au equivalent fwhm ( see their figure 9 ) . given their small extent , such n@xmath0h@xmath1 depletion zones would be easier to detect directly with smaller beams .
for example , our combined n@xmath0h@xmath1 data have a geometrical mean linear resolution of 970 au at the 125 pc distance of oph a , or about half the diameter of the central n@xmath0h@xmath1 depletion zone in the aoh03 models .
the variations of @xmath50(n@xmath0h@xmath1 ) between the oph a objects we observe could be then explained as the result of detecting differing amounts of depletion due to density differences between objects .
man98 determined mean densities for sm1n , sm1 , and sm2 from their 1.3 mm continuum observations ranging from @xmath25 @xmath8 10@xmath47 @xmath10 ( sm2 ) to @xmath24 @xmath8 10@xmath72 @xmath10 ( sm1n ) , i.e. , in the regime where central n@xmath0h@xmath1 depletions were found by aoh03 .
note , however , that the aoh03 models had constant @xmath71 = 10 k , so the possible effect on depletion that warmer @xmath37 may have must also be considered in an intepretation .
table 4 lists @xmath50(n@xmath0h@xmath1 ) , @xmath28(n@xmath0h@xmath1 ) , and @xmath28(h@xmath0 ) for oph a - n6 in addition to those for vla 1623 , sm1n , sm1 , and sm2 , also found using equations 2 and 3 and averaging over coincident beam - sized pixels .
for n6 , we assume @xmath37 = 15 k following awb93 , although they suggested that @xmath37 @xmath56 15 k southeast of sm2 .
the @xmath28(n@xmath0h@xmath1 ) of n6 , 6 @xmath8 10@xmath67 @xmath74 , lies between the higher values of sm1 and sm1n and the lower values of sm2 and vla 1623 . on the other hand , the @xmath28(h@xmath0 ) of n6 ,
@xmath23 @xmath8 10@xmath75 @xmath74 , is actually similar to those of sm1n and sm1 , but appears smaller than those of vla 1623 and sm2 .
relative to oph a - n6 , vla 1623 and sm2 have likely higher mean densities , given their apparently larger @xmath28(h@xmath0 ) and more compact appearance as 1.3 mm continuum maxima . in the context of the aoh03 models ,
their low n@xmath0h@xmath1 column densities are due to n@xmath0h@xmath1 depletion at high density . since vla 1623 has formed a protostar ( or two ; see looney , mundy , & welch 2000 ) but sm2 not formed any , the mean density of vla 1623 should be even larger than that of sm2 .
the weaker n@xmath0h@xmath1 emission of vla 1623 compared to sm2 likely reflects even greater n@xmath0h@xmath1 depletion that should accompany higher density .
( in addition , n@xmath0h@xmath1 depletion toward vla 1623 may be enhanced if grain mantles containing co have returned to the gas phase through evaporation or outflow shocks . )
sm1n and sm1 do not fit easily into this sequence , however , as they have higher @xmath28(n@xmath0h@xmath1 ) but similar @xmath28(h@xmath0 ) .
their higher @xmath28(n@xmath0h@xmath1 ) may reflect their warmer @xmath37 relative to sm2 if n@xmath0 does not easily adsorb onto their warmer dust grains .
co , however , is likely also not adsorbed readily at these @xmath37 values , and its presence in the gas phase likely would reduce @xmath28(n@xmath0h@xmath1 ) rather than increase it .
the relatively high @xmath28(n@xmath0h@xmath1 ) for sm1n and sm1 may be due to internal , dense regions where co has been depleted but not n@xmath0 , i.e. , that are cooler than suggested by their ( average ) @xmath37 but warmer than regions of similar density in sm2 .
since density must increase as a result of evolution towards star formation and @xmath50(n@xmath0h@xmath1 ) appears to decrease above @xmath210@xmath47 @xmath10 , @xmath50(n@xmath0h@xmath1 ) may be useful as an indicator of the relative evolution of several objects forming within the same dense core .
our @xmath50(n@xmath0h@xmath1 ) results alone therefore suggest that oph a - n6 is at an earlier stage relative to sm2 and that sm2 is at an earlier stage relative to vla 1623 .
given the warmer @xmath37 of sm1n and sm1 relative to sm2 , their evolutionary stage relative to n6 , sm2 , and vla 1623 is difficult to judge .
we stress that these results are highly qualitative .
evans et al .
( 2001 ) and zucconi , walmsley , & galli ( 2001 ) have shown recently that bonnor - ebert spheres with radial temperature gradients ( e.g. , with low central @xmath37 ) are better than isothermal models in reproducing the observed dust emission of isolated starless objects , and the @xmath37 values derived by awb93 and used here are likely averages of the actual dust temperature distributions within the oph a objects .
more detailed modeling of the dust continuum emission of all these objects is clearly required to determine better their thermal and density structures and provide more quantitative results , especially if n@xmath0 depletion onto dust grains is strongly dependent on @xmath37 .
variations of the @xmath3 and @xmath6 of n@xmath0h@xmath1 10 in oph a allow us to probe the dynamics of dense gas within the core . complementing the previous discussion on n@xmath0h@xmath1 abundances , the dynamics of dense gas
provide further insight into the relative evolution of objects within oph a. in the following , we consider separately the dynamics of oph a as a whole and of its population of starless objects , as well as those of oph a - n6 relative to sm1n , sm1 , and sm2 .
the effective average velocity dispersion of gas , @xmath40 , within oph a can be estimated from our data and compared to simple models to evaluate the dynamical state of the core .
following bertoldi & mckee ( 1992 ; appendix c ) , @xmath40 can be estimated in several ways .
first , the dispersion of an average spectrum comprised of all line data across the core can be measured and @xmath76 is approximately equal to the quadrature sum of this quantity and the thermal dispersion , i.e. , that of velocities of mean molecular mass expected from thermal motions .
the fwhm resulting from a gaussian fit to such an average spectrum from our oph a data is 0.58 km s@xmath5 , yielding @xmath40 = 0.36 km s@xmath5 if @xmath71 = 20 k. second , the rms of both @xmath3 and @xmath6 can be measured and @xmath76 is approximately equal to the sum of the squares of these quantities and the thermal dispersion . using values from table 1 and again assuming @xmath71 = 20 k , @xmath40 = 0.33 km s@xmath5 , i.e. , within 10% of the first value .
the estimated @xmath40 of oph a is similar to but less than expected from simple virial arguments , i.e. , those that ignore contributions from surface pressure and magnetic fields @xmath56 10 @xmath32 g was found in oph from oh zeeman measurements by troland et al .
( 1996 ) . ] .
for example , the average total one - dimensional velocity dispersion of a virialized core can be estimated by equating 2@xmath62 = 2(3/2)@xmath77 , the magnitude of the total core kinetic energy ( @xmath78 is the total core mass ) , and @xmath79 = @xmath80 , the magnitude of the total core gravitational potential energy ( @xmath81 is the average projected core size , @xmath82 is the gravitational constant , and @xmath83 is a dimensionless parameter accounting for core asphericity and nonuniformity ; see bertoldi & mckee ) .
for oph a , @xmath78 @xmath52 11 m , assuming @xmath37 = 20 k on average and using the 1.3 mm continuum data of man98 over positions with detected n@xmath0h@xmath1 10 emission .
we find @xmath40 = 0.57 km s@xmath5 , assuming this mass and uniformity in density .
( an aspect ratio of @xmath27 and a projected semiminor axis of @xmath20.01 pc are also assumed , yielding @xmath81 = 0.023 pc and @xmath83 = 0.78 ; see fig . 2 of bertoldi & mckee .
) these velocities suggest oph a as a whole is in gravitational virial equilibrium to first order , i.e. , they are within a factor of 2 of those required .
( although these velocities suggest oph a has insufficient support against gravity in the absence of significant magnetic energy , the difference is likely not significant given the uncertainties in our analysis . )
awb93 also concluded the oph a filament was similarly close to gravitational virial equilibrium , based on their submillimeter continuum data and a typical @xmath3 from available line data .
more recently , tachihara et al . found similar conclusions for ( lower density ) star forming c@xmath68o cores in the near - neighborhood of oph a. better determinations of the local magnetic field strengths are clearly needed to evaluate better the global dynamical state of oph a. variations of @xmath6 in the population of objects within oph a provide constraints on some models of clustered star formation .
table 4 lists the @xmath6 of sm1n , sm1 , sm2 , and oph a - n6 , obtained by averaging 4 - 5 beam - sized pixels around each object .
the rms of these values is only 0.10 km s@xmath5 , or @xmath20.4 @xmath84(20 k ) , the isothermal sound speed of molecular gas at @xmath60 = 20 k. the similarities between the @xmath6 of gas associated with the continuum objects and oph a - n6 suggest strongly that all these lie within the same filament , and are not objects in separate cores seen by coincidence near each other towards ophiuchus .
furthermore , the objects do not have substantial radial velocities relative to their neighboring dense gas .
for example , the mean difference in the @xmath6 of a given oph a object and the mean of its surrounding gas is only 0.06 km s@xmath5 , much smaller than the rms of @xmath6 .
this value is only @xmath20.25 @xmath84(20 k ) , i.e. , at the low end of expected mean relative gas velocities in cluster formation models where starless objects move ballistically within a core and accrete mass competitively ( e.g. , see bonnell et al .
1997 , 2001 ) .
the oph a objects , however , are few in number , are found in close proximity , and as a whole may be at a relatively early , slow stage of such an evolution ( e.g. , when gas dominates the gravitational potential ) .
variations of @xmath3 between starless objects in oph a may be also related to these objects being at different stages in the evolution of dense gas towards star formation .
here we consider oph a - n6 as an object , to place its physical characteristics in the context of models relevant to these epochs .
unlike sm1 , sm1n , and sm2 , n6 is not a local 2d maximum of continuum emission ( i.e. , @xmath28(h@xmath0 ) ) , but instead resides on a gradient of continuum emission along the local northwest - southeast direction of the filamentary n@xmath0h@xmath1 10 emission .
note , however , that n6 is at least at a 1d maximum of continuum emission in the direction perpendicular to the filament .
given its 125 pc distance and following our arbitrary definition of its extent , i.e. , where @xmath3 @xmath56 0.30 km s@xmath5 , oph a - n6 has a ( nondeconvolved ) geometrical mean radius of 0.008 pc . from its 1.3 mm continuum emission ( see fig .
1a ) and assuming its dust is isothermal at @xmath37 = 15 k , n6 has a mean column density of 2.5 @xmath8 10@xmath75 @xmath74 and a mass of 0.9 m. ( assuming n6 is a sphere , its mean density is 7.6 @xmath8 10@xmath47 @xmath74 , well above the 2.5 @xmath8 10@xmath9 critical density of the n@xmath0h@xmath1 10 line . ) in addition , n6 has a mean @xmath3 = 0.25 km s@xmath5 ( see table 4 ) , suggesting a mean dispersion for the n@xmath0h@xmath1 10 line of 0.11 km s@xmath5 .
assuming the gas of n6 is isothermal at @xmath60 = 17.4 k ( see 3 ) , the mean nonthermal component @xmath85 = 0.080 km s@xmath5 and @xmath40 = 0.26 km s@xmath5 .
the observed quantities of oph a - n6 are within a factor of 2 of those predicted from simple models of equilibrium spheres .
for example , the mass - radius ratio of a uniform sphere in gravitational virial equilibrium is @xmath86 and that of a critically - stable bonnor - ebert sphere is @xmath87 ( bonnor 1956 ) . given the effective radius and mass of n6 , the mean velocity dispersions predicted from these simple models are 0.31 km s@xmath5 and 0.45 km s@xmath5 respectively , i.e. , more than that of n6 itself by factors of 1.2 and 1.7 respectively .
like oph a as a whole , n6 is in gravitational virial equilibrium to first order .
( again , the differences are likely not significant enough to suggest n6 has insufficent support against gravity . )
interestingly , the mass and radius of n6 can be reproduced simultaneously if the mass - radius ratio @xmath52 @xmath88 , a situation that could occur , e.g. , in virial equilibrium if magnetic energy @xmath44
@xmath52 @xmath89 , or more if surface pressure is also non - zero .
there are three questions about oph a - n6 that simple equilibrium models do not directly address .
first , why does n6 have such narrow line widths that suggest it has relatively little turbulence ?
second , why are these line widths narrower than those in its immediate surroundings ?
third , why is n6 located so close to its neighbor sm2 ?
these questions may be answered simultaneously by considering models of equilibrium spheres embedded in sufficiently turbulent and opaque cores . within such environments ,
turbulent motions could be reduced locally , e.g. , if the spheres were similar or smaller in size than the cutoff wavelength for mhd waves in their surroundings ( mckee 1989 , mckee et al .
( turbulence could instead be damped locally by other processes such as shocks or dissipation ; e.g. , see nakano 1998 . ) myers ( 1999 ; see also myers 1997 ) , following reasoning by larson ( 1985 ) and mouschovias ( 1991 ) , proposed thermally dominated , critically stable bonnor - ebert spheres could exist within turbulent , opaque , and dense cores .
such cores require large column densities ( e.g. , @xmath90 @xmath49 4 ) , so that ionization by cosmic rays dominates over that by ultraviolet radiation , and the coupling between the magnetic field and the gas is minimized .
the embedded objects , called kernels " by myers , would be compressed by external ( turbulent ) core pressure to sizes comparable to the observed separations of young stellar objects within embedded clusters ( e.g. , @xmath20.05 - 0.15 pc ) . a protostellar cluster could then arise from the near - simultaneous collapse of several kernels within a common core .
oph a - n6 may be an excellent candidate for a kernel - like object , given its near - thermal line widths .
furthermore , its 0.9 m mass matches quite well the 0.8 m mean mass predicted for kernels within nearby star forming molecular cores , i.e. , masses obtained by myers using values of @xmath62 , @xmath64 and @xmath91 derived by from harju , walmsley , & wouterloot ( 1993 ) and ladd , myers , & goodman ( 1994 ) . also , the projected separation between n6 and sm2 is 0.025 pc , within a factor of @xmath22 of the @xmath20.05 pc separation of kernels predicted by myers .
( indeed , the mean nearest - neighbor separation of n6 , sm1n , sm1 , sm2 , and vla 1623 , is only 0.023 pc . )
oph a - n6 does not fit completely to a kernel interpretation , however . as noted above
, its 0.008 pc radius is @xmath23 smaller than the 0.024 pc mean radius expected for such kernels .
( note that the radius of a singular isothermal sphere ( shu 1977 ) at @xmath62 = 17.4 k required to make a 0.9 m star is @xmath20.032 pc , a factor of @xmath24 larger than that of n6 . ) despite these caveats , however , the kernel model provides at least some basis for understanding n6 as a highly localized region of low turbulence , unlike simple equilibrium models .
the non - detection of oph a - n6 as a two - dimensional maximum of @xmath28(h@xmath0 ) but its clear detection as a location of maximal @xmath28(n@xmath0h@xmath1 ) suggest that interesting sites possibly related to star formation may be overlooked in maps of thermal continuum emission .
why was n6 not seen , e.g. , by man98 ?
contrasts of @xmath28(h@xmath0 ) in n6 could be minimized if dust at its inner radii are colder than dust at its outer radii ( e.g. , see evans et al . ) , a strong possibility given the high densities and extinctions observed in oph a. note , however , that bouwman , galli , & andr ( 2004 , in preparation ; see also andr et al .
2003 ) have recently shown that 1.3 mm continuum flux and @xmath37 are anticorrelated in starless cores , even those with the high column densities and incident radiation fields of oph a ( e.g. , 85 @xmath92 ; liseau et al .
the oph a objects , especially n6 , must be specifically modeled to evaluate better this possibility .
if oph a - n6 is a kernel - like object , we speculate that the oph a continuum objects may also be or once were kernel - like . in this scenario
, the large @xmath3 of these objects relative to that of oph a - n6 may suggest that they are now in a state of collapse .
for example , n6 has narrow lines close to the thermal line width ; an absence of wings or asymmetrically blue profiles in the line suggest that collapse has not yet occurred.h@xmath1 10 has thermalized within n6 , and @xmath23 @xmath52 @xmath71 , the line will not show infall asymmetry if no gradient in @xmath23 exists . ]
if the continuum objects have had similar physical conditions to n6 earlier , turbulence may have been similarly reduced at that epoch .
therefore , their larger @xmath3 may not be due to substantial turbulent motions in their associated gas .
indeed , how turbulent motions could be reintroduced to these objects after its reduction in the absence of protostars , winds , or outflows is unclear . instead , gas within these objects may still be thermally dominated and the line broadening observed may be due to infall motions stemming from collapse .
the n@xmath0h@xmath1 10 line is only marginally optically thick toward these objects , so the line should remain symmetric , as seen in figure 5 , if emitted from infalling gas ( myers , evans , & ohashi 2000 ) .
we note here again that sm1 , sm1n , and sm2 are found relatively close to ( but are not coincident with ) positions of the largest @xmath3 in oph a ; these extreme line widths may also be caused by infall motions related to these objects .
furthermore , the correlation of @xmath3 vs. @xmath6 found for bright line emission shown in fig .
6a may be also related to infall motions .
the observed fwzis of n@xmath0h@xmath1 10 toward sm1n , sm1 , and sm2 are @xmath21.0 km s@xmath5 , suggesting infall motions on scales of @xmath21000 - 3000 au up to 0.5 km s@xmath5 , similar to those measured directly toward the class 0 objects ngc 1333 iras 4a and 4b by di francesco et al .
further speculating , we note that the variations of @xmath6 in oph a may provide additional evidence for differences in the evolutionary stages of starless objects within oph a , by supporting the idea that oph a - n6 is not collapsing but sm1n , sm1 , and sm2 may be .
n6 appears to have no associated n@xmath0h@xmath1 10 velocity gradients but the continuum objects have small ones ( i.e. , @xmath560.5 km s@xmath5 over @xmath20.01 pc ; see fig .
3b ) . such small velocity gradients could stem from increases in local rotation speeds of dense gas that could accompany inward motions due to angular momentum conservation .
( indeed line broadening by rotational motions may also be contributing to the larger line widths associated with these objects .
note that the gradients in fig .
3b , however , are seen on scales larger than the beam size . )
high - resolution data of oph a in other tracers of quiescent dense gas that require higher levels of excitation than n@xmath0h@xmath1 10 could reveal velocity gradients more dominated by infall or rotation and clarify this issue ( e.g. , see caselli et al .
\1 . combined single - dish and interferometer data of n@xmath0h@xmath1 10 line emission from oph a
are presented that yield a map of dense gas at high linear resolution ( e.g. , @xmath21000 au ) with information retained on large spatial scales .
six maxima of integrated line intensity , oph a - n1 through n6 , have been detected .
n4 and n5 are coincident with the starless objects sm1 and sm1n respectively .
the other maxima , however , are not coincident with known starless or protostellar oph a objects .
moreover , little n@xmath0h@xmath1 10 emission is associated with the starless object sm2 or the class 0 protostar vla 1623 .
this situation is in contrast with observations of isolated cores where dust peaks and integrated n@xmath0h@xmath1 intensity maxima are commonly coincident . \2 . with 1.3 mm continuum data from man98 and @xmath37 estimates from awb93 , @xmath28(n@xmath0h@xmath1 ) , @xmath28(h@xmath0 ) , and @xmath50(n@xmath0h@xmath1 )
were estimated for the oph a objects .
sm1 and sm1n have similar values of @xmath50(n@xmath0h@xmath1 ) @xmath52 2 - 3 @xmath8 10@xmath51 .
sm2 and vla 1623 , however , have values of @xmath50(n@xmath0h@xmath1 ) that are factors of @xmath494 less than those of sm1 and sm1n . the @xmath50(n@xmath0h@xmath1 ) of oph a - n6
is similar to that of sm1 and sm1n , if @xmath37 @xmath52 15 k at that location . for sm2 and vla 1623
, the lower values of @xmath50(n@xmath0h@xmath1 ) are likely due to gas - grain chemistry in cold , dense cores where n@xmath0 adsorbs onto grain surfaces and n@xmath0h@xmath1 is depleted , as suggested by many .
similar depletion may not have occurred at sm1 or sm1n because their dust is warmer , e.g. , @xmath37 = 27 k , perhaps inhibiting n@xmath0 adsoption .
n@xmath0h@xmath1 10 line widths ( @xmath3 ) in oph a vary by a factor of @xmath25 with values @xmath40.30
km s@xmath5 found in fourteen locations . of these ,
only the location associated with the n6 maximum of integrated intensity is both a local minimum of line width and has an extent larger than the beam size .
the minimum @xmath3 in n6 is only 0.193 km s@xmath5 , or just larger than line widths expected from thermal motions alone at expected temperatures .
in contrast , the @xmath3 associated with the sm1 , sm1n , and sm2 objects are @xmath20.5 km s@xmath5 .
n@xmath0h@xmath1 10 line centroids ( @xmath6 ) in oph a vary little across the core , having an rms of only @xmath20.17 km s@xmath5 .
no large - scale gradient is seen across oph a , although small - scale @xmath6 gradients of @xmath40.5 km s@xmath5 over 0.01 pc are associated with sm1 , sm1n , and sm2 , but not with n6 .
the @xmath3 and @xmath6 of oph a as a whole yield a one - dimensional velocity dispersion , @xmath40 , for oph a that is a factor of @xmath22 of that required for gravitational virial equilibrium .
the low @xmath3 and lack of @xmath6 gradient of n6 relative to sm1 , sm1n , and sm2 suggest evolutionary differences .
the one - dimensional velocity of n6 suggest it alone is within a factor of 2 from gravitational virial equilibrium .
n6 has properties ( e.g. , @xmath3 , mass , and short proximity to neighbors ) similar to those predicted for kernels " , i.e. , thermally dominated , critically stable , bonnor - ebert spheres embedded within turbulent cores , although some of its properties ( e.g. , size and no column density maximum ) are dissimilar .
n6 may represent an early stage of the evolution of dense gas towards star formation where significant local reduction of turbulent motions occurs .
the non - detection of n6 in thermal continuum maps of oph a suggests that interesting sites possibly related to star formation may be missed in such maps .
buisson , g. , desbats , l. , duvert , g. , forveille , t. , gras , r. , guilloteau , s. , lucas , r. , & valiron , p. 2002
, continuum and line analysis single - dish system manual , ( class ; iram : grenoble ; http://iram.fr/gs/class/class.html ) cccc name & r.a.(2000 ) & decl .
( 2000 ) & association n1 & 16 26 26.4 & -24 23 07 & a - mm5 ?
n2 & 16 26 26.5 & -24 22 24 & n3 & 16 26 27.2 & -24 24 20 & n4 & 16 26 27.2 & -24 23 50 & sm1 n5 & 16 26 27.4 & -24 23 30 & sm1n n6 & 16 26 31.6 & -24 24 52 & ccccccc object & @xmath37(awb93 ) & @xmath50(n@xmath0h@xmath1 ) & @xmath28(n@xmath0h@xmath1 ) & @xmath28(h@xmath0 ) & @xmath3(n@xmath0h@xmath1 ) & @xmath6 & k & @xmath810@xmath51 & @xmath810@xmath67 @xmath74 & @xmath810@xmath75 @xmath74 & km s@xmath5 & km s@xmath5 vla 1623 & 15 - 20 & @[email protected] & @xmath43 & 6 - 4 & & sm1n & 27 & 3 & 10 & 3 & 0.49 & 3.60 sm1 & 27 & 2 & 8 & 4 & 0.59 & 3.72 sm2 & @xmath420 & @xmath40.9 & 2 & @xmath493 & 0.52 & 3.47 oph a - n6 & 15 ? &
4 & 6 & 3 & 0.25 & 3.47 ccccc name & r.a.(2000 ) & decl .
( 2000 ) & @xmath3 & association a & 16 26 24.8 & -24 22 54 & 0.24 & b & 16 26 26.1 & -24 23 14 & 0.29 & n1 c & 16 26 26.3 & -24 22 24 & 0.26 & n2 d & 16 26 26.3 & -24 24 08 & 0.27 & e & 16 26 26.4 & -24 23 48 & 0.28 & f & 16 26 26.9 & -24 23 10 & 0.29 & g & 16 26 29.3 & -24 24 40 & 0.27 & n1 h & 16 26 30.0 & -24 25 01 & 0.28 & i & 16 26 30.1 & -24 24 05 & 0.28 & j & 16 26 30.6 & -24 25 01 & 0.29 & k & 16 26 30.7 & -24 24 40 & 0.27 &
l & 16 26 30.7 & -24 25 20 & 0.26 & n6 m & 16 26 32.4 & -24 24 33 & 0.21 & n & 16 26 33.4 & -24 24 48 & 0.29 & | we present combined bima interferometer and iram 30 m telescope data of n@xmath0h@xmath1 10 line emission across the nearby dense , star forming core ophiuchus a ( oph a ) at high linear resolution ( e.g. , @xmath21000 au ) .
six maxima of integrated line intensity are detected which we designate oph a - n1 through n6 .
the n4 and n5 maxima are coincident with the starless continuum objects sm1 and sm2 respectively but the other maxima are not coincident with previously - identified objects .
in contrast , relatively little n@xmath0h@xmath1 10 emission is coincident with the starless object sm2 and the class 0 protostar vla 1623 .
the fwhm of the n@xmath0h@xmath1 10 line , @xmath3 , varies by a factor of @xmath25 across oph a. values of @xmath3 @xmath4 0.3 km s@xmath5 are found in 14 locations in oph a , but only that associated with n6 is both well - defined spatially and larger than the beam size .
centroid velocities of the line , @xmath6 , vary relatively little , having an rms of only @xmath20.17 km s@xmath5 .
small - scale @xmath6 gradients of @xmath40.5 km s@xmath5 over @xmath20.01 pc are found near sm1 , sm1n , and sm2 , but not n6 .
the low n@xmath0h@xmath1 abundances of sm2 or vla 1623 relative to sm1 , sm1n , or n6 may reflect relatively greater amounts of n@xmath0 adsorption onto dust grains in their colder and probably denser interiors .
the low @xmath3 of n6 , i.e. , 0.193 km s@xmath5 fwhm , is only marginally larger than the fwhm expected from thermal motions alone , suggesting turbulent motions in the oph a core have been reduced dramatically at this location .
the non - detection of n6 in previous thermal continuum maps suggests that interesting sites possibly related to star formation may be overlooked in such data . |
the set of @xmath26 linear equations with @xmath25 variable weights , @xmath46 , in eq .
[ eq1 ] can be rewritten as a system of @xmath26 linear equations with @xmath25 variables : @xmath47 where now @xmath48 is a @xmath49 matrix of real numbers , and @xmath50 .
notice that the linear system in eq .
[ eq : sistema ] has solutions since the rank of @xmath48 is @xmath51 ( all the equations are separated and each of the variables , @xmath46 , appears in one equation only ) , and the in - degree of all nodes is positive by definition .
hence , there always exists @xmath52 such that eq .
[ eq1 ] is satisfied .
it is convenient to rewrite eq .
[ eq : sistema ] in a form that emphasises the dependence of matrix @xmath48 from @xmath53 .
we choose to label the arcs as follows : @xmath54 , @xmath55 denotes the @xmath56-th arc entering node @xmath11 , where @xmath57 is the in degree of node @xmath11 .
likewise , @xmath58 is the source of arc @xmath54 , while @xmath59 is the corresponding weight . using this notation , the @xmath11-th component of eq .
[ eq : sistema ] can be written as : @xmath60 by direct computation , one positive solution of eq .
[ eq : eq2 ] is given by @xmath61 where @xmath62 , and by continuity there are infinite many solutions such that @xmath59 are all positive . in particular , if for node @xmath11 we have @xmath63 , then the @xmath11-th equation of eq .
[ eq : eq2 ] has a unique solution , while if @xmath64 , there are always infinitely many solutions depending on @xmath65 parameters . summing up , eq .
( [ eq : sistema ] ) has only one solution if all the node in - degrees are equal to one , while there are , in general , infinitely many solutions depending on @xmath66 parameters .
notice that @xmath67 can be different from @xmath9 , meaning that it is also possible to set the value of the largest eigenvalue of the weighted graph . here
, we show that it is not necessary to fix the weights of all the graph links in order to get an arbitrary centrality vector @xmath13 .
in fact , given a subset of links @xmath68 containing at least one incoming link for each node , it is sufficient to assign some positive weights @xmath69 to each @xmath70 , while keeping @xmath71 constant @xmath72 , for instance all equal to @xmath24 , such that the resulting weighted graph has eigenvector centrality equal to @xmath16 . without loss of generality we can assume that the first @xmath73 incoming links of each node @xmath11 belong to @xmath30 , so that the components of eq .
[ eq : eq2 ] can be written as : therefore , since @xmath75 for each @xmath76 , then there is a @xmath77 such that for every @xmath78 @xmath79 and hence , by a similar continuity argument as above , we can ensure that there are infinitely many positive solutions to eq.[eq : eq3 ] .
a _ controlling set _ of graph @xmath7 is any set of nodes @xmath80 such that : @xmath81 this means that , for each node @xmath31 in the graph , at least one of the two following conditions holds : _ a _ ) @xmath82 , or _
@xmath31 is pointed by at least one node in @xmath83 .
we use @xmath84 to denote the size of the controlling set , i.e. the number of nodes contained in @xmath83 . finding the _ minimum controlling set
_ @xmath85 of a graph @xmath7 , i.e. a controlling set having minimal size , is equivalent to computing the so - called _ domination number _ of @xmath7 .
the domination number problem is a well known np - hard problem in graph theory @xcite .
therefore , the size of the minimum controlling set can be determined exactly only for small @xmath26 graphs as those in fig .
[ fig : real_social ] . to investigate larger graphs
we have used two greedy algorithms .
the first algorithm , called top down controller search ( tdcs ) , works as follows .
we initially set @xmath86 .
we select the node @xmath87 with the maximum out - degree in @xmath88 , and mark it as _ controller node_. then , all the nodes in the out - neighbourhood of @xmath87 are marked as _ controlled _ and are removed from @xmath88 , together with @xmath87 itself . in this way , we obtain a new graph @xmath89 , and we store the controller node @xmath87 , together with the list of nodes controlled by @xmath87 . notice that , removing a generic node @xmath31 from @xmath88 , also implies that @xmath89 does not contain any of the links pointing to @xmath31 or originating from it .
the same procedure is iteratively applied to @xmath89 , @xmath90 and so on , until all the nodes of @xmath7 are either marked as controller or as controlled nodes .
the algorithm produces a set @xmath91 , with @xmath92 , which is a controlling set of @xmath7 by construction .
the second algorithm is called bottom up controller search ( bucs ) , and it works as follows .
we set @xmath86 and consider the set @xmath93 containing all the nodes in @xmath88 with minimum in degree . for each node @xmath94
, we consider the set of nodes pointing to @xmath11 and select from this set the node @xmath95 with the maximal out degree .
this node is marked as _
controller_. then we obtain a new graph @xmath89 by removing from @xmath88 all the controller nodes @xmath95 for all @xmath94 , together with all the nodes , marked as _ controlled _ , pointed by them .
the same procedure is iteratively applied to @xmath89 , @xmath90 and so on , until all the nodes of @xmath7 are either marked as controller or as controlled nodes .
if a graph @xmath96 contains isolated nodes , these are marked as _ controller _ and removed from @xmath96 .
the algorithm finally produces a set @xmath91 which is a controlling set of @xmath7 by construction .
we have verified that the controlling sets obtained by both tdcs and bucs for each of the networks considered are much smaller than those obtained by randomly selecting the controlling nodes . moreover , the set of controller nodes found by tdcs is in general different from that obtained on the same network by bucs .
also the sizes of the two controlling sets obtained by the two algorithms are different .
in particular , we have noticed that in assortative ( disassortative ) networks the controlling set produced by tdcs is smaller ( larger ) than that produced by bucs .
boccaletti , s. , latora , v. , moreno , y. , chavez , m. , hwang , d.u .
, complex networks : structure and dynamics .
_ phys . rep .
_ * 424 , * 175 - 308 ( 2006 ) .
arenas , a. , daz - guilera , a. , khurths , j. , moreno , y. , zhou , c. synchronization in complex networks .
_ phys . rep .
_ * 469 , * 93 - 153 ( 2008 ) .
barrat , a. , barthlemy , m. , vespignani , a. _ dynamical processes in complex networks .
_ ( cambridge university press , cambridge , 2008 ) .
fortunato , s. community detection in graphs._phys . rep . _ * 486 , * 75 - 174 ( 2010 ) . bavelas , a. a mathematical model for group structures . _
organ _ * 7 * , 16 ( 1948 ) .
wasserman , s. , faust , k. _ social networks analysis . _ ( cambridge university press , cambridge , 1994 ) .
jeong , h. , mason , s.p . ,
barabsi , a .-
, oltvai , z. n. lethality and centrality in protein networks .
_ nature _ * 411 , * 41 - 42 ( 2001 ) .
albert , r. , jeong , h. , barabsi , a .-
l . error and attack tolerance of complex networks . _
nature _ , * 406 , * 378 - 382 ( 2000 ) .
crucitti , p. , latora , v. , porta , s. centrality measures in spatial networks of urban streets .
e _ * 73 , * 036125 ( 2006 ) .
barthlemy , m. spatial networks , _ phys . rep . _
* 499 , * 1 - 101 ( 2011 ) .
freeman , l.c .
centrality in social network .
conceptual clarification .
_ social networks _ * 1 , * 215 - 239 ( 1979 ) .
barthlemy , m. betweenness centrality in large complex networks .
j. b _ * 38 , * 163 - 168 ( 2004 ) .
perra , n. , fortunato , s. spectral centrality measures in complex networks .
e _ * 78 , * 036107 ( 2008 ) .
bonacich , p. factoring and weighting approaches to status scores and clique identification . _ j. math .
* 2 , * 113 ( 1972 ) .
bonacich , p. , lloyd , p.
eigenvector - like measures of centrality for asymmetric relations .
* 23 , * 191 - 201 ( 2001 ) .
bonacich , p. power and centrality : a family of measures .
j. sociol .
_ * 92 , * 1170 - 1182 ( 1987 ) .
katz , l. a new status index derived from sociometric analysis .
_ psychometrika _ * 18 , * 39 - 43 ( 1953 ) .
brin , s. , page , l. the anatomy of a large - scale hypertextual web search engine .
_ comput .
netw . _ * 30 , * 107 - 117 ( 1998 ) .
delvenne , j .- c . , libert , a .- s . centrality measures and thermodynamic
formalism for complex networks .
e _ * 83 , * 046117 ( 2011 ) .
gfeller , d. , de los rios , p. spectral coarse graining of complex networks .
_ * 99 , * 038701 ( 2007 ) .
fortunato , s. , flammini , a. , random walks on directed networks : the case of pagerank , _ int .
j. bifurcat . chaos _ * 17 , * 2343 - 2353 ( 2007 ) .
hlebec , v .
recall versus recognition : comparison of the two alternative procedures for collecting social network data .
_ developments in statistics and methodology , p. 121 - 129 .
_ ( a. ferligoj , a. kramberger , editors ) metodoloki zvezki 9 , fdv , ljubljana , 1993 , breiger , r. , boorman , s. , arabie , p. an algorithm for clustering relational data with applications to social network analysis and comparison with multidimensional scaling .
_ j. math .
_ * 12 , * 328 - 383 ( 1975 ) .
zachary , w. w. an information flow model for conflict and fission in small groups . _ j. anthropol .
* 33 , * 452 - 473 ( 1977 ) .
west , d.b .
_ introduction to graph theory . _
( prentice - hall , 2@xmath97 edition , nj , 2001 ) .
leskovec , j. , lang , k. , dasgupta , a. , mahoney , m. community structure in large networks : natural cluster sizes and the absence of large well - defined clusters .
_ arxiv.org:0810.1355_ , ( 2008 ) .
albert , r. , jeong , h. , barabsi , a .-
diameter of the world wide web .
_ nature _ , * 401 , * 130 ( 1999 ) .
gleiser , p. , danon , l. community structure in jazz .
complex syst .
_ * 6 * , 565 ( 2003 ) .
barabsi , a .-
albert , r. , emergence of scaling in random networks .
_ science _ * 286 , * 509 ( 1999 ) .
newman , m.e.j . scientific collaboration networks .
shortest paths , weighted networks , and centrality .
* 64 * , 016132 ( 2001 ) .
newman , m.e.j .
finding community structure in networks using the eigenvectors of matrices .
* 74 , * 036104 ( 2006 ) .
guimera , r. , danon , l. , daz - guilera , a. , giralt , f. , arenas , a. self - similar community structure in a network of human interactions .
e _ * 68 , * 065103(r ) ( 2003 ) .
klimmt , b. , yang , y. introducing the enron corpus .
_ ceas conference _ ( 2004 ) .
leskovec , j. , kleinberg , j. , faloutsos , c .
graph evolution : densification and shrinking diameters .
_ acm sigkdd international conference on knowledge discovery and data mining _ ( kdd ) ( 2005 ) .
leskovec , j. , huttenlocher , d. , kleinberg , j. predicting positive and negative links in online social networks .
_ www conference _ ( 2010 ) .
available online at http://topology.eecs.umich.edu/. electric engineering and computer science department , university of michigan , topology project .
colizza , v. , pastor - satorras , r. , vespignani , a. , _ nat .
_ * 3 , * 276 - 282 ( 2007 ) .
watts , d .-
j . , strogatz , s .- h .
collective dynamics of small - world networks .
_ nature _ * 393 , * 440 - 442 ( 1998 ) . milo , r. , itzkovitz , s. , kashtan , n . , levitt , r. , shen - orr , s. , ayzenshtat , i. , sheffer , m. , alon , u. superfamilies of evolved and designed networks .
_ science _ * 303 , * 1548 - 1542 ( 2004 ) .
available online at http://vlado.fmf.uni - lj.si / pub / networks / data/. nelson , d. l. , mcevoy , c. l. , schreiber , t. a. the university of south florida word association , rhyme , and word fragment norms . http://www.usf.edu/freeassociation/ ( 1998 ) .
bogua , m. , pastor - satorras , r. , daz - guilera , a. , arenas , a. models of social networks based on social distance attachment .
e _ , * 70 * , 056122 ( 2004 ) .
leskovec , j. , adamic , l. , adamic , b. the dynamics of viral marketing .
_ acm trans . on the web ( acm tweb )
_ , * 1 , * ( 2007 ) .
richardson , m. , agrawal , r. , domingos , p. trust management for the semantic web .
proceedings of iswc , ( 2003 ) .
ripeanu , m. , foster , i. , iamnitchi , a. mapping the gnutella network : properties of large - scale peer - to - peer systems and implications for system design .
_ ieee internet computing journal _ , ( 2002 ) .
adamic , l.a . , glance , n. the political blogosphere and the 2004 us election .
_ proceedings of the www-2005 workshop on the weblogging ecosystem _ ( 2005 ) .
available online at http://www.orgnet.com/ | spectral centrality measures allow to identify influential individuals in social groups , to rank web pages by their popularity , and even to determine the impact of scientific researches .
the centrality score of a node within a network crucially depends on the entire pattern of connections , so that the usual approach is to compute the node centralities once the network structure is assigned .
we face here with the inverse problem , that is , we study how to modify the centrality scores of the nodes by acting on the structure of a given network .
we prove that there exist particular subsets of nodes , called controlling sets , which can assign any prescribed set of centrality values to all the nodes of a graph , by cooperatively tuning the weights of their out - going links .
we show that many large networks from the real world have surprisingly small controlling sets , containing even less than @xmath0 of the nodes .
these results suggest that rankings obtained from spectral centrality measures have to be considered with extreme care , since they can be easily controlled and even manipulated by a small group of nodes acting in a coordinate way .
modelling social , biological and information - technology systems as complex networks has proven to be a successful approach to understand their function @xcite . among the various aspects of networks which have been investigated so far , the issue of _ centrality _ , and the related problem of identifying the central elements in a network , has remained pivotal since its first introduction .
the idea of centrality was initially proposed in the context of social systems , where it was assumed a relation between the location of an individual in the network and its influence and power in group processes @xcite . since then , various _
centrality measures _ have been introduced over the years to rank the nodes of a graph according to their topological importance .
centrality has found many applications in social systems @xcite , in biology @xcite and in man - made spatial networks @xcite . among the various measures of centrality , such as those based on counting the first neighbours of a node ( degree centrality ) , or the number of shortest paths passing through a node ( betweenness centrality ) @xcite , a particularly important class of measures are those based on the spectral properties of the graph @xcite .
spectral centrality measures include the _ eigenvector centrality _
@xcite , the _ alpha centrality _ @xcite , _ katz s centrality _
@xcite and _ pagerank _
@xcite , and are often associated to simple dynamics taking place over the network , such as various kinds of random walks @xcite .
as representative of the class of spectral centralities , we focus here on eigenvector centrality , which is based on the idea that the importance of a node is recursively related to the importance of the nodes pointing to it . given an unweighted directed graph @xmath1 with @xmath2 nodes and @xmath3 links , described by the @xmath4 adjacency matrix @xmath5 , the eigenvector centrality @xmath6 of @xmath7 is defined as the eigenvector of @xmath8 associated to the largest eigenvalue @xmath9 , which in formula reads @xmath10 @xcite .
if the graph is strongly connected , then the perron - frobenius theorem guarantees that @xmath6 is unique and positive .
therefore , @xmath6 can be normalised such that the sum of the components equals 1 , and the value of the @xmath11-th component represents the centrality score of node @xmath11 , i.e. the fraction of the total centrality associated to node @xmath11 . in this article
we show how to change the eigenvector centrality scores of all the nodes of a graph by performing only local changes at node level . as a first step ( see the methods section ) we have proved that , given any arbitrary positive vector @xmath12 , @xmath13 , and @xmath14 , it is _ always _ possible to assign _ the weights of all the links _ of a strongly connected graph @xmath7 and to construct a new weighted network @xmath15 , with the same topology as @xmath7 and with eigenvector centrality equal to @xmath16 : @xmath17 where @xmath18 is the weighted adjacency matrix of @xmath15 .
this is illustrated in fig .
[ fig : smallgraph ] for a graph with @xmath19 nodes and @xmath20 links . in the original unweighted graph @xmath7 ,
node @xmath21 is the node with the highest eigenvector centrality , followed in order by node @xmath22 , node @xmath23 , and node @xmath24 .
now , if we have the possibility of tuning the weights of each of the five links , we can set any centrality value to the nodes of the graph . in figure
we show , for instance , how to fix the weights of the five links in order to construct : _
i ) _ a weighted network @xmath15 in which all nodes have the same centrality score , and _
ii ) _ even a weighted network @xmath15 in which the centrality ranking is totally reversed with respect to the ranking in @xmath7 . as shown in the example , given a graph @xmath7 , by controlling the weights of all the links , it is always possible to set any arbitrary vector @xmath16 as the eigenvector centrality of the graph .
however , tuning the weights of all the @xmath25 links of a given network is practically unfeasible , especially in large systems .
fortunately , this is not necessary , either .
in fact , in the case of fig . [
fig : smallgraph ] , a weighted graph with all nodes having the same centrality score can also be obtained by changing the weights of only four links , while leaving unchanged the weight of the link from node 1 to node 2 .
more in general , it can be proved that the eigenvector centrality of the whole network can be controlled by appropriately tuning the weights of just @xmath26 of the @xmath25 links .
the only constraint is that the @xmath26 links must belong to a subset @xmath27 such that , for every node @xmath28 , there is a link @xmath29 _ pointing to _
@xmath11 ( see methods section ) .
this is illustrated in fig .
[ fig : real_social ] for three real social networks . in each of the three cases ,
it is possible to set any arbitrary eigenvector centrality by changing only the weights of the red arcs , while keeping unchanged ( and equal to @xmath24 ) the weights of all remaining arcs , shown in yellow . the nodes from which the links in @xmath30 originate are also coloured in red , and are referred to as a _ controlling set _ of the network ( see methods section ) .
what is striking is that , in each of the three networks , the set @xmath30 can be chosen in such a way that all the links in @xmath30 originate from a relatively small subset of nodes .
for instance , the controlling set reported for the student government network of the university of ljubljana contains only two nodes .
this is also a _ minimum controlling set _ , since the graph does not admit another controlling set with a smaller number of nodes .
this finding indicates that only two members of the student government , namely node 2 and node 8 , can in principle set the centrality of all the other members by concurrently modifying the weights of some of their links .
it is in fact reasonable to assume that the weight of the directed link from @xmath11 to @xmath31 , representing in this case the social credit ( in terms of reputation , esteem or leadership acknowledgement ) given by individual @xmath11 to individual @xmath31 , can be strengthen or decreased _ only _ by @xmath11 .
consequently , nodes 2 and 8 can modify at their will the weights of their out - going links and , if these changes are opportunely coordinated , they can largely alter the actual roles of all the other individuals .
analogously , only five monks can control the centrality of the sampson s monk network , while only 4 members of the zachary s karate club network can set the eigenvector centrality of the remaining 30 members .
a question of practical interest is to investigate the size @xmath32 of the minimum controlling set in various complex networks .
when @xmath33 is small with respect to @xmath26 , then the centrality of the network is easy to control .
conversely , when the number of nodes in the minimum controlling set is large , the network @xmath7 is more robust with respect to centrality manipulations .
we have used two greedy algorithms to compute approximations of minimum controlling sets in various real systems ( see methods section ) . in table
[ table ] we report the best approximation for @xmath33 , i.e. the size of the smallest controlling set @xmath34 produced by either of the two algorithms in networks whose sizes range from hundreds to millions of nodes . in the majority of the cases we have found unexpectedly small controlling sets , containing only up to @xmath35 of the nodes of the network . for instance , in the graph of jazz musicians , there exists a controlling set made by just @xmath36 of the @xmath37 musicians .
these @xmath36 individuals alone can , in principle , decide to set the popularity of all the other musicians , enhancing the centrality of some of the nodes and decreasing the centrality of others , just by playing more or less often with some of their first neighbours . among all the networks we have considered
, the one with the smallest controlling set is the wikipedia talk communication network , a graph with 2,394,385 nodes in which just @xmath38 of nodes are able to alter the centrality of the entire system .
the quantities in parenthesis indicate that for this network a set of just @xmath39 of the nodes can control the centrality of @xmath40 of the nodes . [ cols="<,^,^,^,^",options="header " , ] for each real network @xmath7 , we have also computed the typical size @xmath41 of the minimum controlling set in its randomised counterpart ( see the rightmost column in table [ table ] ) . in particular , we have considered a randomisation which preserves the degree sequence of the original graph . in
most of the cases @xmath42 , relevant exceptions being some spatial man - made networks , such as power grids , road networks and electronic circuits , and also the patents citation network .
this fact suggests that , in the absence of other limitations , such as strong spatial / geographic constraints @xcite , the structure of real networks has naturally evolved to favour the control of spectral centrality by a small group of nodes . to better compare the controllability of networks with different sizes
, we report in fig .
[ fig : crand ] the ratio @xmath43 as a function of the number of graph nodes @xmath26 .
the smallest values of the ratio @xmath43 are found for collaboration / communication systems , www and socio - economical networks .
the five most controllable networks are respectively wiki - talk , internet at the as level , movie actors , the stanford world wide web , and the collaboration network of researchers in astrophysics .
these are all networks in which single nodes can tune , at their will , the weights of their out - going links
. a scientist can decide whether to weaken or strengthen the connections to some of the collaborators .
the administrators of an internet autonomous system can control the routing of traffic through neighbouring ass , by modifying peering agreements @xcite . and ,
similarly , the owner of a web page can change the weights of hyperlinks , for instance by assigning them different sizes , colour , shapes and positions in the web page . , the ratio between the sizes of the minimum controlling set in real networks and in their respective randomized versions ( we have considered averages over 100 different realizations ) .
different symbols and colors refer to the six network classes considered in table [ table ] .
the observed ratio is lower than 1 in most of the cases , with the smallest values corresponding to collaboration / communication systems , www and socio - economical networks .
the ratio is equal to 1 in three cases .
the networks with ratio larger than 1 , with the exception of one socio - economical system ( namely epinions ) , are all spatially constrained systems : three electronic circuits , the us power grid , and three road networks . ] in this work , we have shown how a small number of entities , working cooperatively , can set any arbitrary eigenvector centrality for all the nodes of a real complex network .
it is straightforward to extend our results to other spectral centralities , such as @xmath44-centrality and katz s centrality .
similar arguments can also be applied , with some limitations , to pagerank : in this case , the inverse centrality problem has solutions only for some particular choices of @xmath45 .
such findings suggest that rankings obtained from centrality measures should be taken into account with extreme care , since they can be easily controlled and even distorted by a small group of cooperating nodes .
the high controllability of real networks potentially has large social and commercial impact , given that centrality measures are nowadays extensively used to identify key actors , to rank web pages , and also to assess the value of a scientific research . |
the main properties that should be satisfied by any jet definition were already pointed out almost twenty years ago @xcite : the jet algorithm should be 1 ) simple to implement both in experimental analysis and theoretical calculations , 2 ) well defined and yielding a finite cross section at any order of perturbation theory , 3 ) relatively insensitive to hadronization .
many different jet definitions have been proposed in recent years but it turns out that some of them do not strictly meet the above features .
this could in principle lead to serious problems especially when infrared(ir)-safety ( i.e. , the second item in the above list ) is not correctly implemented , since in this case the matching with fixed - order theoretical results would be spoiled and the whole jet - finding procedure would heavily depend on non - perturbative effects ( hadronization , underlying event , pile - up ) .
we can group jet algorithms in two broad classes : iterative cone ( ic ) and sequential recombination ( sr ) .
ic algorithms cluster particles according to their relative distance in _ coordinate - space _ and have been extensively used at past lepton and hadron colliders .
sr algorithms cluster particles according to their relative distance in _ momentum - space _ and are somehow preferred by theorists since they rigorously take into account ir - safety . in
what follows i will try to give an overview of recent developments obtained in both the ic and sr classes . for a complete and extensive treatment of these and other aspects of jet algorithms
i refer the reader to the recent literature on the subject @xcite .
the core structure of any ic algorithm @xcite can be described as follows : choose a _ seed _ particle _ i _ , sum the momenta of all particles _ j _ within a cone of radius _ r _ ( in _ y _ and @xmath0 ) around _ i _
, take the direction of this sum as a new seed , and repeat until the cone is stable ( i.e. , the direction of the _ n_-th seed coincides with the direction of the _ ( n-1)_-th ) .
this procedure , however , may eventually lead to find multiple stable cones sharing the same particles , i.e. the cones may overlap .
the problem can be solved either by preventing the overlap ( progressive removal ( ic - pr ) algorithms ) or through a splitting procedure ( split - merge ( ic - sm ) algorithms ) .
the ic - pr algorithm starts the iteration from the particle with largest-@xmath1 and , once a stable cone is found , it removes all particles inside it .
then the procedure starts again with the hardest remaining particle and go on until no particles are left .
this algorithm is also known as ua1-cone @xcite , since it was first introduced and extensively used by the cern ua1 collaboration .
it is quite easy to see that this algorithm is ir(collinear)-unsafe .
assume that the hardest particle undergoes a collinear splitting @xmath2 with @xmath3 : in this case the ic - pr algorithm would lead to a different number / configuration of jets , since the @xmath1-ordering has been modified by the collinear emission .
the ic - sm algorithm does not rely on any particular ordering instead .
once all stable cones have been found the prescription is to _ merge _ a pair of cones if more than a fraction @xmath4 ( typically @xmath4=0.5 - 0.7 ) of the softer s cone @xmath1 is in common with the harder one , or to _ assign _ the shared particles to the closer cone .
the _ split - merge _ procedure is repeated until there are no overlapping cones left . unfortunately the ic - sm algorithm is ir(soft)-unsafe as well @xcite .
assume that two stable cones are generated starting from two hard partons whose relative distance is between r and 2r : the addition of an extra soft particle in between would act as a new seed and the third stable cone would be found , again leading to a different number / configuration of jets .
a partial solution to this problem was provided by the _ midpoint algorithm _
@xcite : after all possible jets have been found , run the algorithm again using additional _ midpoint _ seeds between each pair of stable cones .
this prescription fixes the ir(soft ) issue of the ic - sm algorithm , since the result is now not dependent on the presence of an extra soft seed in the overlap region , and has been adopted as a recommandation for run ii of the tevatron @xcite . recently @xcite it has been pointed out
that , for particular configurations involving more than two partons , the midpoint algorithm is not able to find all stable cones : for exclusive quantities and/or multi - jet configurations , the midpoint prescription is still ir(soft)-unsafe .
the ir issue is definitely solved by the introduction of seedless algorithms , first proposed in @xcite .
the idea is to identify all possbile subsets of @xmath5 particles in an event and , for each subset @xmath6 , check if the cone defined by the azimuth and rapidity of the total momentum of @xmath6 contains other particles outside the subset : if this is not the case , then @xmath6 defines a stable cone . with this prescription
the jet - finding algorithm is infrared safe at all perturbative orders : the main drawback is that the clustering time ( @xmath7 ) ) leads to extremely slow performances for @xmath8 4 - 5
. the seedless algorithm has been recently improved by the sis - cone ( seedless infrared safe cone ) implementation @xcite , in which the clustering time is sensibly lowered ( @xmath9 , comparable to midpoint ) .
the switching from the midpoint to the seedless cone is expected to have a significant impact only on exclusive quantities ( i.e. , jet mass distribution in multi - jet events ) , while the impact for inclusive observables should be modest since midpoint ir - unsafety only appears at relatively high orders in perturbation theory .
the sr algorithm starts with the introduction of a distance @xmath10 between particles @xmath11 ( where @xmath12 is their distance in the @xmath13 plane and @xmath14 their transverse - momentum ) , and the distance @xmath15 between the particle @xmath16 and the beam .
if @xmath17 then merge @xmath16 and @xmath18 , otherwise call @xmath16 a jet and remove it from the iteration . there are different types of sr algorithms , depending on the value of the integer @xmath19 in the definition of the distances : @xmath20 identifies the _
inclusive-@xmath14 _ algorithm @xcite , @xmath21 defines the _ cambridge - aachen _
algorithm @xcite , while for @xmath22 we have the recently proposed _
algorithm @xcite .
all the sr prescriptions are rigorously ir - safe : any soft parton is first merged with the closest hard parton and only at this point the decision about the merging of two jets is taken , depending exclusively on their opening angle
. moreover , there are no more overlap problems , since any parton is inequivocally assigned to only one jet . a very fast implementation
( clustering time @xmath23 ) of all the above sr algorithms is available ( fastjet @xcite ) : it also includes an interface for the algorithms belonging to the ic class .
another public code providing access to both sr and ic algorithms is spartyjet @xcite .
the issue of ir - safety of a jet algorithm should be seriously taken into account , since multi - jet configurations are sensitive to it and will be far more widespread at the lhc than at previous colliders .
in addition , without an ir - safe prescription , it would not be possible to fully exploit the results provided by the theoretical community involved in nlo multi - leg calculations . several _ fast _ and _ safe _ algorithms ( sis - cone and the sr class ) are now available in public packages , but no definite advantages for a particular algorithm over the others have been found up to now : the use of different prescriptions for physics analysis and a continuous cross - checking of results is thus recommended especially for events with high jet - multiplicity at the lhc
. s. d. ellis , z. kunszt and d. e. soper , phys .
d * 40 * , 2188 ( 1989 ) .
j. e. huth _ et al .
_ , fermilab - conf-90 - 249-e ( 1990 ) . s. d. ellis , j. huston , k. hatakeyama , p. loch and m. tonnesmann , prog . part
. nucl .
phys . *
60 * , 484 ( 2008 ) , and references therein .
c. buttar _
_ , arxiv:0803.0678 [ hep - ph ] , and references therein
. g. arnison _ et al . _
[ ua1 collaboration ] , phys .
b * 123 * , 115 ( 1983 ) .
g. arnison _ et al . _
[ ua1 collaboration ] , phys .
b * 132 * , 214 ( 1983 ) .
m. h. seymour , nucl .
b * 513 * , 269 ( 1998 ) .
s. d. ellis , private communication to the opal collaboration ; d. e. soper and h. c. yang , private communication to the opal collaboration ; l. a. del pozo , university of cambridge phd thesis , ralt-002 , ( 1993 ) ; r. akers _ et al .
_ , opal collaboration , z. phys .
c * 63 * , 197 ( 1994 ) .
g. c. blazey _ et al .
_ , arxiv : hep - ex/0005012 . g. p. salam and g. soyez , jhep * 0705 * , 086 ( 2007 ) .
s. catani , y. l. dokshitzer , m. h. seymour and b. r. webber , nucl .
b * 406 * , 187 ( 1993 ) .
s. d. ellis and d. e. soper , phys .
d * 48 * , 3160 ( 1993 ) .
y. l. dokshitzer , g. d. leder , s. moretti and b. r. webber , jhep * 9708 * , 001 ( 1997 ) . m. wobisch and t. wengler , arxiv : hep - ph/9907280 .
m. cacciari , g. p. salam and g. soyez , jhep * 0804 * , 063 ( 2008 ) .
m. cacciari and g. p. salam , phys .
b * 641 * , 57 ( 2006 ) ; http://www.lpthe.jussieu.fr/@xmath24salam/fastjet/ http://www.pa.msu.edu/@xmath24huston/spartyjet/spartyjet.html . | i provide a very brief overview of recent developments in jet algorithms , mostly focusing on the issue of infrared - safety .
[ 1999/12/01 v1.4c il nuovo cimento ] |
aspirations for safety and security of human lives against crimes and natural disasters motivate us to establish smart monitoring systems to monitor surrounding environment . in this
regard , vision sensors are expected as powerful sensing components since they provide rich information about the outer world . indeed , visual monitoring systems have been already commoditized and are working in practice . typically , in the current systems , various decision - making and situation awareness processes are conducted at a monitoring center by human operator(s ) , and partial distributed computing at each sensor is , if at all , done independently of the other sensors . however , as the image stream increases , it is desired to distribute the entire process to each sensor while achieving total optimization through cooperation among sensors .
distributed processing over the visual sensor networks is actively studied in recent years motivated by a variety of application scenarios @xcite@xcite . among them , several papers address optimal monitoring of the environment assuming mobility of the vision sensors @xcite@xcite , where it is required for the network to ensure the best view of a changing environment @xcite .
the problem is related to coverage control @xcite@xcite , whose objective is to deploy mobile sensors efficiently in a distributed fashion . a typical approach to coverage control
is to employ the gradient descent algorithm for an appropriately designed aggregate objective function .
the objective function is usually formulated by integrating the product of a sensing performance function of a point and a density function indicating the relative importance of the point .
the approach is also applied to visual coverage in @xcite@xcite .
the state of the art of coverage control is compactly summarized in @xcite , and a survey of related works in the computer vision society is found in @xcite . in this paper , we consider a visual coverage problem under the situation where vision sensors with controllable orientations are distributed over the 3-d space to monitor 2-d environment . in the case , the control variables i.e. the rotation matrices must be constrained on the lie group @xmath0 , which distinguishes the present paper from the works on 2-d coverage @xcite@xcite . on the other hand ,
@xcite consider situations similar to this paper .
@xcite take game theoretic approaches which allow the network to achieve globally optimal coverage with high probability but instead the convergence speed tends to be slower than the standard gradient descent approach .
in contrast , @xcite employs the gradient approach by introducing a local parameterization of the rotation matrix and regarding the problem as optimization on a vector space .
this paper approaches the problem differently from @xcite .
we directly formulate the problem as optimization on @xmath0 and apply the gradient descent algorithm on matrix manifolds @xcite .
this approach will be shown to allow one to parametrize the control law for a variety of underactuations imposed by the hardware constraints .
this paper also addresses density estimation from acquired data , which is investigated in @xcite for 2-d coverage .
however , we need to take account of the following characteristics of vision sensors : ( i ) the sensing process inherently includes projection of 3-d world onto 2-d image , and ( ii ) explicit physical data is not provided . to reflect ( i ) , we incorporate the projection into the optimization problem on the embedding manifold of @xmath0 . the issue ( ii ) is addressed technologically , where we present the entire process including image processing and curve fitting techniques . finally , we demonstrate the utility of the present coverage control strategy through simulation of moving objects monitoring .
let us consider a riemannian manifold @xmath1 whose tangent space at @xmath2 is denoted by @xmath3 , and the corresponding riemannian metric , an smooth inner product , defined over @xmath3 is denoted by @xmath4 .
now , we introduce a smooth scalar field @xmath5 defined over the manifold @xmath1 , and the derivative of @xmath6 at an element @xmath2 in the direction @xmath7 , denoted by @xmath8 $ ] .
we see from definition 3.5.1 and ( 3.15 ) of @xcite that the derivative @xmath9 $ ] is defined by @xmath10 = \left.\frac{d f(\gamma(t))}{dt}\right|_{t = 0},\ ] ] where @xmath11 is a smooth curve such that @xmath12 .
in particular , when @xmath1 is a linear manifold with @xmath13 , the derivative @xmath9 $ ] is equal to the classical directional derivative @xmath10 = \lim_{t\to 0}\frac{f(x + t\xi)-f(x)}{t}. \label{eqn : derivative_lin_m}\ ] ] now , the gradient of @xmath6 is defined as follows .
@xcite [ def : grad_m ] given a smooth scalar field @xmath6 defined over a riemannian manifold @xmath1 , the gradient of @xmath6 at @xmath14 , denoted by @xmath15 , is defined as the unique element of @xmath3 satisfying @xmath16\ \ { \forall \xi}\in t_x{{\mathcal m}}.\ ] ] suppose now that @xmath1 is a riemannian submanifold of a riemannian manifold @xmath17 , namely @xmath3 is a subspace of @xmath18 and they share a common riemannian metric .
in addition , the orthogonal projection of an element of @xmath18 onto @xmath3 is denoted by @xmath19 .
then , the following remarkable lemma holds true .
@xcite [ lem : grad_proj ] let @xmath20 be a scalar field defined over @xmath17 such that the function @xmath6 defined on @xmath1 is a restriction of @xmath20 .
then , the gradient of @xmath6 satisfies the equation @xmath21
we consider the situation illustrated in fig .
[ fig : scenario ] where @xmath22 vision sensors @xmath23 are located in 3-d euclidean space .
let the fixed world frame be denoted by @xmath24 and the body fixed frame of sensor @xmath25 by @xmath26 .
we also denote the position vector of the origin of @xmath26 relative to @xmath24 by @xmath27 , and the rotation matrix of @xmath26 relative to @xmath24 by @xmath28 .
then , the pair @xmath29 , called _ pose _ , represents the configuration of sensor @xmath30 . in this paper , each sensor s position @xmath31 is assumed to be fixed , and sensors can control only their orientations @xmath32 .
in addition , we suppose that sensors are localized and calibrated _ a priori _ and @xmath33 is available for control .
we use the notation @xmath33 to describe not only the pose but also a coordinate transformation operator similarly to @xcite .
take two frames @xmath34 and @xmath35 .
let the pose of the frame @xmath35 relative to @xmath34 be denoted by @xmath36 , and the coordinates of a point relative to @xmath34 by @xmath37 .
then , the coordinates @xmath38 of the point relative to @xmath34 are given as @xmath39 let us next define the region to be monitored by a group of sensors @xmath40 .
in this paper , we assume that the region is a subset of a 2-d plane ( fig .
[ fig : scenario ] ) , where the 2-d plane is called the environment and the subset to be monitored is called the mission space .
let the set of coordinates of all points in the environment and the mission space relative to @xmath24 are respectively denoted by @xmath41 and @xmath42 . just for simplicity , we suppose that the world frame @xmath24 is attached so that its @xmath43-plane is parallel to the environment ( fig .
[ fig : scenario ] ) .
then , the set @xmath41 is formulated as @xmath44 with some constant @xmath45 , where @xmath46 is an @xmath30-th standard basis .
suppose that a metric @xmath47 , called a density function , indicating the relative importance of every point @xmath48 is defined over @xmath41 . in this paper , the function @xmath49
is assumed to be small if point @xmath50 is important and to satisfy @xmath51 with a constant @xmath52 such that @xmath53 .
a vision sensor has an image plane containing the sensing array , whose elements , called pixels , provide the numbers reflecting the amount of light incident .
we assume that the image plane is a rectangle as illustrated in fig .
[ fig : image_plane1 ] .
the set of position vectors of all points on the image plane relative to the sensor frame @xmath26 is denoted by @xmath54 .
now , the axes of the sensor frame @xmath26 is assumed to be selected so that its @xmath43-plane is parallel to the image plane and @xmath55-axis perpendicular to the image plane passes through the focal center of the lens
. then , the third element of any point in the set @xmath56 must be equal to the focal length @xmath57 .
we next denote the set of pixels of sensor @xmath58 by @xmath59 and the position vector of the center of the @xmath60-th pixel on the image plane of sensor @xmath30 relative to @xmath26 by @xmath61 .
since @xmath60 in @xmath62 and @xmath63 deffer , we may need to use the notation like @xmath64 but we omit the subscript to reduce notational complexity . in addition , the positions of its vertices relative to @xmath26 are denoted by @xmath65 ( fig . [ fig : image_plane2 ] ) .
when a point on the environment with coordinates @xmath66 relative to @xmath26 is captured by a sensor @xmath30 with @xmath33 , the point is projected onto the image plane as illustrated in fig .
[ fig : proj1 ] .
if the coordinates of the projected point are denoted by @xmath67 , it is well known that the projection is formulated as @xmath68 it is not difficult to show that the inverse map @xmath69 of the map @xmath70 ( fig .
[ fig : proj1 ] ) from @xmath71 to @xmath66 is given by @xmath72 note that , while @xmath70 is independent of @xmath32 , the map @xmath69 depends on @xmath32 and hence we describe @xmath69 as @xmath73 . and @xmath69.,width=245 ] using the map @xmath69 , we denote by @xmath74 the set of coordinates of the field of view ( fov ) of each sensor @xmath30 relative to @xmath24 , which is also a polytope . its @xmath40-polytope representation is trivial , namely it is given by the convex hull of the four points with coordinates @xmath75 relative to @xmath24 ( fig .
[ fig : fov2 ] ) .
the @xmath76-polytope representation is also computed efficiently as follows .
suppose now that @xmath60-th side line segment ( @xmath77 ) specifying the boundary of the image plane connects the vertices @xmath78 and @xmath79 without loss of generality .
then , the line projected onto the environment is also a line segment whose vertices have coordinates @xmath80 and @xmath81 relative to @xmath24 , and hence the line is formulated as @xmath82 where the matrix @xmath83 is derived as @xmath84 from the fact that @xmath85 and @xmath86 are on the line . since the coordinates @xmath87 for any interior @xmath88 of @xmath89 must be inside the fov , a half space specifying the fov is described by the inequality @xmath90 with @xmath91 in the same way , we can find the pair @xmath92 for all @xmath77 . stacking them
allows one to formulate the fov as @xmath93
in this section , we consider a simple case with @xmath94 .
let us first define the objective function to be _ minimized _ by sensor @xmath30 . in this paper , we basically take the concept of coverage control @xcite , where the objective function is defined by a sensing performance function and a density function at a point @xmath95 .
note however that we accumulate the function only at the center of the pixels projected onto the environment @xmath41 in order to reflect the discretized nature of the vision sensors . in the sequel , the sensing performance function and the density function at @xmath95
are denoted by @xmath96 and @xmath97 , respectively .
let us first define a function @xmath98 providing the coordinates in @xmath24 of the point on @xmath41 which is captured by @xmath60-th pixel as @xmath99 then , the objective function takes the form of @xmath100 where @xmath101 is a weighting coefficient .
if we impose a large @xmath102 on the pixel at around the center of the image , the sensor tends to capture the important area at around the image center .
if we need to accelerate computation , replacing @xmath103 in ( [ eqn : obj_single ] ) by its subset is an option . in order to ensure preciseness
, we need to introduce an extended function allowing @xmath104 , but we will not mention it since it can be easily avoided by choosing @xmath105 appropriately . ,
width=245 ] similarly to @xcite , we let the performance function @xmath105 depend only on the distance @xmath106 .
remark however that , differently from @xcite , the third element of @xmath107 is not controllable since the sensor is fixed .
this may cause a problem that penalty of seeing distant area does not work in the case that the element is large enough .
however , the element is not ignorable since it reflects heterogeneous characteristics of vision sensors in the multi - sensor case .
we thus use the weighting distance as @xmath108 with @xmath109 , where @xmath110 is introduced since the distance is scaled by the focal length .
suppose that @xmath111 is set as @xmath112)$ ] .
then , a large @xmath113 imposes a heavy penalty on viewing distant area and a small @xmath113 a light penalty on it .
in particular , when @xmath114 for some @xmath115 , ( [ eqn : perf1 ] ) is rewritten as @xmath116 once the density function @xmath117 is given , the goal is reduced to minimization of ( [ eqn : obj_single ] ) with ( [ eqn : perf2 ] ) under the restriction of @xmath118 . in order to solve the problem , this paper takes the gradient descent approach which is a standard approach to coverage control . for this purpose , it is convenient to define an extension @xmath119 such that @xmath120 if @xmath121 .
we first extend the domain of @xmath122 in ( [ eqn : q_wl ] ) from @xmath0 to @xmath123 as @xmath124 then , the vector @xmath125 is not always on the environment when @xmath126 but the function @xmath127 in ( [ eqn : perf1 ] ) is well - defined even if the domain is altered from @xmath41 to @xmath128 .
we thus denote the function with the domain @xmath128 by @xmath129 , and define the composite function @xmath130 relative to @xmath24 and @xmath26.,width=226 ] we next focus on the term @xmath131 in ( [ eqn : obj_single ] ) and expand the domain of the composite function from @xmath0 to @xmath132 . here ,
since @xmath133 is not always on @xmath41 , we need to design @xmath134 such that @xmath135 if @xmath95 . in this paper
, we assign to a point @xmath136 the density of a point @xmath137 where the operations are illustrated in fig .
[ fig : proj4 ] .
accordingly , the density function is defined by @xmath138 remark that , differently from @xmath127 , the function @xmath117 is not naturally extended and the selection of @xmath139 is not unique . the motivation to choose ( [ eqn : perf6 ] ) will be clear in the next subsection .
consequently , we define the extended objective function @xmath140 from @xmath132 to @xmath141 by using ( [ eqn : perf5 ] ) and ( [ eqn : perf6 ] ) .
let us finally emphasize that @xmath142 holds for any @xmath143 .
in the gradient descent approach , we update the rotation @xmath32 in the direction of @xmath144}^{so(3 ) } h_i$ ] at each time @xmath145 .
this subsection assumes that the density @xmath117 is not given _ a priori _ and that @xmath117 needs to be estimated from acquired vision data as investigated in @xcite .
let us first consider an ideal situation such that the density function is exactly projected onto the image plane , namely @xmath146(q)\ { \forall q
} \in fov_i({r_{wi}}[k ] ) , \label{eqn : psi_phi2}\ ] ] holds with respect to the density @xmath147 over the image plane .
then , the density function value @xmath49 is available at any point in the fov .
we next consider a point @xmath136 which does not always lie on @xmath41 .
then , the value of @xmath148 is also given by the same function as ( [ eqn : psi_phi2 ] ) since @xmath149\circ \phi \circ g_{wi}^{-1}[k ] ( \bar q ) \nonumber\\ \!\!&\!\!=\!\!&\!\ ! \psi \circ
\psi_i \circ g_{wi}^{-1}[k ] \circ g_{wi}[k]\circ \phi \circ g_{wi}^{-1}[k ] ( \bar q ) \nonumber\\ \!\!&\!\!=\!\!&\!\
! \psi \circ \psi_i \circ \phi \circ g_{wi}^{-1}[k ] ( \bar q ) = \psi \circ \psi_i \circ g_{wi}^{-1}[k ] ( \bar q ) .
\nonumber\end{aligned}\ ] ] ensuring the equality is the reason for choosing ( [ eqn : perf6 ] ) .
we next consider estimation of the density @xmath150 on the image since assuming ( [ eqn : psi_phi2 ] ) is unrealistic .
rich literature has been devoted to the information extraction from the raw vision data , and a variety of algorithms are currently available even without expert knowledge @xcite .
for example , it is possible to detect and localize in the image plane specific objects like cars or human faces , and even abstract targets such as everything moving or some environmental changes .
the present coverage scheme is indeed applicable to any scenario such that a nonnegative number @xmath151 reflecting its own importance is assigned to each pixel @xmath115 after conducting some image processing .
however , we mainly focus on a specific scenario of monitoring moving objects on the mission space .
suppose that a sensor captures a human walking from left to right in the image as in fig .
[ fig : snapshot ] .
then , a way to localize such moving objects is to compute optical flows from consecutive images as in fig . [
fig : snapshot ] , where the flows are depicted by yellow lines .
we also let the data @xmath151 be the norm of the flow vector at each pixel .
then , the plots of @xmath151 over the image plane are illustrated by green dots in fig .
[ fig : mix_gauss1 ] .
we next fit the data of @xmath151 by a continuous function defined over @xmath89 and use the function as @xmath150 .
such algorithms are also available even in real time @xcite . similarly to @xcite , we employ the mixed gaussian function known to approximate a variety of functions with excellent precision by increasing the number of gaussian functions , and widely used in data mining , pattern recognition , machine learning and statistical analysis .
[ fig : mix_gauss2 ] shows the gaussian function with @xmath152 computed so as to fit the data in fig .
[ fig : mix_gauss1 ] . of course , using a larger @xmath153 achieves a better approximation as shown in fig .
[ fig : mix_gauss_various ] . as a result
, we obtain a function in the form of @xmath154 over the 2-d image plane coordinates @xmath155 .
note that ( [ eqn : mix_gauss_image2 ] ) is large when @xmath156 captures an important point , which is opposite to the density function .
thus , we define the function @xmath157 where @xmath158 is a positive scalar guaranteeing @xmath159 for all @xmath156 .
it is also convenient to define @xmath160 for all 3-d vectors @xmath161 on the image plane as @xmath162 ( left ) and @xmath163(right).,title="fig:",width=151,height=105 ] ( left ) and @xmath163(right).,title="fig:",width=151,height=105 ] here , we will derive the gradient @xmath144}^{so(3 ) } h_i$ ] , given a rotation @xmath164 \in so(3)$ ] and @xmath150 in ( [ eqn : mix_gauss ] ) .
it is widely known that @xmath165 is formulated as @xmath166 , where @xmath167 is the set of all skew symmetric matrices in @xmath132 .
we also define the operator @xmath168(wedge ) from @xmath128 to @xmath132 such that @xmath169 for the cross product @xmath170 .
the rotational group @xmath0 is known to be a submanifold of a riemannian manifold @xmath123 with @xmath171 and the riemannian metric @xmath172 @xcite .
it is also known that the orthogonal projection @xmath173 of matrix @xmath174 onto @xmath165 in terms of the riemannian metric induced by ( [ eqn : riemannian ] ) is given by @xmath175 see subsection 3.6.1 of @xcite for more details .
now , we have the following theorem , where we use the notation @xmath176 and @xmath177 .
suppose that the objective function @xmath178 is formulated by ( [ eqn : obj_single_fict ] ) with ( [ eqn : perf5 ] ) , ( [ eqn : perf6 ] ) and ( [ eqn : mix_gauss ] ) .
then , the gradient @xmath144}^{so(3 ) } h_i$ ] is given by @xmath179}^{so(3 ) } h_i \!\!&\!\!=\!\!&\!\ !
p_{{r_{wi}}[k]}\left({{\rm grad}}_{{r_{wi}}[k]}^{{{\mathbb r}}^{3\times 3 } } \bar h_i\right ) , \label{eqn : proj2}\\ { { \rm grad}}_{{r_{wi}}[k]}^{{{\mathbb r}}^{3\times 3 } } \bar h_i \!\!&\!\!=\!\!&\!\ ! \tilde \delta_i \eta_i^t({r_{wi}}[k ] ) p_{il}^t , \nonumber\end{aligned}\ ] ] where @xmath180 see appendix [ app:1 ] .
namely , just running the dynamics @xmath181 leads @xmath32 to the set of critical points of @xmath182 .
however , in practice , the vision data is usually obtained at discrete time instants and hence we approximate the continuous - time algorithm ( [ eqn : grad_descent1 ] ) by @xmath183 = { r_{wi}}[k]{\rm exp}\left({r_{wi}}^t[k ] \left(\alpha_k { { \rm grad}}_{{r_{wi}}[k]}^{so(3 ) } h_i\right)\right ) .
\label{eqn : grad_descent2}\end{aligned}\ ] ] see @xcite for the details on the selection of @xmath184 . in the above discussion , we assume that the sensor can take full 3-d rotational motion . however , the motion of many commoditized cameras is restricted by the actuator configurations .
hereafter , we suppose that the sensor can be rotated around two axes @xmath185 ( @xmath186 ) and @xmath187 ( @xmath188 ) , where these vectors are defined in @xmath26 and assumed to be linearly independent of each other .
these axes may depend on the rotation matrix @xmath32 .
for example , in the case of pan - tilt ( pt ) cameras in figs .
[ fig : pan ] and [ fig : tilt ] , which are typical commoditized cameras , the axis of the pan motion ( fig .
[ fig : pan ] ) is fixed relative to @xmath24 , while that of the tilt motion ( fig .
[ fig : tilt ] ) is fixed relative to the sensor frame @xmath26 .
then , only one of the two axes depends on @xmath32 .
note that even when there is only one axis around which the sensor can be rotated , the subsequent discussions are valid just letting @xmath189 .
let us denote a normalized vector @xmath190 ( @xmath191 ) orthogonal to the @xmath192-plane .
then , the three vectors @xmath185 , @xmath193 and @xmath194 span @xmath128 .
thus , any element @xmath195 of @xmath165 can be represented in the form of @xmath196 .
now , we define a _ distribution _ @xmath197 @xcite assigning @xmath118 to the subspace @xmath198 whose dimension is @xmath79 .
the distribution @xmath197 is clearly regular and hence induces a submanifold @xmath199 of @xmath0 @xcite , called integral manifold , such that its tangent space @xmath200 at @xmath201 is equal to ( [ eqn : submanifold ] ) .
the manifold @xmath199 specifies orientations which the camera can take . since @xmath199 is a submanifold of @xmath0 , a strategy similar to theorem 1 is available and we have the following corollary .
suppose that the objective function @xmath178 is formulated by ( [ eqn : obj_single_fict ] ) with ( [ eqn : perf5 ] ) , ( [ eqn : perf6 ] ) and ( [ eqn : mix_gauss ] ) .
then , the gradient @xmath144}^{{\mathcal s}_{ua } } h_i$ ] is given by @xmath179}^{{\mathcal s}_{ua } } h_i \!\!&\!\!=\!\!&\!\ !
p^{ua}_{{r_{wi}}[k]}\left({{\rm grad}}_{{r_{wi}}[k]}^{so(3 ) } h_i\right ) \label{eqn : proj3}\end{aligned}\ ] ] where the orthogonal projection @xmath202 of @xmath203 to @xmath200 is defined by @xmath204 with @xmath205 @xmath206 , where @xmath207 if @xmath208 and @xmath209 if @xmath210 .
we see from this corollary that the gradient @xmath144}^{so(3 ) } h_i$ ] on @xmath0 is utilized as it is and we need only to project it through ( [ eqn : proj4 ] ) .
also , the projection ( [ eqn : proj4 ] ) is successfully parameterized by the vectors @xmath211 and @xmath187 .
in this section , we extend the result of the previous section to the multi - sensor case .
the difference from the single sensor case stems from the overlaps of the fovs with the other sensors as illustrated in fig .
[ fig : overlap ] .
@xcite present sensing performance functions taking account of the overlaps and their gradient decent laws
. however , in this paper , we present another simpler scheme to manage the overlap .
let us first define the set of sensors capturing a point @xmath95 within the fov as @xmath212 where @xmath213 .
we also suppose that , when @xmath214 has multiple elements for some @xmath215 , only the data of the sensor with the minimal sensing performance ( [ eqn : perf1 ] ) among sensors in @xmath214 is employed in higher - level decisions and recognitions .
this motivates us to partition @xmath74 into the two region @xmath216 then , what pixel @xmath60 captures a point in @xmath217 is identified with what it captures a point outside of @xmath42 , whose cost is set as @xmath218 in the previous section , in the sense that both of the data are not useful at all .
this is reflected by assigning @xmath52 to the pixels @xmath219 with @xmath220 .
accordingly , we formulate the function to be minimized by @xmath40 as @xmath221 with @xmath222 remark that ( [ eqn : obj_multi ] ) differs from ( [ eqn : obj_single ] ) only in the set @xmath223 . strictly speaking , to compute the gradient of ( [ eqn : obj_multi ] ) , we need to expand @xmath224 from @xmath225 to @xmath226 . for this purpose , it is sufficient to define @xmath227 from @xmath132 to a subset of @xmath41 .
for example , an option is to define an extension @xmath228 of ( [ eqn : arxiv1 ] ) similarly to ( [ eqn : arxiv2 ] ) , and to let @xmath227 be the convex full of these points .
however , at the time instants computing the gradient with @xmath229 $ ] , the extension @xmath230 for a sufficiently small perturbation @xmath231 is equivalent to the original set @xmath223 irrespective of the selection of @xmath227 except for the pathological case when a pixel is located on the boundary of @xmath232 .
namely , ignoring such pathological cases which do not happen almost surely for ( [ eqn : grad_descent2 ] ) , the gradient can be computed by using the set @xmath233)$ ] instead of its extension .
hence , the gradient is simply given as theorem 1 by just replacing @xmath234 by @xmath223 .
note that the curve fitting process is run without taking account of whether @xmath235 or not , and @xmath52 is assigned to @xmath236 at the formulation of @xmath150 as in ( [ eqn : mix_gauss ] ) .
this is because letting @xmath237 at the curve fitting stage would degrade the density estimation accuracy at around the boundary of @xmath238 .
the remaining issue is efficient computation of the set @xmath234 .
hereafter , we assume that each sensor acquires @xmath239 , i.e. @xmath240 and @xmath241 for all @xmath242 , and its index @xmath243 through ( all - to - all ) communication or with the help of a centralized computer .
the computation under the limited communication will be mentioned at the end of this section .
in addition , we suppose that every sensor stores the set @xmath244 for all @xmath245 which can be computed off - line since the sensor positions are fixed .
s ( left ) and @xmath246s ( right).,width=151,height=94 ] s ( left ) and @xmath246s ( right).,width=151,height=94 ] s ( left ) and @xmath246s ( right).,width=151,height=94 ] s ( left ) and @xmath246s ( right).,width=151,height=94 ] s ( left ) and @xmath246s ( right).,width=151,height=94 ] s ( left ) and @xmath246s ( right).,width=151,height=94 ] then , the set @xmath217 is computed as @xmath247 in polynomial time with respect to @xmath22 .
namely , checking @xmath248 for all @xmath249 provides @xmath223 .
the computation process including image processing , curve fitting and gradient computation is successfully distributed to each sensor but the resulting fovs need to be shared among all sensors to compute @xmath223 .
a way to implement the present scheme under limited communication is to restrict the fov of each sensor so that the fov can overlap with limited number of fovs of the other sensors .
such constraints on the fovs are easily imposed by adding an artificial potential to the objective function but we leave the issue as a future work due to the page constraints .
in this section , we demonstrate the utility of the present approach through simulation using 4 cameras with @xmath250 mm @xmath251 . here
, we suppose that the view of the environment from @xmath24 with @xmath252 m and focal length @xmath253 mm is given as in fig .
[ fig : initial_image ] , and that the mission space @xmath42 is equal to the fov corresponding to the image .
since the codes of simulating the image acquisition and processing are never used in experiments , we simplify the process as follows , and demonstrate only the present coverage control scheme with the curve fitting process . before running the simulation , we compute the optical flows for the images of fig . [
fig : initial_image ] as in fig .
[ fig : of_image ] , and also fitting functions of the data as in fig .
[ fig : curve_fit ] .
the resulting data is uploaded at http://www.fl.ctrl.titech.ac.jp/paper/2014/data.wmv .
then , we segment the image by the superlevel set of the function using a threshold @xmath254 , and assign a boolean variable @xmath78 to @xmath151 if @xmath255 is inside of the set and assign @xmath256 otherwise .
the experimental system is now under construction , and the experimental verification of the total process will be conducted in a future work .
note however that it is at least confirmed that the skipped image acquisition and processing can be implemented within several milliseconds in a real vision system .
.,width=151,height=94 ] .,width=151,height=94 ] .,width=151 ] let the position vectors of cameras be selected as @xmath257 and the length of each side of the image plane be @xmath258 mm and @xmath259 mm .
the other elements of @xmath31 are set as illustrated by the mark @xmath260 in fig .
[ fig : static ] .
the parameters in @xmath261 is set as @xmath262 , @xmath263 , @xmath264 and @xmath265)$ ] .
the curve fitting process is run with @xmath266 and the gradient is computed by evaluating the objective function not at all points in @xmath103 but at 121 points extracted from @xmath103 . in order to confirm convergence of the orientations ,
we first fix the image as in fig .
[ fig : static ] and run the present algorithm from the initial condition @xmath267 .
then , the evolution of the function @xmath261 is illustrated in fig .
[ fig : obj ] , where we compute the value using not the individually estimated density but the data as in fig .
[ fig : curve_fit ] .
we see from the figure that the function @xmath261 is decreasing through the update process and eventually reaches a stationary point .
the final configuration is depicted in the right figure of fig .
[ fig : static ] .
we next start to play the above movie and check adaptability to environmental changes , where the orientations are assumed to be updated at each frame .
then , the evolution of fovs are shown in http://www.fl.ctrl.titech.ac.jp/paper/2014/sim.wmv whose snapshots at times @xmath268 are depicted in fig .
[ fig : snaps ] .
we see from the movie and figures that the cameras adjust their rotations so as to capture moving humans .
the above results show the effectiveness of the present approach .
in this paper , we have investigated visual coverage control where the vision sensors are assumed to be distributed over the 3-d space to monitor the 2-d environment and to be able to control their orientations .
we first have formulated the problem as an optimization problem on @xmath0 .
then , in order to solve the problem , we have presented the entire process including not only the gradient computation but also image processing and curve fitting , which are required to estimate the density function from the acquired vision data . finally , we have demonstrated the effectiveness of the approach through simulation of moving objects monitoring .
for notational simplicity , we describe @xmath164 $ ] by @xmath269 in the sequel . substituting ( [ eqn : perf5 ] ) , ( [ eqn : perf6 ] ) and ( [ eqn : mix_gauss ] ) into ( [ eqn : obj_single_fict ] )
, the objective function to be minimized is formulated as @xmath270 from lemma 1 and the fact that @xmath0 is a submanifold of @xmath132 , we first compute the gradient @xmath273 . from definition 1 and ( [ eqn : derivative_lin_m ] ) , we need to compute the directional derivative @xmath274 $ ] . from linearity of the directional derivative operator @xmath275 ,
it is sufficient to derive @xmath276 $ ] and @xmath277 $ ] .
we first consider @xmath276 $ ] . by calculation
, we have @xmath278 hence , @xmath279 = \lim_{t\to 0}\frac{\tilde h^{l}_i(r + \xi t ) - \tilde h^{l}_i(r)}{t}$ ] is given by @xmath280 \!\!&\!\!=\!\!&\!\ ! \tilde
\eta_i^l(r)\xi p_{il } , \label{eqn : modify2}\\ \eta_i^l(r ) \!\!&\!\!=\!\!&\!\ ! \frac{2 ( ( { \bf e}_3^t r p_{il})p_{il}^tr^tw - \|rp_{il}\|_w^2 { \bf e}_3^t ) } { ( { \bf e}_3^t r p_{il})^3}. \nonumber\end{aligned}\ ] ] let us next consider @xmath281 $ ] .
we first have the equations @xmath282 hence , we also have @xmath283 we also obtain @xmath284 where @xmath285 , @xmath286 and @xmath287 are introduced for notational simplicity . using @xmath288 , we can decompose low and high order terms in @xmath289 as @xmath290 ( [ eqn : app4 ] ) is also simplified as @xmath291 where @xmath292 substituting ( [ eqn : app4 ] ) and ( [ eqn : app6 ] ) into ( [ eqn : app1 ] ) yields @xmath293 let us now compute @xmath294 $ ] . substituting ( [ eqn : app7 ] ) into the definition of the directional derivative ( [ eqn : derivative_lin_m ] ) , i.e. @xmath295 \!\!&\!\!=\!\!&\!\!\lim_{t \to 0}\frac{\bar h^{lj}_i(r + \xi t ) - \bar h^{lj}_i(r)}{t } , \label{eqn : app8}\end{aligned}\ ] ] we have @xmath296 = \frac{\|rp_{il}\|_w^2({\bf e}_3^t r p_{il } ) } { ( { \bf e}_3^tr p_{il})^3 } \left(\frac{d h_i^{lj}}{dt}\right)(0 ) \nonumber\\ \hspace{-1.2 cm } & & \hspace{.2 cm } + 2h_i^{lj}(0 ) \frac { ( { \bf e}_3^tr
p_{il})p_{il}^tr^tw\xi p_{il } -\|rp_{il}\|_w^2{\bf e}_3^t\xi p_{il } } { ( { \bf e}_3^tr p_{il})^3 } \label{eqn : app9}\end{aligned}\ ] ] by calculation , the derivative @xmath297 is given by @xmath298 and hence @xmath299 substituting ( [ eqn : app10 ] ) and definitions of @xmath300 and @xmath301 into ( [ eqn : app9 ] ) yields @xmath302 = \bar \eta_{i}^{lj}(r ) \xi p_{il } \label{eqn : app13}\\ \hspace{-.8cm}&&\eta_{i}^{lj}(r ) = \frac{2e^{-\|b_{lj}\|^2_{\sigma_j } } } { \lambda_i({\bf e}_3^tr p_{il})^3 } \big(({\bf e}_3^tr p_{il})\xi^{lj}_i(r ) -\lambda_i\|rp_{il}\|_w^2{\bf e}_3^t \big ) \nonumber\\ \hspace{-.8cm}&&\xi^{lj}_i(r ) = \|rp_{il}\|_w^2b_{lj}^t \sigma_j ( p_{il } { \bf e}_3^t - \lambda_i i_3)r^t + \lambda_i p_{il}^tr^tw . \nonumber\end{aligned}\ ] ] note that @xmath286 is constant and @xmath303 is independent of the matrix @xmath304 . from ( [ eqn : obj_single_fict3 ] ) , ( [ eqn : modify2 ] ) and ( [ eqn : app13 ] ) , we obtain @xmath305 = \tilde \delta_i \eta_i(r ) \xi p_{il } = { { \rm tr}}\big(\xi^t
\big(\tilde \delta_i \eta_i^t(r ) p_{il}^t \big)\big ) , \nonumber\\ \hspace{-.8cm}&&\eta_i(r ) = \sum_{l \in \tilde { { \cal l}}_i^c(r)}w_{il } \bar{\phi } \eta_i^l+\!\ ! \sum_{l \in \tilde { { \cal l}}_i(r)}\!\ ! w_{il } \big ( \bar{\psi } \eta_i^l - \sum_{j=1}^m \alpha_j \eta_{i}^{lj } \big ) .
\nonumber\end{aligned}\ ] ] from definition [ def : grad_m ] , we have @xmath306 .
combining it with lemma 1 and ( [ eqn : proj_so(3 ) ] ) completes the proof .
b. song , c. ding , a. kamal , j. a. farrell and a. roy - chowdhury , `` distributed camera networks : integrated sensing and analysis for wide area scene understanding , '' _ ieee signal processing magazine _ , vol .
3 , pp . 2031 , 2011 .
t. hatanaka and m. fujita , `` cooperative estimation of averaged 3d moving target object poses via networked visual motion observers , '' _ ieee trans .
automatic control _
3 , pp . 623638 , 2013 .
b. m. schwager , b. j. julian , m. angermann and d. rus , `` eyes in the sky : decentralized control for the deployment of robotic camera networks , '' _ proc . of the ieee _
9 , pp . 15411641 , 2011 .
t. hatanaka , y. wasa and m. fujita game theoretic cooperative control of ptz visual sensor networks for environmental change monitoring _ proc . of 52nd ieee conf . on decision and control
_ , to appear , 2013 j. cortes , s. martinez , and f. bullo , `` spatially - distributed coverage optimization and control with limited - range interactions , '' esaim : control , optimisation & calculus of variations , vol .
691719 , 2005 . | this paper investigates coverage control for visual sensor networks based on gradient descent techniques on matrix manifolds .
we consider the scenario that networked vision sensors with controllable orientations are distributed over 3-d space to monitor 2-d environment .
then , the decision variable must be constrained on the lie group @xmath0 .
the contribution of this paper is two folds .
the first one is technical , namely we formulate the coverage problem as an optimization problem on @xmath0 without introducing local parameterization like eular angles and directly apply the gradient descent algorithm on the manifold .
the second technological contribution is to present not only the coverage control scheme but also the density estimation process including image processing and curve fitting while exemplifying its effectiveness through simulation of moving objects monitoring . |
using data from experiment e791 at fermilab , we reconstruct pairs of charm mesons produced in 500 gev/@xmath2 @xmath3-nucleon interactions , where @xmath10 gev , and use correlations between the mesons to probe two aspects of the hadroproduction of mesons containing a heavy quark : the dynamics of the production of heavy quark - antiquark pairs and the subsequent hadronization of the quarks into hadrons . correlations between the @xmath0 and @xmath11 momenta transverse to the beam direction are sensitive to corrections to the leading - order calculations of the @xmath12 cross section
. correlations between the longitudinal momenta , as well as differences in the production rates of the four types of @xmath13 pairs ( @xmath14 , @xmath15 , @xmath16 , and @xmath17 ) , provide information regarding the role of the remnants of the colliding hadrons in the hadronization process that transforms the charm quarks into charm mesons . in most studies of the hadroproduction of charm particles , distributions for _ single _ charm particles are used to probe the underlying production physics @xcite . the variables used to describe the single particle distributions are the transverse momentum with respect to the beam direction , @xmath18 , and either the rapidity @xmath19 or the feynman scaling variable @xmath20 , where @xmath21 @xmath22 @xmath23 and @xmath24 are the center - of - mass energy and longitudinal momentum of the charm particle and @xmath25 is the total center - of - mass energy .
the center of mass is that of the pion - nucleon system .
such single charm studies are insensitive to correlations between the two charm hadrons in a single event .
we have fully reconstructed @xmath4 @xmath13 pairs .
based on this sample , we present background - subtracted , acceptance - corrected distributions for the following variables : 1 . the invariant mass of the pair of charm mesons , @xmath26 ; 2 .
the square of the vector sum of the transverse momenta , with respect to the beam direction , of the @xmath0 and @xmath11 mesons ( @xmath27 ) ; 3 .
correlations between @xmath28 and @xmath29 , as well as @xmath30 and @xmath31 ; 4 .
@xmath32 and @xmath33 ; 5 .
@xmath34 and @xmath35 ; 6 .
correlations between the squares of the magnitudes of the transverse momenta of the @xmath0 and @xmath11 mesons , @xmath36 and @xmath37 ; 7 .
@xmath38 and @xmath39 ; 8 .
the azimuthal separation between the momentum vectors of the @xmath0 and @xmath11 mesons in the plane perpendicular to the beam direction , @xmath40 ( minimum of @xmath41 and @xmath42 ) ; 9 .
correlations between the azimuthal separation ( @xmath43 ) and the scalar sum and difference of the @xmath0 and @xmath11 transverse momenta , @xmath44 and @xmath45 ; in addition , this paper reports the relative production rates for each type of @xmath13 pair ( @xmath14 , @xmath15 , @xmath16 , and @xmath17 ) , and compares the rapidity correlations for the various @xmath13 pair combinations .
we also investigate the extent to which the observed charm - pair correlations can be duplicated by simply convoluting the observed single charm particle distributions .
in addition , we compare our measured distributions to three sets of theoretical predictions : 1 .
the distributions of @xmath12 pairs from a next - to - leading - order perturbative qcd calculation by mangano , nason and ridolfi@xcite ; 2 .
the distributions of @xmath12 pairs from the pythia / jetset monte carlo event generator@xcite which uses a parton - shower model to include higher - order perturbative effects@xcite ; and 3 . the distributions of @xmath13 pairs from pythia / jetset which uses the lund string model to transform @xmath12 pairs to @xmath13 pairs@xcite
. in table [ tab : tab1 ] , we compare the e791 charm - pair sample to those from other fixed - target experiments ( both hadroproduction and photoproduction ) . the largest previous sample of fully - reconstructed hadroproduced charm pairs used to study correlations is 20 pairs from the cern @xmath3-nucleon experiment na32@xcite .
some studies have been conducted with partially - reconstructed charm hadrons , in which the direction but not necessarily the magnitude of the charm particle momentum is determined directly . na32
partially reconstructed 642 such charm pairs@xcite .
in photoproduction experiments , the largest sample of charm pairs reconstructed is from the e687 data@xcite , with 325 fully - reconstructed and 4534 partially - reconstructed charm pairs . in the e687 partially - reconstructed sample , one @xmath0 meson
is fully reconstructed and the momentum vector of the other charm meson is determined by scaling the momentum vector of low - momentum charged pions from the decays @xmath46 .
[ htbp ] .
summary of fully - reconstructed and partially - reconstructed charm - pair samples from hadroproduction and photoproduction fixed - target experiments . in the last column , we list the physics variables studied in each experiment . the variables are defined in the text . [
cols="<,^,^,^ " , ] [ tab : tab2 ] momentum analysis was provided by two dipole magnets that bent particles in the same direction in the horizontal plane .
the transverse momentum kicks were 212 mev/@xmath2 for the first magnet and 320 mev/@xmath2 for the second magnet .
the centers of the two magnets were 2.8 m and 6.2 m downstream of the last target , respectively .
the @xmath47 aperture of the pole faces of the first magnet was 183 cm by 81 cm by 100 cm and that of the second magnet was 183 cm by 86 cm by 100 cm . two segmented , gas - filled , threshold erenkov counters@xcite provided particle identification over a large range of momenta .
the threshold momenta above which a charged particle emits light were 6 , 20 and 38 gev/@xmath2 for @xmath48 s , @xmath49 s , and @xmath50 s , respectively , for the first counter , and 11 , 36 , and 69 gev/@xmath2 for the second .
the particle identification algorithm correlates the erenkov light observed in a given mirror - phototube segment with the charged particle tracking information .
the algorithm indicates the likelihood that a charged particle of a given mass could have generated the observed erenkov light in the segment(s ) in question .
the electromagnetic calorimeter , which we called the segmented liquid ionization calorimeter ( slic ) , consisted of 20 radiation lengths of lead and liquid scintillator and was 19 m from the target .
layers of scintillator counters 3.17 and 6.24 cm wide were arranged transverse to the beam and their orientations alternated among horizontal and @xmath51 with respect to the vertical direction@xcite .
the hadronic calorimeter consisted of six interaction lengths of steel and acrylic scintillator .
there were 36 layers , each with a 2.5-cm - thick steel plate followed by a plane of 14.3-cm - wide by 1-cm - thick scintillator slats ; the slats were arranged alternately in the horizontal and vertical directions , and the upstream and downstream halves of the calorimeter were summed separately@xcite .
the signals from the hadronic calorimeter as well as those from the electromagnetic calorimeter were used for electron identification .
signals from both calorimeters were used to form the transverse energy requirement in the hardware trigger @xcite .
electron identification was not used in this analysis .
muons were identified by two planes of scintillation counters located behind a total of 15 interaction lengths of shielding , including the calorimeters . the first plane , 22.4 m from the target , consisted of twelve 40-cm - wide by 300-cm - long vertical scintillation counters in the outer region and three counters 60 cm wide in the central region .
the second plane , added for e791 , consisted of 16 scintillation counters 24.2 m from the target .
these counters were each 14 cm wide and 300 cm long , and measured position in the vertical plane .
these counters were equipped with tdc s which provided information on the horizontal position of the incident muons @xcite .
muon identification was not used in this analysis . to minimize biasing the charm data sample ,
the trigger requirements were very loose .
the most significant requirements were that the signal in a scintillation counter downstream of the target be at least four times the most likely signal from one minimum - ionizing particle , and that the sum of the energy deposited in the electromagnetic and hadronic calorimeters , weighted by the sine of the angle relative to the beam , be above a threshold corresponding to 3 gev of transverse energy .
the time for the full hardware trigger decision was about 470 ns .
this trigger was fully 100% efficient for charm decays .
a total of 24,000 channels were digitized and read out in 50 @xmath52s with a parallel - architecture data acquisition system@xcite .
events were accepted at a rate of 9 khz during the 23-second tevatron beam spill .
the typical recorded event size was 2.5 kbytes .
data were written continuously ( during both the 23-second spill and the 34-second interspill periods ) to forty - two exabyte@xcite model 8200 8-mm - tape drives at a rate of 9.6 mbytes / s . over @xmath53 hadronic interactions
were recorded on 24,000 tapes .
the @xmath53 interactions recorded constitute about 50 terabytes of data . event reconstruction and filtering took place over a period of two and a half years at four locations : the university of mississippi , the ohio state university ( moved to kansas state university in 1993 ) , fermi national accelerator laboratory , and o centro brasileiro de pesquisas fsicas , rio de janeiro ( cbpf ) . the first three sites used clusters of commercial unix / risc workstations controlled from a single processor with multiprocessor management software@xcite , while cbpf used acp - ii custom - built single - board computers@xcite . as part of the reconstruction stage ,
a filter was applied which kept @xmath5420% of the events .
this filter was effectively an offline trigger . to pass this filter ,
an event was required to have a reconstructed primary production vertex whose location coincided with one of the target foils .
the event also had to include at least one of the following : 1 .
at least one reconstructed secondary decay vertex of net charge 0 for an even number of decay tracks and @xmath55 for an odd number of decay tracks .
the longitudinal separation of the secondary vertex from the primary had to be at least four sigma for secondary vertices with three or more tracks and at least six sigma for vertices with two tracks , where sigma is the error in the separation , 2 .
at least one reconstructed @xmath56 or @xmath57 candidate whose decay was observed upstream of the first magnet , 3 . and for part of the run , at least one reconstructed @xmath58 candidate . for one - third of the data sample
, several other classes of events were also kept .
these are included in analyses not covered in this paper : 1 .
events in which the net charge of all the reconstructed tracks was negative and their total momentum was a large fraction of the beam momentum .
2 . @xmath56 or @xmath57 candidates that decayed inside the aperture of the first magnet . following the initial reconstruction / filter , which was applied to all events , additional selections of events were made to further divide the large data sample into subsets by class of physics analysis .
the important detector performance characteristics for this analysis are the resolution for reconstructing the positions of both the primary interaction and secondary decay vertices , the efficiency for reconstructing the trajectories of charged particles , the resolution for measuring charged track momenta , and the efficiency and the misidentification rates for identifying charged pions and kaons using information from the erenkov counters .
the resolution for measuring the position of the primary vertex along the beam direction varies from about 240 @xmath52 m for the most downstream target foil to 450 @xmath52 m for the upstream foil .
the variation is due to the extrapolation from the smd system and to multiple scattering in material downstream of the interaction .
the mean number of reconstructed tracks used to fit the primary vertex is seven .
the measured secondary vertex resolution depends on the decay mode , the momentum of the @xmath0 , and the selection criteria .
for example , the vertex resolutions along the beam direction for @xmath59 and @xmath60 are 320 and 395 @xmath52 m , respectively , for a mean @xmath61 momentum of 65 gev/@xmath2 , and worsen by 33 and 36 @xmath52 m for every 10 gev/@xmath2 increase in @xmath61 momentum . the total efficiency , including acceptance , for reconstructing charged tracks is approximately 80% for particles with a momentum greater than 30 gev/@xmath2 and drops to around 60% for particles of momentum 10 gev/@xmath2 .
( this includes the inefficiency in the beam region . ) for tracks which pass through both magnets and have a momentum greater than 10 gev/@xmath2 , the average resolution for measuring charged particle momentum @xmath50 is @xmath62 where @xmath63 stands for the quadratic sum , and @xmath50 is in gev/@xmath2 .
tracks which pass through only the first magnet have a resolution @xmath64 . the mean @xmath0 mass resolution for hadronic decays to two , three and four charged particles varies from 13 to 8 mev/@xmath65 as the decay multiplicity increases .
the mass resolution varies by about a factor of 2 between low and high momentum @xmath0 mesons . in most e791 analyses , the erenkov counters play a very important role @xcite .
however , in this analysis with the two fully reconstructed d - meson decays , the erenkov counters play a minimal role .
we use the erenkov counters for charged kaon identification .
the kaon identification efficiencies and misidentification probabilities vary with longitudinal and transverse momentum and with the signatures required in the v cerenkov counters . for typical particle momenta in the range 20 gev/@xmath2 to 40 gev/@xmath2 , and for the nonstringent criteria used for some of the final states , in this analysis , the erenkov identification efficiency for a kaon ranged from 64% to 72% while the probability for a pion to be misidentified as a kaon ranged from 6% to 12% .
a complete monte carlo simulation of the apparatus was used in this analysis to calculate the efficiency and investigate systematic effects .
the simulation included all relevant physical processes such as multiple interactions and multiple scattering as well as geometrical apertures and resolution effects .
it produced data in the same format as the real experiment .
that monte carlo data was then reconstructed and analyzed with the same software as the real data .
in each e791 event , we search for two charm mesons ( @xmath61 , @xmath66 , @xmath67 or @xmath68 ) decaying to cabibbo - favored final states that can be reconstructed with relatively high efficiency : @xmath69 , @xmath70 , @xmath71 , and the charge conjugate modes .
to optimize the efficiency for reconstructing charm pairs , we search for both @xmath0 candidates simultaneously , rather than searching for the two candidates consecutively . in such a simultaneous search
, we can require that one candidate or the other satisfy a fairly stringent selection criterion based on a particular variable used to discriminate charm decays from background , or that both candidates satisfy less stringent criteria .
we start with a sample of events , each containing two candidate @xmath59 , @xmath72 or @xmath60 combinations with invariant mass between 1.7 and 2.0 gev@xmath73 , and rapidity in the range @xmath74 .
these candidates are found by looping over all reconstructed tracks .
the primary vertex is then refit after removing tracks which are associated with either candidate .
no particle identification requirements are applied at this time .
candidates are rejected if any charged track , the primary vertex or either of the two secondary vertices do not meet minimal fit quality criteria .
the sample of candidate pairs that pass just these criteria is dominated by combinatoric backgrounds . to choose further selection criteria
, we use this sample to represent background . to represent signal , we use reconstructed charm pairs generated with the monte carlo program described at the end of the previous section .
we then search for selection criteria that provide optimal discrimination between signal and background . in order to extract the signal , we use selection criteria defined by discrimination variables ( properties of the candidate event ) and minimum or maximum allowed values for each variable . for candidate pairs with the same final states (
_ i.e. _ , both @xmath75 , both @xmath76 , or both @xmath77 ) , the same discrimination variables and maximum or minimum values are used for both candidate @xmath0 s ; for pairs with different final states , the discrimination variables are allowed to be different for the two candidate @xmath0 s .
the discrimination variables used address the following questions .
is a @xmath0 candidate consistent with originating from the primary interaction vertex ? is the vertex formed by the decay products of a @xmath0 candidate well separated from the primary interaction vertex and not inside a target foil ? do any of the decay products of the @xmath0 candidate appear to originate from the primary interaction vertex or from the other @xmath0 candidate vertex ? is the scalar sum of the squares of the transverse momenta of the @xmath0 candidate decay products , with respect to the @xmath0 candidate trajectory , indicative of a heavy meson decay ? is the erenkov information for the kaon candidate consistent with that for a real kaon ? as an example of the cuts used : the @xmath78 balance cut was 400 mev/@xmath2 , and the secondary vertices were separated from the primary vertices by 8 times the rms - uncertainty in the separation . to optimize the significance of the signal , we repeatedly choose the selection criterion that maximizes @xmath79 while rejecting no more than 5% of the ( monte carlo ) signal .
@xmath80 is the number of signal pairs satisfying the selection criterion , determined by monte carlo simulation , and @xmath81 is the number of background pairs , determined from the data .
this requires properly normalizing the number of signal pairs in the monte carlo to the number of signal pairs in the data .
when the background becomes dominated by pairs with only one true @xmath0 decay , we exclude from the background sample only those pairs in which both @xmath0 candidates lie in a narrow range around the @xmath0 mass .
we iterate the procedure of finding the optimal selection criterion ( always allowing variables to be reused in subsequent iterations ) until the significance of the signal no longer increases .
the selection criteria are optimized separately for each of five decay topologies of @xmath13 pairs : 2 - 2 , 3 - 3 , 2 - 3 , 2 - 4 and 3 - 4 , where each integer represents the number of charged particles in the decay .
we find that selection criteria are more often applied to one @xmath0 candidate or the other , rather than to both , especially early in the optimization procedure . in several cases ,
a criterion will be applied to one of the @xmath0 candidates , and a more stringent criterion involving the same discrimination variable will be applied to the other . after optimizing our selection criteria
, we end up with a sample of 9254 events in the data with both @xmath0 candidates in the mass range 1.7 to 2.0 gev@xmath73 and in the rapidity range @xmath82 to @xmath83 . only pairs
in which the two @xmath0 candidates have opposite charm quantum numbers are included in this sample .
no significant signal for @xmath84 or @xmath85 pairs is observed . in fig .
[ fig : fig4 ] , we plot the mass of the @xmath0 candidate versus the mass of the @xmath11 candidate , for all five types of pairs .
three types of candidate pairs are evident in this scatter plot .
combinatoric background pairs consisting of a fake @xmath0 and a fake @xmath11 candidate are spread over the entire plot .
the density of these points decreases linearly with increasing candidate-@xmath0 mass .
pairs containing one real and one fake @xmath0 candidate appear as horizontal and vertical bands ( called @xmath0 and @xmath11 ridge events , respectively ) . in the center of the plot , we see an enhancement due to the crossing of the two bands and due to real pairs of @xmath0 mesons .
in this section we describe the analysis procedures by which we determine the number of signal events in the full data sample shown in fig . [
fig : fig4 ] , as well as in each bin of the physics variables used to study the charm - pair production .
acceptance corrections include geometric acceptance , relative branching ratios , reconstruction efficiencies , and event selection efficiencies .
the experimental resolution for the @xmath0 mass measurement in the e791 spectrometer depends on both the @xmath0 decay mode and the @xmath20 of the @xmath0 meson , and the mean reconstructed mass depends on the decay mode .
therefore , we fit to the normalized @xmath0 mass defined as @xmath86 where @xmath87 is the measured mass , @xmath88 is the mean measured mass for the particular decay mode of the @xmath0 candidate , and @xmath89 is the experimental resolution for the particular decay mode and @xmath20 of the @xmath0 candidate . in this analysis
we use the maximum likelihood method which assumes we have @xmath90 independent measurements of one or more quantities and that these quantities @xmath91 are distributed according to some probability density function @xmath92 where @xmath93 is a set of unknown parameters to be determined . to determine the set of values @xmath93 that maximizes the joint probability for all events
, we numerically solve the set of equations@xcite : @xmath94 the quantities that we measure for each event are the normalized mass of both the @xmath0 and @xmath1 candidate ; _ i.e. _ , @xmath95 , @xmath96 ) .
the unknown parameters in the maximum likelihood fit are the number of signal events , @xmath80 ; combinatoric events , @xmath97 ; events with one real @xmath0 and one combinatoric background called @xmath0-ridge events , @xmath98 ; events with one real @xmath1 and one combinatoric background called @xmath1-ridge events , @xmath99 ; the slope of the background @xmath100 distribution , @xmath101 ; and the slope of the background @xmath102 distribution , @xmath103 .
that is , the unknown parameters are @xmath104 the terms @xmath100 and @xmath102 refer to @xmath0 or @xmath11 decays into a kaon and @xmath105 pions .
we construct our probability density function using the following two assumptions : ( i ) the normalized mass distribution for background @xmath100 and @xmath102 is linear in @xmath106 and @xmath96 , and ( ii ) the normalized mass distribution of real @xmath0 s and real @xmath1 s is gaussian with mean of 0 and sigma of 1 . under these assumptions , which are correct for our data , the probability density functions normalized to unity in the two - dimensional window defined by @xmath107 and @xmath108 for each class of events is @xmath0-ridge background events : .
= combinatoric background : @xmath109 + + @xmath0-ridge background
events:@xmath110 + + @xmath1-ridge background events : @xmath111 + + signal events : @xmath112 + the distribution for each set of events is @xmath113 .
the overall probability density function is then @xmath114 in this analysis , we use the extended maximum likelihood method ( * ? ? ?
249 ) in which the number of @xmath13 candidates found , @xmath115 , is considered to be one more measurement with a gaussian probability distribution @xmath116 of mean @xmath117 and @xmath118 .
our likelihood function is then @xmath119 to maximize the likelihood , we use the function minimization and error analysis fortran package minuit @xcite .
figure [ fig : fig5 ] shows the function @xmath120 that maximizes the likelihood function for the final sample of @xmath13 candidates from fig . [ fig : fig4 ] with @xmath121 , the mass range used for all fits in this analysis .
the projections of the fit onto the @xmath0 and @xmath11 axes are compared to the data in fig .
[ fig : fig6 ] .
the projected background contains both ridge events ( one real @xmath0 and one combinatoric background ) and events with two combinatoric background candidates .
therefore , the background under the charm - pair signal in the projected distribution is a linear distribution plus a gaussian distribution , shown as the dotted line in the figure .
the net charm - pair signal is shown as the residual after background subtraction .
we determine the size of acceptance and smearing effects with a sample of approximately 7000 monte carlo simulated pairs that pass the same selection criteria as the real data .
the size of the resolution with which we measure each charm - pair physics variable is much smaller than the range over which we bin that variable .
therefore , we ignore smearing effects .
we incorporate acceptance effects in the likelihood function for the fit by replacing the probability @xmath122 for event @xmath123 by @xmath124 where @xmath125 is the weight for event @xmath123 @xcite .
the weight @xmath125 is inversely proportional to the efficiency and is normalized such that @xmath126 , where @xmath127 is the number of @xmath13 candidates in the final sample .
corrections for relative branching fractions are also included in @xmath125 , as described below . by construction
, the mean of the weights is equal to 1 .
the standard deviation of the weights is 1.3 .
the total number of @xmath13 events found in the unweighted fit is @xmath128 . for the weighted likelihood fit ,
we find @xmath129 .
candidates , the fact that the number of signal events is significantly larger for the weighted data sample indicates that , on average , the weights for signal events are larger than for background events .
since we correct for relative efficiencies , not absolute efficiencies , the absolute number of weighted signal events has no significance .
it is only of interest in interpreting the figures . ]
the efficiency depends not only on the detector acceptance but also on the relative branching fractions for the detected decay modes . by correcting for branching fractions ,
the final efficiency - corrected distributions reflect the relative production rates of the four types of @xmath13 pairs ( @xmath14 , @xmath15 , @xmath16 , and @xmath17 ) rather than the relative detected rates .
we use the values @xmath130 , @xmath131 , and @xmath132@xcite .
a minimal independent set of properties that the acceptance could depend on is the decay mode of each of the @xmath0 mesons ( @xmath59 , @xmath72 , or @xmath60 ) , the rapidity @xmath19 , the transverse momentum @xmath78 , and the azimuthal angle @xmath133 of each of the @xmath0 mesons . in principle
, we can use monte carlo simulated events to determine the acceptance for a particular candidate pair .
the problem is the large number of monte carlo events that is needed to span such a large space ( a 48-dimensional space , six variables for each pair of decay modes used ) .
however , the efficiency function can be factorized for each combination of the @xmath0 decay modes , greatly increasing the statistical power of the monte carlo . using the monte carlo simulated events
, we find that the acceptance of the @xmath0 is independent of the @xmath11 , and vice versa , at the level of the statistical precision of the simulation .
this is true in spite of the correlations in the selection criteria described in section [ sec : selection ] . for each one
, however , the shape of the acceptance as a function of @xmath19 depends on both the number of particles in the decay , @xmath134 or @xmath135 , and , at high @xmath19 , whether the candidate decay is a @xmath0 or @xmath11 .
the shape of the acceptance as a function of @xmath78 depends only on the number of particles in the decay , @xmath134 or @xmath135 .
it is also found that the acceptance does not depend on the azimuthal angle @xmath133 of the @xmath0 or @xmath11 .
therefore , the acceptance function factorizes as follows : @xmath136 @xmath137 where the subscripts , superscripts , and functional dependences of the terms b , c , and d are explicit , showing how the factorization is done .
we next determine which of the variables that describe the candidate pair , and for which the acceptance is not uniform , are correlated in the originally generated monte carlo .
such correlations could affect the apparent acceptance from the monte carlo if we simply integrate over a variable that is correlated with the variable for which we are determining the acceptance .
we find that the most significant correlations in the monte carlo generator are between the variables @xmath30 and @xmath31 , where the acceptance is not uniform , and between the variables @xmath138 and @xmath139 , where the acceptance is uniform .
therefore , we can not simply integrate over @xmath30 when determining the acceptance as a function of @xmath31 . instead
, we use the monte carlo to determine the two - dimensional acceptance function @xmath140 which is independent of the monte carlo generated correlations for each of the possible values of @xmath134 and @xmath141 and use this function in eq .
[ eq : accept ] .
( this removes any dependence on the physics assumptions of the monte carlo from this equation . ) because of the uniform @xmath133 acceptance for the observed events , the double variable technique is not needed for @xmath133 .
finally , the weight @xmath125 is calculated for each event such that it is proportional to @xmath142 and @xmath126 , where @xmath127 is the number of unweighted @xmath13 candidates in the final sample . here
@xmath143 is @xmath144 for the charged @xmath0 candidates and @xmath143 is @xmath145 for the neutral @xmath0 candidates except for @xmath14 events where @xmath146 due to the exclusion of 4 - 4 pairs from the final sample .
sources of systematic errors in our measurements include effects associated with the fitting procedure used to obtain the yields , the finite statistics of the monte carlo data sample used to generate the acceptance contributions to the weights , and imperfections in our modeling of the apparatus in the monte carlo .
for all the measured distributions , we compared the data to the two dimensional normalized mass distributions ; in all cases , the fits qualitatively match the data .
( for example , see fig .
[ fig : fig6 ] . )
we also checked the fitting procedure by comparing the yields with those given by a simple counting method . in this method the normalized mass scatterplot was divided into regions corresponding to different combinations of signal , ridge , and combinatoric background events .
the number of signal events was then found by subtracting the properly normalized number of events in the ridge and background regions from the central signal region .
the results are in agreement , but the fitting technique gave smaller statistical errors , as expected .
the effect of the finite statistics in the monte carlo was determined by repeating the fits for the yields in each kinematic bin while varying the weight of each event randomly according to a gaussian whose width corresponded to the statistical error on the weight .
the systematic errors on the yields generated by this process were about @xmath147 of the statistical errors from the fit , which are negligible when added in quadrature .
as demonstrated in figs .
[ fig : fig7 ] to [ fig : fig12 ] , in most cases the weighted and unweighted distributions are very similar . statistical errors associated with modeling the acceptance are most important for events with large weights , but the number of events with large weights is small .
we checked the effect of large weights by generating distributions without the large weight events , with no significant change .
the distributions were also examined with all @xmath77 candidates eliminated , the source of most of the events with a large acceptance correction .
again , the change did not significantly affect the distributions .
another potential source of systematic error is uncertainty in the modeling of the beam - induced inefficiency in the centers of the drift chambers .
the inefficiencies increased as the run progressed , primarily affecting @xmath0 s at large @xmath20 .
since most @xmath0 s in the @xmath13 events are at low to modest @xmath20 , the drift chamber inefficiencies did not have a significant effect on the relative efficiencies used in this analysis .
we checked this by comparing the experimental results that had been efficiency - corrected with monte carlo simulations corresponding to different parts of the run , and found no significant differences . in summary ,
systematic errors were found to be small relative to statistical errors .
in this section , we present the background - subtracted , acceptance - corrected charm - pair distributions from the data and compare them to theoretical predictions .
the appendix contains an extensive discussion of the theoretical predictions for charm - pair distributions . as discussed in sec .
[ sec : selection ] , all distributions experimental and theoretical are obtained after excluding any events in which the center - of - mass rapidity of either @xmath0 meson or either charm quark is less than @xmath82 or greater than @xmath83 . for the experimental results ,
the acceptance - corrected distributions are obtained from maximizing the likelihood function with weighted events as discussed in sec .
[ ssec : accept ] .
the uncorrected distributions are obtained from maximizing the unweighted likelihood function ; the total number of signal @xmath13 events found in the unweighted fit is @xmath128 .
if the two charm mesons in each @xmath13 event are completely uncorrelated , then the charm - pair distributions contain no more information than the single - charm distributions . before comparing the observed distributions to theoretical predictions
, we use two methods to determine whether there exist correlations in the data . in the first method , described in section [ ssec : single ] , we convolute acceptance - corrected single - charm distributions to predict what the charm - pair distributions would be if the @xmath0 and @xmath11 were uncorrelated . comparing these single - charm predictions ,
the measured charm - pair distributions provide one measure of the degree of correlation between the @xmath0 and @xmath11 . in the second method , described in section [ ssec : twod ]
, we look directly for correlations by examining several two - dimensional distributions . for example , by finding the number of signal @xmath13 events per @xmath30 interval in several @xmath31 intervals , we can determine whether the shape of the @xmath30 distribution depends on the value of @xmath31 . in section [ ssec : theory ] , we compare our experimental distributions to the theoretical predictions discussed in sec .
[ sec : theory ] and in the appendix . in section [ ssec : asym ] , we examine integrated production asymmetries among the four types of @xmath13 pairs @xmath14 , @xmath15 , @xmath16 , and @xmath17 and compare our experimental results to the predictions from the pythia / jetset event generator . in fig .
[ fig : fig7 ] we show the measured single - charm distributions for @xmath148 , @xmath19 , @xmath149 and @xmath133 , as defined in sec . [ sec : intro ] .
the single - charm distributions are obtained by fitting the two - dimensional normalized mass distributions for only those @xmath13 pairs in which the value of the single - charm variable for the candidate @xmath0 ( or @xmath11 ) is in the appropriate interval for each bin . in this way , the contribution to the single - charm signal from the @xmath0 and @xmath11 ridge events is excluded .
the distributions shown in fig . [ fig : fig7 ] correspond to single - charm mesons from @xmath13 pairs in which the center - of - mass rapidity of _ both _ charm mesons lies between @xmath82 and @xmath150 .
each distribution shown in fig .
[ fig : fig7 ] is obtained by summing the @xmath0 and @xmath11 distributions .
we have checked and found that the @xmath0 and @xmath11 distributions are the same within statistical errors .. ] the vertical axis of each distribution gives the fraction of signal mesons per variable @xmath151 interval , @xmath152 , where the total number of signal @xmath0 mesons @xmath98 is simply twice the number of signal @xmath13 events . only a very small fraction ( 0% 3% ) of the signal events lie outside any of the ranges used in figs .
[ fig : fig7][fig : fig11 ] . for each single - charm variable @xmath153 , @xmath19 , @xmath154 , and
@xmath133 we obtain two measured charm - pair distributions : the difference in @xmath151 for the two @xmath0 s , @xmath155 , and the sum of the @xmath151 s for the two @xmath0 s , @xmath156 .
( @xmath157 is defined to be the minimum of @xmath158 and @xmath159 , and @xmath160 is defined to be @xmath161 modulo @xmath162 . ) in figs .
[ fig : fig8][fig : fig11 ] , we compare these measured charm - pair distributions to the charm - pair distributions one would generate from the measured single - charm distributions assuming the @xmath0 and @xmath11 are completely uncorrelated , calculated as follows : @xmath163 and @xmath164 where @xmath165 refers to the single - charm distributions shown in fig .
[ fig : fig7 ] . this convolution can not be done with previously reported inclusive single charm distributions @xcite since the inclusive distributions contain events which are excluded from the charm - pair sample , for example , events in which the @xmath20 of the unobserved @xmath0 is outside the acceptance or in which the unobserved charm particle is a charm baryon . if the @xmath0 and @xmath11 in the signal @xmath13 events are completely uncorrelated , then the measured charm - pair distributions for @xmath166 and @xmath167 should agree with the single - charm predictions , because both the charm - pair and single - charm distributions are for @xmath0 mesons with exactly the same restrictions on the rapidity of both @xmath0 mesons in the event . with the exception of the @xmath43 distribution ( fig . [ fig : fig11 ] ) ,
the measured distributions are quite similar to the uncorrelated single - charm predictions , indicating both that the correlation between the @xmath0 and @xmath1 longitudinal momenta is small and that the correlation between the amplitudes of the @xmath0 and @xmath1 transverse momenta is small . the measured @xmath168 and @xmath169 distributions , however , are somewhat more peaked near zero than the single - charm predictions , possibly indicating slight longitudinal correlations .
two other commonly used charm - pair variables are the square of the net transverse momentum of the charm pair , @xmath170 , and the invariant mass of the charm pair , @xmath26 .
the measured distributions and the uncorrelated single - charm predictions for these two variables are shown in fig .
[ fig : fig12 ] . obtaining these single - charm predictions for @xmath171 and @xmath26 is slightly more involved than for the @xmath166 and @xmath167 variables because @xmath171 and @xmath26 are not linear functions of @xmath28 , @xmath29 , @xmath172 , @xmath173 , @xmath36 , and @xmath37 .
rather , in terms of these single - charm variables , @xmath174 @xmath175 where the @xmath0 meson energy @xmath23 is @xmath176 , and @xmath177 is the square of the center - of - mass energy of the colliding hadrons .
we obtain single - charm predictions by randomly generating @xmath178 @xmath13 events in which all three variables ( @xmath20 , @xmath133 , and @xmath154 ) for both @xmath0 mesons from each @xmath13 event are selected independently and randomly from a probability density function that is flat within the bins shown in fig .
[ fig : fig7 ] , and zero elsewhere .
each event is weighted by @xmath179 where @xmath165 refers to the single - charm distributions shown in fig .
[ fig : fig7 ] and @xmath180 is the jacobian determinant of the transformation from the complete and independent set of variables ( @xmath181 , @xmath182 , @xmath172 , @xmath173 , @xmath183 , and @xmath184 ) to the set ( @xmath181 , @xmath182 , @xmath172 , @xmath173 , @xmath185 , and @xmath186 ) . specifically , @xmath187 @xmath188 the measured distribution for @xmath26 agrees quite well with the uncorrelated single - charm prediction .
the measured distribution for @xmath171 , however , is steeper than the uncorrelated single - charm prediction , indicating the presence of correlations between @xmath189 and @xmath190 .
the dashed histogram in fig .
[ fig : fig12 ] demonstrates that this lack of agreement is not due to the correlations between @xmath138 and @xmath173 evident in the @xmath43 distribution in fig .
[ fig : fig11 ] .
this latter prediction is obtained by assuming that @xmath28 and @xmath29 are uncorrelated and that @xmath36 and @xmath37 are uncorrelated , but that @xmath172 and @xmath173 are correlated as shown in fig . [
fig : fig11 ] .
the correlations in @xmath171 should reflect similar correlations in @xmath43 , @xmath36 , and @xmath37 , since @xmath171 is a function of these variables .
the fact that the disagreement in fig .
[ fig : fig12 ] is not so readily explained is a sign that the correlations can be subtle , and that there are additional correlations among the variables . in the following section
we investigate correlations between @xmath43 , @xmath191 , and @xmath192 in more detail . a direct method for investigating whether the variables @xmath193 and @xmath194 are correlated is to determine the number of @xmath13 signal events per @xmath193 interval for a series of @xmath194 intervals .
such two - dimensional distributions show whether the @xmath193 distribution depends on the value of @xmath194 , and vice - versa .
given the limited number of @xmath13 pairs , we use coarse binning to see statistically meaningful effects . in figs .
[ fig : fig13 ] , [ fig : fig14 ] and [ fig : fig15 ] , we show the results for @xmath195 , @xmath19 and @xmath196 , respectively . in part
( a ) of each figure , we show the number of acceptance - corrected @xmath13 signal events reconstructed in nine @xmath197 bins three @xmath193 bins times three @xmath194 bins .
note that the three bin sizes are not equal . from the information in this two - dimensional plot , several normalized one - dimensional plots are created , facilitating our ability to detect differences in the shapes of the distributions .
in particular , plot ( b ) in each figure shows the @xmath193 distribution for each @xmath194 bin , @xmath198 , where @xmath199 is the number of events in the relevant bin , and @xmath200 is the total number of events in the three @xmath193 bins .
this normalization is chosen so that the integral over each @xmath193 distribution equals one .
the symbols are defined in plot ( a ) .
similarly , plot ( c ) in each figure shows the normalized @xmath201 distribution for each @xmath193 bin
. if the three sets of points in the figures ( b ) and ( c ) are statistically consistent , there are no significant correlations .
lastly , plots ( d)(f ) simply rearrange the information shown in ( b ) and ( c ) . plot ( d )
shows the normalized @xmath193 and @xmath194 distributions for the first @xmath194 and @xmath193 bin , respectively ; plot ( e ) shows results for the second bins ; and plot ( f ) shows results for the third bins . in ( d)-(f ) , agreement of the two sets of points implies that correlations in @xmath194 are the same as correlations in @xmath193 .
the two - dimensional plots show the actual number of acceptance - corrected @xmath13 signal events in each bin , whereas the one - dimensional plots , proportional to @xmath202 , take into account the variation in bin size .
figure [ fig : fig13 ] indicates some correlation between @xmath28 and @xmath29 .
in particular , the first - bin distributions are more peaked at low @xmath20 than the second- and third - bin distributions of figures ( b ) and ( c ) .
this result is consistent with fig .
[ fig : fig8 ] , discussed above , which shows that the measured @xmath168 distribution is somewhat steeper than the single - charm prediction . because @xmath20 and @xmath19 are highly correlated , fig .
[ fig : fig14 ] shows the same trends as fig .
[ fig : fig13 ] .
figure [ fig : fig15 ] indicates that @xmath36 and @xmath37 are also slightly correlated the second - bin @xmath36 and @xmath37 distributions are enhanced in the second bin ; and the third - bin @xmath36 and @xmath37 distributions are enhanced in the third bin of figures ( b ) and ( c ) .
this result should be compared with fig .
[ fig : fig10 ] , which is consistent with no correlation . correlations are also seen in fig .
[ fig : fig12 ] which shows that the measured @xmath171 distribution is somewhat steeper than the uncorrelated single - charm prediction . in all three figures ( [ fig : fig13][fig : fig15 ] ) , the shapes of the three @xmath193 distributions are remarkably similar to the shapes of the respective @xmath194 distributions as seen in figures ( d)(f ) . in fig . [
fig : fig16 ] we investigate whether the separation in azimuthal angle between the @xmath0 and @xmath1 is correlated to the amplitude of the transverse momenta of the @xmath0 and @xmath1 .
in particular , we determine the number of signal @xmath13 events per @xmath157 interval in @xmath203 intervals and the number of signal @xmath13 events per @xmath157 interval in @xmath204 intervals .
although we find no significant correlation between @xmath157 and @xmath204 , we find that @xmath43 and @xmath45 are quite correlated .
the @xmath43 distribution is more peaked at large @xmath45 and the @xmath45 distribution is flatter at large @xmath43 .
a theoretical explanation for these correlations is discussed in the following section . in this section , we compare all the acceptance - corrected distributions discussed in the previous two sections ( figs .
[ fig : fig7][fig : fig16 ] ) to three sets of theoretical predictions : the distributions of @xmath12 pairs from a next - to - leading - order ( nlo ) perturbative qcd calculation by mangano , nason and ridolfi ( mnr)@xcite ; the distributions of @xmath12 pairs from the pythia / jetset monte carlo event generator@xcite which uses a parton - shower model to include higher - order perturbative effects@xcite ; and the distributions of @xmath13 pairs from the pythia ( version 5.7)/jetset ( version 7.4 ) monte carlo event generator@xcite which uses the lund string model to transform @xmath12 pairs to @xmath13 pairs@xcite . for all theoretical predictions ,
we use the default parameters suggested by the respective authors , which are discussed in the appendix .
all distributions are obtained after excluding any candidates in which the center - of - mass rapidity of either the @xmath0 or @xmath1 ( or , for the mnr calculation , the @xmath2 or @xmath205 ) is less than @xmath82 or greater than 2.5 .
lack of agreement between an experimental charm - pair distribution and a theoretical prediction can arise if the theory does not model the correlations between the two charm particles correctly .
however , it can also arise if the theory models the correlations correctly but does not correctly model the single - charm distributions . hence , before comparing the experimental charm - pair distributions to theory , we first compare the acceptance - corrected single - charm distributions ( @xmath148 , @xmath19 , @xmath149 and @xmath133 ) to theory in fig .
[ fig : fig17 ] . for the longitudinal momentum distributions
@xmath148 and @xmath19 the experimental results and theoretical predictions based on the default parameters do not agree .
the experimental distributions are most similar to the nlo and pythia / jetset @xmath12 distributions , but are narrower than all three .
the difference between the pythia / jetset @xmath12 and the pythia / jetset @xmath13 longitudinal distributions shows the effect of the hadronization scheme that color - attaches one charm quark to the remnant beam and the other to the remnant target , broadening the longitudinal distributions .
the experimental @xmath154 distribution agrees quite well with all three theoretical distributions .
however , the pythia / jetset @xmath12 distribution is somewhat flatter ; the pythia / jetset @xmath13 distribution is somewhat steeper . as expected , both the theoretical and experimental @xmath133 distributions are consistent with being flat . the experimental and theoretical @xmath168 and @xmath206 distributions ( fig .
[ fig : fig18 ] ) and @xmath169 and @xmath207 distributions ( fig .
[ fig : fig19 ] ) do not agree with theoretical predictions derived from default parameters .
this may be a reflection of the disagreement between the measured single - charm longitudinal distributions and theoretical models .
as with the single - charm distributions , the experimental results are much closer to the two @xmath12 predictions than to the pythia / jetset @xmath13 prediction , but narrower than all three predictions .
the pythia / jetset hadronization scheme introduces a strong correlation between the @xmath0 and @xmath1 which significantly broadens the @xmath169 distribution .
that is , hadronization tends to pull the @xmath13 apart , due to color string attachment to the incident hadronic remnants . as we show in fig .
[ fig : fig20 ] , the pythia / jetset @xmath13 @xmath169 distribution is broader than the prediction we obtain by using the predicted single - charm distributions and assuming they are uncorrelated , as calculated using eqn .
[ eq : qdv ] .
in contrast , the experimental @xmath169 distribution is slightly narrower than its uncorrelated single - charm prediction ( fig .
[ fig : fig9 ] ) . in figs .
[ fig : fig21][fig : fig23 ] , we compare experimental distributions to theoretical predictions for the following transverse variables : @xmath204 , @xmath45 , @xmath43 , @xmath208 , and @xmath171 . any observed discrepancy between theory and data for the @xmath209 , @xmath157 , and @xmath185 distributions is noteworthy because the single - charm @xmath196 and @xmath133 experimental distributions agree well with theory . an observed discrepancy , therefore , must derive from the theory modeling the correlation between @xmath189 and @xmath190 incorrectly . if @xmath189 and @xmath190 were completely uncorrelated , then the single - charm predictions ( figs .
[ fig : fig10][fig : fig12 ] ) would provide good estimates for these three distributions . at the opposite extreme , if @xmath189 and @xmath190 were completely anticorrelated as in the leading - order perturbative qcd prediction then the @xmath210 distribution would be a delta function at @xmath211 ; the @xmath185 distribution a delta function at @xmath212 ; and the @xmath157 distribution a delta function at @xmath213 .
both the experimental distributions and the three sets of theoretical predictions lie between these extremes .
none of the three experimental distributions , however , is as steep as any of the theoretical predictions .
the next - to - leading - order predictions are the steepest that is , the next - to - leading - order calculation predicts the most correlation between @xmath214 and @xmath215 as a model for @xmath189 and @xmath190 .
thus hadronization and higher - order perturbative effects smear out the @xmath12 correlations .
the pythia / jetset hadronization scheme broadens the @xmath157 distribution , bringing it closer to the experimental results .
the same hadronization scheme also narrows the @xmath171 and @xmath44 distributions , which makes them disagree even more with the experimental results .
one mechanism which would broaden the @xmath171 and @xmath44 distributions as well as the @xmath43 distribution ( bringing all into better agreement with the experimental results ) is to increase the intrinsic transverse momentum of the colliding partons in the beam and target hadrons ( see the appendix , secs . [
ssec : npeffects ] and [ ssec : itm ] ) .
an improved theoretical understanding may involve adding terms higher than nlo to calculations , although other authors find good agreement by choosing appropriate values for nonperturbative parameters @xcite . in fig .
[ fig : fig23 ] , we compare the experimental charm - pair invariant mass distribution to the pythia / jetset prediction . the experimental @xmath26 distribution is steeper than the theoretical predictions .
this is similar to the experimental single - charm @xmath148 ( or @xmath19 ) distributions , which are also steeper than the theoretical predictions .
( see fig .
[ fig : fig17 ] . ) in addition , the correlations introduced by the pythia / jetset hadronization scheme broaden the invariant mass distribution . in figs .
[ fig : fig24][fig : fig27 ] , we examine the same two - dimensional experimental distributions discussed in section [ ssec : twod ] .
we now compare these experimental results to the three sets of theoretical predictions . in each figure , the top row shows the nlo perturbative qcd @xmath12 prediction ; the middle row , the pythia / jetset @xmath12 prediction ; and the bottom row , the pythia / jetset @xmath13 prediction . the experimental data points and errors are repeated in each row .
the longitudinal distributions , @xmath148 and @xmath19 , are shown in figs .
[ fig : fig24 ] and [ fig : fig25 ] .
the three theoretical predictions are quite different .
the nlo @xmath12 predictions show no significant correlation between @xmath28 and @xmath29 ( or between @xmath30 and @xmath31 ) and the @xmath28 and @xmath29 distributions are quite similar .
the pythia / jetset @xmath12 predictions show a slight correlation and the @xmath28 and @xmath29 distributions are somewhat different . due to the pythia / jetset hadronization scheme ,
the pythia / jetset @xmath13 prediction shows the strongest correlation between @xmath28 and @xmath29 .
interestingly , in the pythia / jetset @xmath13 prediction , @xmath28 and @xmath29 are anticorrelated ; in the pythia / jetset @xmath12 prediction they are positively correlated .
the correlation patterns in the experimental results , although inconsistent with any of the theoretical predictions , are closest to the pythia / jetset @xmath12 predictions . in fig .
[ fig : fig26 ] , we investigate the correlations between @xmath183 and @xmath216 .
the three theoretical predictions and the experimental results all show similar trends .
although all the distributions are broader than the leading - order perturbative qcd prediction a delta function at @xmath217 they all shows signs of an enhancement in the @xmath218 bins .
the pythia / jetset @xmath12 distributions and the pythia / jetset @xmath13 distributions are very similar and resemble the experimental results more so than the nlo @xmath12 distributions .
all of the theoretical third - bin distributions are significantly flatter than the experimental third - bin distributions .
in contrast to the longitudinal distributions , all the @xmath183 distributions are very similar to the respective @xmath184 distributions . in fig .
[ fig : fig27 ] , we investigate correlations between @xmath43 and @xmath45 . for the @xmath43 dependence ,
a leading - order perturbative qcd calculation predicts a delta function at @xmath219 .
we expect perturbative predictions to be more accurate as the energy scale @xmath220 of the partonic hard scattering increases : @xmath221 that is , we expect the @xmath43 distribution to be more peaked at @xmath222 for @xmath13 events with larger @xmath45 . this behavior is clearly evident in our experimental distributions as well as in all three theoretical predictions .
the theoretical @xmath43 distributions , however , for all three @xmath45 bins , are significantly steeper than the respective experimental distributions .
the nlo @xmath12 @xmath43 distributions are the steepest .
the experimental and theoretical @xmath45 distributions are in fairly good agreement , with the @xmath45 distribution broadening as @xmath43 increases .
no significant correlation between @xmath43 and @xmath204 or between @xmath43 and @xmath169 is seen in the data or theory . in table
[ tab : tab3 ] , we compare the experimental yields for each type of @xmath13 pair to the predictions from the pythia / jetset event generator and to a naive spin - counting model .
the experimental results are obtained by maximizing the weighted likelihood function where the weights account for both acceptance effects and the relative branching fractions of the reconstructed decay modes ( sec .
[ ssec : accept ] ) . again , for both data and pythia / jetset predictions , the results are for pairs in which the rapidities of both the @xmath0 and @xmath11 lie between @xmath82 and @xmath83 .
the pythia / jetset and naive spin - counting models both assume that vector @xmath223 production is three times more likely than pseudoscalar @xmath0 production due to the number of spin states and that contributions from higher spin states are negligible .
they also use the known @xmath224 branching fractions , @xmath225 and @xmath226 , to determine @xmath0 production .
the differences between the pythia / jetset and naive spin - counting model come from the pythia / jetset hadronization scheme in particular , the rate of coalescence . as discussed in sec .
[ ssec : hadro ] , a charm quark tied to a valence quark by a low - mass string can coalesce with that valence quark into a meson .
this will tend to increase the rate of @xmath227 and @xmath228 production in the forward region for the e791 @xmath229 beam .
since @xmath230 decays to @xmath66 , production of @xmath66 will also be enhanced .
this effect is seen in table [ tab : tab3 ] where the number of pairs that contain a @xmath67 is reduced while the number of pairs that contain a @xmath231 or @xmath66 is increased in the pythia / jetset model relative to the naive model .
both models agree with data as far as the relative ordering but predict too many @xmath14 pairs and too few @xmath17 pairs .
[ htbp ] lccc & & & + + @xmath14 & @xmath232 & 0.572 & @xmath233 + @xmath15 & @xmath234 & 0.184 & @xmath235 + @xmath16 & @xmath236 & 0.184 & @xmath237 + @xmath17 & @xmath238 & 0.060 & @xmath239 + as shown in fig .
[ fig : fig28 ] , in the pythia / jetset hadronization scheme , the correlation between @xmath30 and @xmath31 is quite different for each of the four types of @xmath13 pairs . in fig .
[ fig : fig29 ] , we investigate whether this is also true for data .
given the limited size of our data sample , we can only search for gross asymmetries in the ( @xmath30 , @xmath31 ) distributions .
we obtain the four plots in fig .
[ fig : fig29 ] by bisecting the two - dimensional ( @xmath30 , @xmath31 ) distributions along the following four lines ( @xmath240 ) : @xmath241 , @xmath242 , @xmath243 , and @xmath244 .
these four lines are indicated by dashed lines in fig .
[ fig : fig28 ] . to search for possible differences in asymmetries , we determine whether the fraction of signal events on one side of a given line depends on the type of @xmath13 pair .
specifically , for both theory and data , we show in fig .
[ fig : fig29 ] @xmath245 where @xmath246 and @xmath200 is the number of signal @xmath13 events of type @xmath123 .
the pythia / jetset @xmath247 distribution is fairly flat , indicating no significant differences among the four @xmath13 types for the @xmath207 distribution .
the pythia / jetset @xmath248 , @xmath249 , and @xmath250 distributions , however , indicate significant differences , all of which are easily interpreted in terms of the pythia / jetset coalescence mechanism discussed in sec .
[ sec : theory ] .
unfortunately , our experimental errors are of the same order as the degree of differences in the pythia / jetset predictions .
the experimental @xmath249 distribution , for example , is consistent with the pythia / jetset prediction , but it is also consistent with being flat .
similarly , the experimental @xmath251 distribution is fairly consistent with the flat pythia / jetset prediction , but it also shows some indication of a difference between @xmath14 and @xmath17 .
the most significant difference between the experimental results and the pythia / jetset predictions occurs for the @xmath13 types @xmath14 and @xmath17 in the @xmath250 distribution . both theory and data indicate a difference between @xmath14 and @xmath17 ; however , we find experimentally that @xmath252 , whereas the pythia / jetset model finds @xmath253 .
we fully reconstructed @xmath254 true @xmath13 pairs after all background subtractions .
this is the largest such sample of charm pairs used in an analysis of the hadroproduction of @xmath255 to date .
the full reconstruction of the final states of both @xmath0 mesons offers several advantages over some of the previous studies that have used partially reconstructed @xmath0 candidates .
we are able to correct the data for both detector inefficiencies and for the branching fractions of the observed decays so that the acceptance - corrected distributions represent the produced mixture of @xmath0 mesons , rather than the detected mixture . because the final states are fully reconstructed , we are able to calculate both the magnitude and direction of the @xmath0 momenta .
therefore , we are able to thoroughly investigate the degree of correlation between both the transverse and longitudinal components of the momenta , with respect to the beam direction , of the @xmath0 and @xmath11 . we have compared all the measured acceptance - corrected distributions to predictions of the fully differential next - to - leading - order calculation for @xmath12 production by mangano , nason and ridolfi @xcite , as well as to predictions from the pythia / jetset monte carlo event generator @xcite for @xmath12 @xcite and @xmath13 production @xcite .
our measurements indicate that the transverse momenta of the @xmath0 and @xmath1 in charm - pair events are correlated in several ways .
( see secs .
[ ssec : twod ] and [ sssec : twodd ] . )
the square of the amplitudes of the @xmath0 and @xmath1 transverse momenta are slightly correlated ( fig . [
fig : fig15 ] ) .
the directions of the @xmath0 and @xmath1 in the plane transverse to the beam axis are significantly correlated ( fig . [
fig : fig11 ] ) . the separation in azimuthal angle , @xmath43 ,
is significantly correlated with the sum of the squares of the @xmath0 and @xmath11 transverse momenta , @xmath203 ( fig .
[ fig : fig16 ] ) .
these features have been observed by several other experiments @xcite . using the default parameters ,
the three models yield the same trend in correlations as we find in data .
the models also predict that the relative transverse angles of the @xmath0 and @xmath11 are more correlated than we find in data ( fig .
[ fig : fig22 ] ) .
these results provide an opportunity to tune the default parameters , or add higher order terms in the models , to obtain better agreement with data @xcite .
our measurements indicate that the longitudinal momenta of the @xmath0 and @xmath11 from charm - pair events are slightly correlated .
the measured @xmath168 and @xmath169 distributions ( figs .
[ fig : fig8][fig : fig9 ] ) are somewhat narrower than what one would predict from the observed single - charm predictions assuming no correlations .
the @xmath181 ( @xmath256 ) distribution depends on the value of @xmath182 ( @xmath257 ) , and vice - versa ( figs .
[ fig : fig13][fig : fig14 ] ) .
the single - charm @xmath148 and @xmath19 distributions from the three theoretical models do not agree with each other or with the measured distributions ( fig .
[ fig : fig17 ] ) .
the three models predict different correlations between the charm and anticharm longitudinal momenta ( figs .
[ fig : fig24][fig : fig25 ] ) the next - to - leading - order calculation predicts no significant correlation ; the pythia / jetset @xmath12 prediction indicates a slight positive correlation ; and the pythia / jetset @xmath13 prediction indicates a strong negative correlation .
the @xmath13 data agree best with the pythia / jetset @xmath12 prediction .
the disagreement between the models and the data might be corrected by adjusting the non - perturbative parameters in the models , or by adding higher order terms .
the relative yields of the four types of charm pairs , @xmath14 , @xmath15 , @xmath16 , and @xmath17 , as calculated in the pythia / jetset event generator , agree with data as far as their ordering but disagree with regard to number of pairs produced , predicting too many @xmath14 pairs and too few @xmath17 pairs .
( see table [ tab : tab3 ] . )
we studied the degree to which longitudinal correlations depend on the type of @xmath13 pair in data and in the pythia / jetset event generator .
although we see differences between data and the event generator ( fig .
[ fig : fig29 ] ) , the statistical uncertainties on the measured correlations are too large to make any conclusive statements . the charm - pair distributions presented in this paper provide an opportunity to extend our understanding of charm production beyond what was previously possible with single - charm and lower - statistics or partially reconstructed charm - pair distributions .
the measured distributions and observed correlations can be compared to the predictions of models , testing assumptions in the models and providing discrimination among different values for the free parameters in the models .
some comparisons have been made in the paper and in the appendix . before comparing the measurements to predictions
, we looked for correlations directly in the charm - pair data by comparing the observed @xmath13 pair distributions with the convolution of the measured single - charm distributions assuming no correlations .
we find that the charm - pair distributions are quite similar to the convoluted single - charm distributions , indicating little correlation between the two charm mesons in an event , with the exception of the distributions for @xmath43 and @xmath171 .
the @xmath43 distribution shows clear evidence of correlations , and the @xmath171 distribution is steeper than the uncorrelated single - charm prediction .
in addition , the data are consistent with possible small correlations in the @xmath168 and @xmath169 pair distributions , which are somewhat more peaked near zero than the single - charm convolutions . in the comparisons of the measured and predicted charm - pair distributions ,
we observe less correlation between transverse momenta and different correlations between longitudinal momenta than theoretical models predict , for the default values of parameters in the models .
work by other authors suggests a different set of parameters might provide better agreement @xcite .
both the single - charm and charm - pair distributions agree best with the predictions for charm quark ( rather than @xmath0 meson ) production , possibly caused by an accidental cancellation of color - dragging and fragmentation effects .
also , the @xmath43 distribution is more similar to the prediction of the nlo theory at higher @xmath45 . in the appendix
, we investigate the sensitivity of single - charm and charm - pair distributions to various theoretical assumptions .
we conclude that the predictions depend not only on unknown parameters such as the mass of the charm quark and the intrinsic transverse momentum of the partons that collide to form the @xmath255 pair , but also on the values of the renormalization and factorization scales .
the measurements reported here , and the charm - pair measurements from photoproduction experiments , should allow the free parameters in the theoretical models to be further constrained .
we gratefully acknowledge the assistance of the staffs of fermilab and of all the participating institutions .
this work was supported by the brazilian conselho nacional de desenvolvimento cientfico e tecnolgico , conacyt ( mexico ) , the israeli acadamy of sciences and humanities , the u. s. department of energy , the u.s.israel binational science foundation and the u. s. national science foundation .
this work was performed at the fermi national accelerator laboratory , which is operated by the universities research association , under contract de - ac02 - 76ch03000 with the u.s .
department of energy .
in sec . [ sec : theory ] , we introduced the theoretical framework used to describe the hadroproduction of @xmath13 pairs . in particular , we discussed : * the leading - order perturbative qcd description of the hadroproduction of @xmath12 pairs ; * higher - order perturbative corrections to the leading - order calculation ; * the addition of intrinsic transverse momentum to the hard - scattering partons that collide to form the @xmath12 pair ; and ,
lastly , * the hadronization of @xmath12 pairs to observable @xmath13 pairs . using this framework
, we investigate how sensitive single - charm and charm - pair distributions are to various theoretical assumptions .
all predictions discussed in this appendix are for a 500 gev/@xmath2 @xmath3 beam incident on a nuclear target the same beam - target as the data from experiment e791 reported in this paper .
the pythia / jetset event generator depends on many parameters .
unless otherwise mentioned , we use the default settings for all parameters . the next - to - leading order perturbative qcd calculation depends on the following six parameters : * the mass of the charm quark , @xmath258 , * the beam and target parton distribution functions , @xmath259 and @xmath260 , respectively , * @xmath261 , the free parameter that must be determined experimentally , which roughly defines the mass scale below which quarks and gluons do not behave as independent , free partons that is , below which perturbative qcd calculations are no longer valid , and * the renormalization and factorization scales , @xmath262 and @xmath263 .
the pairs of pion and nucleon parton distribution functions considered in this section , obtained from the cern computer library package pdflib @xcite , are listed in table [ tab : tab4 ] .
parton distribution functions depend on the fraction @xmath264 of the hadron momentum carried by the hard - scattering parton , on both the factorization and renormalization scales , and on @xmath261 . in the parton distribution functions accessible from pdflib
, the renormalization scale is defined to be the same as the factorization scale .
lclcccc set & & name & @xmath265 $ ] & @xmath266 [ mev ] & order & ref .
+ + ( 1 ) & @xmath259 & smrs - p2 & 5 & 190 & nlo & @xcite + & @xmath260 & hmrs - b ( 4.90 ) & 5 & 190 & nlo & @xcite + + ( 2 ) & @xmath259 & grv - p & 0.3 & 200 & nlo & @xcite + & @xmath260 & grv & 0.3 & 200 & nlo & @xcite + + ( 3 ) & @xmath259 & smrs - p2 & 5 & 190 & nlo & @xcite + & @xmath260 & hmrs - b ( 8.90 ) & 5 & 100 & nlo & @xcite + + ( 4 ) & @xmath259 & abfkw - p3 & 2 & 281 & nlo & @xcite + & @xmath260 & hmrs - b ( 8.90 ) & 5 & 300 & nlo & @xcite + + ( 5 ) & @xmath259 & ow - p1 & 4 & 200 & lo & @xcite + & @xmath260 & cteq 2l & 4 & 190 & lo & @xcite + for each parton distribution function listed in table [ tab : tab4 ] , we specify the square of the minimum factorization scale allowed , @xmath267 ; whether the evolution equations were calculated to leading - order ( lo ) or to next - to - leading order ( nlo ) and the value of @xmath268 used in the fit . querying pdflib for the value of a parton distribution function at a scale below @xmath269 gives undefined results .
the default set of parton distribution functions for the pythia / jetset event generator is set ( 5 ) in table [ tab : tab4 ] ; the default suggested by the authors of hvqmnr is set ( 1 ) .
when possible , we choose pion and nucleon distribution functions that are fit assuming similar values for @xmath261 . for all predictions shown below , the @xmath261 used in the next - to - leading order calculation of the partonic cross section is defined to be the same as the @xmath261 used to extract the nucleon parton distribution function @xmath260 . the degree
to which the charm - pair distributions are sensitive to variations in @xmath262 and @xmath263 gives an indication of how important higher - order corrections are ; that is , an indication of how much ( or little ) we can trust the @xmath270 calculation . in general , one tries to minimize higher - order contributions by choosing @xmath262 and @xmath263 to be of the same order as the energy scale @xmath220 of the hard - scattering process .
however , this scale can not be defined unambiguously .
one reasonable choice is @xmath271 the default setting for the pythia / jetset event generator is @xmath272 , leading to factorization scales as low as the mass of the charm quark , @xmath258 , which by default is set to 1.35 gev . the parton distribution functions used by the pythia / jetset event generator , however , are only defined for scales above 2 gev .
this problem is handled by setting the parton distribution function to @xmath273 for all factorization scales less than @xmath269 .
the suggested default for the hvqmnr program is @xmath274 and @xmath275 .
given their suggested default for the mass of the charm quark of @xmath276 gev , this choice ensures that the factorization scale will never go below the minimum allowed scale , @xmath277 gev . in figs .
[ fig : fig30][fig : fig35 ] we show single - charm and charm - pair distributions for a wide range of theoretical assumptions .
when obtaining these theoretical predictions , we only allow charm - pair events in which both charm rapidities are greater than @xmath82 and less than @xmath83 . for the hvqmnr generator , which does not hadronize the @xmath12 pair to charmed mesons ,
the cut is on the charm quark rapidities . for the pythia / jetset generator ,
the cut is on the @xmath0 meson rapidities . in table
[ tab : tab5 ] , we show which generator ( hvqmnr or pythia / jetset ) and what theoretical assumptions are used in each figure . the same set of single - charm and charm - pair distributions are shown in each figure .
each charm particle in a charm - pair event can be described using three variables .
a common choice of independent variables for single - charm analyses is @xmath20 , @xmath154 , and @xmath133 .
we ignore the latter variable because all theoretical predictions give a flat @xmath133 distribution .
although @xmath148 and @xmath19 are very correlated , we show predictions for both distributions . for each single - charm variable @xmath151
, we obtain predictions for two charm - pair distributions : @xmath278 and @xmath279 .
( @xmath157 is defined to be the minimum of @xmath280 and @xmath281 . ) as with the single - charm @xmath133 variable , we ignore the charm - pair @xmath160 variable because all theoretical predictions give a flat @xmath160 distribution .
we do not , however , ignore the @xmath157 distribution which is very sensitive to theoretical assumptions .
two other commonly used charm - pair distributions that we examine are the square of the transverse momentum of the charm - pair , @xmath282 , and the invariant mass of the charm - pair , @xmath283 . the vertical axis of each distribution shown in figs .
[ fig : fig30][fig : fig35 ] gives the fraction of single - charm ( charm - pair ) events per variable @xmath151 interval , @xmath284 , where @xmath90 is the total number of single - charm ( charm - pair ) events generated .
the number of single - charm events generated is , of course , just twice the number of charm - pair events generated . in fig .
[ fig : fig30 ] , we compare the complementary methods used by the hvqmnr program and the pythia / jetset event generator to include higher - order perturbative corrections to the leading - order partonic cross section . as discussed in the previous section , the pythia / jetset event generator ,
beginning with leading - order matrix elements , uses parton showers to include higher - order perturbative effects , whereas the hvqmnr program calculates the next - to - leading order @xmath12 cross section . to more
directly compare these two approaches , we change three of the default pythia / jetset settings @xmath258 , @xmath259 and @xmath260 to match the default hvqmnr settings .
( see table [ tab : tab5 ] . )
we obtain pythia / jetset @xmath12 distributions assuming no intrinsic transverse momentum for the interacting partons , as well as assuming @xmath285 gev which is the pythia / jetset default .
as argued by t. sj " ostrand , the intrinsic transverse momentum may , at least in part , be seen as a replacement for gluon emission that is truncated in the parton shower approach due to the introduction of an energy scale below which the parton shower evolution is stopped@xcite . in fig .
[ fig : fig30 ] , we also show the hvqmnr leading - order distributions to emphasize which distributions are , and which are not , sensitive to higher - order corrections .
llcccccclc & & & & & @xmath258 & @xmath286 & & + generator & & pdf & @xmath287 & @xmath288 & [ gev ] & [ gev ] & fig./ps & + + mnr & nlo & ( 1 ) & 1.0 & 2.0 & 1.5 & 0 & [ fig : fig30 ] & * solid * + mnr & lo & ( 1 ) & 1.0 & 2.0 & 1.5 & 0 & & dashed + p / j & @xmath12 refers to the default pythia / jetset @xmath12 distributions .
] & ( 1 ) & 1.0 & 1.0 & 1.5 & 0 & y & dotted + p / j & @xmath12 & ( 1 ) & 1.0 & 1.0 & 1.5 & 0.44 & y & solid + + mnr & nlo & ( 1 ) & 1.0 & 2.0 & 1.5 & 0 & [ fig : fig31 ] & solid + mnr & nlo & ( 1 ) & 1.0 & 2.0 & 1.2 & 0 & & dashed + mnr & nlo & ( 1 ) & 1.0 & 2.0 & 1.8 & 0 & & dotted + + mnr & nlo & ( 1 ) & 1.0 & 2.0 & 1.5 & 0 & [ fig : fig32 ] & * solid * + mnr & nlo & ( 2 ) & 1.0 & 2.0 & 1.5 & 0 & & dashed + mnr & nlo & ( 3 ) & 1.0 & 2.0 & 1.5 & 0 & & dotted + mnr & nlo & ( 4 ) & 1.0 & 2.0 & 1.5 & 0 & & solid + + mnr & nlo & ( 2 ) & 0.5 & 0.5 & 1.5 & 0 & [ fig : fig33 ] & solid + mnr & nlo & ( 2 ) & 1.0 & 1.0 & 1.5 & 0 & & dashed + mnr & nlo & ( 2 ) & 1.5 & 1.5 & 1.5 & 0 & & dotted + + p / j & @xmath13 refers to the default pythia / jetset @xmath13 distributions . ] & ( 5 ) & 1.0 & 1.0 & 1.35 & 0.44 & [ fig : fig34]/y & * solid * + p / j & @xmath12 & ( 5 ) & 1.0 & 1.0 & 1.35 & 0 & y & dashed + p / j & @xmath12 & ( 5 ) & 1.0 & 1.0 & 1.35 & 0.44 & n & dotted + p / j & @xmath13 & ( 5 ) & 1.0 & 1.0 & 1.35 & 0 & n & solid + + p / j & @xmath13 & ( 5 ) & 1.0 & 1.0 & 1.35 & 0.44 & [ fig : fig35]/y & * solid * + p / j & @xmath13 & ( 5 ) & 1.0 & 1.0 & 1.35 & 0.7 & y & dashed + p / j & @xmath13 & ( 5 ) & 1.0 & 1.0 & 1.35 & 1.0 & y & dotted + p / j & @xmath13 & ( 5 ) & 1.0 & 1.0 & 1.35 & 1.5 & y & solid + figure [ fig : fig30 ] shows that higher - order perturbative corrections do not significantly affect the shapes of most of the single - charm and charm - pair distributions .
that is , the hvqmnr leading - order and next - to - leading predictions for all distributions are very similar except for the @xmath209 , @xmath157 , and @xmath289 distributions . in the leading - order calculation , these latter distributions are delta functions at 0 ( gev/@xmath290 , @xmath222 , and 0 ( gev/@xmath290 , respectively because the leading - order charm and anticharm quark are back - to - back in the plane transverse to the beam axis . the next - to - leading order predictions and the parton shower prediction are also quite similar .
the @xmath209 and @xmath157 pythia / jetset distributions with no intrinsic transverse momentum included , indicate that the parton shower evolution is playing a very small role . the @xmath157 parton shower distribution , in particular , is closer to the leading - order delta - function prediction than to the next - to - leading order prediction . adding intrinsic transverse momentum , with @xmath285 gev ,
brings the pythia / jetset prediction very close to the next - to - leading order hvqmnr prediction . in fig .
[ fig : fig31 ] , we investigate the degree to which the single - charm and charm - pair distributions are sensitive to variations in the mass of the charm quark .
all distributions are obtained from hvqmnr nlo calculations using the default values for all parameters except for @xmath258 .
higher - order effects play a larger role as the charm - quark mass decreases because the ratio @xmath291 decreases , where @xmath220 is the energy scale of the interaction ( eq .
[ eq : q2 ] ) . for the lightest charm - quark mass ( @xmath292 gev/@xmath65 ) ,
the single - charm @xmath148 and @xmath196 distributions are steepest because the outgoing charm quark can more easily radiate gluons ; the single - charm @xmath19 distribution is less peaked near @xmath293 ; and the invariant mass of the charm - pair is significantly steeper than the higher mass predictions .
the increase in higher - order effects for smaller @xmath258 is also evident in the @xmath157 distribution , which is flattest for @xmath292 gev/@xmath65 . in fig .
[ fig : fig32 ] , we investigate the degree to which the single - charm and charm - pair distributions are sensitive to variations in the parton distribution functions and @xmath261 .
all distributions are obtained from hvqmnr nlo calculations using the default values for all parameters except for the parton distribution functions .
we examine predictions for four pairs of pion and nucleon parton distribution functions , sets ( 1 ) through ( 4 ) defined in table [ tab : tab4 ] . at fixed - target energies ,
the dominant contribution to the @xmath12 cross section is from gluon fusion . in fig .
[ fig : fig36 ] , we compare the gluon distribution functions @xmath294 for sets ( 1 ) through ( 4 ) . by energy conservation
, the energy of the two colliding partons must be at least twice the mass of a charm quark to produce a @xmath12 pair ; that is , @xmath295 where @xmath10 gev is the center - of - mass energy of the colliding hadrons . hence , for each set
, the pion and nucleon functions are obtained after imposing the constraint @xmath296 .
we impose this constraint because we want to investigate how the four sets compare in the region of @xmath264 that we explore , not in the very low @xmath264 region where the functions are largest .
although the four sets of parton distribution functions differ significantly , the single - charm and charm - pair distributions shown in fig . [ fig : fig32 ] are not very sensitive to these differences .
the sensitivity of the @xmath157 distribution is due to the variation in @xmath261 in sets ( 1 ) through ( 4 ) ( see table [ tab : tab4 ] ) . as the value of @xmath261 increases ,
the ratio @xmath291 decreases , where @xmath220 is the energy scale of the interaction ( eq .
[ eq : q2 ] ) , causing higher - order effects to play a larger role .
hence , the flattest @xmath43 distribution results from using set ( 4 ) ( @xmath297 gev ) ; the steepest @xmath43 distribution , from using set ( 3 ) ( @xmath298 gev ) . in fig .
[ fig : fig33 ] , we investigate the degree to which the single - charm and charm - pair distributions are sensitive to variations in the renormalization and factorization scales .
all distributions are obtained using the hvqmnr nlo calculation .
we set the two arbitrary scales equal to each other , @xmath299 , and obtain distributions for @xmath300 , @xmath220 , and @xmath301 , where @xmath220 gives the energy scale of the interaction ( eq .
[ eq : q2 ] ) .
we use the grv parton distribution functions for both the pion and the nucleon ( set ( 2 ) in table [ tab : tab4 ] ) , which have been evolved down to @xmath302 = 0.3 @xmath303 . with this choice
, the factorization scale @xmath52 can go as low as @xmath304 without going below @xmath269 . as mentioned
, the degree to which the distributions are sensitive to variations in the renormalization and factorization scales gives an indication of how much ( or little ) we can trust the @xmath270 calculation .
as expected , the distributions that are most sensitive to variations in @xmath52 are those distributions that are trivial at leading - order : @xmath209 , @xmath157 , and @xmath289 .
the smaller the factorization and renormalization scales , the broader these distributions are .
that is , the higher - order @xmath305 terms play a larger role , compared to the leading - order @xmath306 terms , as renormalization and factorization scales decrease .
the sensitivity of these scales indicates that the model may have several ways of obtaining accurate predictions , both by adjusting the model s parameters , and by adding higher order terms . in fig .
[ fig : fig34 ] , we look separately at the effects of parton showers , intrinsic transverse momentum , and hadronization . all distributions are obtained using the pythia / jetset event generator . the distributions obtained using the default settings ( * solid * ) include all three effects .
we compare these default distributions to three sets of distributions that are obtained by including : * _ only _ hadronization , but no parton shower evolution or intrinsic transverse momentum ( solid ) ; * _ only _ the parton shower evolution , but no intrinsic transverse momentum or hadronization ( dashed ) ; * _ only _ intrinsic transverse momentum , but no hadronization or parton shower evolution ( dotted ) . for the longitudinal momentum distributions ( @xmath148 , @xmath307 , @xmath308 , @xmath19 , @xmath309 , @xmath310 ) , the most important factor is whether or not hadronization is included .
the two sets of distributions that include hadronization effects are quite similar ; the two sets of distributions that do not include hadronization effects are similar ; but the latter two sets of distributions differ significantly from the former two sets . in the pythia / jetset hadronization model ,
the broadening of the longitudinal momentum distribution is the result of color - connecting the charm quark to a valence antiquark ( or diquark ) from one of the colliding hadrons and the anticharm quark to a valence quark from the other colliding hadron .
all three higher - order effects broaden the leading - order delta function prediction for the @xmath157 distribution .
the broadening due to the parton shower evolution , however , is significantly smaller than the broadening due to either the hadronization process or the addition of intrinsic transverse momentum ( @xmath285 gev@xmath311 ) .
the latter two effects broaden the @xmath157 distribution by roughly the same amount .
all three higher - order effects also broaden the leading - order delta function prediction for the @xmath289 distribution . in this case , however , the broadening due to the parton shower evolution is larger than the broadening due to either hadronization effects or the addition of intrinsic transverse momentum ( @xmath285 gev@xmath311 ) . in fig .
[ fig : fig35 ] , we investigate the degree to which the single - charm and charm - pair distributions are sensitive to variations in the amount of intrinsic transverse momentum added to the hard - scattering partons that collide to form a @xmath12 pair .
all distributions are obtained from the pythia / jetset event generator , with default settings for all parameters except for the width of the gaussian intrinsic transverse momentum distribution , @xmath286 .
when intrinsic transverse momentum is included , the hard - scattering partons from the colliding hadrons are no longer necessarily moving parallel to the colliding hadrons .
the plane that is transverse to the axis of the parton - parton collision which can not be determined experimentally is no longer the same as the plane that is transverse to the beam axis .
hence , including intrinsic transverse momentum smears the leading - order prediction @xmath312 . not surprisingly , the distributions that are most sensitive to variations in @xmath286 are those transverse distributions that are trivial at leading - order : @xmath204 , @xmath43 , and @xmath289 .
as the width @xmath286 increases , these distributions become flatter . the longitudinal momentum distributions @xmath148 , @xmath307 , @xmath308 , @xmath19 , @xmath309 , and @xmath310
are relatively insensitive to all variations considered above , except for inclusion or omission of the pythia / jetset hadroproduction hadronization ( fig .
[ fig : fig34 ] ) .
the steepness of the invariant mass distribution is also sensitive to whether or not hadronization is included , as well as to the mass of the charm quark ( fig .
[ fig : fig31 ] ) .
therefore , the measured distributions for these physics variables , discussed in sec .
[ sec : results ] , provide a test of the pythia / jetset hadronization model in particular , a test of the string topology scheme that color - connects the charm quark to a valence quark from one of the colliding hadrons and the anticharm quark to a valence quark from the other colliding hadron . the transverse distributions @xmath204 , @xmath45 , @xmath43 , and @xmath289 are sensitive to almost all variations considered above because they are sensitive to the degree of correlation between the charm and anticharm transverse momenta . varying @xmath258 ( fig .
[ fig : fig31 ] ) , @xmath261 ( fig .
[ fig : fig32 ] ) , or @xmath262 ( fig .
[ fig : fig33 ] ) in the next - to - leading order calculation changes the definition of the running coupling constant @xmath313 , which is approximately proportional to @xmath314 . as the coupling constant increases
that is , as @xmath258 decreases , @xmath261 increases , or @xmath262 decreases higher - order effects play a larger role , and consequently the charm and anticharm transverse momenta become less correlated .
the other methods we discussed for including higher - order effects were parton showers , intrinsic transverse momentum , and hadronization . in sec .
[ sec : results ] , we quantify the degree of correlation between the transverse momenta of the @xmath0 and @xmath1 mesons from our @xmath13 data sample . the sensitivity of the nlo predictions to the arbitrary renormalization and factorization scales ( fig .
[ fig : fig33 ] ) indicates that higher - order perturbative corrections are important . in principle
, one could determine the sets of theoretical parameters that generate predictions that are in good agreement with the full range of experimental results .
the set of fit values chosen , however , may not be unique .
the fit values of ( _ e.g. _ , the mass of the charm quark ) would depend on the values of the renormalization and factorization scales .
for example , if a renormalization scale of @xmath315 , rather than @xmath220 , is assumed , then a smaller value for @xmath286 or a larger value for @xmath258 could each be used to fit the data .
b. andersson , g. gustafson , g. ingelman and t. sj " ostrand , phys .
rep . * 97 * , 31 ( 1983 ) ; t. sj " ostrand , computer physics commun . *
39 * , 347 ( 1986 ) ; t. sj " ostrand and m. bengtsson , computer physics commun . *
43 * , 367 ( 1987 ) .
see for example , the doubly cabibbo supressed decay @xmath316 and @xmath317 mass difference papers- e. m. aitala _ et al .
_ , phys .
b*404 * , 187 ( 1997 ) , and e. m. aitala _ et al .
_ , phys .
b*379 * , 292 ( 1996 ) , respectively .
w. t. eadie , d. drijard , f. e. james , m. roos , b. sadoulet , _ statistical methods in experimental physics _
( north holland , 1971 ) , p. 174 ; a. g. frodesen , o. skjeggestad , h. tofte , _ probability and statistics in particle physics _ ( universitetsforlaget , norway , 1979 ) , p. 40 | we present a study of correlations between @xmath0 and @xmath1 mesons produced in 500 gev/@xmath2 @xmath3-nucleon interactions , based on data from experiment e791 at fermilab .
we have fully reconstructed @xmath4 charm meson pairs to study correlations between the transverse and longitudinal momenta of the two @xmath0 mesons and the relative production rates for different types of @xmath0 meson pairs .
we see slight correlations between the longitudinal momenta of the @xmath0 and the @xmath5 , and significant correlations between the azimuthal angle of the @xmath0 and the @xmath5 .
the experimental distributions are compared to a next - to - leading - order qcd calculation and to predictions of the pythia / jetset monte carlo event generator .
we observe less correlation between transverse momenta and different correlations between longitudinal momenta than these models predict for the default values of the model parameters .
better agreement between data and theory might be achieved by tuning the model parameters or by adding higher order perturbative terms , thus contributing to a better understanding of charm production .
the relative production rates for the four sets of charm pairs , @xmath6 , @xmath7 , @xmath8 , @xmath9 , as calculated in the pythia / jetset event generator with the default parameters , agree with data as far as the relative ordering , but predict too many @xmath8 pairs and too few @xmath9 pairs . |
star clusters are an important tool for understanding the formation and evolution of a galaxy .
being an aggregate of many stars , they are bright enough to be observed at great distances .
they span a very broad range of age , from a few myrs to more than 10 gyrs , encapsulating the whole history of a galaxy .
age distribution of star clusters in a galaxy tells us about the formation and destruction history of the clusters and about the evolution of the galaxy as well .
spatial distribution of the star clusters delineates the structure of a galaxy , and their velocity distribution reveals the kinematics of stellar population and the mass of the host galaxy .
recent studies for star clusters in nearby late type galaxies have revealed details of the star clusters , showing that the populations of star clusters are diverse among the galaxies ( see @xcite and references therein ) .
local group galaxies in the nearby universe are ideal targets , due to their proximity , for studying in detail the properties of star clusters .
there are three spiral galaxies in the local group : our galaxy , m31 and m33 .
while star clusters in our galaxy and m31 have been extensively studied , the star clusters in m33 received relatively less attention .
m33 ( = ngc 598 ) , the triangulum galaxy , is a late type spiral galaxy ( scd ) .
it is located at @xmath13 kpc from us @xcite .
its large angular size and inclination of @xmath14 @xcite make it particularly suitable for studies of its stellar contents .
its mass , @xmath15 @xcite , falls in the intermediate range among the local group galaxies .
surveys for star cluster in m33 have been rather sporadic .
early photometric surveys by @xcite and @xcite identified about 25 star cluster candidates .
@xcite reported the discovery of 58 possible star cluster candidates , including most of the objects studied by @xcite .
@xcite presented the most comprehensive catalog of 250 non - stellar objects in m33 , including @xmath16 photometry of 60 star cluster candidates .
later @xcite presented @xmath0 photometry for 71 additional star cluster candidates .
all these surveys were based on the photographic plates so that they were limited rather to the outer part of m33 . the first survey for m33 star clusters based on ccd imaging
was given by @xcite using the @xmath0 images from direct project @xcite .
they presented a list of 51 globular cluster candidates . while they covered the central region of m33 where previous surveys could not examine in detail , their survey was limited to the bright star clusters of @xmath17 mag , and did not cover the entire area of m33 .
the hubble space telescope ( hst ) offers a great advantage over the ground - based telescopes for star cluster survey in m33 . at the distance of m33
, star clusters can be easily identified without ambiguity with hst images .
therefore , star cluster samples from hst images are much less likely to be contaminated by other extended sources such as a background galaxy or an h ii region .
@xcite searched for star clusters in m33 with hst data for the first time .
they reported a discovery of 162 star clusters including 131 previously unknown ones in the 55 fields of hst / wfpc2 images .
recently , @xcite have found 33 star clusters and 51 possible candidates from the @xmath18 image of one hst / acs field .
@xcite also reported a discovery of 24 star clusters including 11 new ones from the @xmath19 images of two hst / acs fields in m33 .
however , the areas of m33 covered by the previous hst - based surveys are much smaller than the entire area of m33 .
therefore , it is needed to increase the number of hst fields to make as complete a sample of star clusters as possible .
the sample of star clusters established from the hst images also serves as a template for ground - based star cluster surveys .
we searched for new m33 star clusters using the additional hst / wfpc2 images that became available after @xcite s work .
we combined our result with previous m33 star cluster catalogs based on the hst images @xcite , and obtained homogeneous @xmath0 photometry for the star clusters from the deep ccd imaging data taken at the canada - france - hawaii - telescope ( cfht ) ( park et al .
2007 , in preparation ) in this paper , we present the result of m33 star cluster survey with the hst / wfpc2 images . throughout this paper
, we will refer the star clusters from the catalog of @xcite with b designation , and ones from @xcite with cbf .
cs designation means the star clusters from @xcite , mkkss designation means the ones from @xcite , md the ones from @xcite .
finally , sbghs designation means the star clusters from @xcite .
the paper is composed as follows .
section 2 describes the hst / wfpc2 and cfht data used in this study . in section 3 , star cluster search method and the integrated photometry procedures on cfht images
are described .
section 4 presents various properties of the star clusters : color - magnitude diagrams ( cmd ) , color - color diagrams ( ccd ) , and spatial distribution of star clusters .
finally , summary and conclusion are given in the last section .
we used wfpc2 images of m33 in the hst archive that were not used in the previous studies by @xcite .
the images were obtained using various filter combinations of @xmath20 , @xmath21 , @xmath22 , @xmath23 , @xmath24 , and @xmath25 .
we selected images only with exposure times longer than 100 sec either in @xmath26 or @xmath27 .
the number of fields thus selected is 24 , as listed in table [ fld_list ] .
the locations of the fields are displayed on fig .
[ fld_location ] , along with the locations of hst / wfpc2 fields for previous star cluster surveys @xcite .
nine of our fields are partially overlapped with previous survey regions .
hst / wfpc2 image sets used by @xcite were also downloaded from the archive , and images of previously known star clusters on these image sets were utilized as references for the visual search on the new image sets .
bright star clusters in m33 are easily resolved on the hst / wfpc2 images so that we selected star clusters using visual inspection of each image .
we considered extended objects with the hint of resolved stars as star clusters .
since the hst / wfpc2 image sets used in this study have different depths , and they cover different environments in m33 with different filters , we could not put a completeness limit of the survey for each image .
the detection of star clusters depends not only on the brightness of the star clusters , but also on their sizes , morphologies , and environments .
experience from the visual investigation tells us that star clusters with @xmath28 can be identified in the shallowest images of the current hst / wfpc2 image sets . from the investigation of hst / wfpc2 fields
, we have found 104 star clusters , 32 of which are new ones .
the 32 new star clusters are listed in table [ newcl_list ] , and their @xmath26-band grayscale images are shown in fig .
[ newcl ] . during the investigation
, we identified eight cbf star clusters that were redundantly identified and registered under two different ids : cbf-15 ( cbf-45 ) , cbf-20 ( cbf-126 ) , cbf-22 ( cbf-91 ) , cbf-53 ( cbf-87 ) , cbf-54 ( cbf-85 ) , cbf-56 ( cbf-156 ) , cbf-59 ( cbf-95 ) , and cbf-60 ( cbf-94 ) .
the two star clusters in each pair have not only similar coordinates , but also similar @xmath26 magnitudes with differences @xmath29 mag .
we inspected carefully the hst / wfpc2 images of the regions where eight pairs are located , and we found neither two star clusters of similar magnitudes in the regions , nor any possible substructures ( i.e. multiple ob groups in a cluster ) in the clusters that @xcite might have recognized as two separate star clusters .
therefore , we regard these eight cases as redundant identifications of the same star cluster . throughout this paper
, we will present the total number of cbf samples as 154 instead of 162 . from these 154 cbf samples
, we identified 45 star clusters on our image sets .
one of our fields partially overlaps the region where @xcite searched for star clusters with an hst / acs image , where they found 33 star clusters and 51 star cluster candidates .
14 star clusters and 18 star cluster candidates fall on our hst / wfpc2 image .
we could identify 13 star clusters except cluster i d 15 in @xcite s list that seems to be too faint to be seen on our @xmath30 image .
we also inspected the 18 star cluster candidates on our image .
however , they are so compact in size that it was impossible to tell whether they are star clusters , except for cluster i d 83 in @xcite s list .
we classified this one as a star cluster after careful inspection of the image .
in addition , we found that 12 star cluster candidates from previous ground - based surveys @xcite are real star clusters .
the coordinates , photometric properties , and cross identification among the various studies for these 71 previously known clusters are listed in table [ othercl_list ] .
we make a catalog of 242 star clusters in m33 , combining our result with the previous results based on hst - based surveys @xcite .
it is the largest catalog of m33 star clusters available to date which includes only the star clusters confirmed in the hst images .
it consists of 32 newly found star clusters in this study , 12 previous star cluster candidates that were confirmed to be the genuine ones in this study , 154 star clusters from @xcite , 32 star clusters out of 33 star clusters in @xcite , one star cluster from the star cluster candidate list of @xcite , and 11 new star clusters from the list of @xcite .
the positions and the photometry of all m33 star clusters not included in tables [ newcl_list ] and [ othercl_list ] are listed in table [ prevcl_list ] .
all star clusters in the combined catalog were observed with hst in different conditions with different purpose .
the combination of the filters used for the hst observation varies depending on the field . to secure homogeneous photometry data for all clusters , we used the integrated photometry of these clusters derived from the ground - based images given in park et al .
( 2007 , in preparation ) instead of deriving the photometry from the hst images .
park et al .
( 2007 ) presented @xmath0 integrated aperture photometry of clusters and cluster candidates in the @xmath31 field of m33 based on the ccd images taken with cfh12k mosaic camera at the canada - france - hawaii telescope ( cfht ) .
the typical seeing of the data is @xmath32 arcsec for all filters , and the point sources of @xmath33 can be measured with errors smaller than @xmath34 .
details of the photometry are given in park et al .
( 2007 ) , and we give a brief description here .
examination of an aperture growth curve of isolated bright clusters on cfh12k ccd images indicates that most of star cluster light is contained within the aperture of @xmath35 arcsec .
however , using such an aperture would cause a large scatter in the measurement of some clusters , because star clusters are often located in the crowded field in m33 so that some neighbor objects happen to be inside the same aperture .
for this reason , we used an aperture of @xmath36 arcsec for the measurement of color , although the aperture of @xmath37 arcsec was used for @xmath26 magnitude measurement to measure the total light of the star clusters .
some of the star clusters we missed happened to fall on the gaps between the chips in mosaic camera , and some of them happen to be located in so crowded a region that we could not derive their photometry from the cfht images . to supplement the photometry data for star clusters that could not be measured with cfh12k data ,
aperture photometry was carried out on the hst / wfpc2 images .
we made photometry on the fields that were observed with more than two filters since instrumental magnitude can not be converted to standard system without color information . for most of fields ,
two hst / wfpc2 images of identical pointings exist per filter .
fields observed with only one exposure were not included in the measurement .
two images were combined with combine task in iraf with crreject option to remove the cosmic rays .
cosmetic defects were masked with data quality images that accompany the scientific images .
the centers of the clusters were determined visually with iraf / imexamine task .
then iraf / phot task was used to measure the instrumental magnitudes of star clusters on the combined images . to be consistent with previous hst - based studies @xcite , aperture of @xmath38 arcsec
was adopted for the magnitude measurement , while aperture of @xmath39 arcsec for color measurement .
local median values were measured as sky background level in the annulus of inner radius of @xmath40 arcsec and outer radius of @xmath41 arcsec around each star cluster .
the cte correction was applied to the instrumental magnitudes according to the prescription given at dolphin ( 2002 ) s web site .
however , we did not correct for the geometrical distortions in the wfpc2 ccd .
the instrumental magnitudes were converted to standard system using the relation also found at the same web site iteratively . in the end , we derived the photometry of 227 out of 242 star clusters with both cfh12k and hst / wfpc2 images .
we compared our photometry of the clusters with previous photometry .
there are 154 clusters in common with @xcite , 16 clusters in common with @xcite , 12 clusters in common with @xcite , and 11 clusters in common with @xcite .
note that all photometry was carried out with different aperture sizes : @xcite and @xcite used nominal aperture of @xmath42 arcsec for magnitude measurement , while @xcite used apertures of @xmath43 and @xmath44 arcsec , respectively .
@xcite used an aperture of @xmath45 arcsec .
however , aperture sizes in all cases were determined to contain as much as possible the light from a star cluster , so the resultant aperture magnitudes for a star cluster should be more or less the same . fig .
[ photdiff ] shows the comparison of our photometry of the clusters with previous photometry .
large scatter made it difficult to see the difference in magnitude , but for bright star clusters ( @xmath46 ) , there is little magnitude differences between this and other studies : @xmath47(this study@xcite ) @xmath48 with @xmath49 , @xmath47(this study@xcite ) @xmath48 with @xmath50 , @xmath47(this study@xcite ) @xmath51 with @xmath52 , and @xmath47(this study@xcite ) @xmath48 with @xmath53 , respectively .
for fainter sources , however , our measurements get systematically fainter than the other space based measurements . given that most of the star clusters are located in the region near the m33 center where crowding is severe and
sky varies significantly , the sky background level was likely to be overestimated in ground - based images due to the unresolved faint stars , which would lead to a fainter estimation of source brightness .
both the @xmath54 and @xmath55 colors show good agreement with all the previous studies down to the faintest magnitude in our samples .
the differences between this study and other photometry are @xmath56(this study@xcite)@xmath57 with @xmath58 , @xmath59(this study@xcite)@xmath60 with @xmath52 , @xmath56(this study@xcite)@xmath61 with @xmath52 , @xmath59(this study@xcite)@xmath62 with @xmath63 , @xmath56(this study@xcite)@xmath64 with @xmath65 , and @xmath59(this study@xcite)@xmath66 with @xmath65 . the only exception to
this tendency is the @xmath55 difference between ours and those of @xcite , which turned out to be : @xmath59(this study@xcite ) @xmath67 with @xmath52 . to get the intrinsic color of the star clusters ,
the photometric measurements must be corrected for reddening .
we estimated the foreground reddening for each star cluster using the @xmath68@xmath54 diagram for early type stars in the field covering the cluster .
we used the @xmath69 photometry of stars in the area of m33 covered in the local group survey given by @xcite .
the @xmath70 derived for each cluster is listed in the fourth column of table [ newcl_list ] , [ othercl_list ] , and [ prevcl_list ] .
@xmath70 values we derived for m33 star clusters range from @xmath71 to @xmath72 . for about 130 star clusters ,
the value of @xmath70 turned out to be @xmath73 , and for about 100 star clusters the values of @xmath70 were found to be @xmath74 .
@xcite obtained @xmath70 toward their 60 star clusters either from the photometry of neighboring individual stars by applying reddening free parameters or by using @xmath70 values from @xcite .
the @xmath70 values range @xmath75 , and most of them are found to be @xmath76 .
@xcite also derived reddening for their 10 globular cluster samples from the shapes of the red giant branches ( rgb ) in the color - magnitude diagrams of resolved stars .
their @xmath77 estimates are below @xmath34 , except for @xmath78 for cluster i d 131 ( = cs - h10 ) . @xcite also obtained @xmath77 for 17 star clusters in the same way as @xcite , which @xmath77 ranges from @xmath79 .
since our error in @xmath70 measurement is as large as @xmath71 , it is quite difficult to see the consistency between our result and other results .
but the fact that @xmath70 distribution shows peak at @xmath80 , a typical value found in previous studies @xcite , indicated that the result obtained in this study is in general consistent with previous studies .
fig . [ cmd ] displays the integrated @xmath81 and @xmath82 color - magnitude diagrams ( cmd ) of the star clusters in m33 .
absolute magnitudes of the star clusters were derived for the adopted distance modulus of @xmath83 @xcite and the visual extinction to reddening ratio @xmath84 @xcite . for comparison , we display the cmd for the star clusters in the lmc @xcite in the panel ( c ) , and the cmds for the open clusters @xcite and globular clusters @xcite in our galaxy in panel ( d ) .
we adopted a value for the foreground reddening of @xmath85 @xcite and the distance modulus of @xmath86 for the lmc clusters .
several features are noted in fig .
first , the star clusters in all three galaxies are roughly separated into blue and red groups with a color boundary of @xmath87 in the @xmath81 cmd .
however , this separation for the m33 clusters in the @xmath82 cmd is not as clear as in the @xmath81 cmd .
second , the dominant population in m33 is blue clusters , as known before @xcite .
most of the new clusters found or the clusters confirmed in this study are also blue clusters .
third , the brightest blue clusters in m33 are as bright as @xmath88 mag , which is similar to the value for the galactic open clusters , but about 1.5 magnitude fainter than that for the lmc .
fourth , the blue envelope for the blue clusters in m33 in the @xmath89@xmath3 cmd is slightly redder than that for the lmc blue clusters , as seen from the normalized @xmath3 color histograms in the panel ( a ) and ( c ) .
fifth , there are only a small number of red clusters with @xmath90 in m33 .
the @xmath3 color range of the red clusters in m33 , @xmath91 , is similar to that for the galactic globular clusters , but is much larger than that for the lmc red clusters , @xmath92 .
we found only one new red cluster , and confirmed two known red clusters .
cluster i d 85 ( i d 27 in @xcite ) , one of the two red clusters confirmed in this study , happens to be the brightest ( @xmath93 and @xmath94 ) among the known red clusters in m33 .
the reddest cluster is cluster i d 89 , which has integrated colors @xmath95 , and @xmath96 .
it is interesting to note that there are only a few red star clusters that are fainter than @xmath97 in m33 , while there are many such faint red star clusters seen in the lmc .
the lack of faint red clusters in our catalog of m33 clusters may be due to incompleteness of our cluster search or may be an intrinsic feature of the m33 star cluster system .
while we selected hst / wfpc2 images on the basis of @xmath26 or @xmath27 exposure times , many of resultant image sets were obtained with combinations of @xmath26 and shorter wavelength passbands such as @xmath98 and @xmath99 , rather than the combinations of @xmath26 and @xmath27 .
so there is a possibility that our star cluster detection might have been biased toward blue star clusters .
however , it may be an intrinsic feature , considering that @xcite already showed that m33 red star clusters with @xmath100 has log normal type luminosity function ( lf ) , while intermediate color clusters with @xmath101 show no turnover down to the completeness limit in their lf .
[ ccd ] shows the integrated @xmath3@xmath9 color - color diagram ( ccd ) of m33 star clusters , along with @xmath3 and @xmath9 histograms of m33 star clusters .
galactic globular clusters are also plotted for comparison .
we overlaid the theoretical evolutionary path for the single stellar population ( ssp ) @xcite for @xmath7 , @xmath8 that was reported to give the best overall fit to their m33 star cluster samples which younger than @xmath102 @xcite . for comparison , evolutionary path of ssp for @xmath103
, @xmath104 is also overlaid .
most of the new clusters in m33 including the ones from @xcite are located along the sequence that is consistent with the theoretical evolutionary path for @xmath7 , @xmath8 , while a few of them are in the redder side in the @xmath9 color .
this is similar to the case of previously known clusters in m33 and also to the case of the star clusters in m51 @xcite .
some bright clusters among these redder clusters are consistent with the theoretical evolutionary path for @xmath103 , @xmath104 .
however , the rest of them can not be explained with any other models .
the continuous distribution of star clusters along the model line indicates that m33 star clusters have been formed continuously from the epoch of the first star cluster formation until recent time .
the @xmath9 distribution shows only one peak at @xmath105 and monotonically decrease toward red end , while the @xmath3 distribution shows two distinct peaks at @xmath106 and @xmath107 .
the @xmath3 distribution is similar to that given by @xcite , except that the red peak at @xmath107 in our result , is 0.1 bluer than their result , @xmath108 .
the blue peak corresponds to the age of @xmath109 year and the red one to the age of @xmath110 year , if we assume the metallicity of the star clusters , @xmath7 .
based on the integrated photometry of the clusters , we can select old globular cluster candidates from our catalog .
the @xmath3 color distribution shows the blue end of red star cluster population is located at @xmath111 , which is similar to that of galactic globular clusters .
although we do not see any similar red peak at @xmath9 color distribution of m33 star clusters , it is reasonable to set the color criterion for @xmath9 color as well , in accordance with the @xmath9 distribution of galactic globular clusters .
the color ranges for most of galactic globular clusters in fig .
[ ccd ] are @xmath1 and @xmath112 .
there are 29 star clusters with this color range in our catalog , which are old globular cluster candidates in m33 .
ten of them are newly found in this study .
the globular cluster candidates are listed in table .
[ gccand_list ] , along with their integrated photometry and cross identifications with previous catalogs .
nine out of ten globular clusters in @xcite satisfy our color selection criterion .
one globular cluster , c38 in @xcite ( = cbf-114 ) , is not included because its color @xmath113 is a little bluer than the blue limit of the criterion .
our list of 29 globular cluster candidates also includes three oldest star clusters in the list of @xcite , clusters i d 58 , 131 , and 228 that are all estimated to be older than @xmath114 .
we divided the m33 star cluster sample into three groups according to their @xmath3 color for further analysis : blue star clusters with @xmath115 , intermediate color star clusters with @xmath5 , and red star clusters with @xmath116 .
[ spatial_dist ] displays the spatial distribution of each color group .
we also plotted for comparison the locations of h ii regions @xcite that trace well the spiral arm structures of m33 .
since hst / wfpc2 observations covered in our catalog do not cover the entire area of m33 , the spatial distribution of star clusters can not be used for the study of spatial structure of the m33 cluster system .
[ spatial_dist ] shows that the red clusters are found relatively more in the outer region of m33 than the blue and intermediate color clusters , and that many of the blue stars are located in the hii regions .
note the regions investigated in this study include three fields far east from m33 center ( see fig .
[ fld_location ] for locations of these fields . ) . had any star clusters been found in these regions , it would give a clue on the extent of star cluster system in m33 .
however , no star clusters were found in these regions with our image sets . with the largest sample of star clusters available , we derived the luminosity function ( lf ) of star clusters , displaying it in fig .
[ lf](a ) .
although we did not estimate the completeness of our survey , it appears that our survey is reasonably complete for clusters brighter than @xmath46 .
therefore , only the bright part of lf is considered in further discussion . in fig .
[ lf](a ) the lf for the blue star clusters shows a peak at @xmath10 mag , and decreases as the magnitude gets fainter , while the intermediate color star cluster lf shows a peak at the fainter magnitude @xmath11 mag .
the lf for the red star clusters shows also a peak at @xmath12 mag , although the number of the clusters is small .
we fitted the lfs for @xmath117 mag using a single gaussian function , obtaining @xmath118 mag and @xmath119 for the blue star clusters , @xmath120 mag and @xmath121 for the intermediate color star clusters , and @xmath122 mag and @xmath123 for the red star clusters .
@xcite presented the lf of the m33 star clusters for two color groups : @xmath124 ( intermediate color star clusters ) and @xmath125 ( red star clusters ) .
so we derived the cluster lfs for the same color groups , as shown in fig .
[ lf](b ) .
we also plotted the cluster lfs given by @xcite in fig .
[ lf](c ) for comparison .
we adjusted the lfs given by @xcite applying the same distance modulus for m33 ( we adopted @xmath126 , while @xcite adopted @xmath127 ) , shifting the magnitude bins of @xcite s lf 0.16 magnitude brighter . in fig .
[ lf](b ) the lfs for the intermediate color and red star clusters derived according to the @xmath9 color show peaks at @xmath12 mag , which is roughly similar to those derived according to the @xmath3 color in fig .
[ lf](a ) .
however , the lfs for the intermediate color and red star clusters derived in this study are somewhat different from those given by @xcite in fig .
[ lf](c ) .
our lfs contain a larger number of faint star clusters than those given by @xcite .
the peak magnitude in the lf for the red star clusters derived in this study is about 0.3 magnitude fainter than that of @xcite , @xmath128 mag , while the peak magnitude in the lf for the intermediate color star clusters in this study is similar to that of @xcite s , @xmath129 mag .
@xcite sampled brighter star clusters more than in our survey .
@xcite combined their catalog based on hst / wfpc2 data with the previous catalogs of m33 star clusters to increase the number of the sample for deriving the lfs .
we used only star cluster samples obtained with hst observation .
previous ground based surveys covered a wider area in m33 with shallower depth and poorer resolution compared with the hst survey so that they would include preferentially a larger number of bright and large clusters than the hst surveys .
for example , the blue lf for @xcite samples included in this study with @xmath4 has a peak at the same magnitude to that for blue samples displayed in fig .
[ lf](a ) .
thus it is natural that our lfs include a higher portion of faint star clusters than those of @xcite .
we have carried out a survey of star clusters in m33 using the hst / wfpc2 images in the hst archive that were not covered in the previous studies .
we have found 104 stars clusters including 32 new clusters and 12 known clusters confirmed in this study from 24 fields of hst / wfpc2 images .
we combined 44 star clusters from this study , 154 star clusters from @xcite , 33 star clusters from @xcite , and 11 star clusters from @xcite , to build a catalog of 242 star clusters .
this is the most comprehensive catalog available to date which contains only the confirmed genuine star clusters in m33 .
we present the integrated photometry of these star clusters derived from the @xmath0 images taken with the cfh12k ccd camera at the cfht .
photometric properties of the star clusters are summarized as follows : * the integrated color - magnitude diagram of star clusters shows that there are two distinct populations in m33 : a large number of blue star clusters and a small number of red star clusters .
the cmd of the m33 star clusters is in general similar to that of the lmc star clusters .
* we divide the cluster sample into three groups according to their @xmath3 color : blue star clusters with @xmath115 , intermediate color star clusters with @xmath5 , and red star clusters with @xmath116 .
most of the new clusters found or confirmed in this study are blue clusters .
there are 29 old globular cluster candidates with @xmath1 and @xmath2 , ten of which are newly found in this study .
* most of the new clusters in m33 are located along the theoretical evolutionary path for @xmath7 , @xmath8 in @xmath3@xmath9 color - color diagram , while a few of them are in the redder side in the @xmath9 color . *
the red clusters are found relatively more in the outer region of m33 than the blue and intermediate color clusters , and that many of the blue stars are located in the hii regions .
* the luminosity function ( lf ) for the blue star clusters shows a peak at @xmath10 mag , while the intermediate color star cluster lf shows a peak at the fainter magnitude @xmath11 mag .
the lf for the red star clusters shows also a peak at @xmath12 , although the number of the clusters is small .
authors thank narae hwang for many fruitful discussions and comments during the work in this study .
this work was supported in part supported by the abrl(r14 - 2002 - 058 - 01000 - 0 ) .
this is based in part on the w. -k .
park s ph.d . thesis in seoul national university .
bedin , l. r. , piotto , g. , baume , g. , momany , y. , carraro , g. , anderson , j. , messineo , m. , & ortolani , s. 2005 , , 444 , 831 bica , e. , claria , j. j. , dottori , h. , santos , j. f. c. , jr . , & piatti , a. e. 1996 , , 102 , 57 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 chandar , r. , bianchi , l. , & ford , h. c. 1999a , , 122 , 431 chandar , r. , bianchi , l. , & ford , h. c. 1999b , , 517 , 668 chandar , r. , bianchi , l. , & ford , h. c. 2001 , , 366 , 498 christian , c. a. , & schommer , r. a. 1982 , , 49 , 405 christian , c. a. , & schommer , r. a. 1988 , , 95 , 704 corbelli , e. 2003 , , 342 , 199 harris , w. e. 1996 , , 112 , 1487 hiltner , w. a. 1960 , , 131 , 163 hodge , p. w. , balsley , j. , wyder , t. k. , & skelton , b. p. 1999
, , 111 , 685 hwang , n. , & lee , m. g. 2007 , , submitted kaluzny , j. , stanek , k. z. , krockenberger , m. , sasselov , d. d. , tonry , j. l. , & mateo , m. 1998 , , 115 , 1016 kim , m. , kim , e. , lee , m. g. , sarajedini , a. , & geisler , d. 2002 , , 123 , 244 kron , g. e. , & mayall , n. u. 1960 , , 65 , 581 lata , s. , pandey , a. k. , sagar , r. , & mohan , v. 2002 , , 388 , 158 lee , m. g. 2006 , bull . astr .
india , 34 , 99 ma , j. , zhou , x. , chen , j. -s . , wu , h. , jiang , z. -j .
, xue , s. -j .
, & zhu , j. 2002 , , 2 , 197 massey , p. , armandroff , t. e. , pyke , r. , patel , k. , & wilson , c. 1995 , , 110 , 2715 massey , p. , olsen , k. a. g. , hodge , p. w. , strong , s. b. , jacoby , g. h. , schlingman , w. , & smith , r. c. 2006 , , 131 , 2478 melnick , j. , & dodorico , s. 1978 , , 34 , 249 mochejska , b. j. , kaluzny , j. , krockenberger , m. , sasselov , d. d. , & stanek , k. z. 1998 , , 48 , 455 sarajedini , a. , geisler , d. , schommer , r. , & harding , p. 2000 , , 120
, 2437 sarajedini , a. , barker , m. k. , geisler , d. , harding , p. , & schommer , r. 2007 , , 133 , 290 sarajedini , a. , & mancone , c. l. 2007 , , 134 , 447 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 stanek , k. z. , kaluzny , j. , krockenberger , m. , sasselov , d. d. , tonry , j. l. , & mateo , m. 1998 , , 115 , 1894 zaritsky , d. , elston , r. , & hill , j. m. 1989 , , 97 , 97 cccccccc 1 & 6694 & @xmath130s & @xmath131s & @xmath132s & & 0(8 ) + 2 & 8805 & & & @xmath133s & & 2(6 ) + 3 & 8207 & @xmath134s & @xmath135s & @xmath136s & & 1(10 ) + 4 & 8207 & @xmath134s & @xmath135s & @xmath136s & & 3(7 ) + 5 & 8207 & @xmath134s & @xmath135s & @xmath136s & & 1(5 ) + 6 & 8805 & & & @xmath137s , @xmath138s & & 0(5 ) + 7 & 8018 & & & @xmath139s , @xmath130s & @xmath139s & 4(13 ) + 8 & 6431 & & @xmath140s & @xmath141s & @xmath142s , @xmath143s & 5(13 ) + 9 & 9127 & @xmath141s & @xmath135s & @xmath144s & & 3(0 ) + 10 & 8805 & & & @xmath133s & & 3(6 ) + 11 & 8805 & & & @xmath133s & & 6(7 ) + 12 & 7909 & & & @xmath145s & & 0(17 ) + 13 & 8207 & @xmath134s & @xmath135s & @xmath136s & & 0(0 ) + 14 & 9127 & @xmath141s & @xmath135s & @xmath144s & & 4(7 ) + 15 & 8207 & @xmath134s & @xmath135s & @xmath136s & & 2(0 ) + 16 & 5237 & @xmath146s & & @xmath147s & @xmath147s & 0(0 ) + 17 & 9127 & @xmath144s , @xmath141s & @xmath135s & @xmath144s & & 2(0 ) + 18 & 8207 & @xmath134s & @xmath135s & @xmath136s & & 1(1 ) + 19 & 8061 & @xmath148s & & @xmath149s & & 1(1 ) + 20 & 8207 & @xmath134s & @xmath135s & @xmath136s & & 0(0 ) + 21 & 9479 & & & @xmath150s & @xmath145s & 0(0 ) + 22 & 9127 & & & @xmath150s & @xmath145s & 0(0 ) + 23 & 9127 & & & @xmath150s & @xmath145s & 0(0 ) + 24 & 8207 & @xmath134s & @xmath135s & @xmath151 & & 0(0 ) + rccrrrc 1 & 1 32 36.36 & 30 36 48.61 & @xmath152 & @xmath153 & @xmath154 & @xmath155 + 2 & 1 33 11.74 & 30 38 56.14 & @xmath156 & @xmath157 & @xmath158 & @xmath159 + 3 & 1 33 12.67 & 30 38 55.35 & @xmath160 & @xmath161 & @xmath162 & @xmath163 + 4 & 1 33 14.97 & 30 39 04.70 & @xmath164 & @xmath165 & @xmath166 & @xmath167 + 5 & 1 33 19.51 & 30 35 30.19 & @xmath168 & @xmath169 & @xmath170 & @xmath159 + 6 & 1 33 20.25 & 30 34 50.27 & @xmath171 & @xmath172 & @xmath173 & @xmath159 + 7 & 1 33 21.35 & 30 31 13.02 & @xmath174 & @xmath175 & @xmath176 & @xmath159 + 8 & 1 33 22.36 & 30 35 30.20 & @xmath177 & @xmath178 & @xmath179 & @xmath159 + 9 & 1 33 23.15 & 30 52 42.41 & @xmath180 & @xmath181 & @xmath182 & @xmath167 + 10 & 1 33 26.58 & 30 31 30.82 & @xmath183 & @xmath184 & @xmath185 & @xmath159 + 11 & 1 33 27.38 & 30 41 48.39 & @xmath186 & @xmath187 & @xmath188 & @xmath155 + 12 & 1 33 30.06 & 30 31 48.66 & @xmath189 & @xmath190 & @xmath191 & @xmath159 + 13 & 1 33 30.23 & 30 41 18.89 & @xmath192 & @xmath193 & @xmath194 & @xmath155 + 14 & 1 33 30.36 & 30 37 43.41 & @xmath195 & @xmath196 & @xmath197 & @xmath167 + 15 & 1 33 30.38 & 30 36 41.45 & @xmath198 & @xmath199 & @xmath200 & @xmath159 + 16 & 1 33 30.41 & 30 36 00.04 & @xmath201 & @xmath202 & @xmath203 & @xmath159 + 17 & 1 33 31.32 & 30 37 34.10 & @xmath204 & @xmath205 & @xmath206 & @xmath167 + 18 & 1 33 32.16 & 30 30 18.01 & @xmath207 & @xmath208 & @xmath209 & @xmath159 + 19 & 1 33 32.46 & 30 35 38.30 & @xmath210 & & @xmath211 & @xmath167 + 20 & 1 33 36.54 & 30 50 35.79 & @xmath212 & @xmath213 & @xmath214 & @xmath167 + 21 & 1 33 38.19 & 30 43 24.51 & @xmath215 & @xmath216 & & @xmath167 + 22 & 1 33 41.44 & 30 51 10.51 & @xmath217 & @xmath218 & @xmath219 & @xmath220 + 23 & 1 33 58.09 & 30 37 29.68 & @xmath221 & @xmath222 & @xmath223 & @xmath155 + 24 & 1 33 58.49 & 30 38 20.96 & @xmath224 & @xmath225 & @xmath226 & @xmath155 + 25 & 1 33 58.67 & 30 35 26.10 & @xmath227 & @xmath228 & @xmath229 & @xmath159 + 26 & 1 34 14.44 & 30 34 17.56 & @xmath230 & @xmath231 & @xmath232 & @xmath159 + 27 & 1 34 14.96 & 30 33 53.66 & @xmath233 & @xmath234 & @xmath235 & @xmath159 + 28 & 1 34 15.21 & 30 33 15.11 & @xmath236 & @xmath237 & @xmath238 & @xmath159 + 29 & 1 34 15.74 & 30 33 41.02 & @xmath239 & @xmath240 & @xmath241 & @xmath159 + 30 & 1 34 18.08 & 30 31 58.93 & @xmath242 & @xmath243 & @xmath244 & @xmath159 + 31 & 1 34 42.79 & 30 37 20.21 & @xmath245 & @xmath246 & @xmath247 & @xmath159 + 32 & 1 34 44.60 & 30 37 32.32 & @xmath248 & @xmath249 & @xmath250 & @xmath251 + rccrrrcl 33 & 1 32 34.45 & 30 37 42.09 & @xmath252 & @xmath253 & @xmath254 & @xmath251 & cbf-143 + 34 & 1 33 22.12 & 30 40 26.15 & @xmath255 & @xmath256 & @xmath257 & @xmath159 & cbf-59 + 35 & 1 33 22.26 & 30 40 59.64 & @xmath258 & & @xmath259 & @xmath159 & cbf-60 + 36 & 1 33 22.35 & 30 30 14.67 & @xmath260 & @xmath261 & @xmath262 & @xmath159 & cs - h36 + 37 & 1 33 23.84 & 30 40 26.18 & @xmath263 & @xmath264 & @xmath265 & @xmath159 & cbf-96 + 38 & 1 33 24.81 & 30 33 55.25 & @xmath266 & @xmath267 & @xmath268 & @xmath159 & cbf-18 + 39 & 1 33 25.60 & 30 29 57.03 & @xmath269 & @xmath270 & @xmath271 & @xmath159 & cs - u126 , mkkss-5 + 40 & 1 33 25.96 & 30 36 24.58 & @xmath272 & @xmath273 & @xmath274 & @xmath159 & mkkss-6 + 41 & 1 33 26.34 & 30 41 06.96 & @xmath275 & @xmath276 & @xmath277 & @xmath155 & cbf-92 + 42 & 1 33 26.46 & 30 41 11.84 & @xmath278 & @xmath279 & @xmath280 & @xmath155 & cbf-93 + rccrrrcl 105 & 1 32 38.83 & 30 39 17.77 & @xmath281 & @xmath282 & & @xmath159 & cbf-162 + 106 & 1 32 44.28 & 30 40 12.25 & @xmath283 & @xmath284 & @xmath285 & @xmath159 & cbf-161 , cs - u88 + 107 & 1 33 01.16 & 30 35 44.65 & @xmath286 & @xmath287 & @xmath288 & @xmath251 & cbf-39 + 108 & 1 33 02.39 & 30 34 44.16 & @xmath289 & @xmath290 & @xmath291 & @xmath167 & cbf-40 + 109 & 1 33 08.10 & 30 28 00.12 & @xmath292 & @xmath293 & @xmath294 & @xmath167 & cbf-86 , cs - u140 + 110 & 1 33 10.08 & 30 29 56.51 & @xmath295 & @xmath296 & @xmath297 & @xmath167 & cbf-89 + 111 & 1 33 13.84 & 30 29 05.27 & @xmath298 & @xmath299 & @xmath300 & @xmath155 & cbf-53 + 112 & 1 33 13.88 & 30 29 44.87 & @xmath301 & @xmath302 & @xmath303 & @xmath167 & cbf-88 + 113 & 1 33 14.28 & 30 28 22.78 & @xmath304 & @xmath305 & @xmath306 & @xmath159 & cbf-54 , cs - u137 , md-8 + 114 & 1 33 15.15 & 30 32 53.26 & @xmath307 & @xmath308 & @xmath309 & @xmath251 & cbf-144 + rrrrcl 28 & @xmath236 & @xmath237 & @xmath238 & @xmath159 & + 39 & @xmath269 & @xmath270 & @xmath271 & @xmath159 & cs - u126 , mkkss-5 + 45 & @xmath310 & @xmath311 & @xmath312 & @xmath155 & cbf-90 , cs - u77 , md-17 + 49 & @xmath313 & @xmath314 & @xmath315 & @xmath155 & cbf-90 + 58 & @xmath316 & @xmath317 & @xmath318 & @xmath319 & sbghs-10 + 76 & @xmath320 & @xmath321 & @xmath322 & @xmath159 & cbf-49 , mkkss-25 + 84 & @xmath323 & @xmath324 & @xmath325 & @xmath167 & cbf-29 + 93 & @xmath326 & @xmath327 & @xmath328 & @xmath159 & cbf-118 + 94 & @xmath329 & @xmath330 & @xmath331 & @xmath159 & cbf-119 + 106 & @xmath283 & @xmath284 & @xmath285 & @xmath159 & cbf-161 , cs - u88 + 111 & @xmath298 & @xmath299 & @xmath300 & @xmath155 & cbf-53 + 113 & @xmath304 & @xmath305 & @xmath306 & @xmath159 & cbf-54 , cs - u137 , md-8 + 114 & @xmath307 & @xmath308 & @xmath309 & @xmath251 & cbf-144 + 130 & @xmath332 & @xmath333 & @xmath334 & @xmath167 & cbf-108 , sbghs-6 + 131 & @xmath335 & @xmath336 & @xmath337 & @xmath155 & cbf-106 , sbghs-4 + 146 & @xmath338 & @xmath339 & @xmath340 & @xmath159 & cbf-61 , cs - u49 , md-24 , mkkss-21 + 149 & @xmath341 & @xmath342 & @xmath343 & @xmath159 & cbf-62 + 157 & @xmath344 & @xmath345 & @xmath346 & @xmath159 & cbf-138 + 163 & @xmath347 & @xmath348 & @xmath349 & @xmath159 & cbf-104 , cs - h38 , md-25 , mkkss-22 + 189 & @xmath350 & @xmath351 & @xmath352 & @xmath167 & cbf-28 + 191 & @xmath353 & @xmath354 & @xmath355 & @xmath159 & cbf-98 , mkkss-31 + 196 & @xmath356 & @xmath357 & @xmath358 & @xmath167 & b-5 + 206 & @xmath359 & @xmath360 & @xmath361 & @xmath159 & cbf-116 , mkkss-36 + 214 & @xmath362 & @xmath363 & @xmath364 & @xmath159 & cbf-11 + 228 & @xmath365 & @xmath366 & @xmath367 & @xmath159 & cbf-70 , cs - c03 , md-47 , sbghs-13 + 234 & @xmath368 & @xmath369 & @xmath370 & @xmath159 & cbf-75 + 236 & @xmath371 & @xmath372 & @xmath373 & @xmath159 & cbf-74 , cs - h24 + 240 & @xmath374 & @xmath375 & @xmath376 & @xmath167 & cbf-82 + 241 & @xmath377 & @xmath378 & @xmath379 & @xmath167 & cbf-81 , cs - c20 , md-55 + | we present the result of a survey for star clusters in m33 using the hst / wfpc2 archive images .
we have found 104 star clusters , including 32 new ones , in the images of 24 fields that were not included in the previous studies . combining these with previous data in the literature
, we increase the number of m33 star clusters found in the hst images to 242 .
we have derived @xmath0 integrated photometry of these star clusters from the ccd images taken with cfh12k mosaic camera at cfht .
integrated color - magnitude diagrams of the m33 star clusters are found to be similar in general to those of star clusters in the large magellanic cloud , except that m33 has a much lower fraction of blue star clusters .
we find 29 red star clusters with @xmath1 and @xmath2 , which are old globular cluster candidates .
we divide the cluster sample into three groups according to their @xmath3 color : blue star clusters with @xmath4 , intermediate color star clusters with @xmath5 , and red star clusters with @xmath6 .
most of the new clusters in m33 are located along the sequence that is consistent with the theoretical evolutionary path for @xmath7 , @xmath8 in the @xmath3@xmath9 diagram , while a few of them are in the redder side in the @xmath9 color .
the red clusters are found relatively more in the outer region of m33 than the blue and intermediate color clusters , and that many of the blue stars are located in the hii regions .
the luminosity function for the blue star clusters shows a peak at @xmath10 mag , while that for the intermediate color star cluster shows a peak at the fainter magnitude @xmath11 mag .
the luminosity function for the red star clusters shows also a peak at @xmath12 mag , although the number of the clusters is small . |
in 1965 , birch , chowla , hall , and schinzel @xcite asked a question which soon became famous:[bchs ] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ let @xmath7 and @xmath8 be two coprime polynomials with complex coefficients ; what is the possible minimum degree of the difference @xmath9 ? _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ it is reasonable to suppose that @xmath10 and @xmath11 have the same degree and the same leading coefficients .
let us take @xmath12 , @xmath13 , so that @xmath14 .
then the following was conjectured in @xcite:[init - problem ] 1
. for @xmath9 one always has @xmath15 .
this bound is sharp : that is , it is attained for infinitely many values of @xmath16 .
the first conjecture was proved the same year by davenport @xcite .
the second one turned out to be much more difficult and remained open for 16 years : in 1981 stothers @xcite showed that the bound is in fact attained not only for infinitely many values of @xmath16 but for all of them .
the above problem may be generalized in various ways .
the following one was considered in 1995 by zannier @xcite .
let @xmath17 be two partitions of @xmath18 , @xmath19 and let @xmath2 and @xmath3 be two coprime polynomials of degree @xmath18 having the following factorization pattern : @xmath20 in these expressions we consider the multiplicities @xmath21 and @xmath22 , @xmath23 , @xmath24 as being given , while the roots @xmath25 and @xmath26 are not fixed , though they must all be distinct . the problem is to find the minimum possible degree of the difference @xmath27 . in his paper , zannier proved the following .
let @xmath28 denote the greatest common divisor of the numbers @xmath29 . if @xmath30 then @xmath31 and this bound is always attained . if , on the other hand , @xmath32 , then a weaker bound @xmath33 is valid , and it is also attained .
[ def : dz - triple ] let @xmath34 $ ] be coprime polynomials with factorization pattern ( [ eq : p - and - q ] ) , @xmath35 , while the degree of the polynomial @xmath36 equals @xmath37 .
then the triple @xmath38 is called a _ davenport
zannier triple _ , or , in a more concise way , a _
dz - triple_. the main subject of this paper is a study of dz - triples _ defined over _ @xmath5 , that is , the triples @xmath39 $ ] .
the paper is organized as follows .
a preliminary work is carried out in section [ sec : dessins ] .
first , we show that bound ( [ eq : main - bound ] ) follows from the riemann
hurwitz formula for the function @xmath40 .
then we reduce the problem about polynomials to a problem about _ weighted bicolored plane trees_. a weighted bicolored plane tree is a _ plane tree _
( `` plane '' means that the cyclic order of branches around each vertex is fixed ) whose vertices are colored in black and white in such a way that the ends of each edge have opposite colors , and whose _ edges are endowed with positive integral weights_. the _ degree of a vertex _ is defined as the sum of the weights of the edges incident to this vertex .
the sum of the weights of all the edges is called the _ total weight _ of the tree .
we show that a dz - triple with prescribed factorization pattern ( [ eq : p - and - q ] ) exists if and only if there exists a weighted bicolored plane tree of the total weight @xmath41 having @xmath42 black vertices of degrees @xmath43 and @xmath44 white vertices of degrees @xmath45 . as a corollary ,
we give a spectacularly simple proof of stothers s 1981 result for the squares and cubes , namely , the attainability of the lower bound @xmath46 where @xmath47 and @xmath48 , see example [ ex : stothers ] .
the results of section [ sec : dessins ] as well as the framework of the whole paper are based on the theory of _ dessins denfants _ ( see , e.g. , chapter 2 of @xcite , or a collection of papers @xcite , or a recent book @xcite ) .
this theory establishes a correspondence between simple combinatorial objects , graphs drawn on two - dimensional surfaces , and a vast world of riemann surfaces , algebraic curves , number fields , galois theory , etc . in section [ sec :
existence ] we prove the existence theorem for weighted bicolored plane trees .
namely , we show that a necessary and sufficient condition for the existence of a tree with the above characteristics is inequality ( [ qwert ] ) .
the attainability of bound is deduced from this result . in section 4
we establish bound ( [ no - qwert ] ) and its attainability in the case when inequality is not satisfied .
although bounds and ( [ no - qwert ] ) and their attainability were proved by u.zannier , we reprove these results here for the sake of completeness , and also in order to show how the pictorial language clarifies and simplifies the exposition .
in sections [ sec : unitrees ] and [ sec : galois ] we study dz - triples defined over @xmath5 .
this case is the most interesting one since by specializing @xmath49 to a rational value one may obtain an important information concerning differences of integers with given factorization patterns .
this subject is actively studied in number - theoretic works : see , for example , a recent paper by beukers and stewart @xcite ( 2010 ) and the bibliography therein .
our approach here is based on the following corollary of the theory of dessins denfants which gives a sufficient ( though not necessary ) condition for a dz - triple to be defined over @xmath5 : the triple is defined over @xmath5 if there exists _ exactly one _ weighted bicolored plane tree with the degrees of black vertices equal to @xmath43 , and the degrees of white vertices equal to @xmath45 .
we will call such trees _
unitrees_. in section [ sec : unitrees ] we prove the main result of the paper , namely , a complete classification of unitrees : see theorem [ th : main ] .
the formulation of this theorem is rather long so we do not enunciate it in the introduction .
we mention only that the class of unitrees consists of ten infinite series of trees and ten sporadic trees which do not belong to the above series . while the results of section [ sec : unitrees ] may be considered as conclusive , section [ sec : galois ] represents only first steps in a far - ranging programme of study of the galois action on weighted plane trees and of combinatorial invariants of this action .
this approach permits to find yet more dz - triples defined over @xmath5 and also to study dz - triples over other number fields .
note that essentially all previously found examples of dz - triples over @xmath5 correspond either to unitrees or to the trees constructed in section [ sec : galois ] .
we have also found quite a few new examples of dz - triples .
finally , in section [ sec : further ] we mention some further possible developments of the subject .
this paper deals only with the combinatorial aspect of the whole construction .
the computation of the corresponding dz - triples is postponed to a separate publication ( see @xcite ) since the techniques used for this purpose are very different form the ones used in this paper .
in particular , a great deal of symbolic computations as well as certain polynomial identities are required . for an individual unitree
, the computation of the corresponding dz - triple is a difficult task but the verification of the result is easy . indeed ,
when polynomials @xmath50 ( together with their appropriate factorizations ) are given , it is immediate to observe that their coefficients are rational , and the only thing to verify is that @xmath51 is indeed equal to @xmath4 . the situation becomes significantly more complicated for infinite series of trees since in this case the proof may become rather elaborate .
let @xmath17 be two partitions of @xmath18 , @xmath52 , @xmath53 , @xmath54 , and let @xmath55 $ ] be two polynomials of degree @xmath18 with the factorizations @xmath56 we suppose all @xmath57 , @xmath58 , @xmath59 to be distinct .
let the difference @xmath36 have the following factorization : @xmath60 our goal is to minimize @xmath61 ; obviously , @xmath62 consider the following rational function of degree @xmath18 : @xmath63 note that @xmath64 [ def : crit - value ] a point @xmath65 is called _ critical value _ of a rational function @xmath66 if the equation @xmath67 has multiple roots .
the expressions written above for the function @xmath68 provide us with at least three critical values of @xmath66 : * @xmath69 , provided that not all @xmath21 are equal to 1 ; * @xmath70 , provided that not all @xmath22 are equal to 1 ; and * @xmath71 , if only we do not consider the trivial case @xmath72 ; if @xmath73 then @xmath66 has a multiple pole at infinity . denote @xmath74 the other critical values of @xmath66 , if there are any , and let @xmath75 be the number of preimages of @xmath76 , @xmath77 ; by the definition of a critical value , @xmath78 .
[ lem : r ] the number @xmath79 of distinct roots of the polynomial @xmath51 is @xmath80 in fact , equality ( [ eq : r ] ) is a particular case of the riemann
hurwitz formula , but for the sake of completeness we give its proof here .
[ [ proof . ] ] proof .
+ + + + + + let us draw a _ star - tree _ with the center at 0 and with its rays going to the critical values @xmath81 , see figure [ fig : star ] .
considered as a map on the sphere , this tree has @xmath82 vertices , @xmath83 edges , and a single outer face with its `` center '' at @xmath84 .
now let us take the preimage of this tree under @xmath66 .
we will get a graph drawn on the preimage sphere which has @xmath85 edges since each edge is `` repeated '' @xmath18 times in the preimage .
its vertices are the preimages of the points @xmath86 , so their number is equal to @xmath87 . what occurs to the faces ?
if we puncture at @xmath84 the single open face in the image sphere , we get a punctured disk without any ramification points inside .
the only possible unramified covering of a punctured open disk is a disjoint collection of punctured disks ; their number is equal to the number of poles of @xmath66 , namely , @xmath88 ( @xmath79 roots of @xmath51 and @xmath84 ) . inserting a point into each puncture
we get @xmath88 simply connected open faces in the preimage sphere . the fact that they are simply connected implies that the graph drawn on the preimage sphere is connected .
thus , the preimage of our star - tree is a plane map .
what remains is to apply euler s formula : @xmath89 which leads to .
@xmath90 notice that in order to prove lemma [ lem : r ] , instead of the tree of figure [ fig : star ] , we could take any other plane map with vertices at the critical values ( see e.g. the proof of proposition [ prop : thom ] below ) .
we have @xmath91 the proof follows from ( [ eq : r ] ) and ( [ eq : r > r ] ) .
@xmath90 note that @xmath61 can not be negative ; therefore , when @xmath92 the latter bound can not be attained . in this case
one can attain the bound @xmath93 , that is , the polynomial @xmath51 can be made equal to a constant .
this situation is studied in more detail in section [ sec : weak - bound ] .
equation provides us with guidelines of how to get the minimum degree of @xmath51 .
[ prop : lower - bound ] bound is attained if and only if the following conditions are satisfied : * @xmath94 . *
the number @xmath95 of the critical values of @xmath66 other than @xmath96 , is equal to zero , so that the sum @xmath97 in the right - hand side of is eliminated altogether . the tree of figure [ fig : star ]
is then reduced to merely the segment @xmath98 $ ] . *
all the roots of @xmath51 are simple , that is , @xmath99 , so that @xmath100 .
another formulation of the same condition : the partition @xmath101 , @xmath102 , which corresponds to the multiplicities of the poles , has the form of a hook : @xmath103 . the conditions which imply the existence of such a function @xmath66 will be obtained in section [ sec : existence ] .
considering rational functions with only three critical values brings us into the framework of the theory of _
dessins denfants_. here we give a brief summary of this theory ( only in a planar setting ) ; the missing details , proofs , and bibliography can be found , for example , in @xcite , chapter 2 .
a rational function @xmath104 is called _ belyi function _ if it does not have critical values outside the set @xmath105 .
for such a function , the tree considered in the proof of lemma [ lem : r ] is reduced to the segment @xmath98 $ ] .
let us take this segment , color the point 0 in black and the point 1 in white , and consider the preimage @xmath106)$ ] ; we will call this preimage a _
dessin_. the dessin @xmath106)$ ] is a connected graph drawn on the sphere , and its edges do not intersect outside the vertices .
therefore , @xmath107 may also be considered as a plane map .
this map has a bipartite structure : black vertices are preimages of @xmath108 , and white vertices are preimages of @xmath109 .
the degrees of the black vertices are equal to the multiplicities of the roots of the equation @xmath110 , and the degrees of the white ones are equal to the multiplicities of the roots of the equation @xmath111 .
the sum of the degrees in both cases is equal to @xmath112 , which is also the number of edges .
the map @xmath107 being bipartite , the number of edges surrounding each face is even .
it is convenient , in defining the face degrees , to divide this number by two .
[ def : face - degree ] we say that _ an edge is incident to a face _ if , while remaining inside this face and making a circuit of it in the positive ( trigonometric ) direction , we follow the edge from its black end toward the white one .
thus , only half of the edges surrounding a face are incident to it .
moreover , each edge is incident to exactly one face .
the _ degree of a face _ is equal to the number of edges incident to it . according to this definition and to the remarks preceding it , every edge is incident to one black vertex , to one white vertex , and to one face .
the sum of the face degrees is equal to @xmath112 .
inside each face there is a single pole of @xmath66 , and the multiplicity of this pole is equal to the degree of the face .
the triple @xmath113 of partitions @xmath114 which correspond to the degrees of the black vertices , of the white vertices , and of the faces of a dessin , is called a _
passport _ of the dessin .
[ def : comb - orbit ] a set of the dessins having the same passport is called a _ combinatorial orbit _ corresponding to this passport .
the construction which associates a map to a belyi function works also in the opposite direction .
two bicolored plane maps are _ isomorphic _ if there exists an orientation preserving homeomorphism of the sphere which transforms one map into the other , respecting the colors of the vertices .
let @xmath115 be a bicolored map on the sphere .
then , the sphere may be endowed with a complex structure , thus becoming the riemann complex sphere , and a representative of the isomorphism class of @xmath115 can be drawn as a dessin @xmath107 obtained via a belyi function .
the following statement is a particular case of the classical riemann s existence theorem : [ prop : belyi - exist ] for every bicolored plane map @xmath115 there exists a dessin @xmath107 isomorphic to @xmath115 , that is , @xmath106)$ ] where @xmath66 is a belyi function .
the function @xmath116 is unique up to a linear fractional transformation of the variable @xmath49 . of course , when we draw a map we do not respect the specific geometric form of the corresponding dessin .
we are content with the fact that such a dessin exists .
now proposition [ prop : lower - bound ] may be reformulated in purely combinatorial terms : [ prop : attain ] the lower bound is attained if and only if there exists a bicolored plane map with the passport @xmath113 in which the partitions @xmath52 and @xmath53 are given , and @xmath117 has the form @xmath118 where @xmath109 is repeated @xmath119 times . in geometric terms , all the faces of our map except the outer one must be of degree 1 . recall that the _ number _ of faces , which is equal to @xmath88 , is prescribed by euler s formula .
[ ex : stothers ] let us look once again at the problem posed by birch et al . in @xcite ( see page ) . in order to show that if @xmath47 , @xmath48 , and @xmath9 , then the lower bound @xmath120 is attained , we must construct a map with the following properties : all its black vertices are of degree 3 ; all its white vertices are of degree 2 ; and all its finite faces are of degree 1 . in order to simplify our pictures we sometimes use the following convention . [
con : without - white ] when all the white vertices are of degree 2 , it is convenient , in order to simplify a graphical representation of such maps , to draw only black vertices and to omit the white ones , considering them as being implicit .
in such a picture , a line connecting two black vertices contains an invisible white vertex in its middle , and is thus not an edge but a union of two edges .
the construction of the maps we need to solve the above problem about @xmath121 is trivial : first we draw a tree with all internal nodes being of degree 3 , and then attach loops to its leaves : see figure [ fig : stothers ] .
we see in this example a remarkable efficiency of the pictorial representation of problems concerning polynomials .
if this representation was known in 1965 , the proof of the conjecture would have taken 16 minutes instead of 16 years .
as it was told in proposition [ prop : belyi - exist ] , a belyi function @xmath122 corresponding to a dessin is defined up to a linear fractional transformation of @xmath49 . in this family of equivalent belyi functions
it is always possible to find one whose coefficients are algebraic numbers .
if we act simultaneously on all the coefficients of such a function by an element of the _ absolute galois group _ @xmath123 , that is , by an automorphism @xmath124 of the field @xmath125 of algebraic numbers , or , in other words , if we replace all the coefficients @xmath25 of @xmath66 by their algebraically conjugate numbers @xmath126 , we obtain once again a belyi function .
furthermore , one can prove that in such a way the action of @xmath123 on belyi functions descends to an action on dessins .
there exist many combinatorial invariants of this action , the first and the simplest of them being the passport of the dessin .
thus , a combinatorial orbit ( see definition [ def : comb - orbit ] ) may constitute a single galois orbit , or may further split into a union of several galois orbits .
every combinatorial orbit is finite , and therefore every galois orbit is also finite .
one of the most important notions concerning the galois action on dessins is that of the field of moduli .
let @xmath107 be a dessin , and let @xmath127 be its stabilizer .
since the orbit @xmath107 is finite , the group @xmath128 is a subgroup of finite index in @xmath123 .
let @xmath129 be the maximal normal subgroup of @xmath123 contained in @xmath128 . according to the galois correspondence between subgroups of @xmath123 and algebraic extensions of @xmath5
, there exists a number field @xmath130 corresponding to @xmath131 .
this field is called the _ field of moduli _ of the dessin @xmath107 . by construction ,
this field is unique : a dessin can not have two different fields of moduli . below we list some properties of the fields of moduli .
let @xmath132 be an orbit of the galois action on dessins .
* the field of moduli @xmath130 is the same for all the elements of the orbit . *
the degree of @xmath130 as an extension of @xmath5 is equal to @xmath133 . * the coefficients of belyi functions corresponding to the dessins
@xmath134 , if they are chosen as algebraic numbers , always belong to a finite extension @xmath135 of @xmath130 . *
the action of the group @xmath123 on the orbit @xmath136 coincides with the action of @xmath137 . *
the action of @xmath138 on belyi functions may change a position of a dessin @xmath134 on the complex sphere but does not change its combinatorial structure ; in other words , as a map , the dessin in question remains the same . in the absolute majority of cases
the situation is much simpler : the field of moduli of an orbit is the smallest number field to which the coefficients of the corresponding belyi functions belong .
however , in some specially constructed examples we need a larger field @xmath139 to be able to find belyi functions . there exists a simple _ sufficient condition _ which ensures that the coefficients do belong to @xmath130 , see @xcite : this condition is the existence of a _ bachelor_. [ def : bachelor ] a _ bachelor _ is a black vertex ( a white vertex ; a face ) such that there is no other black vertex ( no other white vertex ; no other face ) of the same degree .
[ rem : bachelor ] if a dessin contains several bachelors then up to three of them can be placed at rational points , that is , at points in @xmath140 , and this will not prevent the belyi function for the dessin in question to be defined over the field of moduli . for the dessins we study in this paper
a bachelor always exists : it is the outer face ( since all the other faces are of degree 1 ) .
recalling that the degree of @xmath130 is equal to the size of the orbit we may conclude the following : summarizing what was stated above we may affirm the following : [ prop : in - q ] if for a given passport @xmath113 , where the partition @xmath117 is of the form @xmath118 , there exists a unique bicolored plane map , then there exists a corresponding belyi function with rational coefficients , and therefore there also exists a dz - triple with rational coefficients . note that proposition [ prop : attain ] , which concerns the existence , is of the `` if and only if '' type , while proposition [ prop : in - q ] , which concerns the definability over @xmath5 , provides only an `` if''-type condition .
though the weighted trees are , in our opinion , natural and interesting objects to be studied for their own sake , in our paper they are used as a merely technical tool which is easy to manipulate . in figure
[ fig : map->tree ] , left , it is shown how a typical bicolored map whose all finite faces are of degree 1 looks like .
( recall that according to definition [ def : face - degree ] a face of degree 1 is surrounded by two edges , but only one of these edges is incident to the face . )
it is convenient to symbolically represent such a map in a form of a tree ( see figure [ fig : map->tree ] , right ) by replacing several multiple edges which connect neighboring vertices , by a single edge with a weight equal to the number of these multiple edges . in this way ,
the operations of cutting and gluing subtrees , exchanging the weights between edges , etc .
, become easier to implement and to understand .
[ def : weighted ] a _ weighted bicolored plane tree _ , or a _ weighted tree _ , or just a _ tree _ for short , is a bicolored plane tree whose edges are endowed with positive integral _
weights_. the sum of the weights of the edges of a tree is called the _ total weight _ of the tree .
the _ degree _ of a vertex is the sum of the weights of the edges incident to this vertex .
the _ weight distribution _ of a weighted tree is a partition @xmath141 , @xmath142 where @xmath143 is the number of edges , and @xmath144 , @xmath145 are the weights of the edges . leaving aside the weights and considering only the underlying plane tree , we speak of a _ topological tree_. weighted trees whose weight distribution is @xmath146 will be called _
ordinary trees_. ordinary trees correspond to _ shabat polynomials _ : these are particular cases of belyi functions , with a single pole at infinity .
we call a _ leaf _ a vertex which has only one edge incident to it , whatever is the weight of this edge . by abuse of language
, we will also call a leaf this edge itself .
the adjective _ plane _ in this definition means that our trees are considered not as mere graphs but as plane maps .
more precisely , this means that the cyclic order of branches around each vertex of the tree is fixed , and changing this order will in general give a different tree .
_ all the trees considered in this paper will be endowed with the `` plane '' structure _ ; therefore , the adjective `` plane '' will often be omitted .
the adjective _ plane _ in this definition means that our trees are considered not as mere graphs but as plane maps .
more precisely , this means that the cyclic order of branches around each vertex of the tree is fixed , and changing this order will in general give a different tree . _
all the trees considered in this paper will be endowed with the `` plane '' structure _ ; therefore , the adjective `` plane '' will often be omitted .
[ def : iso ] two weighted trees are _
isomorphic _ if the underlying bicolored plane maps are isomorphic .
in other words , they are isomorphic if there exists a color - preserving bijection between the vertices of the trees and a bijection between the edges which respects the incidence of edges and vertices , the cyclic order of the edges around each vertex , and which also respects the weights of the edges .
[ def : pass - tree ] the pair @xmath147 of partitions
@xmath17 of the total weight @xmath18 of a tree , corresponding to the degrees of the black vertices and of the white vertices of a weighted tree , is called a _ passport _ of this tree .
the total weight of the tree shown in figure [ fig : map->tree ] is @xmath148 ; its passport is @xmath149 ; the face degree distribution is @xmath150 , and the weight distribution is @xmath151 .
we must not confuse weighted trees with `` weighted maps '' .
the weighted tree on the left , and the `` weighted map '' on the right of figure [ fig : tree - vs - graph ] have the same set of black and white vertex degrees : @xmath152 , but the face degree partitions of the underlying maps are different : @xmath153 for the map represented by the tree , and @xmath154 for the map on the right .
in particular , the corresponding dessins can not belong to the same galois orbit . through the whole paper
, we speak exclusively about weighted trees .
now propositions [ prop : lower - bound ] and [ prop : attain ] may be reformulated as follows : [ th : attain ] let @xmath17 be two partitions of @xmath18 having @xmath42 and @xmath44 parts , respectively . then the lower bound is attained if and only if there exists a weighted tree with the passport @xmath147 .
in this section we study the following question : for a given pair of partitions @xmath17 , does there exist a weighted tree of the total weight @xmath18 with the passport @xmath147 ?
equivalently , does there exist a rational function with three critical values , and with the multiplicities of the preimages of these critical values being , first , two given partitions @xmath17 , and then , the third partition being equal to @xmath118 ? this question is a particular case of a more general problem of _ realizability _ of ramified coverings : does there exist a ramified covering of a given riemann surface with the given `` local data '' ( that is , with given multiplicities of the preimages of ramification points ) ?
the problem goes back to the classical paper by hurwitz @xcite ( 1891 ) .
though many particular cases are well studied , the problem in its full generality remains unsolved . among numerous publications dedicated to the realizability
we would like to mention early works by husemoller @xcite ( 1962 ) and thom @xcite ( 1965 ) ; an important paper by edmonds , kulkarni , and stong @xcite ( 1984 ) ; and recent publications @xcite ( 2009 ) , @xcite ( 2008 ) , and @xcite ( 2009 ) . the main result of this section is the following theorem ( recall that @xmath155 denotes the greatest common divisor of the numbers @xmath29 ) : [ th : realize ] let @xmath17 be two partitions of @xmath18 , @xmath52 , @xmath53 , and
let @xmath156 .
then a weighted tree with the passport @xmath147 exists if and only if @xmath157 by theorem [ th : attain ] the attainability of the bound ( coinciding with ) follows from this statement .
the attainabilty of the bound in the case when condition ( [ eq : realize ] ) is not satisfied will be established in section [ sec : weak - bound ] .
theorem [ th : realize ] and theorem [ th : weak ] below are equivalent to the main result ( theorem 1 ) of zannier @xcite . in his paper ,
zannier remarks that it would be interesting to apply the theory of dessins denfants to this problem in a more direct way , and mentions a remark by g.jones that such an approach might produce a simpler proof .
this is indeed the case , as we will see in this section . beside that
, this theory enables us to find a huge class of dz - triples over @xmath5 ( in a way , `` almost all '' of them ) , see section [ sec : unitrees ] ; and it also gives us a more direct access to galois theory , see section [ sec : galois ] .
we have already had a first glimpse of the power of the `` dessin method '' in example [ ex : stothers ] .
a _ forest _ is a disjoint union of trees .
[ prop : forest - exist ] any pair @xmath147 of partitions of @xmath18 can be realized as a passport of a forest of weighted trees .
[ [ proof.-1 ] ] proof .
+ + + + + + if there are two equal parts @xmath158 in the partitions @xmath159 and @xmath160 , we make a separate edge with the weight @xmath161 and proceed with the new passport @xmath162 , where @xmath163 and @xmath164 are obtained from @xmath159 and @xmath160 by eliminating their parts @xmath21 and @xmath22 , respectively .
if there are no equal parts , suppose , without loss of generality , that there are two parts @xmath165 . then we do the following ( see figure [ fig : forest ] ) : * make an edge with the weight @xmath166 ; * consider the new passport @xmath162 where @xmath164 is obtained from @xmath160 by eliminating the part @xmath22 , and @xmath163 is obtained from @xmath159 by replacing @xmath21 with @xmath167 ; * construct inductively a forest @xmath168 of the total weight @xmath169 corresponding to the passport @xmath162 ; by definition , this forest must have a black vertex of degree @xmath170 ; * glue the edge of weight @xmath171 to the forest @xmath168 by fusing two vertices , as is shown in figure [ fig : forest ] , and get a forest @xmath172 corresponding to @xmath147 ( since @xmath173 ) .
the proposition is proved .
[ th : existence ] suppose that @xmath175 . then the passport @xmath147 can be realized by a weighted tree if and only if @xmath94 . [ [ proof.-2 ] ] proof
. + + + + + + according to proposition [ prop : forest - exist ] we may suppose that we already have a forest corresponding to the passport @xmath147 .
now suppose that there are two edges of weights @xmath171 and @xmath176 , @xmath177 , which belong to different trees .
then we may stitch them together by the operation shown in figure [ fig : gluing - edges ] .
the degrees of the vertices in the new , connected figure are the same as in the old , disconnected one .
figure [ fig : gluing - trees ] shows that the operation works in the same way when there are subtrees attached to the ends of the adjoined edges .
we repeat this stitching operation until it becomes impossible to continue .
the latter may happen in two ways .
either we have got a connected tree , and then we are done . or there are no more edges with different weights while the forest remains disconnected . then , taking into account that @xmath175 , we conclude that the weights of all edges are equal to 1 , that is , we have got a forest consisting of @xmath178 ordinary trees . in this case , the number of vertices @xmath179 equals @xmath180 and therefore is strictly greater than @xmath181 , which contradicts the condition of the theorem .
@xmath90 note that the side edges in figures [ fig : gluing - edges ] and [ fig : gluing - trees ] have the same weight @xmath171 .
we will use this property while performing the operation inverse to stitching , namely , the ripping of a connected tree in two , in the proof of proposition [ prop : chain - sts ] ( see figure [ fig : unglue ] ) .
now suppose that @xmath182 .
[ lem : d>1 ] the degrees of all vertices of a forest are divisible by @xmath183 if and only if the weights of all edges are also divisible by @xmath184 . [ [ proof.-3 ] ] proof .
+ + + + + + in one direction this is evident : the degrees of the vertices are sums of weights , and therefore , if all the weights are multiples of @xmath184 , then the same is true for the degrees . in the opposite direction , if all the vertex degrees are divisible by @xmath184 , then it is true , in particular , for the leaves .
cut any leaf off the tree to which it belongs , and the statement is reduced to the same one for a smaller forest .
@xmath90 thus , dividing by @xmath184 all the vertex degrees , that is , all the elements of the partitions @xmath159 and @xmath160 , we return to the situation of theorem [ th : existence ] , with the same numbers @xmath42 and @xmath44 , and with the total weight equal to @xmath185 .
this finishes the proof of theorem [ th : realize ] .
@xmath90 we hope the reader will appreciate the simplicity of the above proof : number theorists have been approaching this result for 30 years ( 19651995 ) .
once again , the credit goes to the pictorial representation of polynomials with the desired properties .
when condition ( [ eq : realize ] ) of theorem [ th : realize ] is satisfied then , according to theorem [ th : attain ] , the main bound ( [ eq : main - bound ] ) is attained . if this condition is _ not satisfied _ , then the following holds : [ th : weak ] let @xmath156 , and let @xmath32 .
then @xmath186 and this bound is attained .
we will need the following proposition which was proved in 1965 by thom @xcite , and then reproved in many other publications .
for the reader s convenience we provide a short proof based on
dessins denfants " theory following @xcite , corollary 1.6.9 .
[ prop : thom ] let @xmath187 be a set of @xmath188 partitions @xmath189 of number @xmath18
. denote by @xmath190 the number of parts of @xmath191 , @xmath192 .
let @xmath193 be arbitrary complex numbers .
then a necessary and sufficient condition for the existence of a polynomial @xmath194 $ ] of degree n , whose all finite critical values are contained in the set @xmath195 , with the multiplicities of the roots of the equations @xmath196 corresponding to the partitions @xmath191 ,
@xmath192 , is the following equality : @xmath197 [ [ proof.-4 ] ] proof .
+ + + + + + for purely aesthetic reasons , instead of taking a tree with the vertices @xmath198 , as we did in the proof of lemma [ lem : r ] , let us take a jordan curve @xmath199 on the @xmath200-plane passing through the points @xmath198 , and let @xmath201 be its preimage @xmath202 .
then @xmath201 is a tree - like map often called _ cactus _ : it does not contain any cycles except @xmath18 `` copies '' of @xmath199 glued together at the vertices which are preimages of @xmath203 ; the number of copies of @xmath199 glued together at a vertex is equal to the multiplicity of the corresponding root of the equation @xmath196 ; see figure [ fig : cactus ] .
equation ( [ eq : thom ] ) may then be interpreted as euler s formula for the cactus since the cactus has @xmath204 vertices , @xmath205 edges , and @xmath181 faces ( @xmath18 copies of @xmath199 and the outer face ) .
we leave it to the reader to verify that another proof of the necessity of formulas can be deduced from the fact that the sum @xmath206 in the second equality in ( [ eq : thom ] ) represents the degree of the derivative @xmath207 .
these observations prove that conditions ( [ eq : thom ] ) are necessary .
notice that the partition @xmath208 may be eliminated from @xmath209 ( if it belongs to it ) , and may also be added to it , and this does not invalidate equalities ( [ eq : thom ] ) .
the proof that ( [ eq : thom ] ) is also sufficient is divided into two parts .
the first part is purely combinatorial and consists in constructing a cactus ( at least one ) with the vertex degrees corresponding to @xmath209 .
the second part is just a reference to riemann s existence theorem which relates combinatorial data to the complex structure , as it was already the case in proposition [ prop : belyi - exist ] .
the proof of the _ existence of a cactus _ in question is similar to that of proposition [ prop : forest - exist ] ; namely , it consists in `` cutting a leaf '' . here
a _ leaf _ means a copy of @xmath199 which is attached to @xmath201 at a single vertex ( see figure [ fig : cut - leaf ] ) .
this cutting operation must be carried out not with the cactus itself ( since it is not yet constructed ) but with its passport @xmath209 : it is easy to verify that implies that all partitions @xmath210 except maybe one contain a part equal to 1 .
we eliminate these parts , and diminish by 1 a part in the remaining partition . in this way
we obtain a valid passport @xmath211 of degree @xmath212 ; then we construct inductively a smaller cactus ; and then glue to it an @xmath18th copy of @xmath199 .
we leave details to the reader .
@xmath90 note that for rational functions , and even for laurent polynomials , a similar statement is not valid , see @xcite : conditions based on the euler formula remain necessary but they are no longer sufficient .
see also example [ ex : non - realize ] .
another approach to the proof of proposition [ prop : thom ] is to use an enumerative formula due to goulden and jackson @xcite which gives the number of cacti corresponding to a given list of partitions @xmath209 .
let us write a partition @xmath213 in the power notation : @xmath214 where @xmath215 is the number of parts of @xmath216 equal to @xmath217 , so that @xmath218 ( here @xmath42 is the total number of parts in @xmath216 ) , and @xmath219 .
denote @xmath220 then the following is true : [ prop : gj ] for a given @xmath187 satisfying conditions ( [ eq : thom ] ) we have @xmath221 where the sum is taken over the cacti @xmath201 with the passport @xmath209 , and @xmath222 is the size of the automorphism group of @xmath201 . now in order to prove proposition [ prop : thom ] it suffices to remark that the right - hand side of formula ( [ eq : gj ] ) is always positive .
@xmath90 taking @xmath223 in proposition [ prop : thom ] we may put the critical values @xmath224 and @xmath225 to 0 and 1 , and replace the jordan curve @xmath199 passing through these points by the segment @xmath98 $ ] . then a cactus becomes an _
ordinary _ bicolored plane tree with the passport @xmath226 . in this case
the number of trees ( with the weights @xmath227 ) is also given by formula ( [ eq : gj ] ) .
this fact will be useful in the future : in order to verify that a given ordinary tree is a unitree we can just compute the number given by ( [ eq : gj ] ) .
consider first the case @xmath228 , so that @xmath92 .
in this case the inequality ( [ eq : weak ] ) is trivial : it is reduced to @xmath93 .
thus , we only need to prove that this bound is attained .
we have @xmath229 , therefore @xmath230 .
let us take an arbitrary partition @xmath231 having @xmath232 parts , and also take @xmath233 and @xmath234 .
then for @xmath235 conditions ( [ eq : thom ] ) are satisfied .
hence , there exists a polynomial @xmath236 satisfying all the conditions of proposition [ prop : thom ] , with three critical values @xmath237 which may be chosen arbitrarily . taking @xmath238 and @xmath239
we obtain two polynomials which factorize as in ( [ eq : p , q ] ) and whose difference is @xmath240 thus , the obvious lower bound @xmath93 is indeed attained .
let us now consider the case @xmath182 . in this case
we must prove both the bound ( [ eq : weak ] ) and its attainability .
we have @xmath241 and @xmath242 . therefore , @xmath243 factors into @xmath184 factors @xmath244 , where @xmath245 runs over the @xmath184-th roots of unity . if one of the factors , which we may without loss of generality assume to be @xmath246 , has degree @xmath247 , then the leading coefficients of @xmath66 and @xmath248 coincide .
hence , the leading coefficients of @xmath66 and @xmath249 for @xmath250 do not coincide , and all the remaining @xmath251 factors have the degree exactly equal to @xmath185 .
this gives us the inequality @xmath252 according to the first part of this proof , the bound @xmath253 is attained , and therefore the bound ( [ eq : weak ] ) is also attained .
this finishes the proof of theorem [ th : weak ] .
@xmath90 notice that the above reasoning may be used for deducing the attainabilty of the bound in the case when condition ( [ eq : realize ] ) is satisfied and @xmath183 , from the case @xmath254 .
indeed , it follows from the case of coprime @xmath159 and @xmath160 that @xmath255 and hence @xmath256 + ( d-1)\cdot\frac{n}{d } { \,=\,}(n+1)-(p+q).\ ] ] let us take @xmath257 , @xmath258 , and @xmath259 , so that @xmath260 , @xmath261 , and @xmath262
. then we have @xmath263 but this bound can not be attained .
the correct answer is given by theorem [ th : weak ] : @xmath264 and , indeed , there is only one plane map with two black vertices of degrees 4 and 2 , respectively , and with three white vertices of degree 2 , see figure [ fig : small ] .
this map has two finite faces , but one of them is not of degree 1 .
the sum of degrees of the finite faces is 3 .
[ ex : non - realize ] let us take @xmath159 and @xmath160 such that @xmath182 and @xmath265 .
let us also take @xmath119 and @xmath118 .
then the passport @xmath113 satisfies the euler relation : there are @xmath179 vertices , @xmath18 edges , and @xmath88 faces , so that @xmath266 { \,=\,}2.\ ] ] however , a plane map with these data does not exist .
the principal vocation of this paper is to use combinatorics for the study of polynomials . but
here in this particular example we , essentially , deduce a non - trivial statement about plane maps from a trivial property of polynomials .
namely , we deduce the non - existence of certain maps from the fact that the degree of a polynomial can not be negative .
this section contains the main results of the paper : here we give a complete classification of the passports satisfying the following property : a weighted bicolored plane tree having this passport is unique .
as we have explained before , in proposition [ prop : in - q ] , belyi functions corresponding to such trees are defined over @xmath5
. ordinary unitrees were classified by adrianov in 1989 .
however , his initial proof was never published since it looked too cumbersome .
then , in 1992 , appeared the paper @xcite by goulden and jackson with the enumerative formula ( [ eq : gj ] ) , which opened a possibility for another proof , by carefully analizing this formula and looking for the cases in which it gives a number @xmath267 ( recall that ( [ eq : gj ] ) counts each tree @xmath201 with the weight @xmath268 ) .
our situation is more difficult for two reasons .
first , we deal with weighted trees ; and , second , we do nt have an enumerative formula at our disposal . for these ( and many other ) reasons such a formula would be very welcome . to be more specific ,
we need a formula which would give us , in an explicit way , the number of the weighted bicolored plane trees _ corresponding to a given passport_. an additional difficulty here ensues from the fact that the same passport may correspond not only to trees but also to forests . [ ass : simple - passport ] in the remaining part of the paper we will consider only the passports @xmath147 such that @xmath175 and @xmath94 . according to lemma [ lem : d>1 ] , the case @xmath269
is reduced to this one . indeed , starting from a tree @xmath270 with @xmath182 we can obtain a tree @xmath271 with @xmath272 by dividing the weights of all edges of @xmath270 by @xmath184 , and it is easy to see that @xmath270 is a unitree if and only if @xmath271 is a unitree .
recall that the face partition @xmath117 is defined by @xmath147 and is always equal to @xmath118 where @xmath119 .
a weighted bicolored plane tree such that there is no other tree with the same passport is called a _
unitree_. the _ diameter _ of a tree is the length of the longest path in this tree .
the classification of unitrees is summarized in the following theorem : [ th : main ] up to an exchange of black and white and to a multiplication of all the weights by @xmath183 , the complete list of unitrees consists of the following @xmath273 cases : * five infinite series a , b , c , d , e of trees shown in figures [ fig : stars - a ] , [ fig : chains - b ] , [ fig : brushes - c ] , [ fig : brushes - d ] , and [ fig : brushes - e ] , involving two integral weight parameters @xmath171 and @xmath170 which are supposed to be coprime @xmath274thus either @xmath275 , or @xmath276 .
note that * * for the diameter @xmath277 , only the trees of the types @xmath8 and @xmath278 exist ; * * for the diameter @xmath279 , the trees of types @xmath8 , @xmath107 , and @xmath278 exist ; * * for the diameter @xmath280 , the trees of types @xmath8 , @xmath201 , and @xmath278 exist .
* five infinite series f , g , h , i , j shown in figures [ fig : diameter-4 ] and [ fig : diameter-6 ] .
* ten sporadic trees k , l , m , n , o , p , q , r , s , t shown in figures [ fig : sporadic-5 ] , [ fig : sporadic-6 ] , [ fig : sporadic-8 ] , and [ fig : sporadic-10 ] .
the above series are not disjoint . for example
, the trees of the series @xmath201 with @xmath281 also belong to the series @xmath8 . if @xmath282 , @xmath201 becomes a part of @xmath283 , up to a renaming of variables ; etc .
the list of ordinary unitrees compiled by adrianov in 1989 consists of the following cases : the series @xmath7 , @xmath8 , and @xmath201 with @xmath284 ; the series @xmath285 , @xmath131 , and @xmath286 ; and the sporadic tree @xmath3 .
notice that quite a few of our trees have all their white vertices being of degree 2 , and thus , according to convention [ con : without - white ] , we can make these vertices implicit and draw the pictures as usual plane maps .
the corresponding maps are shown in figure [ fig : trees2maps ] .
the strategy of the proof of theorem [ th : main ] is as follows .
we propose various transformations of trees changing the trees themselves while preserving their passports : this is a way to show that the combinatorial orbit of a given tree consists of more that one element .
the trees which survive such a surgery are ( a ) those to which the transformation in question can not be applied , and ( b ) those for which the transformed tree turns out to be isomorphic to the initial one . in this way
we gradually eliminate all the trees which are not unitrees .
then , at the final stage , we show that all the trees which have passed through all the sieves are indeed unitrees .
the proof ends on page .
[ def : rooted ] a tree with a distinguished leaf edge is called _ rooted tree _ , and the distinguished edge itself
is called its _ root_. two rooted trees are isomorphic if there exists an isomorphism which sends the root of one of the trees to the root of the other one .
[ def : branch ] let a vertex @xmath287 of a tree @xmath270 be given .
then a _ branch _ of @xmath270 attached to @xmath287 is a rooted tree which is a subtree of @xmath270 containing a single edge incident to @xmath287 .
this edge is the root of the branch .
the following characterization of unitrees eliminates a vast amount of possibilities .
[ lem : branches ] all branches going out of a vertex of a unitree , except maybe one , are isomorphic as rooted trees .
this property must be true for every vertex of a unitree .
[ [ proof.-5 ] ] proof .
+ + + + + + let us call a vertex _ central _ if it is obtained by the following procedure .
we cut off all the leaves of the tree ; then do the same with the remaining smaller tree , then again , etc . in the end , what remains is either a single vertex , or an edge . in the first case , there is a single central vertex ; in the second case , there are are two of them , one black and one white .
obviously , _ any isomorphism of trees sends a central vertex to a central one _ , and if there are two central vertices , it sends the black central vertex to the black one , and the white , to the white one . on the other hand , according to definition [ def : iso ] , any isomorphism which sends a vertex @xmath287 to itself must preserve the cyclic order of the branches attached to @xmath287 .
thus , the property affirmed in this lemma , namely , that all the branches except maybe one are isomorphic , is valid for a central vertex or vertices , since otherwise an operation of _ exchanging _ of two non - isomorphic branches attached to the central vertex would change the cyclic order of branches around this vertex .
further , observe that the operation of exchanging of two non - isomorphic branches attached to a vertex of a _ rooted _ tree always changes this tree unless all the branches attached to this veretx , except maybe the branch containing the root , are isomorphic . indeed
, the introduction of a root makes a cyclic order on branches around any vertex into a _ linear order _ on the branches incident to it and not containing the root . the only possibility to make this linear order invariant under the operation of exchanging the branches is to make them all equal .
now , if a vertex @xmath287 of a tree @xmath270 is not central then it belongs to a branch @xmath288 attached to a central vertex .
this branch itself is a rooted tree and , in case if the condition of the lemma is not satisfied , the operation of exchanging of the branches changes @xmath288 .
however , changing the branch @xmath288 would mean changing a single branch of @xmath270 attached to its central vertex .
this would make @xmath270 not isomorphic to itself .
thus , in this case @xmath270 would not be a unitree .
@xmath90 sometimes , we can change not the topology of the tree but the distribution of its weights , while remaining in the same combinatorial orbit .
let us first formulate a statement which is entirely obvious : [ lem : wd ] if a passport @xmath147 corresponds to a unitree then the corresponding weight distribution @xmath289 ( see definition [ def : weighted ] ) is determined by @xmath159 and @xmath160 in a unique way .
[ lem : weight - exchange ] let @xmath290 be the weights of three successive edges of a unitree , as in figure [ fig : weight - exchange ] , left . if @xmath291 , then either @xmath292 , or @xmath293 .
similarly , if @xmath294 , then either @xmath292 , or @xmath295 .
[ [ proof.-6 ] ] proof .
+ + + + + + rotating if necessary the tree under consideration , without loss of generality we may assume that @xmath291 .
if @xmath177 then we can construct the tree shown in figure [ fig : weight - exchange ] , right , replacing the weight @xmath176 with @xmath296 , the weight @xmath170 with @xmath297 , and exchanging the places of the subtrees @xmath298 and @xmath299 .
we see that the vertex degrees of the new tree are the same as in the initial one , while the weights of two edges have changed , unless @xmath293 .
thus , if @xmath177 but @xmath300 , then exactly two parts of the weight distribution @xmath289 have changed , which contradicts lemma [ lem : wd ] .
@xmath90 [ cor : adj - edges ] if in a unitree there are two adjacent edges of the same weight @xmath171 , and at least one of them is not a leaf , then @xmath301 . [ [ proof.-7 ] ] proof .
+ + + + + + an edge which is not a leaf must be the middle edge of a path of length 3 , see figure [ fig : adj - edges ] . according to lemma [ lem : weight - exchange ] we have either @xmath302 or @xmath303 . if there is an edge of weight @xmath200 attached to the middle edge of the path , like in figure [ fig : adj - edges ] , then we have either @xmath304 or @xmath305 , so the possible values for @xmath200 are @xmath171 , @xmath306 , or @xmath307 .
dealing in the same way with the other edges of the tree we see that the weights of all of them are multiples of @xmath171 . according to assumption [ ass : simple - passport ]
this means that @xmath301 .
@xmath90 [ prop : chain - sts ] suppose that a unitree contains a path of three successive edges having the weights @xmath171 , @xmath170 , @xmath171 .
then the only possible weights for all the edges of this tree are @xmath171 , @xmath170 , or @xmath296 .
[ [ proof.-8 ] ] proof .
+ + + + + + let us make an operation inverse to the one used in the proof of theorem [ th : existence ] , that is , `` rip '' the tree along the edge of the weight @xmath170 , as in figure [ fig : unglue ] .
now suppose that in one of the subtrees @xmath308 there exists an edge of a weight @xmath309 , and try to stitch the two trees together in a different way . 1 .
suppose that in one of the subtrees @xmath310 or @xmath298 there exists an edge of a weight @xmath311 .
stitch it to the edge of the weight @xmath296 by the procedure explained in the proof of theorem [ th : existence ] .
1 . if @xmath312 then the weights of the four edges participating in the operations of ripping and stitching , instead of being @xmath313 become @xmath314 .
removing from the two sets the coinciding elements @xmath171 and @xmath49 , we get , on the one hand , @xmath315 , and , on the other hand , @xmath316 .
these sets coincide only when @xmath302 or @xmath317 .
if @xmath318 then , instead of @xmath313 , we obtain @xmath319 .
these two sets can not coincide at all since @xmath49 is greater than every term in the second set .
2 . suppose that in one of the subtrees @xmath320 or @xmath299 there exists an edge of a weight @xmath321 .
stitch it to the edge of the weight @xmath171 by the same procedure as above . 1 .
if @xmath322 then , instead of @xmath313 , we get @xmath323 . removing @xmath324 from both sets we get , on the one hand , @xmath325 , and on the other hand , @xmath326 .
these sets coincide only when @xmath327 .
if @xmath328 then the new set of weights is @xmath329 ; this set can not coincide with @xmath313 since @xmath296 is greater than every term in the second set .
thus , the hypothesis that there exists an edge of a weight @xmath309 leads to a contradiction .
the proposition is proved .
@xmath90 notice that the operation of ripping and stitching introduced in the proof of proposition [ prop : chain - sts ] often leads to another tree even in the case when the weights of all the edges of the tree under consideration are @xmath171 , @xmath170 , or @xmath296 .
below we will often use this operation and call it @xmath330-operation .
[ prop : chain - st - sum ] suppose that a unitree contains a path of three successive edges having the weights @xmath171 , @xmath170 , @xmath296 , and suppose also that @xmath275 .
then the edge of the weight @xmath296 is a leaf .
[ [ proof - of - proposition - propchain - st - sum . ] ] proof of proposition [ prop : chain - st - sum ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + take the tree shown in figure [ fig : sts - i ] , left , and exchange the subtrees @xmath298 and @xmath299 . obviously , both trees in the figure have the same passport .
suppose that the edge of the weight @xmath296 is not a leaf , so that the subtree @xmath299 of the left tree is non - empty . according to lemma [ lem : branches ] , the edges of @xmath136 which are adjacent to the vertex @xmath44 have the same weight .
denote this weight by @xmath49 . by lemma [ lem : weight - exchange ] , the possible values of @xmath49 are either @xmath170 or @xmath331 .
in the first case , we get a sub - path containing three edges of the weights @xmath170 , @xmath296 , @xmath170 , and , according to proposition [ prop : chain - sts ] , the only possible edge weights in such a tree could be @xmath170 , @xmath296 , and @xmath331 .
but this contradicts the supposition that we have already an edge of the weight @xmath171 with @xmath275 .
if , on the other hand , @xmath332 , then there are at least three _ non - isomorphic _ subtrees attached to the vertex @xmath42 in the right tree , since there are edges of three different weight @xmath171 , @xmath170 , and @xmath331 incident to this vertex .
this situation violates lemma [ lem : branches ] .
@xmath90 if @xmath333 then both arguments in the above proof are no longer valid , though it is still difficult to make left and right trees of figure [ fig : sts - i ] isomorphic . ( recall that if @xmath333 then @xmath284 , see corollary [ cor : adj - edges ] . ) however , it is possible to construct unitrees having the paths of length 4 with the weights @xmath334 ( see the series @xmath335 , figure [ fig : diameter-4 ] ) , and the paths with the weights @xmath336 ( see the tree @xmath2 , figure [ fig : sporadic-6 ] ) . [
prop : chain - st - sum1 ] suppose that a unitree contains a path of three successive edges having the weights @xmath171 , @xmath170 , @xmath296 , and suppose also that the vertex adjacent to the edges of weights @xmath171 and @xmath170 has the valency @xmath296 .
then the edge of the weight @xmath296 is a leaf .
[ [ proof.-9 ] ] proof .
+ + + + + + keeping notation of figure [ fig : sts - i ] , it is enough to observe that if the subtree @xmath298 is empty , then the two trees can not be isomorphic since the right one has more leafs than the left one .
this statement remains valid also for @xmath333 .
@xmath90 a brush is a chain with two bunches of leaves attached to its ends : see formal definition below
. typical representatives of brushes are the trees shown in figure [ fig : brushes - e ] . in this section
we classify all brush unitrees .
[ def : crossroad ] a vertex of a tree is _ profound _ if , after having removed all the leaves from the tree , this vertex does not become a leaf .
a vertex of a tree is a _ crossroad _ if it is profound and has at least three branches going out of it .
a tree is called a _ brush _ if it does not contain crossroads .
[ prop : uni - brush ] a brush unitree belongs either to one of the series @xmath7 , @xmath8 , @xmath201 , @xmath107 , @xmath278 ( figures [ fig : stars - a ] , [ fig : chains - b ] , [ fig : brushes - c ] , [ fig : brushes - d ] , and [ fig : brushes - e ] ) , or to the series @xmath285 with the degree of the central vertex @xmath223 , or to the series @xmath335 with the degree of the central vertex @xmath337 ( figure [ fig : diameter-4 ] ) . [ [ proof.-10 ] ] proof .
+ + + + + + in this part of the proof we only eliminate brush trees which are not unitrees . the uniqueness of the remaining brush trees will be proved later , in section [ sec : uniqueness ] . when all the edges of a tree are leaves we get the series @xmath7 consisting of stars ( figure [ fig : stars - a ] ) .
according lemma [ lem : branches ] , only one of the leaves may have a weight different from the others .
for the trees of diameter three , lemma [ lem : weight - exchange ] leads to two possible patterns .
one of them corresponds to the series @xmath201 ( figure [ fig : brushes - c ] ) ; the other one is shown in figure [ fig : not - c2 ] , left .
we see that when @xmath338 we can transform the left tree of figure [ fig : not - c2 ] into the right one .
the new tree has the same passport but is not isomorphic to the initial one .
thus , the pattern shown in figure [ fig : not - c2 ] , left , is not a unitree .
if @xmath339 then this pattern is a particular case of the series @xmath340 , so it is a unitree .
now let us consider first the trees of diameter @xmath341 , and after that return to the diameter 4 case .
suppose that a tree contains two adjacent edges of weights @xmath171 and @xmath170 which are not leaves .
it follows from lemma [ lem : weight - exchange ] and proposition [ prop : chain - st - sum1 ] that the weights @xmath171 and @xmath170 alternate along all the path connecting vertices from which leafs grow .
furthermore , since this path contains at least three edges , it follows from proposition [ prop : chain - sts ] that the only possible weight of a leaf which is not obtained by the further alternance of @xmath171 and @xmath170 is @xmath296 .
now look at figure [ fig : brush - pere ] , where an @xmath330-operation is applied to a brush tree having @xmath342 leaves of weight @xmath171 on one of its ends .
the tree thus obtained , shown on the right , is distinct from the initial one since it contains a crossroad .
a similar surgery can be made if there are @xmath343 leaves of the weight @xmath171 or @xmath170 ( depending on the parity of the diameter ) on the right end of the tree .
thus , for the diameters @xmath277 only the types @xmath8 and @xmath278 survive .
namely , if a bunch of leaves at an end of the tree contains two or more leaves then the weight of these leaves is @xmath296 .
the above argument fails for the brush trees of diameter 4 : indeed , this time the operation shown in figure [ fig : brush - pere ] does not create a crossroad .
therefore , the diameter 4 case needs a special consideration .
let us take a diameter of the tree , that is , a chain of length 4 going from one of its ends to the other . by lemma [ lem : weight - exchange ]
the sequence of the weights of its three first edges can be either @xmath171 , @xmath170 , @xmath171 , or @xmath171 , @xmath170 , @xmath296 , or @xmath296 , @xmath170 , @xmath171 .
consider first the case @xmath171 , @xmath170 , @xmath296 , so that the edge of the weight @xmath296 is not a leaf . in this case by proposition [ prop : chain - st - sum ] we necessarily have @xmath284 and a tree either belongs to the series @xmath335 , where the degree of the central vertex is @xmath337 , or has the form shown in figure [ fig : no-3 ] on the left . in the last case , however , the tree under consideration is not a unitree , which can be seen by a transformation shown on the right .
assume now that the sequence of weights starts with @xmath171 , @xmath170 , @xmath171 .
using lemma [ lem : weight - exchange ] again we see that it must be a part of one of the following three possible sequences : either @xmath344 , or @xmath345 , or @xmath346 . for the latter one , taking @xmath347 , @xmath348 , we find the already considered above case @xmath349 read from right to left .
two other forms are shown in figure [ fig : diam4-pere ] on the left .
the first of these forms ( above , left ) can be transformed in a way shown on the right .
the new tree is not equal to the initial one , unless either it belongs to the type @xmath8 ( that is , @xmath281 ) , or @xmath284 , which is a particular case of the series @xmath285 , where the degree of the central vertex is @xmath223 . for the second form ( below , left ) ,
the operation shown on the right can not be applied when @xmath339 , that is , when the tree belongs to the series @xmath340 ; and it does not change the tree when @xmath223 and @xmath350 , which corresponds to the series @xmath107 .
finally , if the sequence of weights of edges of a diameter starts as @xmath296 , @xmath171 , @xmath170 , then either a tree belongs to the series @xmath351 ( or @xmath352 ) , or the sequence of weights of the edges of a diameter is @xmath296 , @xmath171 , @xmath170 , @xmath353 . in the latter case , however , proposition [ prop : chain - st - sum ] yields that @xmath354 , implying that a tree is the one shown in figure [ fig : no-3 ] on the left .
@xmath90 from now on
we will assume that the trees we consider are not brushes , that is , they contain at least one crossroad .
recall that a crossroad is a profound vertex at which three or more branches meet , see definition [ def : crossroad ] .
a typical tree with a crossroad is shown in figure [ fig : crossroad ] . according to lemma [ lem : branches ] , all the branches attached to the crossroad , except maybe one , are isomorphic as rooted trees , with their root edges ( shown by thick lines in the figure ) being incident to the crossroad .
we call these branches _ repeating _ ; in the figure they are denoted by the same letter @xmath355 ; the subtrees of @xmath355 attached to the root are denoted by @xmath356 .
the subtrees @xmath356 , are non - empty since otherwise the vertex to which @xmath355 and @xmath357 are attached would not be profound . the number of branches of the type @xmath355 can be two or more , but the majority of the transformations given below involve only two branches ; therefore , in the majority of pictures we will draw only two repeating branches . by convention
, we suppose that _ the branch @xmath357 is always non - empty_. if all the branches meeting at the crossroad are isomorphic to @xmath355 and might therefore be all considered as repeating , we take an arbitrary one of them and , somewhat artificially , declare it to be the `` non - repeating '' branch @xmath357 .
the subtree @xmath358 has a right to be empty .
the roots of repeating branches are adjacent edges which are not leaves .
therefore , according to corollary [ cor : adj - edges ] , their weights must be equal to 1 .
finally , the _ height _ of a repeating or non - repeating branch is the distance from its root vertex ( that is , the crossroad ) to its farthest leaf . in the previous section we classified the brush unitrees , which are by definition unitrees without crossroads . in this section
we establish a complete list of all possible unitrees whose repeating branches all have the height 2 .
more precisely , we assume that for _ any _ crossroad of a unitree under consideration the repeating branches are of height 2 .
[ prop : repdiam2 ] a unitree whose all repeating branches are of height @xmath359 belongs to one of the types @xmath285 , @xmath335 , @xmath131 , or @xmath130 . [ [ proof.-11 ] ] proof .
+ + + + + + first of all observe that weights of leaves of repeating branches can not be equal to 2 since otherwise the transformation shown in figure [ fig : proof - rep - brush-2 ] leads to a non - isomorphic tree ( the tree on the right has fewer leaves than the one on the left ) .
thus , the weights of these leaves are equal to 1
. therefore , according to lemma [ lem : weight - exchange ] the root edge of @xmath357 can only be of weight 1 or 2 . the case @xmath360 ( that is , when @xmath357 consists of a single edge , see figure [ fig : crossroad ] ) is a particular case of the series @xmath285 and @xmath335 .
suppose then that @xmath361 .
if the non - repeating branch is of the height @xmath359 , that is , if it is a root edge with a bunch of leaves attached to it , then these leaves could be of weight 1 or 2 when the root edge is of weight 1 , and they could be of weight 1 or 3 when the root edge is of weight 2 . however , the leaves of the weights 2 and 3 are impossible , as two transformations of figure [ fig : nonrep-12 ] show . in this figure ,
we take all the repeating branches but one and re - attach them to one of the leaves of the non - repeating branch . these transformations always change the trees : on the left , there appears a non - leaf of weight 2 , and on the right , there appears a non - leaf of weight 3 .
thus , all the leaves of the non - repeating branch must be of weight 1 .
in addition , when the root edge of the non - repeating branch is of weight 2 this branch can not be isomorphic to the repeating branches .
this observation implies that the degree of the black vertex lying on this branch must be equal to the degrees of the black vertices of repeating branches since otherwise an exchange of leaves between repeating and non - repeating branches could be possible .
thus , the only remaining possibilities are the trees of the types @xmath285 and @xmath335 , see figure [ fig : diameter-4 ] .
consider now the case of a non - repeating branch of height @xmath362 .
first suppose that the vertex @xmath44 , which is the nearest neighbor of the crossroad vertex @xmath42 when we move along the non - repeating branch , is itself a crossroad . according to our supposition
, the tree does not have repeating branches of the height greater than 2 .
hence , the repeating branches growing out of @xmath44 are of height 2 .
but then a leaf @xmath363 of such a branch can be interchanged with a repeating branch @xmath364 growing out of @xmath42 , see figure [ fig : rep - to - rep ] .
this operation would create at least three different trees attached to @xmath42 : one of them would be of height 1 , another one of height 2 , and a third one of height 4 .
thus , this possibility is ruled out .
suppose next that the vertex @xmath44 is not a crossroad .
then the tree looks like the one in figure [ fig : rep-2-nonrep-3 ] , top left , where @xmath310 is non - empty .
a priori , there are four possibilities for the values @xmath365 , namely , @xmath366 , @xmath367 , @xmath368 , and @xmath369 .
the case @xmath369 can be immediately ruled out since the edge of weight 3 should be a leaf by proposition [ prop : chain - st - sum ] , but we have supposed that @xmath370 . the cases @xmath371 or @xmath368 can be treated together .
when @xmath372 we can re - attach @xmath310 to one of the leaves of the repeating branches , as is shown in the same figure on the top right . among the branches attached to the vertex @xmath42 of the tree
thus obtained there is only one branch of a height greater than 2 : it is @xmath373 .
therefore , all the remaining branches are repeating , so we may conclude that @xmath301 ( the case @xmath374 is impossible ) , and all the repeating branches have only one leaf .
thus , the tree looks like the one on the bottom left in figure [ fig : rep-2-nonrep-3 ] , where two possibilities may occur : either @xmath375 ; or @xmath376 , and then , according to proposition [ prop : chain - st - sum1 ] , @xmath377 , so the edges of the weight @xmath376 are leaves . in the first case
, we can exchange the repeating branches attached to the vertices @xmath42 and @xmath79 . therefore
, they all must be equal , and we get a tree of the type @xmath131 , see figure [ fig : diameter-6 ] . in the second case
, we can interchange one of the leaves of weight 2 with two repeating branches attached to @xmath42 . the only tree which does not change after this transformation is the one which has exactly one leaf of weight 2 and exactly two repeating branches attached to @xmath42 , that is , the tree @xmath130 , see figure [ fig : sporadic-5 ] .
there remains the last case to be ruled out : when the tree shown in figure [ fig : rep-2-nonrep-3 ] , top left , has @xmath301 and @xmath378 ; see also figure [ fig : outrule ] , left . in this case
we can interchange the subtree @xmath310 with all but one repeating branches , see the tree on the right of figure [ fig : outrule ] .
we see that @xmath310 must consist of several copies of the branch @xmath364 since otherwise @xmath310 should consist of copies of the longer branch at @xmath42 , and we would get repeating branches of the height greater than 2
. then , we may take the left tree of figure [ fig : outrule ] , cut all the repeating branches form @xmath42 and re - attach them to one of the leaves of @xmath310 .
this operation will necessarily produce a different tree since the only edge of the weight 2 will be now at distance 1 from the leaf while it was at distance 2 in the initial tree .
therefore , this case is impossible .
proposition [ prop : repdiam2 ] is proved .
@xmath90 if a unitree has a crossroad from which grow repeating branches of height @xmath380 , then these branches `` start '' either with a path @xmath109 , @xmath171 , @xmath381 , or with a path @xmath109 , @xmath170 , @xmath109 , where @xmath171 and @xmath170 here may be equal to either 1 or 2 ( see figures [ fig:1.t.t+1 ] and [ fig:1.t.1 ] ) . in this subsection
we classify unitrees which have no crossroads of the second type .
we start with the following lemma .
[ lem : xxx ] if a unitree has a repeating branch of the type @xmath379 , then this branch has one of the two forms shown in figure [ fig : xxx ] .
furthermore , in the second case the unitree is necessarily the tree @xmath2 .
[ [ proof.-12 ] ] proof .
+ + + + + + first of all observe that the subtrees @xmath310 and @xmath320 in figure [ fig:1.t.t+1 ] can be interchanged , and if one of them was empty while the other was not , this operation would change the number of leaves , so that the tree in question could not be a unitree .
we will show now that the assumption that both trees @xmath310 and @xmath320 are not empty also leads to a contradiction ( so that , in fact , both of them are empty )
. since the tree @xmath288 is isomorphic to a subtree of @xmath364 and is therefore distinct from @xmath364 , if @xmath310 is not empty , then it consists of a certain number of copies of @xmath364 or of @xmath288 .
the first case is impossible since @xmath310 is a subtree of @xmath364 .
therefore , @xmath310 consists of a certain number of copies of @xmath288 implying that @xmath320 is a proper subtree of @xmath382 now , interchanging @xmath310 and @xmath320 in every repeating branch we may prove in the same way that that @xmath310 is a proper subtree of @xmath320 , implying the contradiction that we need .
thus , @xmath310 and @xmath320 are empty . in particular , @xmath298 is merely a collection of leaves of weight @xmath381 .
assume now that @xmath374 .
then our tree must look as in figure [ fig : proof - rep - brush-12 ] , top left , where the number of repeating branches at the vertex @xmath42 might be two or more .
let us take two of these branches , apply the transformation shown on top right , and see what takes place at the vertex @xmath44 . according to lemma [ lem : branches ] , all the subtrees growing out of this vertex , except maybe one , must be isomorphic
. this can only happen when @xmath339 , and the subtree growing from @xmath44 to the left is isomorphic to the one growing from @xmath44 to the right .
therefore , before the transformation there were exactly two ( and not more ) repeating branches at @xmath42 , and the subtree @xmath357 was reduced to a single leaf of weight 3 .
the resulting situation is shown in figure [ fig : proof - rep - brush-12 ] , bottom left . in this case , our transformation can still be applied , but it leads to a tree isomorphic to the initial one .
the unitree thus obtained is @xmath2 , see figure [ fig : sporadic-6 ] . @xmath90
[ prop : repsts+1 ] a unitree which has at least one crossroad of type @xmath379 but no crossroads of type @xmath383 belongs to one of the types @xmath199 , @xmath135 , @xmath384 , @xmath115 , @xmath385 , @xmath2 , @xmath51 , or @xmath386 .
[ [ proof.-13 ] ] proof .
+ + + + + + in view of lemma [ lem : xxx ] we may assume that the repeating branches have the form shown in figure [ fig : xxx ] on the left .
suppose first that the number of the repeating branches is three or more , and apply the transformation shown in figure [ fig : rep - branch-112 ] , that is , interchange the positions of a leaf of weight 2 and of a pair of repeating branches . if the number of the repeating branches was more than three then the principle `` all branches except maybe one are isomorphic '' would be violated at the vertex @xmath42 .
the same principle would be violated at the vertex @xmath44 if the number of leaves in a repeating branch was more than two .
therefore , the number of repeating branches is three , and our transformation looks as is shown in figure [ fig : rep - branch-112 ] , bottom .
if the number of leaves in a repeating branch is two , then applying once again the same principle at the point @xmath44 , we arrive at the tree @xmath385 .
assume now that this number is equal to one .
then the height of @xmath357 is less than two , since otherwise the new tree would have more crossroads than the initial one .
furthermore , if @xmath357 is a bunch of leaves of weight two , then we could transfer all these leaves to the vertex @xmath44 changing the tree .
therefore , @xmath357 is empty , and we arrive at the tree @xmath387 suppose next that the number of repeating branches is two .
the starting edge of the non - repeating branch @xmath357 is either of weight 2 , and then , according to proposition [ prop : chain - st - sum1 ] , it is a leaf , and we get the tree @xmath199 ( figure [ fig : sporadic-6 ] ) , or it is of weight 1 . in the latter case
it does not have to be a leaf , though this situation imposes another constraint : the repeating branches must have only one leaf , otherwise the transformation shown in figure [ fig : rep - br - r1 ] can be applied , producing three non - isomorphic branches growing from the crossroad .
what remains is to study more attentively the structure of the non - repeating branch @xmath357 .
here we consider the following cases : * the height of @xmath357 is 1 , that is , @xmath360 . *
the height of @xmath357 is 2 . *
the height of @xmath357 is 3 or more , and @xmath357 starts with a path having the weights @xmath109 , @xmath171 , @xmath109 where @xmath171 is equal to 1 or 2 . *
the height of @xmath357 is 3 or more , and @xmath357 starts with a path having the weights @xmath109 , @xmath171 , @xmath381 where @xmath171 is equal to 1 or 2 .
the height of @xmath357 equal to 1 case is trivial : we get the tree @xmath135 ( figure [ fig : sporadic-6 ] ) .
the height of @xmath357 equal to 2 case is illustrated in figure [ fig : nonrep - height-2 ] .
if the weight of the leaves of the non - repeating branch is equal to 2 then , whatever is their number , the reattachment shown on the left changes the tree since the new tree has one leaf less than the initial one .
if the weight of the leaves of the non - repeating branch is equal to 1 then the reattachment shown on the right also changes the tree unless there is only one leaf in the non - repeating branch .
the latter case gives us the tree @xmath115 ( figure [ fig : sporadic-6 ] ) .
assume now that the height of the non - repeating branch is @xmath362 , and this branch contains a path with the weights 1 , @xmath171 , 1 , see the upper tree in figure [ fig : nonrep.1.s.1 ] .
first of all , we remark that the subtree @xmath310 may be interchanged with a chain of length 2 attached to the vertex @xmath42 .
then , according to the principle `` all branches except maybe one are isomorphic '' , two situations may occur .
first , we could thus create two repeating branches @xmath364 attached to the vertex @xmath44 , see the tree in the middle .
but such a tree would contain repeating branches of the type @xmath388 which contradicts our supposition .
the other possibility is that @xmath310 is equal to the chain which was attached to @xmath42 .
then we get a vertex @xmath79 ( see the lower tree ) which is of degree 2 and is incident to two edges of weights 1 and 1 . therefore , according to proposition [ prop : chain - st - sum1 ] , the edge of weight @xmath171 , which is not a leaf , can not have weight 2 ; hence , @xmath301 . finally , we affirm that @xmath389 , otherwise it could be reattached to the vertex @xmath42 and we would get three different trees attached to @xmath44 . the resulting tree is shown in figure [ fig : nonrep.1.s.1 ] , bottom . if @xmath390 we get the tree @xmath386 ( figure [ fig : sporadic-8 ] ) . if @xmath391 then , again according to the principle `` all branches except maybe one are isomorphic '' , @xmath298 must be equal to a chain of weights 1 , 1 , 2 , and we get the tree @xmath51 ( figure [ fig : sporadic-8 ] ) , since otherwise a tree would contain repeating branches of the type @xmath388 .
( note that in the last case we obtain the tree @xmath392 which is considered in proposition [ prop : repsts+2 ] which treats the case of repeating branches of the type @xmath383 . ) finally , consider the case when the non - repeating branch is of height @xmath362 and contains a path with the weights 1 , @xmath171 , @xmath381 , see figure [ fig : nonrep.1.s.s+1 ] .
we affirm that in this case @xmath301 and all the three subtrees @xmath310 , @xmath298 , @xmath320 are empty , so that we get the tree @xmath384 of figure [ fig : sporadic-6 ] ( and what we call `` non - repeating branch '' is in this case equal to the repeating branches ) . indeed
, the tree contains a path with the weights 1 , 1 , 1 ; therefore , according to proposition [ prop : chain - sts ] , the only possible weights are 1 and 2 , so @xmath301 .
now , if @xmath370 then it could be reattached to the vertex @xmath44 thus producing a tree with one more leaf .
therefore , @xmath393 . then , @xmath320 is also empty since @xmath310 and @xmath320 can be interchanged . finally , if @xmath391 then there are two possibilities . either @xmath298 is a bunch of leaves of weight 2 ; but then it can be reattached to the vertex @xmath42 . or @xmath298 is a number of copies of the long branch growing out of the vertex @xmath79 ; but then , once again , we would create repeating branches of the type @xmath383 .
proposition [ prop : repsts+1 ] is proved .
@xmath90 in this subsection we classify unitrees which have crossroads of the type @xmath383 .
[ prop : repsts+2 ] a unitree which has at least one crossroad of type @xmath383 belongs to one of the types @xmath286 , @xmath3 , or @xmath392 .
[ [ proof.-14 ] ] proof .
+ + + + + + first of all , observe that by lemma [ lem : branches ] the subtree @xmath298 is a collection of copies of the subtree @xmath364 , and the subtree @xmath310 is a collection of copies of the subtree @xmath288 ( see figure [ fig:1.t.1 ] ) .
further , an @xmath330-operation , applied to the first tree of figure [ fig:1.t.1 ] gives the second tree shown in this figure .
this image implies that there are only two repeating branches growing from the vertex @xmath42 , otherwise the tree would certainly change .
now , looking at the vertex @xmath44 of the second tree of figure [ fig:1.t.1 ] we see that either @xmath394 or @xmath395 . if @xmath395 then the initial tree would look like the third tree of the same figure
. then we could once again apply an @xmath330-transformation and make the long branch even longer , and one of the repeating branches , shorter ( see the fourth tree of the figure ) , which would give us three different branches attached to @xmath42 .
hence , @xmath394 . in particular , we have proved that whenever a unitree has a crossroads of type @xmath383 the corresponding non - repeating branch is a subtree of the repeating branch .
now , it follows from @xmath394 that the first tree of figure [ fig:1.t.1 ] has a ( unique ) center at the vertex @xmath42 while the second one has a ( unique ) center at the vertex @xmath79 .
hence , the vertex @xmath42 of the first tree must correspond to the vertex @xmath79 of the second one , and thus we must have @xmath372 and @xmath396 .
therefore , the tree has the form shown in figure [ fig:1.t.1-bis ] , with the same number @xmath397 of branches growing out of the vertices @xmath176 and @xmath287 . if @xmath389 then we get a tree of the type @xmath286 .
thus we may assume that @xmath398 is non - empty .
observe first that @xmath350 . indeed , if @xmath178 then @xmath288 is a repeating branch .
furthermore , since @xmath398 is non - empty , @xmath288 is either @xmath399-branch or @xmath383-branch .
the first case is impossible by lemma [ lem : xxx ] , while the second case is impossible since , as we have shown in the previous paragraph , here the corresponding non - repeating branch should be a subtree of @xmath288 , and this is not so .
if @xmath320 is a collection of leaves , then the transformation of figure [ fig : proof - rep - brush-2 ] shows that all leaves are of weight 1 .
moreover , @xmath320 contains not more than one leaf since otherwise we could transfer all the other leaves to the vertex @xmath287 , which would change the tree .
therefore , in this case we get the tree @xmath3 . finally , if @xmath320 is not a collection of leaves , then @xmath373 is a repeating branch of height at least 3 , which , as above , is necessarily of type @xmath399 , and lemma [ lem : xxx ] implies that @xmath373 has the form shown in figure [ fig : xxx ] on the left , where the number of leaves is equal to one since otherwise we could transport all the leaves but one to the vertex @xmath400 .
therefore , in this case we get the tree @xmath392 .
proposition [ prop : repsts+2 ] is proved .
@xmath90 our main tool will be `` cutting and gluing leaves '' , though these operations will be carried out not with the trees themselves but with their passports ; the trees must be kept in mind for an intuitive understanding of the proof .
we do not repeat every time that `` the same reasoning remains valid if we interchange black and white '' .
\(a ) there is only one black vertex ( see figure [ fig : stars - a ] ) ; therefore , all white vertices must be adjacent to it , which means that they are leaves .
the uniqueness is evident .
the uniqueness proofs for the cases from ( b ) to ( e ) all follow the same lines .
if the structure of a passport implies the existence of a leaf of degree , say , @xmath171 , then the only way to construct a corresponding tree is to glue an edge of the weight @xmath171 to a vertex of the opposite color of a tree having one less edge .
furthermore , in the initial ( bigger ) tree this edge can only be attached to a vertex of degree bigger than @xmath171 .
if the smaller tree is a unitree ( and usually it is by induction ) , and if there is essentially one way to attach the new edge to it , then the bigger tree is also a unitree . in certain cases
more than one way of attaching a new edge may exist , but they all lead to isomorphic trees .
\(b ) let us consider , for example , the case of an odd length , and examine not the tree itself but its passport @xmath147 .
for this tree , @xmath159 and @xmath160 are the same : @xmath401 .
the passport implies that there are two vertices of degree @xmath171 , a black one and a white one , while all the other vertices , both black and white , are of degree @xmath296 .
these latter vertices can not be leaves since otherwise there should exist vertices of a bigger degree to which such leaves would be attached .
a tree must have at least two leaves .
we conclude that there are exactly two leaves , and they are vertices of degree @xmath171 .
they are connected to the tree by edges of the weight @xmath171 .
the degree of the vertex to which such a leaf is attached is @xmath296 , and its color is opposite to the color of the leaf .
now let us cut off one of these leaves , for example , the white one .
then we get a tree with one less edge and with the passport @xmath402 and @xmath403 .
this passport corresponds to the chain tree of smaller ( and even ) length .
we may inductively suppose that this tree is a unitree .
the vertex of degree @xmath170 of this tree is a leaf .
now , we must make an operation which would simultaneously fulfill the following three goals : * it re - attaches back a white leaf of weight @xmath171 to the smaller tree ; * it makes the black vertex of degree @xmath170 in the smaller tree to disappear ; * it makes to appear an additional , @xmath16th black vertex of degree @xmath296 , to the already existing @xmath404 ones .
it is clear that the only way to do all that is to attach this white leaf of weight @xmath171 to the black vertex of degree @xmath170 .
this operation re - creates the initial chain - tree .
the proof for an even length repeats the previous one almost word to word , only the leaves are now of the same color and of degrees @xmath171 and @xmath170 .
the base of induction is a tree consisting of a single edge , which is obviously unique .
\(c ) the passport of a tree of the type @xmath201 is @xmath405 , @xmath406 .
we affirm that there exists a leaf of degree @xmath171 . indeed
, a tree must have at least two leaves , and the vertex of the biggest degree can not be a leaf .
the biggest degree is either @xmath407 , or @xmath408 , or both .
suppose that we have a black leaf of degree @xmath171 .
then it has to be attached to the only white vertex of degree bigger than @xmath171 , which is the white vertex of degree @xmath408 .
cut this leaf off .
we get a smaller tree , with one less edge , with @xmath409 being replaced with @xmath410 .
this tree is a smaller instance of @xmath201 which may supposed to be a unitree by induction .
( note that in particular cases it can also be of type @xmath8 , or even of @xmath7 , the latter one when @xmath409 was equal to 1 . )
now we no longer work with the passports but with the trees .
we know the smaller tree since it is unique , and we must re - attach the previously cut - off black leaf to a white vertex of this smaller tree . here
two cases may take place . 1 .
if @xmath275 , or even if @xmath284 but @xmath411 , the initial ( bigger ) tree did not have a white vertex of degree @xmath306 . therefore , we can not attach the cut - off leaf to a white vertex of degree @xmath171 .
hence , the only vertex to which it can be attached is the white vertex of degree @xmath412 .
2 . if @xmath284 and @xmath350 then the smaller tree is the star with all its leaves being of degree 1
. then we may re - attach the leaf to any one of them , the resulting tree will be the same .
there is an additional subtlety here .
the planar structure of our trees means that we must choose not only a vertex to which we attach a new edge .
we must also choose an angle between neighboring edges incident to the vertex of attachment , and to insert the leaf into the angle between these edges .
if there are @xmath95 edges incident to a vertex , there also are @xmath95 angles between them , and therefore @xmath95 ways of placing the new edge .
but , obviously , in our case all these ways give the same plane tree , see figure [ fig : brushes - c ] .
\(d ) two black vertices of degree @xmath413 can not be leaves .
therefore , there exists a leaf of degree @xmath171 or @xmath296 . cut it off , and we get either @xmath201 or @xmath340 .
indeed , if the cut - off ( white ) leaf was of degree @xmath171 , and was attached to one of the black vertices of degree @xmath413 ( there are no other black vertices ) , then the passport of the smaller tree becomes @xmath414 , @xmath415 .
this passport corresponds to the pattern @xmath340 , with @xmath350 and the length of the chain equal to 4 .
the uniqueness of the corresponding tree will be proved in a moment . the only way to glue a leaf of degree @xmath171 to this tree and to create a vertex of degree @xmath413 instead of vertex of degree @xmath296 is to glue this leaf to the vertex of degree @xmath296 . if , on the other hand , the cut - off ( white ) leaf was of degree @xmath296 , and was attached to one of the black vertices of degree @xmath413 , then the passport of the smaller tree becomes @xmath416 , @xmath417 .
this passport corresponds to a tree of type @xmath201 , with @xmath223 and @xmath350 .
the uniqueness of such a tree was proved above . then the only way to glue back the leaf of degree
@xmath296 is to glue it to the black vertex of degree @xmath171 of the smaller tree .
it is easy to see that in both cases we get the same tree @xmath107 .
\(e ) the proof is similar to the cases considered above , so we will shorten our presentation . consider first the cases @xmath283 and @xmath351 .
all the vertices except two are of degree @xmath296 ; the two remaining ones are of degrees @xmath418 and @xmath419 for @xmath283 , and @xmath418 and @xmath420 for @xmath351 . without loss of generality
we may suppose that @xmath418 is the bigger of the two ; therefore , it can not be a leaf . for @xmath351
, the `` second best '' vertex can not be a leaf either since it has the same color . for @xmath283 ,
if @xmath421 , the vertex of degree @xmath419 might in principle be a leaf .
whatever is the case , there exists a leaf of degree @xmath296 . cut it off , and we obtain a smaller tree , with the possible pattern transitions as follows : @xmath422 ; @xmath423 ; @xmath424 ; or @xmath425 , the latter two maybe with renaming the variables .
now , for the cases @xmath340 and @xmath352 the situation is similar .
all the vertices except two are of degree @xmath296 .
the vertex of the biggest degree can not be a leaf .
therefore , there exists a leaf of degree @xmath171 or @xmath296 . cut it off , and we get a smaller tree , with the possible pattern transitions as follows : @xmath426 ; @xmath427 ; @xmath428 ; @xmath429 , or we may arrive to the patterns @xmath7 or @xmath8 .
what remains now is to see that there is only one way to re - attach the cut - off leaf to the smaller unitree .
( f , h , i , q ) the trees @xmath285 , @xmath131 , @xmath286 , @xmath3 are ordinary ; therefore , the enumerative formula ( [ eq : gj ] ) can be applied . if @xmath430 , a tree of the series @xmath285 is asymmetric , and therefore its contribution to ( [ eq : gj ] ) is 1 .
now , formula ( [ eq : gj ] ) in this case gives 1 ; therefore , there is no other tree with this passport . when @xmath431 , a tree of the series @xmath285 is symmetric , with the rotational symmetry of order @xmath16 .
therefore , its contribution to ( [ eq : gj ] ) is @xmath432 .
now , the formula itself gives @xmath432 ; therefore , there is no other tree in this case either . for the trees of the series @xmath131 , formula ( [ eq : gj ] )
gives 1 when @xmath433 , and gives @xmath434 when @xmath435 .
this corresponds to the symmetry order of these trees : they are asymmetric when @xmath433 , and symmetric of order 2 when @xmath435 .
the trees of the series @xmath286 are asymmetric , and formula ( [ eq : gj ] ) gives 1 .
the tree @xmath3 is asymmetric , and formula ( [ eq : gj ] ) gives 1 .
\(g ) the tree has @xmath436 vertices and hence @xmath437 edges .
since the total weight is @xmath436 , there exists exactly one edge of weight 2 while all the other edges are of weight 1 .
the only white vertex to which the edge of weight 2 can be attached is the vertex of degree @xmath16 , since all the other white vertices are of degree 1 .
the rest is obvious .
\(j ) all white vertices are of degree 2 ; therefore , a weight of an edge can only be 1 or 2 .
there are only three black vertices , their degrees being 4 , @xmath438 , @xmath438 .
therefore , the black vertices can not be leaves since such leaves could not be attached to a white vertex of degree 2 ; thus , all the leaves are white .
a white vertex which is not a leaf must have two black neighbors ; therefore , there are exactly two white vertices which are not leaves : they are `` intermediate '' white vertices between the black ones .
the edges incident to them are both of weight 1 .
the black vertex of degree 4 can not have two incident edges of weight 2 since these edges should be leaves , and such a tree would not be connected ; it can not have four incident edges of weight 1 either since such a tree should need more than three black vertices . therefore , the weights of the edges attached to this vertex must be 2 , 1 , 1 , and the edge of weight 2 is a leaf .
the rest is obvious .
\(p ) all black vertices are of degree 5 , all white ones are of degree 3
. therefore , black vertices can not be leaves .
arguing now as in the case ( j ) we conclude that there are exactly three white leaves , implying easily that there exists only one tree with this passport .
( k , l , m , n , o , r , s , t ) the proof of all these cases follows the same pattern .
let us take , for example , the tree @xmath385 .
its passport is @xmath439 .
therefore , the number of vertices is @xmath440 and the number of edges is 13 , while the total weight is @xmath441 .
thus , the overweight is 7 , and it must be distributed among the edges .
now , no edge can have a weight greater than 2 since the degrees of white vertices are all equal to 2 .
therefore , the tree @xmath385 has seven edges of weight 2 .
moreover , _ all of them are leaves _ ; indeed , if something were attached to the white end of such an edge then this white end would have a degree greater than 2 .
the same reasoning may be carried out for all the above cases , with their respective overweights and numbers of leaves of weight 2 .
now , let us cut off all the leaves of weight 2 .
what remains is an ordinary tree , and we must verify that it is a unitree .
usually it is immediately obvious since the ordinary tree in question is very small ; otherwise , we may apply formula ( [ eq : gj ] ) , or else we may remark that such a tree belongs to one of the previously established cases .
for example , for the tree @xmath392 what remains after cutting off the leaves of weight 2 is the tree @xmath3 .
the last step consists in proving that there is only one way to glue back to this ordinary unitree the leaves of weight 2 that were previously cut off .
for example , in the case @xmath385 the ordinary unitree has black vertices of degrees 3 , 1 , 1 , 1 , and we have , by gluing to them seven edges of weight 2 , made these degrees equal to 5 , 5 , 5 , 5 .
obviously , there is only one way to do that .
in fact , in certain cases there are several ways of gluing but they give the same result because of a symmetry of the underlying ordinary tree . for the tree @xmath135 , there is an additional condition we must satisfy : white leaves can only be attached to black vertices .
theorem [ th : main ] is proved . @xmath90
[ end - proof ]
the theory of dessins denfants studies combinatorial invariants of the galois action on dessins .
these invariants have various levels of generality .
the most general ( and the most simple ) one is the passport ; the subject of this paper is precisely the case when the passport alone guarantees the definability of a dessin over @xmath5 .
but our exposition would be incomplete if we did not mention several other galois invariants which lead to further examples of dessins and dz - triples defined over @xmath5 .
the following proposition is obvious .
[ prop : composition ] let @xmath116 and @xmath442 be two rational functions such that : * @xmath66 is a belyi function , with the corresponding dessin @xmath443 ; * @xmath444 is a function @xmath274not necessarily a belyi one@xmath445 , all of whose critical values are either vertices or face centers of @xmath443 .
then the function @xmath446 obtained as a composition @xmath447 is a belyi function . if , furthermore , both @xmath66 and @xmath444 are defined over @xmath5 , then , obviously , the same is true for @xmath285 .
the above proposition gives us a very general method of constructing belyi functions with all its finite poles being simple , or , in other words , belyi functions corresponding to weighted trees . [ cor : composition ] suppose that the functions @xmath66 and @xmath444 of the above proposition satisfy the following properties : * the dessin @xmath443 corresponds to a weighted tree , that is , all its finite faces are of degree @xmath109 ; * @xmath444 is a polynomial all of whose critical values except infinity are vertices of @xmath443 .
then all the finite faces of the dessin @xmath448 corresponding to the belyi function @xmath449 are of degree @xmath109 . if , furthermore , both @xmath66 and @xmath444 are defined over @xmath5 , then , obviously , the same is true for @xmath285 . [ [ proof.-15 ] ] proof . + + + + + + since @xmath444 is a polynomial , the only poles of @xmath450 , except infinity , are the preimages of the simple poles of @xmath66 , i.e. , the preimages of the centers of the small faces of @xmath443 .
since @xmath444 is not ramified over these simple poles , they remain simple for @xmath285 , and each of them is `` repeated '' @xmath451 times .
[ ex : composition-1 ] consider the following functions : @xmath452 here @xmath66 is a belyi function corresponding to the upper left dessin in figure [ fig : composition ] , and @xmath176 is a belyi function corresponding to the lower left dessin . substituting @xmath453 in @xmath66 we obtain a belyi function @xmath285 corresponding to the dessin shown on the right of figure [ fig : composition ] .
it is obvious that the _ combinatorial _ orbit of the dessin @xmath448 consists of more than one element : for example , the petals attached to the vertices of degrees 9 and 6 can be cyclically arranged in many different ways .
still , @xmath454 by construction .
note that the dessins @xmath443 and @xmath455 serving as building blocks for the above example both belong to the classification we have established in section [ sec : unitrees ] : they both correspond to unitrees , and it is their passports that guarantee that they are defined over @xmath5 .
we do nt have a simple way of constructing more general examples over @xmath5 ( when , e.g. , the polynomial @xmath176 has more than two critical values with prescribed positions at vertices of @xmath443 ) .
[ ex : composition-2 ] another example , based on the same function @xmath66 , is as follows .
we have @xmath456 so the white vertices of the dessin @xmath443 ( which are not shown explicitly in figure [ fig : composition ] ) are the roots of @xmath457 , that is , they are equal to @xmath458 .
now , the critical values of the polynomial @xmath459 that is , the values of @xmath287 at the roots of @xmath460 , are equal to exactly @xmath458 .
therefore , the composition @xmath461 is once again a belyi function , and all its poles except infinity are simple .
the corresponding dessin is shown in figure [ fig : composition-2 ] .
it is obvious that the dessin @xmath462 is not the only one having the passport @xmath463 .
for example , the two `` vertical '' edges can both be put above , or both can be put below the horizontal axis , or they can be cut off and attached to the leftmost white vertex of degree 2 ( the one on the loop ) , or to the rightmost one ( the one on the horizontal segment ) .
nevertheless , the dessin thus obtained is defined over @xmath5 by construction .
now the dessins @xmath448 and @xmath462 , being defined over @xmath5 , may themselves serve for a similar construction : if , for example , @xmath400 is a polynomial with coefficients in @xmath5 whose critical values are vertices of @xmath462 , then the function @xmath464 is a belyi function corresponding to a dessin @xmath465 , all of whose finite faces are small . in such a composition , only @xmath66 has to be a belyi function while the subsequent terms may have more than three critical values .
the group of the orientation preserving automorphisms of a plane tree is always cyclic .
if it is @xmath466 then the belyi function for the corresponding map is @xmath467 where @xmath66 is the belyi function for the map corresponding to a single branch of the tree ( the vertex of this branch , which will become the center of the symmetric tree , must be put to the origin ) . among the unitrees classified in section [ sec :
unitrees ] , the trees @xmath384 and and @xmath51 are symmetric ( of order 3 and 2 respectively ) .
some elements of the infinite series may also be symmetric ( for special values of parameters ) .
we leave it to the reader to see that the series @xmath131 and @xmath286 are compositions ( although the composition in this case is not reduced to a rotational symmetry ) , and that a multiplication of all the weights of the edges of a tree by a factor @xmath184 can be represented as a composition with the following belyi function : @xmath468 [ def : primitive ] a permutation group of degree @xmath18 acting on a set @xmath469 , @xmath470 , is called _ imprimitive _ if the set @xmath469 can be subdivided into @xmath95 disjoint _ blocks _ @xmath471 of equal size @xmath472 , where @xmath473 , such that an image of a block under the action of any element of the group is once again a block . a permutation group which is not imprimitive
is called _
primitive_. a primitive permutation group not equal to @xmath474 or @xmath475 is called _ special_. it is known ( see @xcite ) that a covering is a composition of two or more coverings of smaller degrees _ if and only if _
its monodromy group is imprimitive .
thus , a covering which is not a composition has a primitive monodromy group .
however , in a vast majority of cases this group is equal to either @xmath474 or @xmath475 , for a very simple reason : a permutation group generated by a randomly chosen pair of permutations is @xmath474 or @xmath475 with a probability close to 1 .
this is why special groups are of a particular interest : since _ the monodromy group is a galois invariant _ , such a group gives an additional invariant in the absence of the composition . in the case of weighted trees ,
that is , in the case of coverings realized by belyi functions with all poles except one being simple , the monodromy group must contain a permutation of the cycle structure @xmath476 : it is the monodromy permutation corresponding to a loop around infinity on the riemann sphere .
motivated by our study of weighted trees , gareth a. jones classified all special permutation groups containing such a permutation , see @xcite ( 2012 ) .
in particular , it is shown that in all such cases @xmath477 .
this property is based on two results .
the first is an old theorem by jordan ( 1871 ) @xcite stating that a primitive group containing a permutation with the cycle structure @xmath476 is @xmath478-transitive .
the second is the complete list of multiply transitive groups : it is based on the classification theorem of finite simple groups .
the classification due to jones looks as follows ( we use standard notation for projective , cyclic and affine groups and for the mathieu groups ) : [ th : jones ] let @xmath335 be a primitive permutation group of degree @xmath18 not equal to @xmath474 or @xmath475 .
suppose that @xmath335 contains a permutation with cycle structure @xmath476 .
then @xmath477 , and one of the following holds : * @xmath479 and either * * @xmath480 with @xmath481 prime , or * * @xmath482 with @xmath483 and @xmath484 for some prime power @xmath485 , or * * @xmath486 , @xmath487 or @xmath488 with @xmath489 , @xmath490 or @xmath491 respectively . * @xmath492 and either * * @xmath493 with @xmath494 and @xmath495 for some prime power @xmath485 , or * * @xmath496 or @xmath497 with @xmath498 for some prime @xmath499 , or * * @xmath500 , @xmath501 or @xmath502 with @xmath503 , @xmath504 or @xmath505 respectively . * @xmath506 and @xmath507 with @xmath508 for some prime power @xmath485 .
[ ex : special ] there are six weighted trees of degree @xmath509 with the passport @xmath510 .
one may expect that their common moduli field would be an extension of @xmath5 of degree 6 . however , this is not the case .
five trees out of six have the monodromy group @xmath511 , while the remaining one , shown in figure [ fig : primitive-1 ] ( left ) has the monodromy group @xmath512 .
therefore , this tree is defined over @xmath5 .
the other trees form a single galois orbit over a field of degree 5 . in order to establish that the group in question is indeed @xmath512
we may proceed as follows .
first , we draw the bicolored map represented by this tree , as it is done in figure [ fig : primitive-1 ] , right . then , we label the edges of the map and write down two permutations : the first one represents the cyclic order of the edges around its black vertices in the counterclockwise direction , while the second one represents the cyclic order of the edges around their white vertices in the same direction . in our case
these permutations are @xmath513 the permutation corresponding to the faces is @xmath514 so that @xmath515 .
the cycle @xmath516 can be read in the picture by going around the outer face .
note that the cycle structures of @xmath517 , @xmath518 and @xmath516 are @xmath519 , @xmath520 and @xmath521 , respectively .
then , the monodromy group is @xmath522 using maple it is easy to find out that @xmath523 , and the only transitive subgroup of @xmath511 of order 336 is @xmath512 ( see , for example , @xcite ) .
for the other five trees with the same passport the same maple package shows that the size of their monodromy group is @xmath524 , so the group in question is @xmath511 .
[ ex : special - bis ] one more example is shown in figure [ fig : primitive-2 ] .
there are five trees with the passport @xmath525 .
one of them is symmetric and therefore forms a galois orbit containing a single element and thus defined over @xmath5 .
three trees have the monodromy group @xmath511 ; they form a cubic galois orbit .
finally , the remaining tree shown in figure [ fig : primitive-2 ] has the monodromy group @xmath512 .
therefore , it forms a galois orbit in itself and is thus defined over @xmath5 . a dual to a map is usually constructed as follows .
first , one puts a new vertex inside every face of the initial map : this vertex is called `` center '' of the face . then , the centers of the adjacent faces are connected by edges in such a way that every edge of the initial map is crossed in its `` middle point '' by a new edge .
a dual of a dual is the initial map . for the _ bicolored maps _ a specific variant of the above construction
is used , when only black vertices are considered as vertices , while the white vertices play the role of the edge midpoints .
an association is thus made between the faces of the initial map and _ black vertices _ of the dual map .
the white vertices belong to both maps .
more exactly , a center of a face is connected by edges with all the white verices lying on the border of the face : see an example in figure [ fig : dual ] where the initial map is shown in an unbroken line , and its dual , in a dashed line ; the black vertices of the dual map are designated by the little squares . from the point of view of belyi functions , if @xmath122 is a belyi function for the initial map then @xmath526 is a belyi function for its dual .
indeed , @xmath527 interchanges 0 and @xmath84 while leaving 1 untouched .
therefore , the former poles become roots ( i.e. , black vertices ) , and vice versa . [
def : autodual ] a bicolored map is called _ self - dual _ if it is isomorphic to its dual map . of course
, the fact that a map is self - dual does not mean that @xmath528 where @xmath66 is its belyi function .
it means that @xmath529 where @xmath530 a linear fractional transformation of the variable @xmath49 .
the self - duality is an invariant of the galois action : if a function satisfies an algebraic relation while the other function does not satisfy the same relation , they can not belong to the same galois orbit .
a weighted tree represents a map whose all faces except one are of degree 1 .
therefore , its dual map must have all its black vertices except one being of degree 1 .
this can only happen if the dual map corresponds to a weighted tree of diameter 4 : it has a black vertex of a degree greater than 1 ( its central vertex ) , while all its black leaves are of degree 1 .
therefore , if we are interested in self - dual maps which correspond to weighted trees then we must consider only the trees of diameter 4 .
the condition on the branches of such trees in order for them to be dual to each other is shown in figure [ fig : dual - branches ] .
now we are ready to give an example where the self - duality plays the role of a galois invariant .
let us take two integers @xmath42 and @xmath44 , @xmath531 , and consider the following passport of degree @xmath532 : * there is a black vertex of degree @xmath179 ( the center ) , and @xmath533 black vertices of degree 1 ( the leaves ) ; * there are two white vertices , of degrees @xmath534 and @xmath535 respectively ; * the above data imply that the trees have @xmath179 edges , and therefore the outer face is of degree @xmath179 , the same as the degree of the central black vertex .
there are exactly @xmath534 trees with this passport .
their general appearance is shown in figure [ fig : autodual ] . here
the parameters take the following values : @xmath536 while @xmath537 among all these trees , only one is self - dual : it corresponds to the values @xmath538 , @xmath539 , @xmath540 , and @xmath541 .
therefore , this tree is defined over @xmath5 .
( in this example both branches are dual to themselves .
an attempt to make one branch dual to the other leads to the equality @xmath542 , but we have supposed that @xmath531 . ) all the five trees with the passport @xmath525 considered in example [ ex : special - bis ] are self - dual .
therefore , for them the self - duality can not serve as an additional galois invariant leading to a splitting of the combinatorial orbit into two ( or more ) galois orbits .
the world of dessins denfants is rich with various specific cases .
let us consider , for example , the set of dessins shown in figure [ fig : zannier-5 ] .
they constitute a combinatorial orbit for the passport @xmath543 where @xmath544 , @xmath545 , and @xmath546 .
we might navely suppose that this combinatorial orbit also constitutes a galois orbit ; if this were the case , this orbit would be defined over a field of degree 4 ( since it has four elements ) . however , the reality is more complicated and , in fact , more exciting .
namely : the dessin @xmath517 is the only one having a rotational symmetry of order 3 around a black vertex .
therefore , the singleton @xmath547 constitutes a galois orbit .
two dessins @xmath518 and @xmath516 are the only ones which have rotational symmetry of order 2 , the center being a white vertex ( we recall that the white vertices , being all of degree 2 , are not shown explicitly in the picture ) .
therefore , the set @xmath548 must also be taken apart from the combinatorial orbit .
there are two _ a priori _ possibilities : @xmath518 and @xmath516 may make two galois orbits , both defined over @xmath5 , or they may make a single galois orbit defined over a quadratic field . but any map whose belyi function is defined over a real field must be axially symmetric since it remains invariant under the complex conjugation .
this observation excludes the possibility of two orbits over @xmath5 , and it also excludes a real quadratic field .
we may conclude that the set @xmath548 constitutes a single galois orbit defined over an imaginary quadratic field .
the dessin @xmath184 is not symmetric and does not have any other specific combinatorial features .
but it remains solitary , and therefore it constitutes a galois orbit all by itself . since the orbits @xmath547 and @xmath549 consist of a single element ,
their belyi functions are defined over @xmath5 .
thus , the dessin @xmath184 is defined over @xmath5 for no other reason than the fact that _ it remains alone after all the other galois orbits are taken away_. this combinatorial orbit is markworthy for the reason that different authors returned to it , or to some of its elements , many times .
the belyi function for @xmath517 was computed by birch and already appeared in @xcite ( 1965 ) ; the one for @xmath184 was computed 35 years later by elkies @xcite ( 2000 ) .
all the four belyi functions were independently computed by shioda @xcite ( 2004 ) . in particular , he found out that the orbit @xmath550 is defined over the field @xmath551 .
shioda had already used as a starting point the above combinatorial orbit ; the other authors have apparently made a `` blind '' search .
our combinatorial approach does not make the computational part of the work any easier .
its advantage is elsewhere .
it consists in the fact that , before any computation , we may be sure of the following .
* there exist exactly four non - equivalent belyi functions with the passport @xmath552 ; here `` non - equivalent '' means that they can not be obtained from one another by a linear fractional change of variables .
* belyi functions corresponding to @xmath517 and @xmath184 are defined over @xmath5 .
* belyi function corresponding to @xmath517 is a rational function in @xmath553 ( because of the threefold symmetry of the dessin @xmath517 ) .
* belyi functions for the orbit @xmath548 are defined over an imaginary quadratic field .
more examples similar to this one are given below .
all the examples in this section are borrowed from the paper @xcite by beukers and stewart , which was one of the sources of inspiration for our study . in their paper , the authors consider only the case of powers of polynomials .
namely , they look for polynomials @xmath7 and @xmath8 , _ defined over _ @xmath5 , for which the degree @xmath554 attains its minimum
. the degrees of polynomials in question are @xmath555 , @xmath556 where the parameter @xmath79 may be greater than 1 .
the passport of the corresponding tree is @xmath557 .
the authors find , as we do , several infinite series of dz - triples ( which they call davenport pairs ) , and several sporadic examples .
the first such example , for which @xmath558 , corresponds to our sporadic tree @xmath385 .
the next one , @xmath559 , corresponds to the sporadic tree @xmath2 .
however , the subsequent examples do not correspond to anything we have found up to now .
what is going on ?
it turns out that here we encounter once again the phenomenon that we already explained in section [ sec : galois - split ] .
there exist two trees corresponding to the passport @xmath560 : they are shown in figure [ fig : split-7 - 3 ] .
we see that one of the trees is symmetric , with the symmetry of order 3 , while the other one is not .
therefore , this combinatorial orbit splits into two galois orbits , and hence both trees are defined over @xmath5 .
the left - hand one corresponds to the example given in @xcite .
note that an axial symmetry is not a galois invariant .
the situation for the passports @xmath561 and @xmath562 is similar to the previous one .
for the first passport there are two trees , and one of them is symmetric while the other is not ( see figure [ fig : split-8 - 3 ] ) ; therefore , both are defined over @xmath5 . for the second passport
there are three trees ( see figure [ fig : split-10 - 3 ] ) .
one of them is symmetric with the symmetry of order 2 ; one is symmetric with the symmetry of order 3 ; and one is asymmetric .
therefore , all the three trees are defined over @xmath5 . in both cases `` sporadic '' polynomials given in @xcite
correspond to asymmetric trees .
the next example given in @xcite corresponds to the passport @xmath563 .
this time , there are three trees : one of them is symmetric with the symmetry of order 2 ; another one is symmetric with the symmetry of order 4 ; the third one is asymmetric .
all the three are therefore defined over @xmath5 . for the passport @xmath564
there are four trees .
one of them is symmetric with the symmetry of order 5 ; two are symmetric with the symmetry of order 2 ; the remaining tree is asymmetric
. therefore , the combinatorial orbit containing four trees splits into three galois orbits .
the asymmetric tree corresponds to the sporadic example given in @xcite .
we leave it to the reader to draw the trees in question .
[ ex : why ] all known combinatorial invariants fail to explain why the tree with the passport @xmath565 shown in figure [ fig : why ] is defined over @xmath5 .
the trees corresponding to this passport have 13 edges , so the outer face is of degree 13 , which is prime .
this implies that this tree can not be a composition .
indeed , the outer face of a tree @xmath448 corresponding to a composition @xmath450 can only be ramified over the outer face of the tree @xmath443 , so the degree of the outer face of @xmath448 should be the product of the degree of the outer face of @xmath443 and of @xmath451 .
but 13 can not be a product of two integers .
now , the monodromy group can not be special because of theorem [ th : jones ] .
the tree can not be self - dual either since its diameter is greater than 4 , etc . for the moment
, this example is the only one of its kind .
however , one can not hope to reduce the whole body of galois theory to combinatorics .
note nevertheless that the direction of `` diophantine invariants '' ( see the next section ) for the weighted trees remains entirely unexplored .
this example is also borrowed from the paper by beukers and stewart @xcite .
there are no trees in their paper ; the dz - triple corresponding to this example , as well as several other sporadic triples , are found by a brute force computation using grbner bases .
here we discuss briefly some possibilities for further research . to begin with , there are quite a few results known for ordinary trees , which might eventually be generalized to weighted trees
. it would be very interesting to find an enumerative formula which would count the number of weighted trees having a given passport .
however , this problem may turn out to be very difficult because of the fact that the same passport can be realized by a tree and by a forest .
therefore , an inclusion - exclusion procedure might be necessary , preventing a nice closed formula of goulden
jackson s type ( see formula ( [ eq : gj ] ) ) .
right now we can prove only much more modest results , namely , enumerate the weighted trees by their weight and number of edges .
we formulate these results without proof .
let us call a tree with a distinguished edge _ edge - rooted_. [ th : generating ] let @xmath566 be the number of edge - rooted weighted bicolored plane trees of weight @xmath18 .
then the generating function @xmath567 is equal to @xmath568 the asymptotic formula for the numbers @xmath566 is @xmath569 let @xmath570 be the number of edge - rooted weighted bicolored plane trees of weight @xmath18 with @xmath95 edges .
then the generating function @xmath571 is equal to @xmath572 the following is an explicit formula for the numbers @xmath570 : @xmath573 where @xmath574 is the @xmath95th catalan number .
the sequence @xmath566 is listed in the on - line encyclopedia of integer sequences @xcite as the entry a002212 .
it has many interpretations ; the one corresponding to the weighted trees is submitted by roland bacher .
the problem is formulated as follows : _ for a given @xmath578 , classify all passports and corresponding weighted trees such that there exist exactly @xmath95 trees having this passport .
_ in our paper , we have solved this problem for @xmath579 . the following result for ordinary trees
was proved in @xcite .
it does not provide a classification , but nevertheless gives some important information concerning a general pattern for the eventual classifications . * the series of chain trees ( only for @xmath579 ) ; * a finite number of series of diameter @xmath279 ; * a finite number of series of diameter @xmath580 ; * a finite number of sporadic orbits whose elements have at most @xmath581 edges . our results for weighted trees and for @xmath579 fall into line with this pattern
, only the chains must be replaced by brushes , and the bound @xmath581 must be increased
. it would be interesting to see if a similar theorem is valid for the general case .
for the weighted trees , we have seen similar examples in the previous section : they correspond to the passports were @xmath560 and @xmath589 .
an infinite series of weighted trees with the same property is shown in figure [ fig : sym - and - not ] .
it would be interesting to produce a complete classification of such cases .
the condition of _ planarity _ , being added to theorem [ th : jones ] , imposes strong constraints on the numbers of vertices and faces of the corresponding maps . in @xcite ,
the complete list of ordinary special trees is compiled .
this list is finite : it contains 48 trees , the biggest degree being 31 .
we are pretty sure that the list of special weighted trees is also finite and can be drawn up .
the following two examples may be found in @xcite .
the first one is due to adrianov . consider the following passport for ordinary trees : @xmath590 ( here @xmath591 and @xmath584 ) .
it is easy to see that there exist exactly two trees having this passport , and neither of them is symmetric .
a simple computation shows that these trees are defined over the field @xmath592 where @xmath593 .
now , if we take , for example , @xmath594 and @xmath595 , choosing @xmath16 and @xmath409 in such a way that @xmath42 and @xmath44 become positive and not equal , we get @xmath596 to be a perfect square . therefore ,
both trees become defined over @xmath5 , and this splitting of the combinatorial orbit into two galois orbits does not have any specific combinatorial reason : it is due to certain diophantine relations between vertex degrees . once again
, it would be interesting to extend this scheme to weighted trees .
the next example is maybe the most spectacular one .
we consider the ( ordinary ) trees corresponding to the passport @xmath597 ( here @xmath598 and @xmath584 ) .
it is easy to see that there exist exactly three ordinary trees having this passport .
therefore , they are defined over a cubic extension of @xmath5 ; the cubic polynomial generating this field may be written explicitly .
now , we ask the following question : is it possible that this polynomial has a rational root ?
if yes , then the combinatorial orbit in question will split into two galois orbits , one defined over @xmath5 , and the other one quadratic .
it turns out that the search for polynomials having a rational root can be reduced to the search for rational points on a particular elliptic curve .
the curve in question contains infinitely many rational points .
we have computed the first 11 solutions .
the smallest one corresponds to trees having @xmath599 edges ( @xmath600 , @xmath601 ) ; the 11th solution corresponds to trees having @xmath602 edges .
similar constructions certainly must also exist for weighted trees .
let us revisit the initial problem about the minimum degree of the polynomial @xmath603 , see page .
when there are no dz - triples defined over @xmath5 , we may relax the condition of the @xmath61 being the least possible and thus obtain solutions with bigger @xmath61 but , in return , defined over @xmath5 .
two example of this kind are shown in figures [ fig : deg - relaxed ] and [ fig : deg9-instead-8 ] . in the first one ,
@xmath604 but @xmath605 instead of 7 since one of the faces is of degree 3 instead of 1 . in the second example , @xmath606 but @xmath605 instead of 8 since , instead of two black vertices of degree 3 we have here one black vertex of degree 6 .
now let us look at the second example .
though a computation of the belyi function in this case is not difficult , it is still interesting to analyze this example in purely combinatorial terms .
the map shown on the right in figure [ fig : deg9-instead-8 ] is a unimap @xmath171 of figure [ fig : trees2maps ] , which is also equal to the unitree @xmath386 in figure [ fig : sporadic-8 ] .
therefore , it is defined over @xmath5 .
its black vertex of degree 2 is a bachelor ( definition [ def : bachelor ] ) ; therefore , it can be placed at any rational position ( remark [ rem : bachelor ] ) , for example , at the point @xmath607 .
then it remains to insert @xmath553 instead of @xmath49 in its belyi function , and we get a belyi function for the bigger `` triple '' dessin .
this example shows that the possibilities of the combinatorial approach to this problem are far from being exhausted . besides the maps with all finite faces of degree 1 , there are other classes of maps for which the question of uniqueness is interesting .
one such example is the class of maps with exactly two faces : their belyi functions are laurent polynomials .
the existence questions for such maps were completely settled in @xcite ; the uniqueness remains to be studied .
some new galois phenomena related to the non - existence of bachelors appear in this case , see @xcite .
fedor pakovich is grateful to the bordeaux - i university , france , and alexander zvonkin is grateful to the center for advanced studies in mathematics of the ben - gurion university of the negev , israel , for their mutual hospitality .
fedor pakovich is also grateful to the max - planck - institut fr mathematik , bonn , where the most part of this paper was written .
rational points near curves and small non - zero @xmath612 via lattice reduction , in : wieb bosma , ed . ,
`` algorithmic number theory '' , lect .
notes in comp .
, vol . * 1838 * , springer - verlag , 2000 , 3363 .
surface branched covers and geometric 2-orbifolds , _ trans .
soc . _ , 2009 , vol . * 361 * , no .
11 , 58855920 .
the grothendieck theory of dessins denfants .
london math .
lecture notes series , vol .
* 200 * , cambridge univ .
press , 1994 . | a _ weighted bicolored plane tree _ ( or just _ tree _ for short ) is a bicolored plane tree whose edges are endowed with positive integral weights .
the degree of a vertex is defined as the sum of the weights of the edges incident to this vertex .
using the theory of _ dessins denfants _ , which studies the action of the absolute galois group on graphs embedded into riemann surfaces ,
we show that a weighted plane tree is a graphical representation of a pair of coprime polynomials @xmath0 $ ] such that : ( a ) @xmath1 , and @xmath2 and @xmath3 have the same leading coefficient ; ( b ) the multiplicities of the roots of @xmath2 ( respectively , of @xmath3 ) are equal to the degrees of the black ( respectively , white ) vertices of the corresponding tree ; ( c ) the degree of the difference @xmath4 attains the minimum which is possible for the given multiplicities of the roots of @xmath2 and @xmath3 .
moreover , if a tree in question is uniquely determined by the set of its black and white vertex degrees ( we call such trees _ unitrees _ ) , then the corresponding polynomials are defined over @xmath5 .
the pairs of polynomials @xmath6 such that the degree of the difference @xmath4 attains the minimum , and especially those defined over @xmath5 , are related to some important questions of number theory .
dozens of papers , from 1965 @xcite to 2010 @xcite , were dedicated to their study .
the main result of this paper is a complete classification of the _ unitrees _ which provides us with the most massive class of such pairs defined over @xmath5 .
we also study combinatorial invariants of the galois action on trees , as well as on the corresponding polynomial pairs , which permit us to find yet more examples defined over @xmath5 . in a subsequent paper
we compute the polynomials @xmath6 corresponding to all the unitrees . |
in this article we consider the following linear stochastic partial differential equation of fractional orders both in time and space variables : @xmath4 \displaystyle \left.\frac{\partial^k}{\partial t^k } u(t , x)\right|_{t=0}=u_k(x ) , & \qquad 0\le k\le { \left\lceil \beta \right\rceil}-1 , \:\ : x\in\rr^d , \end{cases}\end{aligned}\ ] ] with @xmath5 and @xmath3 $ ] , where @xmath6 is the smallest integer greater than or equal to @xmath7 . here and throughout the paper we denote @xmath8 , @xmath9 .
we limit our consideration to the above parameter ranges of @xmath7 and @xmath10 since we plan to use some particular properties of the corresponding _ fox h - functions _ which will be proved only for these parameter ranges .
now let us give more detailed explanation on the terms appearing in the above equation .
the fractional derivative in time @xmath11 is understood in the _ caputo _ sense : @xmath12 \displaystyle \frac{{\ensuremath{\mathrm{d}}}^m}{{\ensuremath{\mathrm{d}}}t^m}f(t ) & \text{if $ \beta = m$}\ , , \end{cases}\ ] ] where @xmath13 .
@xmath14 is the laplacian with respect to spatial variables and @xmath15 is the fractional laplacian .
@xmath16 is a zero mean gaussian noise with the following covariance structure @xmath17 where both ( possibly generalized ) functions @xmath18 and @xmath19 are assumed to be nonnegative and nonnegative definite .
we denote by @xmath20 the fourier transformation measure of @xmath21 .
namely , @xmath22 this fourier transform is understood in distributional sense ( see section 2 ) .
when @xmath23 and @xmath24 , this noise @xmath1 reduces to the _ space - time white noise_. @xmath25 and @xmath26 are some real valued parameters .
the given initial conditions @xmath27 are assumed to be continuous and bounded functions .
the product @xmath28 in equation is the wick one ( see e.g. @xcite ) .
so , the equation will be understood in the skorohod sense .
let us point out that some of our results can also be extended to nonlinear equation ( namely , replace @xmath28 in by @xmath29 for a lipschitz nonlinear function @xmath30 ) .
however , we limit ourselves to this linear case for two reasons : one is to simplify the presentation and to better explain the ideas and the other one is that we want to use the chaos expansion method .
the deterministic counterparts of equation have received a lot of attention and are called anomalous diffusions .
they appeared in biological physics and other fields .
equation is an anomalous diffusion in a gaussian noisy environment .
more detailed motivations for the study of this type of equations are given in @xcite .
here , we briefly mention some recent results . when @xmath31 is an integer , @xmath32 , the equation has been studied by many authors , see for example , @xcite .
the work by chen and dalang @xmath33 deals with the case where @xmath34 , @xmath35 $ ] .
khoshnevisan and foondun @xcite and song @xcite study a similar equation with the @xmath10-stable generator @xmath36 replaced by a general lvy generator .
when @xmath37 , @xmath38 , @xmath39 is replaced by a general elliptic operator , and @xmath1 is a fractional noise , the equation was studied in @xcite . when @xmath37 , @xmath38 and @xmath1 is a fractional noise , the smoothed equation @xmath40\ ] ] ( see for a generalization )
was studied in @xcite . in a series of papers @xcite , nane and
his coauthors studied the case @xmath41 $ ] .
the case @xmath37 corresponds to the slow diffusion ( subdiffusion ) . for the fast diffusion case ( super diffusion ) ,
i.e. , @xmath42 , there have been only a few works . the first author of this paper studies in @xcite the smoothed equation with @xmath32 , @xmath43 and with space - time white noise .
the corresponding non - smoothed equation is studied recently in @xcite .
both papers @xcite deal with the nonlinear equation , i.e. , @xmath44 with @xmath45 being a lipschitz function . to study equation the important tools that we shall use are the fundamental solutions corresponding to its deterministic counterpart .
let us briefly describe them .
there are two fundamental solutions @xmath46 such that the solution @xmath47 to the following deterministic equation ( the deterministic counterpart of ) @xmath48 \displaystyle \left.\frac{\partial^k}{\partial t^k } u(t , x)\right|_{t=0}=u_k(x ) , & \qquad 0\le k\le { \left\lceil \beta \right\rceil}-1 , \:\ : x\in\rr^d , \end{cases}\end{aligned}\ ] ] is represented by @xmath49 where @xmath50 is a continuous and bounded function on @xmath51 , and @xmath52 is a continuous and bounded function on @xmath53 . in equation and throughout the paper , we denote @xmath54 equations and say that @xmath55 and @xmath56 are the fundamental solutions corresponding respectively to the initial conditions and the inhomogeneous term of equation . for some parameter ranges of @xmath10 and @xmath7 ,
the fundamental solutions have been studied in @xcite . in section 3.1
we shall study them for all @xmath5 and @xmath3 $ ] .
in particular , we shall obtain some new properties such as the positivity of the fundamental solutions @xmath56 and @xmath55 .
equation motivates us to study the mild solution to ( see e.g. definition [ d : sol ] below ) , namely , the solution to the following stochastic integral equation : @xmath57 as in the classical case , the above equation can be studied by using the it - wiener chaos expansion .
to this end we need to understand well the two fundamental solutions @xmath55 and @xmath56 .
in particular , we need their nonnegativity and some heat kernel like estimates . the nonnegativity of some @xmath55 s is known . however , since @xmath56 is the _ riemann - liouville _ fractional derivative of @xmath55 , its nonnegativity is a challenging problem .
there have been only few results : as proved in lemma 25 of @xcite , @xmath58 with @xmath42 is nonnegative if and only if @xmath59 . the one dimensional case is proved in @xcite , namely , @xmath60 , and hence @xmath61 , is nonnegative either if @xmath62 , or if @xmath63 $ ] and @xmath64 . in this paper , we shall show the nonnegativity of @xmath56 in the following three cases : @xmath65,\quad & & \beta\in ( 1/2,1 ) , \quad & & d\in\nn , \\
& \alpha\in ( 0,2],\quad & & \beta\in ( 1,\alpha ) , \quad & & d=1,\\ & \alpha=2 , \quad & & \beta\in ( 1,2 ) , \quad & & d=2,3 .
\end{alignedat } \right.\ ] ] this includes the above mentioned results as special cases .
let us also point out that for the smoothed spde , only the fundamental solution @xmath55 is needed , which is usually more regular than the fundamental solution @xmath56 .
when @xmath66 and @xmath38 , to show the solution of is square integrable , it is assumed in @xcite and @xcite that the covariance of noise satisfies the following conditions : ( i ) @xmath67 is locally integrable ; \(ii ) @xmath68 is satisfied ( see also @xcite ) .
for the existence and uniqueness of the solution to the general equation , dalang s condition will be replaced by the following condition : @xmath69 it is obvious that if it is formally set @xmath38 and @xmath66 , then is reduced to the usual dalang s condition .
the remaining part of the paper is organized as follows .
we first specify the noise structure and present the definition of the solution in section [ sec : pre ] .
the main results are theorem [ t : exuni ] on the existence and uniqueness of the mild solution and theorem [ t : rietz ] on the moment bounds of the solution stated in section [ sec : main ] .
the proof of these two theorems are based on some properties of the fundamental solutions represented in terms of the fox h - functions .
these results themselves are of particular interest and importance .
we also list them as theorem [ t : pde ] and theorem [ t : stable ] in section [ sec : main ] .
the properties of the fundamental solutions ( theorem [ t : pde ] ) are proved in section [ sec : pde ] by using the fox h - functions . in section [ sec : stable ] , we obtain an expression of the density function for the @xmath70-dimensional spherically symmetric @xmath10-stable distribution - an auxiliary result ( theorem [ t : stable ] ) which is used in the proof of theorem [ t : pde ] .
the existence and uniqueness result ( theorem [ t : exuni ] ) of the solution to is proved in section [ sec : exuni ] . in section [ sec : rietz ] , we prove the explicit moment bounds when @xmath19 is the riesz kernel .
our main results ( theorem [ t : exuni ] ) assume that the fundamental solutions are nonnegative . however , when @xmath71 and when the dimension is high , the nonnegativity of the fundamental solution @xmath56 is not known yet . in this case , we shall show in theorem [ beta.2 ] the existence and uniqueness of the solution of for some specific gaussian noise whose covariance function @xmath19 is the riesz kernel .
finally , in appendix [ sec : h ] we collect some knowledge on the fox h - function which we need in this paper .
let us start by introducing some basic notions on fourier transforms .
the space of real - valued infinitely differentiable functions on @xmath53 with compact support is denoted by @xmath72 or @xmath73 .
the space of schwartz functions is denoted by @xmath74 or @xmath75 .
its dual , the space of tempered distributions , is denoted by @xmath76 or @xmath77 .
the fourier transform is defined with the normalization @xmath78 so that the inverse fourier transform is given by @xmath79 .
similarly to @xcite , on a complete probability space @xmath80 we consider a gaussian noise @xmath81 encoded by a centered gaussian family @xmath82 , whose covariance structure is given by @xmath83 where @xmath84 and @xmath85 are nonnegative definite functions and the fourier transform @xmath86 such that @xmath87 is a tempered measure , that is , there is an integer @xmath88 such that @xmath89 . throughout the paper ,
we assume that @xmath67 is locally integrable and we denote @xmath90 let @xmath91 be the completion of @xmath92 endowed with the inner product @xmath93 where @xmath94 refers to the fourier transform with respect to the space variable only .
the mapping @xmath95 defined on @xmath92 can be extended to a linear isometry between @xmath91 and the gaussian space spanned by @xmath81 .
we will denote this isometry by @xmath96 notice that if @xmath97 and @xmath98 are in @xmath91 , then @xmath99 .
we will denote by @xmath100 the derivative operator in the sense of malliavin calculus .
that is , if @xmath101 is a smooth and cylindrical random variable of the form @xmath102 with @xmath103 , @xmath104 ( namely @xmath50 and all its partial derivatives have polynomial growth ) , then @xmath105 is the @xmath91-valued random variable defined by @xmath106 the operator @xmath100 is closable from @xmath107 into @xmath108 and we define the sobolev space @xmath109 as the closure of the space of smooth and cylindrical random variables under the norm @xmath110+\ee[\|df\|^2_{\mathcal{h}}]}\,.\ ] ] we denote by @xmath111 the adjoint of the derivative operator given by the duality formula @xmath112 for all @xmath113 and any element @xmath114 in the domain of @xmath111 .
the operator @xmath111 is also called the _ skorohod integral _ because in the case of the brownian motion , it coincides with an extension of the it integral introduced by skorohod .
we refer to nualart @xcite for a detailed account of the malliavin calculus with respect to a gaussian process . with
the skorohod integral introduced , the definition of the solution to equation can be stated as follows .
[ d : sol ] let @xmath55 and @xmath56 be the fundamental solutions defined by and .
an adapted random field @xmath115 such that @xmath116<+\infty$ ] for all @xmath117 is a _ mild solution _ to , if for all @xmath118 , the process @xmath119}(s ) : \ :
s\ge0,\ : y\in \mathbb{r}^d \right\}\ ] ] is skorohod integrable ( see ) , and @xmath120 satisfies @xmath121 almost surely for all @xmath122 , where @xmath123 is defined by . the main ingredient in proving the existence and uniqueness of the solution is the wiener chaos expansion , to which we now turn .
suppose that @xmath124 is a square integrable solution to equation .
then for all fixed @xmath117 the random variable @xmath125 admits the following wiener chaos expansion @xmath126 where for each @xmath117 , @xmath127 is a symmetric element in @xmath128 .
then , as in @xcite , to show the existence and uniqueness of the solution it suffices to show that for all @xmath117 we have @xmath129 for technical reasons
( see below ) , we will assume , throughout the paper , the following properties on @xmath19 : * @xmath130 $ ] is a continuous function , where @xmath131 $ ] is the usual one - point compactification of @xmath132 . * @xmath133 if and only if @xmath134 or @xmath135 and @xmath133 when @xmath134 . with these two assumptions , according to lemma 5.6 in @xcite , for any borel probability measures @xmath136 and @xmath137 , the following identity holds , @xmath138 in particular , the above result can be applied to the case when @xmath139 and @xmath140 for two nonnegative functions @xmath141 and @xmath142 .
our first result is concerned with the fundamental solutions to stated in the following theorem .
we need the two parameter _ mittag - leffler function _ @xmath143 : @xmath144 where @xmath145 is the real part of the complex number @xmath10 .
when @xmath146 , we also write @xmath147 . the @xmath148-functions appearing in the following theorem and their properties
are given in the appendix .
[ t : pde ] the fundamental solutions to are given by @xmath149 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right)}\end{aligned}\ ] ] and @xmath150 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right)}\,.\end{aligned}\ ] ] if @xmath151 , then @xmath152 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right)}.\end{aligned}\ ] ] the fourier transforms of the fundamental solutions are given by the following : @xmath153 moreover , we have the following results on the positivity of the fundamental solutions .
a. if @xmath154 $ ] , @xmath155 and @xmath156 $ ] , then both @xmath157 and @xmath158 are nonnegative ; b. if @xmath42 , @xmath159 , and @xmath32 , then both @xmath157 and @xmath158 are nonnegative ; c. if @xmath42 , @xmath43 and @xmath160 $ ] , then all @xmath157 , @xmath158 and @xmath161 are nonnegative . the proof of this theorem is given in section [ sec : pde ] . here are some known special cases : 1 .
when @xmath32 and @xmath162 , it is proved in @xcite and in @xcite , respectively , that @xmath163 ( d/2,1),(1,1)\end{subarray}\right)},\end{aligned}\ ] ] and @xmath164 ( d/2,1),(1,1)\end{subarray}\right)},\end{aligned}\ ] ] which correspond to our @xmath165 and @xmath166 , respectively .
the equivalence is clear by applying property [ prop : reduction ] . for @xmath167 , see also ( * ? ? ?
* chapter 6 ) .
when @xmath32 and @xmath168 , it is proved in @xcite that @xmath169 ( d/2,1),(1,1)\end{subarray}\right)},\end{aligned}\ ] ] which corresponds to our @xmath58 with @xmath170 .
3 . in @xcite , the fundamental solution
@xmath171 has been studied for all @xmath172 and @xmath43 . from the mellin - barnes
integral representation ( 6.6 ) of @xcite , we see that the reduced green function of @xcite can be expressed by using the fox h - function : @xmath173 ( 1,1),\:(1,\alpha),\:(1,\frac{\alpha-\theta}{2})\end{subarray}\right ) } , \quad x\in\rr,\end{aligned}\ ] ] where @xmath10 and @xmath7 have the same meaning as in this paper and @xmath174 is the skewness : @xmath175 . for the symmetric @xmath10-stable case , i.e. , @xmath176
, this expression can be simplified by using the definition of the fox h - function and the fact that ( see , e.g. , @xcite ) @xmath177 hence , @xmath178 ( 1/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right)},\quad x\in\rr.\end{aligned}\ ] ] this implies that the fundamental solution in ( * ? ? ? * ( 1.3 ) ) @xmath179 ( 1/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right)}\ ] ] corresponds to our @xmath180 with @xmath170 .
the proof of the nonnegativity part in theorem [ t : pde ] requires a representation of the spherically symmetric @xmath10-stable distribution from the fox h - function , which is of interest by itself .
the one - dimensional case can be found in @xcite ; see remark [ r : stable ] below .
[ t : stable ] let @xmath181 be a centered , @xmath70-dimensional spherically symmetric @xmath10-stable random variable with @xmath182 $ ] .
then the characteristic function and the density of @xmath181 are , respectively , @xmath183 and @xmath184 ( d/2,\alpha/2),\:(1,\alpha/2)\end{subarray}\right)},\quad x\in\rr^d.\end{aligned}\ ] ] the proof of this theorem is given in section [ sec : stable ] .
[ r : stable ] when @xmath43 , the formula yields a result in @xcite . in particular , as proved in @xcite ( see ) , when @xmath43 , we have @xmath185 ( 1,\alpha),\:(1,\alpha/2)\end{subarray}\right)}= \pi^{-1/2 } |x|^{-1 } { h_{1,2}^{1,1}\left((|x|/2)^\alpha\:\middle\vert\ : \begin{subarray}{l}(1,1)\\[0.4em ] ( 1/2,\alpha/2),\:(1,\alpha/2)\end{subarray}\right)},\ ] ] where the second equality is due to and the definition of the fox h - function .
the following is one of the main theorems of the paper .
[ t : exuni ] assume the following conditions : 1 .
@xmath186 is nonnegative ; 2 .
@xmath187 and @xmath182 $ ] ; 3 .
@xmath18 is locally integrable ; 4 .
@xmath20 satisfies dalang s condition ; 5 .
the initial conditions are such that for all @xmath188 , @xmath189 } |j_0(s , y)|<+\infty.\end{aligned}\ ] ] then relation holds for each @xmath117 .
consequently , equation admits a unique mild solution in the sense of definition [ d : sol ] . the proof of this theorem is given in section [ sec : exuni ] . from theorem [ t : pde ]
, it follows that the three cases in satisfy the above assumptions ( 1 ) and ( 2 ) .
moreover , if @xmath190 for the first two cases in , or if @xmath191 and @xmath192 is a constant for the last case in , then by lemma [ l : j0 ] below , @xmath193 hence the assumption ( 5 ) is also satisfied .
the dalang condition imposes a further restriction on the possible values of @xmath194 due to the spatial correlation @xmath21 .
when the noise @xmath16 is a space - time white noise , i.e. , @xmath23 and @xmath24 , then dalang s condition becomes @xmath195 this condition implies that @xmath196 . in particular ,
if @xmath32 and @xmath43 , then reduces to @xmath197 which recovers the condition in @xcite and ( * ? ? ?
* section 5.2 ) .
if @xmath34 and @xmath43 , then this condition becomes @xmath198 which recovers the condition in @xcite .
the methodology used in the proof of theorem [ t : exuni ] can also be used to study the following equation @xmath199\,,\ ] ] with the same initial conditions as . here
@xmath200 is the _ riemann - liouville fractional integral _ of order @xmath7 ( with an abuse of the notation @xmath7 ) : @xmath201 due to the fractional integral in equation which plays a smoothing role , the mild formulation for the solution can be expressed by using @xmath157 only , namely , @xmath202 then , using the same procedure as in the proof of theorem [ t : exuni ] , we have the following result .
[ t : exuni - i ] assume the conditions ( 3 ) and ( 5 ) in theorem [ t : exuni ] and the other conditions are replaced by the following : 1 .
@xmath203 is nonnegative ; 2 .
@xmath204\cup ( 3/2,2)$ ] and @xmath182 $ ] ; 3 .
@xmath20 satisfies @xmath205 then relation holds for each @xmath117 .
consequently , the smoothed equation admits a unique mild solution in the sense of definition [ d : sol ] with @xmath56 replaced by @xmath55 .
the condition ( 1 ) is different and is usually easier to verify than the condition ( 1 ) in theorem [ t : exuni ] . when @xmath206 $ ] , the condition becomes @xmath207 which is also weaker than ( since @xmath208 ) .
when @xmath209 , the condition becomes @xmath210 which is also weaker than ( since @xmath211 ) .
the proof is essentially the same as that for theorem [ t : exuni ] , the only change in the proof worthy to be pointed out is that instead of computing the integral @xmath212 we now need to compute the integral @xmath213 the integrability condition of the above equation at zero and at infinity implies that @xmath214 and @xmath204\cup ( 3/2,2]$ ] ( which is equivalent to @xmath215 ) , respectively .
note that this condition on @xmath7 is more restrictive than the condition @xmath168 in @xcite .
when the noise @xmath16 is a space - time white noise(namely @xmath216 ) , then dalang s condition becomes @xmath217 in particular , if @xmath32 and @xmath43 , then this condition reduces to @xmath218 .
if @xmath34 and @xmath43 , then this condition becomes , which recovers the condition in @xcite .
in this subsection we give some upper bounds for the @xmath219-th moment and the lower bound of the second moment of the solution for some specific choice of the covariance kernel .
[ t : rietz ] assume the following conditions : 1 .
the initial conditions satisfy condition ( 5 ) of theorem [ t : exuni ] ; 2 .
@xmath220 satisfies one of the three conditions in ; 3 .
@xmath221 , @xmath222 with @xmath223 then the solution @xmath224 to satisfies that for all @xmath225 , @xmath226 \leq c^p \widehat{c}_t^p \exp \left(t ( c_{\kappa } c_t \widetilde{c } c _ * ( { 2}/{\nu})^{\kappa/\alpha } ( 2\pi)^{-d})^{\frac{\alpha}{2\alpha \beta- \alpha -\beta \kappa } } p ^{\frac{2\alpha\beta-\beta \kappa}{2\alpha \beta -\alpha -\beta \kappa } } \right)\ , , \label{e.p - moment}\ ] ] where @xmath227 and @xmath228 are defined in , @xmath229 , and @xmath230 and @xmath231 appears in the fourier transform of @xmath232 , i.e. , @xmath233 . in particular , if @xmath18 is the dirac delta function and if the initial data @xmath234 is a constant and @xmath235 when @xmath236 , then for some constant @xmath237 , @xmath238 \geq c \ : { u_0 ^ 2 } \exp \left ( t \ : ( c_{\kappa } \widetilde{c } ( 4\pi)^{-d } c_*\ : ( 2/\nu)^{\kappa/\alpha})^{\frac{1}{2\beta -1 -\beta \kappa/\alpha } } \right)\,.\ ] ] the proof of this theorem is given in section [ sec : rietz ] .
the same method can be used to obtain the moment bound for the solution to the smoothed equation .
when @xmath34 and @xmath32 , equation is reduced to the multiplicative stochastic heat equation ( 1.1 ) considered in @xcite . in this case the exponent of @xmath219 in becomes @xmath239 which is the same as in ( * ? ? ?
* theorem 6.1 , inequality ( 6.1 ) ) ( with @xmath240 ) .
if we assume @xmath241 , then @xmath242 .
the exponent of @xmath243 in is @xmath244 which is the same exponent of @xmath243 as in @xcite , inequality ( 6.1 ) .
hence , we conjecture that the bound is sharp
. assume the conditions ( 1 ) and ( 2 ) of theorem [ t : rietz ] , and assume 1 .
@xmath221 , @xmath222 with @xmath245 then the solution @xmath224 to the smoothed equation satisfies that for all @xmath225 , @xmath246 \leq
c^p \widehat{c}_t^p \exp \left(t \ : \left[c_{\kappa } c_t \bar{c } c_{\ # } ( 2/\nu)^{\kappa/\alpha } \right]^{\frac{\alpha}{2 \alpha { \left\lceil \beta \right\rceil } -\alpha -\beta \kappa } } p^{\frac{2\alpha { \left\lceil \beta \right\rceil } -\beta \kappa}{2 \alpha { \left\lceil \beta \right\rceil}- \alpha -\beta \kappa } } \right)\,,\ ] ] where @xmath229 , @xmath228 is defined in , @xmath247 and @xmath231 is as defined in theorem [ t : rietz ] . in particular , if @xmath18 is the dirac delta function and if the initial data @xmath234 is a constant and @xmath235 when @xmath236 , then for some constant @xmath237 , @xmath248 \geq c \ : { u_0 ^ 2 } \exp \left ( t\ : \left[c_{\kappa } \bar{c } ( 4\pi)^{-d } c_{\ # } ( 2/\nu)^{\kappa/\alpha } \right]^{\frac{1}{2\beta -1 -\beta \kappa/\alpha } } \right)\ , . \ ] ] the proof of this theorem is a line - by - line change of the proof of theorem [ t : rietz ] , and we leave it to the interested reader . when @xmath249 and @xmath251 , we could not show the nonnegativity of @xmath252 for high dimension ( @xmath250 ) ( see theorem [ t : pde ] ( b ) ) .
however , with a slightly different approach , it is possible to obtain similar results to theorem [ t : exuni ] for the riesz kernel case .
[ beta.2 ] assume the conditions ( 2 ) , ( 3 ) and ( 5 ) of theorem [ t : exuni ] , and assume 1 .
@xmath221 , @xmath222 with @xmath253 then relation holds for each @xmath117 .
consequently , equation admits a unique mild solution in the sense of definition [ d : sol ] . this theorem is proved in section [ sec:1beta2 ] .
it is easy to see that the condition @xmath221 with @xmath254 implies dalang s condition .
condition @xmath255 is to guarantee that @xmath256 is a positive function .
the proof of theorem [ t : pde ] will be based on following lemmas .
[ l : fz ] the function @xmath203 has the fourier transform given by .
the proof needs relation between the mittag - leffler function and the fox h - function .
we first note some special cases .
the case where @xmath257 $ ] , @xmath32 and @xmath155 can be found in ( * ? ? ?
* section 4 ) or @xcite . for @xmath257
$ ] and for general @xmath258 , one can simply replace @xmath259 by @xmath260 in the argument of ( * ? ? ?
* section 4 ) and then use to obtain .
the case where @xmath43 , @xmath168 , and @xmath261 is proved by @xcite . for the general case , @xmath262 solves @xmath263 hence , the fourier transform of @xmath262 satisfies @xmath264 this equation can be solved explicitly ( see , e.g. , ( * ? ? ?
* theorem 7.2 , on p. 135 ) ) as @xmath265 which gives immediately when @xmath266 .
when @xmath267 , the integral can be evaluated by ( * ? ? ?
* ( 1.99 ) ) to give @xmath268 this completes the proof of lemma [ l : fz ] . [ l : zexp ] the function @xmath203 can be expressed in .
following lemma [ l : fz ] , we need to compute the inverse fourier transform of . instead of finding the inverse fourier transform
, it turns out that it is easier to verify that the fourier transform of is equal to the right hand side of .
let now @xmath55 be defined by .
* case i @xmath43.*notice that @xmath269 is an even function .
we have that @xmath270 ( 1/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right ) } \cos(x \xi).\ ] ] write the @xmath271 function in the fox h - function form by and then apply property [ prop : power ] to both fox h - functions and property [ prop : frac ] to the fox h - function coming from @xmath271 : @xmath272 ( 1/2,1/2),\:(1,1/\alpha),\:(1,1/2)\end{subarray}\right)}\\ & \times { h_{2,0}^{0,1}\left(\frac{2}{x |\xi|}\:\middle\vert\ : \begin{subarray}{l}(1,1/2),\;(1/2,1/2)\\[0.4em ] \midrule \end{subarray}\right)}.\end{aligned}\ ] ] now we will apply theorem [ t : hconvh ] .
notice that both condition ( 1 ) : @xmath273 , @xmath274 , @xmath275 , and condition : @xmath276 of theorem [ t : hconvh ] hold .
hence , theorem [ t : hconvh ] implies that @xmath277 ( 1/2,1/2),\ : ( 1,1/\alpha),\:(1,1/2)\end{subarray}\right)}.\ ] ] then apply properties [ prop : power ] and [ prop : reduction ] to simplify the above expression : @xmath278 ( 1/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right)}\\ & = t^{{\left\lceil \beta \right\rceil}-1}{h_{1,2}^{1,1}\left(2^{-1}\nu t^\beta |\xi|^\alpha\:\middle\vert\ : \begin{subarray}{l}(0,1)\\[0.4em ] ( 0,1),\:(1-{\left\lceil \beta \right\rceil},\beta)\end{subarray}\right)}\,.\end{aligned}\ ] ] this proves the lemma when @xmath43 .
* case ii @xmath250.*because the function @xmath279 is a radial function , it is known that ( see , e.g. , ( * ? ? ?
* theorem 3.3 on p. 155 ) ) , @xmath280 ( 1/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right ) } j_{(d-2)/2}(x|\xi| ) ( |\xi| x)^{-d/2},\ ] ] where @xmath281 is the bessel function of the first kind
. then we can apply theorem [ t : hankel ] .
similar to the previous case , all conditions are satisfied with the condition @xmath282 .
hence , @xmath283 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2 ) \end{subarray}\right)}\\ & = t^{{\left\lceil \beta \right\rceil}-1 } { h_{1,2}^{1,1}\left(2^{-1}\nu t^\beta |\xi|^\alpha\:\middle\vert\ : \begin{subarray}{l}(0,1)\\[0.4em ] ( 0,1),\:(1-{\left\lceil \beta \right\rceil},\beta)\end{subarray}\right)}\,,\end{aligned}\ ] ] where the second equality is due to property [ prop : reduction ] .
this completes the proof of lemma [ l : zexp ] .
note that the case @xmath250 can be also proved by writing the bessel function in the fox h - function form by and then applying theorem [ t : hconvh ] .
we leave the details for interested readers .
[ l : yexp ] the fundamental solutions @xmath186 and @xmath171 are given by and , respectively .
we first prove the expression for @xmath284 . by section 2 of @xcite , we know that @xmath186 is the riemann - liouville fractional derivative in @xmath243 of @xmath203 of order @xmath285 .
notice that @xmath286 for @xmath287 .
denote the _ riemann - liouville derivative _ of order @xmath162 by @xmath288 , i.e. , @xmath289 by property [ prop : frac ] , @xmath290 ( 0,1),\:(1-{\left\lceil \beta \right\rceil},\beta)\end{subarray}\right)}\,.\ ] ] because @xmath291 , we can apply theorem [ t : rl - d ] , @xmath292 ( 0,1),\:(1-{\left\lceil \beta \right\rceil},\beta),\:(1-\beta,\beta)\end{subarray}\right)}.\end{aligned}\ ] ] then we use properties [ prop : reduction ] and [ prop : power ] to simplify the above expression to obtain .
the expression for @xmath293 can be proved in a similar way .
[ l : fy ] the fourier transforms of @xmath186 and @xmath171 are given by and , respectively .
we first consider @xmath284 . from lemma
[ l : fz ] and the proof of lemma [ l : yexp ] it follows @xmath294 ( 0,1),\:(1-{\left\lceil \beta \right\rceil},\beta)\end{subarray}\right)}\,.\ ] ] because @xmath282 , we can apply theorem [ t : rl - d ] to obtain that @xmath295 ( 0,1),\:(1-{\left\lceil \beta \right\rceil},\beta),\:(1-\beta,\beta)\end{subarray}\right)}\,.\ ] ] this is simplified to by properties [ prop : reduction ] and [ prop : power ] .
the identity can be obtained in a similar way .
[ l_:nonneg ] for all @xmath296 and @xmath297 , the following h - function is nonnegative : @xmath298 ( 1,1)\end{subarray}\right)}\ge 0\ , , \quad \text{for all $ \
x\in \rr$}\ , .
\end{aligned}\ ] ] we only need to prove that the following function is nonnegative @xmath299 ( 1,1)\end{subarray}\right ) } , \quad x\in\rr.\ ] ] by theorem [ t : laplace ] and equation , the laplace transform of @xmath50 is equal to @xmath300 by @xcite , we know that the above mittag - leffler function @xmath301 is completely monotonic if and only if @xmath302
. then the bernstein theorem ( see , e.g. , ( * ? ? ? * theorem 12a ) ) implies that the function @xmath303 is nonnegative .
[ l : nonneg ] the nonnegative statements in theorem [ t : pde ] hold true .
we first prove the case ( a ) . in this case ,
@xmath154 $ ] .
when @xmath34 , @xmath55 and @xmath56 coincide and they are alpha stable densities .
hence , we need only consider the case @xmath162 . because @xmath304 for all @xmath287 , and noticing that @xmath305 ( see , e.g. , theorem 3.8 of @xcite ) , we see that @xmath306 hence , it suffices to show the nonnegativity of @xmath186 .
notice that by property [ prop : power ] , @xmath307 ( d/2,\alpha/2),\:(1,\alpha/2)\end{subarray}\right ) } { h_{1,1}^{1,0}\left((ts)^{-\beta}\:\middle\vert\ : \begin{subarray}{l}(\beta,\beta)\\[0.4em ] ( 1,1)\end{subarray}\right)}\\ = & \beta^{-2 } \int_0^\infty\frac{{\ensuremath{\mathrm{d}}}s}{s}\ : { h_{1,2}^{1,1}\left(\frac{|x|^{\alpha/\beta}}{2^{(\alpha-1)/\beta}\nu^{1/\beta}}s\:\middle\vert\ : \begin{subarray}{l}(1,1/\beta)\\[0.4em ] ( d/2,\alpha/2\beta),\:(1,\alpha/2\beta)\end{subarray}\right ) } { h_{1,1}^{1,0}\left(\frac{1}{ts}\:\middle\vert\ : \begin{subarray}{l}(\beta,1)\\[0.4em ] ( 1,1/\beta)\end{subarray}\right)}.\end{aligned}\ ] ] now we check conditions in theorem [ t : hconvh ] .
condition ( 1 ) of theorem [ t : hconvh ] is satisfied because @xmath308 and @xmath309 .
condition of theorem [ t : hconvh ] holds because @xmath310 hence , theorem [ t : hconvh ] implies that @xmath311 ( d/2,\alpha/2\beta),\:(1,1/\beta),\:(1,\alpha/2\beta)\end{subarray}\right)}.\ ] ] then by property [ prop : power ] , we see that @xmath312 ( d/2,\alpha/2),\:(1,\alpha/2)\end{subarray}\right ) } { h_{1,1}^{1,0}\left((ts)^{-\beta}\:\middle\vert\ : \begin{subarray}{l}(\beta,\beta)\\[0.4em ] ( 1,1)\end{subarray}\right)}.\end{aligned}\ ] ] by lemma [ l_:nonneg ] , the second h - function in the above equation is nonnegative .
on the other hand , theorem [ t : stable ] tells us that the first h - function is nonnegative .
thus , @xmath158 is nonnegative . as for the case ( b ) , it is known from @xcite that @xmath58 is nonnegative for @xmath59 . by the same argument as in the proof of ( a ) , @xmath167 is also nonnegative . finally , for the case ( c ) , it is proved in @xcite that @xmath180 is nonnegative . by the same reason as in the proof of ( a )
, @xmath61 and @xmath313 are fractional integrals of @xmath314 of orders @xmath315 and @xmath316 , respectively .
therefore , both @xmath61 and @xmath313 are nonnegative as well .
the proof of lemma [ l : nonneg ] is now complete .
the theorem [ t : pde ] follows from the above lemmas .
the characteristic function of @xmath181 is proved in ( * ? ? ?
* ( 7.5.3 ) on p. 211 ) . for the density @xmath317
, we need to compute the inverse fourier transform . from (
* ( 7.5.5 ) ) this inverse transform is @xmath318 by and , we have that @xmath319 ( d/2,1),\:(1,1)\end{subarray}\right)}\ ] ] and @xmath320 ( 0,1/\alpha)\end{subarray}\right)}.\ ] ] hence , @xmath321 ( d/2,\alpha/2),\:(1,\alpha/2)\end{subarray}\right ) } { h_{0,1}^{1,0}\left(t\:\middle\vert\ : \begin{subarray}{l}\midrule\\[0.4em ] ( 0,1/\alpha)\end{subarray}\right ) } { \ensuremath{\mathrm{d}}}t.\ ] ] application of theorem [ t : hconvh ] to evaluate the above integral yields the theorem .
recall that @xmath123 defined by is the solution to the homogeneous equation . using an iteration procedure as in @xcite , we have @xmath322 where @xmath323 and @xmath324 denotes a permutation of @xmath325 such that @xmath326 .
fix @xmath327 and @xmath328 , set @xmath329 then we have that @xmath330^{2n } } { \ensuremath{\mathrm{d}}}s{\ensuremath{\mathrm{d}}}r\int_{\mathbb{r}^{2nd}}{\ensuremath{\mathrm{d}}}y{\ensuremath{\mathrm{d}}}z\ : f_n(s , y , t , x)f_n(r , z , t , x)\prod_{i=1}^n\lambda(y_i - z_i)\prod_{i=1}^n\gamma(s_i - r_i).\end{gathered}\ ] ] where @xmath331 , the differentials @xmath332 , @xmath333 and @xmath334 are defined similarly .
set @xmath335 . using the fourier
transform and cauchy - schwartz inequality together with , we obtain that @xmath336^{2n}}\int_{\mathbb{r}^{nd}}\mathcal{f}g_n(s , \cdot , t , x)(\xi ) \overline{\mathcal{f}g_n(r , \cdot , t , x)(\xi)}\mu({\ensuremath{\mathrm{d}}}\xi)\prod_{i=1}^{n}\gamma(s_i -r_i){\ensuremath{\mathrm{d}}}s{\ensuremath{\mathrm{d}}}r\\ \notag \leq & \frac{\widehat{c}_t^2 \ : n!}{(2\pi)^{nd } } \int_{[0,t]^{2n}}\left(\int_{\mathbb{r}^{nd } } \big(\mathcal{f}g_n(s , \cdot , t , x)(\xi)\big)^2\mu({\ensuremath{\mathrm{d}}}\xi)\right)^{1/2 } \\
\notag & \times \left(\int_{\rr^{nd}}\big(\mathcal{f}g_n(r , \cdot , t , x)(\xi)\big)^2 \mu({\ensuremath{\mathrm{d}}}\xi)\right)^{1/2 } \prod_{i=1}^{n}\gamma(s_i -r_i){\ensuremath{\mathrm{d}}}s{\ensuremath{\mathrm{d}}}r \,,\end{aligned}\ ] ] where the constant @xmath228 is defined in .
thus , thanks to the basic inequality @xmath337 and the fact that @xmath18 is locally integrable , we obtain @xmath338^{2n } } \int_{\mathbb{r}^{nd}}|\mathcal{f}g_n(s , \cdot , t , x)(\xi)|^2\mu({\ensuremath{\mathrm{d}}}\xi)\prod_{i=1}^{n}\gamma(s_i -r_i ) { \ensuremath{\mathrm{d}}}s { \ensuremath{\mathrm{d}}}r\\ & \leq \frac{\widehat{c}_t^2\ : c_t^n n!}{(2\pi)^{nd } } \int_{[0,t]^n}{\ensuremath{\mathrm{d}}}s\int_{\mathbb{r}^{nd}}|\mathcal{f}g_n(s , \cdot , t , x)(\xi)|^2\mu({\ensuremath{\mathrm{d}}}\xi)\,,\end{aligned}\ ] ] where the constant @xmath227 is defined in . furthermore , from the fourier transform of @xmath339 we can check that @xmath340 ^ 2\,,\end{gathered}\ ] ] where we have set @xmath341 . as a consequence ,
@xmath342 ^ 2 \mu({\ensuremath{\mathrm{d}}}\xi_{\sigma(i ) } ) , \label{e_:ynonneg}\end{aligned}\ ] ] where
we have used the fact that @xmath343 and that @xmath56 and @xmath256 are nonnegative to get rid of the supremum in @xmath344 .
therefore , using fourier transform again we have @xmath345 where @xmath346 denotes the simplex @xmath347 by the change of variables @xmath348 for @xmath349 and @xmath350 , we see that @xmath351 where @xmath352 we take @xmath353 which will be chosen later , and let @xmath354 let @xmath355 be a subset of @xmath356 and @xmath357 . then we have @xmath358 where @xmath359 . for the indices @xmath360 in the set @xmath355 , for some constant @xmath361 ( one may choose @xmath362 ) @xmath363 now using the inclusion @xmath364 with @xmath365 we obtain that @xmath366 furthermore , one can bound the integral over @xmath367 in the following way @xmath368 then make the change of variables @xmath369 to obtain @xmath370 where @xmath371 note that the integrability of the above quantity at zero and at infinity implies that @xmath196 and @xmath214 , respectively .
thus we have the following bound .
@xmath372 where @xmath373 , and @xmath374 and @xmath375 are defined in . observing the trivial inequality @xmath376 , we have @xmath377 choosing @xmath378 sufficiently large so that @xmath379 yields @xmath380 this proves , and thus the existence and uniqueness of the solution .
[ l : j0 ] suppose that the initial conditions @xmath381 are constant .
then under the three cases of , we have that @xmath382$},\\ u_0 + t^{\beta-1 } u_1 & \text{if $ \beta\in ( 1,2)$}. \end{cases } \end{aligned}\ ] ] by theorem [ t : pde ] , we know that under the first two cases of , the fundamental solutions are nonnegative and hence , @xmath383(0 ) , \ ] ] which is equal to the right hand side of . as for the last case in , because @xmath55 is still nonnegative , the contribution by @xmath384 can be computed in the same way .
however , we do not know whether @xmath385 is nonnegative , and thus we can not use the fourier transform arguments to compute the contribution by @xmath386 . instead , we compute it directly : @xmath387 ( d/2,1),\ : ( 1,1)\end{subarray}\right)}{\ensuremath{\mathrm{d}}}x,\ ] ] where @xmath388 then by theorem [ t : laplace ] , we have the following laplace transform : @xmath389 ( d/2,1),\ : ( 1,1)\end{subarray}\right)}{\ensuremath{\mathrm{d}}}x = { h_{2,2}^{1,2}\left(2\nu t^\beta z^2\:\middle\vert\ : \begin{subarray}{l}(1-d/2,1),\:(0,1)\\[0.4em ] ( 0,\alpha),\:(0,\beta)\end{subarray}\right)}.\ ] ] then by theorem [ t : asypzero ] , @xmath390 putting these identities together , we have that @xmath391 this completes the proof of lemma [ l : j0 ] . since @xmath221
, we have @xmath392 , for some coefficient @xmath231 ; see , e.g. , @xcite .
we begin with the upper bound . by the hypercontractivity property of the @xmath393-th chaos , i.e. @xmath394 on the other hand , from the proof of theorem [ t : exuni ] ( see )
it follows @xmath395 where @xmath228 is defined in , @xmath396 and @xmath397 is defined in . according to the property of the mittag - leffler function at zero and infinity , if @xmath398
, then the above constant @xmath399 is finite .
then , under the condition that @xmath400 ( this condition implies @xmath398 ) , the integration over @xmath333 can be evaluated explicitly ; see ( * ? ? ?
* lemma 4.5 ) .
hence , @xmath401 where @xmath402.denote @xmath403 thus we obtain @xmath404 this bound together with the hypercontractivity implies that @xmath405 therefore , @xmath406 where @xmath407 then by the fact that @xmath408 for @xmath409 , @xmath410 for some positive constant @xmath411 , where in the last step , we have used the asymptotic property of the mittag - leffler function ( see , e.g. , ( * ? ? ? * theorem 1.3 ) ) .
now we consider the special case when @xmath18 is the dirac delta function . by lemma [ l : j0 ] and
the assumptions on the initial conditions we have @xmath412 from the proof of theorem [ t : exuni ] ( see and ) , we see that @xmath413^n}{\ensuremath{\mathrm{d}}}s \int_{\rr^{nd } } { \ensuremath{\mathrm{d}}}\xi \ : \prod_{i=1}^n ( s_{\sigma(i+1)}-s_{\sigma(i)})^{2\beta-2}\\ & \times e_{\beta,\beta}^2 \left(-2^{-1}\nu ( s_{\sigma(i+1)}-s_{\sigma(i)})^{\beta } |\xi_{\sigma(i ) } + \cdots + \xi_{\sigma(1 ) } |^{\alpha } \right ) |\xi_i|^{\kappa - d } \\
= & \frac { { u_0 ^ 2 } c_{\kappa}^n}{(2\pi)^{nd } } \int_{t_n(t ) } { \ensuremath{\mathrm{d}}}s\int_{\rr^{nd}}{\ensuremath{\mathrm{d}}}\xi \ : \prod_{i=1}^n ( s_{i+1}-s_{i})^{2\beta-2}\\ & \times e_{\beta,\beta}^2 \left(-2^{-1}\nu ( s_{i+1}-s_{i})^{\beta } |\xi_{i } + \cdots + \xi_{1 } |^{\alpha } \right ) |\xi_i|^{\kappa - d } \,.\end{aligned}\ ] ] then by the change of variable @xmath414 and replacing @xmath415 by @xmath416 , we obtain that @xmath417 where @xmath418 . then with another change of variable @xmath419 , and by the same reasoning as before ,
we obtain that @xmath420 therefore , by the asymptotic property of the mittag - leffler function , @xmath421 \geq & \sum_{n=0}^{\infty } \frac { { u_0 ^ 2 } \left(c_{\kappa } \widetilde{c } ( 4\pi)^{-d } c _ * \right)^n t^{n(2\beta -1 -\frac{\beta \kappa}{\alpha } ) } ( 2/\nu)^{\frac{\kappa n } {
\alpha}}}{\gamma(n(2\beta -1 -\frac{\beta \kappa}{\alpha})+1 ) } \\ \geq & c\ : { u_0 ^ 2 } \exp \left ( ( c_{\kappa } \widetilde{c } ( 4\pi)^{-d } c _ * ( 2/\nu)^{\kappa/\alpha } ) ^{\frac{1}{2\beta -1 -\beta \kappa/\alpha } } t \right)\,,\end{aligned}\ ] ] for some positive constant @xmath422 .
this completes the proof of theorem [ t : rietz ] . in this section
, @xmath423 denotes a positive constant , possibly dependent on @xmath424 .
[ est.h ] assume that @xmath168 , @xmath425 and @xmath155 .
then there is a nonnegative constant @xmath426 such that for all @xmath427 , @xmath428 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right ) } \right| \leq c_{\alpha , \beta , d } \ : \frac{z^\zeta}{z^{\zeta+1}+1},\qquad\text{for all $ z \ge 0$.}\ ] ] we first note that condition is satisfied .
because @xmath282 , we can apply theorem [ t : asyinfty ] to obtain that @xmath429 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right ) } { = } o(1/z ) ; \qquad z \rightarrow \infty.\ ] ] as for small @xmath430 , note that the poles of @xmath431 are @xmath432 and those of @xmath433 are @xmath434 to find the leading term when @xmath435 , we need to find the first nonvanishing residue of @xmath436 at poles @xmath437 , where @xmath438 * case i. * when @xmath439 and @xmath440 , then the leading pole ( @xmath441 ) in @xmath442 does not coincide with the first two poles ( @xmath443 ) of @xmath444 .
hence , the asymptotic expansion in theorem [ t : asypzero ] ( 1 ) implies that @xmath429 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right ) } { = } h_{10}^ * z^{d/\alpha } + h_{20}^ * z + h_{11}^ * z^2 + o(z^2 ) + o(z^{d/\alpha } ) ; \qquad z \rightarrow 0.\ ] ] here @xmath445 are defined in .
notice that @xmath446 due to the presence of the parameter @xmath447 .
hence , @xmath429 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right ) } { = } o(z^2)+o(z^{d/\alpha } ) ; \qquad z \rightarrow 0.\ ] ] * case ii . *
now we consider the case when @xmath448 .
the first pole in @xmath437 is @xmath449 , which is of order @xmath450 .
now we compute the mentioned residue : @xmath451 & = \lim_{s\rightarrow -1}\left[(s+1)^2 \ch_{2,3}^{2,1}(s ) z^{-s}\right]'\\ & = \lim_{s\rightarrow -1}\left[(s+1)^2 \ch_1^*(s ) \ch_2^*(s ) z^{-s}\right]'\\ & = \lim_{s\rightarrow -1 } z^{-s } \left[\ch_1^*(s)'\ch_2^*(s)+\ch_1^*(s)\ch_2^*(s ) ' - \ch_1^*(s)\ch_2^*(s ) \log z \right],\end{aligned}\ ] ] where @xmath452 is defined in and @xmath453 now simple calculations show that @xmath454 where @xmath455 is the digamma function and the last limit is due to ( 5.7.6 ) and ( 5.7.1 ) of @xcite .
thus , @xmath456 hence , @xmath457=\frac{\beta\ : z}{\gamma(1+d/2)}.\ ] ] therefore , by the definition of the fox h - function , @xmath429 ( d/2,d/2),\:(1,1),\:(1,d/2)\end{subarray}\right ) } { = } o(z);\quad z \rightarrow 0.\ ] ] * case iii . * as for the case @xmath458 , the first pole in @xmath437 is @xmath449 , which is a simple pole . as calculated before , the residue at this pole is vanishing , @xmath459 .
hence , we need to consider the next pole at @xmath460 , which is a pole of order @xmath450 . use the asymptotic expansion to obtain that @xmath461 ( d/2,\alpha/2),\:(1,1),\:(1,\alpha/2)\end{subarray}\right ) } { = } o(z^2\log z ) ; \qquad z \rightarrow 0.\end{aligned}\ ] ] finally , because @xmath462 , by theorem [ t:1.2(3 ) ] , our h - function is a continuous function for @xmath463 . with this
, we complete the proof of lemma [ est.h ] .
[ basic.ineq-1 ] for all @xmath3 $ ] , @xmath155 and @xmath464 , one can find @xmath465 and a nonnegative constant @xmath466 ( independent of @xmath467 ) such that @xmath468 where @xmath469 we divide the integral domain into @xmath470 and @xmath471 . over the domain @xmath470 ,
we have @xmath472 the last inequality is valid since we can choose @xmath473 sufficiently close to @xmath474 so that @xmath475 . on the other hand , over the domain @xmath476 , we have @xmath477 note that the above constant @xmath466 does not depend on @xmath467 .
[ basic.ineq ] assume @xmath464 .
then for all @xmath478 and @xmath479 , we have that @xmath480 where @xmath466 does not depend on @xmath481 and @xmath482 , and @xmath483 we use the notation @xmath484 in the previous lemma . by lemma [ est.h ] and the expression of @xmath56 through fox h - function , we see that for any @xmath485 , there is a constant @xmath486 such that @xmath487 therefore , by lemma [ basic.ineq ] , we have @xmath488 by symmetry , we also have @xmath489 now from the fact that @xmath490 and @xmath491 implies @xmath492 , the lemma follows .
the following lemma is from ( * ? ? ?
* theorem 3.5 ) . [ stint ]
let @xmath493 be the simplex defined in .
then for all @xmath494 , it holds that @xmath495^{h } { \ensuremath{\mathrm{d}}}s = \frac{\gamma(1+h)^{n}}{\gamma(n(1+h)+1 ) } t^{n(1+h)}\,.\ ] ] following the same notation and arguments as the proof of theorem [ t : exuni ] until , we have @xmath496^{2n } } { \ensuremath{\mathrm{d}}}s{\ensuremath{\mathrm{d}}}r\int_{\mathbb{r}^{2nd}}{\ensuremath{\mathrm{d}}}y{\ensuremath{\mathrm{d}}}z\ : g_n(s , y , t , x)g_n(r , z , t , x)\prod_{i=1}^n\lambda(y_i - z_i)\prod_{i=1}^n\gamma(s_i - r_i).\end{gathered}\ ] ] furthermore , by cauchy - schwarz inequality , we obtain @xmath497 applying lemma [ basic.ineq ] to the above two integrals , we have @xmath498 where @xmath499 with @xmath500 hence , @xmath501^{2n } } \prod_{i=1}^n \gamma(s_i - r_i ) ( \phi(s)\phi(r))^\theta { \ensuremath{\mathrm{d}}}s { \ensuremath{\mathrm{d}}}r\\ & \leq \frac{c^n_{\alpha , \beta , d , \nu , \kappa}}{n ! } \frac{1}{2 } \int_{[0,t]^{2n } } \prod_{i=1}^n \gamma(s_i - r_i ) \left(\phi(s ) ^{2\theta } + \phi(r)^{2\theta } \right ) { \ensuremath{\mathrm{d}}}s { \ensuremath{\mathrm{d}}}r\\ & = \frac{c^n_{\alpha , \beta , d , \nu , \kappa}}{n ! } \int_{[0,t]^{2n } } \prod_{i=1}^n \gamma(s_i - r_i )
\phi(s ) ^{2\theta } { \ensuremath{\mathrm{d}}}s { \ensuremath{\mathrm{d}}}r\\ & \leq \frac{c^n_{\alpha , \beta , d , \nu , \kappa } c_t^n}{n ! } \int_{[0,t]^n } \phi(s)^{2\theta } { \ensuremath{\mathrm{d}}}s\\ & = c^n_{\alpha , \beta , d , \nu , \kappa } c_t^n \int_{t_n(t ) } \phi(s)^{2\theta } { \ensuremath{\mathrm{d}}}s\\ & = \frac{c^n_{\alpha , \beta , d , \nu , \kappa } c_t^n \gamma(2\theta + 1)^n t^{(2\theta+1)n}}{\gamma((2\theta+1)n+1)}\,,\end{aligned}\ ] ] where @xmath227 is defined in .
therefore , @xmath502 and @xmath503 converges if @xmath504 .
finally , the condition @xmath504 , which is equivalent to @xmath505 , guarantees both condition @xmath506 in lemma [ stint ] and the assumption @xmath507 used in lemma [ basic.ineq ] .
this completes the proof of theorem [ beta.2 ] .
[ d : h ] let @xmath508 be integers such that @xmath509 .
let @xmath510 be complex numbers and let @xmath511 be positive numbers , @xmath512 and @xmath513 .
let the set of poles of the gamma functions @xmath514 does nt intersect with that of the gamma functions @xmath515 , namely , @xmath516 for all @xmath512 and @xmath517 . denote @xmath518 the _ fox h - function _
@xmath519\end{aligned}\ ] ] is defined by the following integral @xmath520 where an empty product in means @xmath316 and @xmath521 in is the infinite contour which separates all the points @xmath522 to the left and all the points @xmath523 to the right of @xmath521 .
moreover , @xmath521 has one of the following forms : 1 .
@xmath524 is a left loop situated in a horizontal strip starting at point @xmath525 and terminating at point @xmath526 for some @xmath527 ; 2 .
@xmath528 is a right loop situated in a horizontal strip starting at point @xmath529 and terminating at point @xmath530 for some @xmath527 ; 3 .
@xmath531 is a contour starting at point @xmath532 and terminating at point @xmath533 for some @xmath534 .
.,title="fig:",width=192].,title="fig:"].,title="fig : " ] + see figure [ f : l ] for some illustrating paths . according to ( * ? ? ?
* theorem 1.1 ) , the integral exists , for example , when @xmath535 or when @xmath536 we call the poles @xmath523 ( see ) of @xmath537 , @xmath538 , _ simple _ , if @xmath539 similarly , the poles @xmath522 ( see ) of @xmath540 , @xmath541 , _ simple _ , if @xmath542 [ t:1.2(3 ) ] [ conti.h ] let @xmath543 , @xmath544 then @xmath545 is analytic on @xmath546 .
[ prop : reduction ] if @xmath547 and @xmath548 , then @xmath549 ( b_j,\beta_j)_{1,q-1},\:(a_1,\alpha_1)\end{subarray}\right ) } = { h_{p-1,q-1}^{m , n-1}\left(z\:\middle\vert\ : \begin{subarray}{l}(a_i,\alpha_i)_{2,p}\\[0.4em ] ( b_j,\beta_j)_{1,q-1}\end{subarray}\right ) } .\ ] ] and similarly , if @xmath550 and @xmath551 , then @xmath552 ( b_j,\beta_j)_{1,q}\end{subarray}\right ) } = { h_{p-1,q-1}^{m-1,n}\left(z\:\middle\vert\ : \begin{subarray}{l}(a_i,\alpha_i)_{1,p-1}\\[0.4em ] ( b_j,\beta_j)_{2,q}\end{subarray}\right ) } .\ ] ] [ prop : frac ] it holds that @xmath553 ( b_j,\beta_j)_{1,q}\end{subarray}\right ) } = { h_{q , p}^{n , m}\left(z\:\middle\vert\ : \begin{subarray}{l}(1-b_j,\beta_j)_{1,q}\\[0.4em ] ( 1-a_i,\alpha_i)_{1,p}\end{subarray}\right)}.\ ] ] [ prop : power ] for @xmath554 , it holds that @xmath549 ( b_j,\beta_j)_{1,q}\end{subarray}\right ) } = k \ : { h_{p , q}^{m , n}\left(z^k\:\middle\vert\ : \begin{subarray}{l}(a_i , k\alpha_i)_{1,p}\\[0.4em ] ( b_j , k\beta_j)_{1,q}\end{subarray}\right)}.\ ] ] the following two theorems are some basic results on the asymptotic expansions of the fox h - functions at zero and infinity . [
t : asyinfty ] suppose @xmath545 satisfies either @xmath555 or @xmath556 .
if the poles of @xmath557 do not coincide , i.e. , holds , then when @xmath558 , we have @xmath559 where @xmath560\frac{\beta_j}{\alpha_i}\right ) \prod^n_{j=1 , i\neq j}\gamma\left(1-a_j-[1-a_i+k]\frac{\alpha_j}{\alpha_i}\right ) } { \prod^p_{j = n+1}\gamma\left(a_j+[1-a_i+k]\frac{\alpha_j}{\alpha_i}\right ) \prod^q_{j = m+1}\gamma\left(1-b_j-[1-a_i+k]\frac{\beta_j}{\alpha_i}\right ) } .\ ] ] see section 1.5 of @xcite . is asymptotic expansion 1.5.1 of @xcite .
[ t : asypzero ] suppose @xmath545 satisfies either @xmath561 , @xmath462 or @xmath556 . when @xmath562 , we have the following two cases : 1 .
if the poles @xmath522 are simple ( see ) , then @xmath563 2 .
if the poles @xmath522 ( see ) of @xmath514 , @xmath541 , coincide : @xmath564 then @xmath565^i,\ ] ] here @xmath566\frac{\beta_i}{\beta_j}\right ) \prod^n_{i=1}\gamma\left(1-a_i+[b_j+l]\frac{\alpha_i}{\beta_j}\right ) } { \prod^p_{i = n+1}\gamma\left(a_i-[b_j+l]\frac{\alpha_i}{\beta_j}\right ) \prod^q_{i = m+1}\gamma\left(1-b_i+[b_j+l]\frac{\beta_i}{\beta_j}\right ) } ; \ ]
] @xmath567 is summation over @xmath568 such that the @xmath522 do not coincide ; @xmath569 is the summation over @xmath568 such that @xmath522 coincide with order @xmath570 ; @xmath571^{(n^*_{jl}-1-n)}[\mathcal{h}^*_2(b_{jl})]^{(n - i)},\ ] ] where @xmath572 and @xmath573 are orders of derivative ; @xmath574 where @xmath575 is the pole with order @xmath576 see section 1.8 of @xcite . is asymptotic expansion 1.8.1 of @xcite . is asymptotic expansion 1.8.2 of @xcite .
the following theorem is a simplified version of theorems 2.9 and 2.10 in @xcite , which is sufficient for our use .
we need another parameter for the fox h - function : @xmath577 [ t : hconvh ] let @xmath578 and @xmath579 be the constants @xmath580 defined in , and for the following two fox h - functions : @xmath581 ( b_j,\beta_j)_{1,q}\end{subarray}\right ) } \quad\text{and}\quad { h_{p , q}^{m , n}\left(x\:\middle\vert\ : \begin{subarray}{l}(d_i,\delta_i)_{1,p}\\[0.4em ] ( c_j,\gamma_j)_{1,q}\end{subarray}\right)},\ ] ] respectively .
denote @xmath582 with the convention that @xmath583 . if either of the following four conditions holds 1 .
@xmath584 and @xmath585 and @xmath586 ; 2 .
@xmath587 and @xmath588 and @xmath589 ; 3 .
@xmath590 , @xmath591 and @xmath592 ; 4 .
@xmath593 , @xmath594 and @xmath595 ; 5 .
@xmath590 , @xmath591 and @xmath593 , @xmath594 , and if @xmath596 then , for all @xmath463 , @xmath597 , @xmath598 ( b_j,\beta_j)_{1,m},\ : ( c_j,\gamma_j)_{1,q},\ : ( b_j,\beta_j)_{m+1,q}\end{subarray}\right ) } \\ & \hspace{4em } = \int_0^\infty { h_{p , q}^{m , n}\left(z t\:\middle\vert\ : \begin{subarray}{l}(a_i,\alpha_i)_{1,p}\\[0.4em ] ( b_j,\beta_j)_{1,q}\end{subarray}\right ) } { h_{p , q}^{m , n}\left(\frac{x}{t}\:\middle\vert\ : \begin{subarray}{l}(d_i,\delta_i)_{1,p}\\[0.4em ] ( c_j,\gamma_j)_{1,q}\end{subarray}\right)}\frac{{\ensuremath{\mathrm{d}}}t}{t}.\end{aligned}\ ] ] by property [ prop : frac ] , @xmath599 ( c_j,\gamma_j)_{1,q}\end{subarray}\right ) } = { h_{q , p}^{n , m}\left(\frac{t}{x}\:\middle\vert\ : \begin{subarray}{l}(1-c_j,\gamma_j)_{1,q}\\[0.4em ] ( 1-d_i,\delta_i)_{1,p}\end{subarray}\right)}.\end{aligned}\ ] ] if both @xmath595 and @xmath592 , then one can apply theorem 2.9 of @xcite with @xmath600 , @xmath601 , @xmath602 , and with the following replacements : @xmath603 , @xmath604 , @xmath605 , @xmath606 , @xmath607 , @xmath608 .
when @xmath595 but @xmath274 and @xmath586 , because both @xmath430 and @xmath609 are real numbers , theorem 2.9 of @xcite still holds ( see the paragraph before theorem 2.10 of @xcite on p.59 ) . therefore , we have proved the theorem under conditions ( 1 ) or ( 2 ) . if either of conditions ( 3)(5 ) holds , we apply theorem 2.10 of @xcite in the same way . note that the parameters @xmath20 for both fox h - functions in are equal .
the following result on the laplace transform of the fox h - function is used in this paper .
denote @xmath610 [ t : laplace ] suppose that @xmath462 .
assume that @xmath611 , @xmath612 and @xmath613 are such that @xmath614 then @xmath615 ( b_j,\beta_j)_{1,q}\end{subarray}\right)}\right](t ) = \frac{1}{t^{w+1 } } { h_{p+1,q}^{m , n+1}\left(\frac{a}{t^\sigma}\:\middle\vert\ : \begin{subarray}{l}(-w,\sigma),\:(a_i,\alpha_i)_{1,p}\\[0.4em ] ( b_j,\beta_j)_{1,q}\end{subarray}\right)}.\ ] ] we also need the following convolution result , which is related to the hankel transform of the fox h - function . ( corollary 2.5.1 of @xcite)[t : hankel ] let @xmath462 or @xmath616 and @xmath617 .
let @xmath618 , @xmath619 and @xmath613 be such that @xmath620 and @xmath621 then for all @xmath612 and @xmath622 , @xmath623 ( b_j,\beta_j)_{1,q}\end{subarray}\right)}{\ensuremath{\mathrm{d}}}t\\ & = \frac{1}{2\sigma x}\left(\frac{2}{a}\right)^{\frac{w+1}{\sigma } } { h_{p+2,q}^{m , n+1}\left(\frac{b}{x^\tau}\left(\frac{2}{a}\right)^{\tau/\sigma}\:\middle\vert\ : \begin{subarray}{l}\left(1-\frac{w+1}{2\sigma}-\frac{\eta}{2},\frac{\tau}{2\sigma}\right),(a_i,\alpha_i)_{1,p},\left(1-\frac{w+1}{2\sigma}+\frac{\eta}{2},\frac{\tau}{2\sigma}\right)\\[0.4em ] ( b_j,\beta_j)_{1,q}\end{subarray}\right)},\quad x>0.\end{aligned}\ ] ] recall that the riemann - liouville fractional derivative of order @xmath624 is defined in .
the following results are used several times in this paper , which is a special case of part ( i ) of theorem 2.8 of @xcite .
[ t : rl - d ] let @xmath625 , @xmath626 and @xmath613 .
if @xmath462 and holds , then @xmath627 ( b_j,\beta_j)_{1,q}\end{subarray}\right)}\right](x ) = x^{w-\alpha } { h_{p+1,q+1}^{m , n+1}\left(x^\sigma\:\middle\vert\ : \begin{subarray}{l}(-w,\sigma),(a_i,\alpha_i)_{1,p}\\[0.4em ] ( b_j,\beta_j)_{1,q},(-w+\alpha,\sigma)\end{subarray}\right)}.\ ] ] the mittag - leffler function is a special case of the fox h - function ( see ( 2.9.27 ) of @xcite ) : @xmath628 ( 0,1),\:(1-\mu,\rho)\end{subarray}\right)},\qquad \rho>0 , \ : \mu\in\mathbb{c}.\end{aligned}\ ] ] the @xmath629 function can be represented as ( see ( * ? ? ?
* ( 2.9.8 ) ) ) : @xmath630 ( 0,1),\ : ( 1/2,1)\end{subarray}\right)}.\end{aligned}\ ] ] the bessel function of the first kind @xmath631 is equal to ( see ( 2.9.18 ) of @xcite ) @xmath632 \left(\frac{a+\eta}{2},1\right),\:\left(\frac{a-\eta}{2},1\right)\end{subarray}\right)}.\end{aligned}\ ] ] another special case is ( see ( 2.9.4 ) of @xcite ) @xmath633 ( b,\beta)\end{subarray}\right ) } & = \frac{1}{\beta}z^{b/\beta}\exp\left(-z^{1/\beta}\right).\end{aligned}\ ] ]
le chen thanks francesco mainardi for some useful discussion and thanks anatoly n. kochubei for pointing out pskhu s paper @xcite . | this paper studies the linear stochastic partial differential equation of fractional orders both in time and space variables @xmath0 , where @xmath1 is a general gaussian noise and @xmath2 , @xmath3 $ ] .
the existence and uniqueness of the solution , the moment bounds of the solution are obtained by using the fundamental solutions of the corresponding deterministic counterpart represented by the fox h - functions . along the way , we obtain some new properties of the fundamental solutions . |
there is currently much effort devoted to applications of soft - collinear effective theory ( scet ) @xcite to exclusive @xmath0 decays @xcite .
scet provides a systematic framework in which to discuss qcd factorization theorems for these processes @xcite and power corrections using the language of effective field theory .
the hadronic states relevant to exclusive decays such as @xmath3 or @xmath4 contain highly energetic , collinear partons inside the final - state light mesons , and soft partons inside the initial @xmath0 meson .
understanding the intricate interplay between soft and collinear degrees of freedom is a challenge that one hopes to address using the effective theory .
this interplay is relevant even in simpler processes such as semileptonic @xmath0 decays near @xmath5 , which are described in terms of heavy - to - light form factors at large recoil .
power counting in scet is based on an expansion parameter @xmath6 , where @xmath7 is a large scale ( typically @xmath8 in @xmath0 decays ) , and @xmath9 is of order the qcd scale .
a complication in scet is that different components of particle momenta and fields may scale differently with the large scale @xmath10 . to make this scaling explicit one introduces two light - like vectors @xmath11 and @xmath12 satisfying @xmath13 and @xmath14 . typically , @xmath11 is the direction of an outgoing fast hadron ( or a jet of hadrons ) .
any 4-vector can then be decomposed as @xmath15 where @xmath16 .
the light - like vectors @xmath17 are defined by this relation .
the relevant scet degrees of freedom describing the partons in the external hadronic states of exclusive @xmath0 decays are soft and collinear , where @xmath18 for soft momenta and @xmath19 for collinear momenta . here and below we indicate the scaling properties of the components @xmath20 .
the corresponding effective - theory fields and their scaling relations are @xmath21 ( soft heavy quark ) , @xmath22 ( soft light quark ) , @xmath23 ( soft gluon ) , and @xmath24 ( collinear quark ) , @xmath25 ( collinear gluon ) .
the collinear quark is described by a 2-component spinor subject to the constraint @xmath26 .
short - distance fluctuations in exclusive @xmath0 decays are usually characterized by two different large scales : the hard scale @xmath27 associated with off - shell fluctuations of the heavy quark , and the `` hard - collinear '' scale @xmath28 arising in interactions involving both soft and collinear degrees of freedom . in order to disentangle the physics associated with these two scales
it is sometimes useful to perform the matching of full qcd onto the low - energy effective theory in two steps , by going through an intermediate effective theory ( called scet@xmath29 ) containing `` hard - collinear '' modes with virtualities @xmath30 as dynamical degrees of freedom . in a second step scet@xmath29
is matched onto the final theory ( called scet@xmath31 ) containing near on - shell soft and collinear partons only . in this paper
we are concerned with the structure of this final effective theory . at leading order in power counting the effective strong - interaction lagrangian of scet@xmath31
splits up into separate lagrangians for the soft and collinear fields .
this property implies factorization of many processes involving soft and collinear partons at leading power in @xmath32 .
factorization is however not guaranteed for quantities that vanish at leading power , such as heavy - to - light form factors at large recoil .
it was argued in @xcite that at subleading order in @xmath32 interactions between soft and collinear particles occur , which can violate factorization .
as illustrated in figure [ fig : messenger ] , these interactions can be mediated by the exchange of a short - distance hard - collinear mode , or by a long - distance `` messenger particle '' with momentum scaling @xmath33 , where the transverse momentum components can be at most of @xmath34 .
we will argue in the present work that these two exchange mechanisms contribute under different kinematic conditions .
in particular , the hard - collinear exchange requires the presence of collinear particles in the initial state and so is irrelevant for scet applications to @xmath0 decays .
the scaling of the long - distance messenger particle is such that it can couple to both soft and collinear fields without taking them far off - shell .
it was left open in @xcite whether the messenger exchange should be described in terms of a new field in the effective theory , or by considering the exchange particle as a soft ( or collinear ) field subject to certain constraints on some of its momentum components . [ cols="^,^,^,^ " , ] translated to momentum space , this particular position dependence of the fields enforces that the minus component of the total collinear momentum flowing into the vertex exits again through collinear lines .
an analogous statement holds for the plus component of the soft momenta .
the momentum - conservation @xmath35-functions associated with the above vertices therefore reads @xmath36 , where @xmath37 is the sum of all collinear and @xmath38 the sum of all soft momenta ( ingoing minus outgoing ) connected to the vertex . as classical scattering processes ,
these interactions are rather exceptional : the operator ( [ lsc1 ] ) , for example , contributes only to the forward scattering of a collinear quark and a soft quark .
when inserted into loop diagrams , such exceptional momentum configurations in general do not give rise to non - zero contributions . to see this ,
let us analyze the one - loop qcd diagram shown in figure [ fig : hardcoll ] using the strategy of regions .
one finds that the graph receives a contribution from the region where all propagators are hard or hard - collinear . in the effective theory
this contribution is represented by the first diagram on the right - hand side .
there are also contributions from regions where two propagators of the loop are soft ( or collinear ) and all others hard or hard - collinear .
these correspond to effective - theory diagrams ( not shown ) where a collinear or soft gluon is emitted from the weak vertex and absorbed by one of the quarks .
however , no contribution arises that would correspond to the last diagram in figure [ fig : hardcoll ] , which involves the interaction ( [ lsc2 ] ) denoted by the black square .
first , it is impossible to assign a loop momentum and external momenta in the full - theory diagram such that the exceptional momentum configuration corresponding to the last graph ( where the gluon connected to the heavy quark is soft and that connected to the outgoing light quark is collinear ) would be enforced .
this configuration would require the loop momentum to scale like @xmath39 , in which case the denominators of the qcd loop integral can be simplified as @xmath40 where @xmath41 and @xmath42 .
these denominator structures precisely correspond to the multipole - expanded form of the operator ( [ lsc2 ] ) .
note that the plus component of the loop momentum does not propagate into soft lines , while its minus component does not propagate into collinear lines . as a result , the integration over @xmath43 vanishes if the components @xmath44 of the external soft momenta all have the same sign , because then the contour can be deformed away from the poles .
similarly , the integration over @xmath45 vanishes if the components @xmath46 of the external collinear momenta all have the same sign .
the same conclusion is reached upon analyzing a general loop diagram involving ( [ lsc1 ] ) or ( [ lsc2 ] ) : non - zero contributions from these interactions can only occur in processes which involve soft and collinear particles in both the initial and final state , and these interactions therefore do not contribute in decays of @xmath0 mesons .
this is a consequence of the coleman
norton theorem @xcite , which states that on - shell singularities in feynman diagrams correspond to classically allowed scattering processes .
the above conclusion also applies to the analysis in @xcite .
the non - perturbative soft - collinear rescattering mechanism introduced there should not be described in terms of ( [ lsc1 ] ) , but rather through the long - distance messenger exchange interaction ( [ mess1 ] ) . as a final note ,
let us mention that the same arguments ensure that there are no induced self - interactions of the type ( [ ssc ] ) among only soft or only collinear fields , i.e. , the effective lagrangians @xmath47 and @xmath48 are not altered by integrating out the soft - collinear modes .
the corresponding correlation functions @xmath49 etc . vanish , since the contour can be closed avoiding the poles in the complex plane .
this ensures that soft - collinear fields can only be exchanged in graphs involving both soft and collinear fields .
we have argued that the version of soft - collinear effective theory appropriate for the discussion of exclusive @xmath0 decays such as @xmath50 , @xmath3 , and @xmath4 should include a new mode in addition to soft and collinear fields , which can interact with both soft and collinear particles without taking them far off - shell .
the relevance of these messenger fields has been demonstrated using the example of a triangle diagram with one soft and one collinear external momentum .
we have then developed the formalism incorporating the corresponding quark and gluon fields into the soft - collinear effective lagrangian . in the strong - interaction sector of the effective theory
the leading - order interactions of soft - collinear fields with soft or collinear fields can be removed using field redefinitions .
the remaining interactions are power suppressed .
we have explicitly worked out the leading terms involving quark fields .
we stress that this decoupling in the qcd sector of the theory does not necessarily imply that soft - collinear modes can be ignored at leading order in power counting
. this would only be true in cases where the decoupling transformation leaves external operators such as weak - interaction currents invariant . when the messenger fields are `` integrated out '' in the functional integral they give rise to highly non - local induced interactions between soft and collinear fields , which appear at the same order in power counting as short - distance interactions among these fields induced by the exchange of hard - collinear modes .
however , only the long - distance interactions induced by soft - collinear exchange graphs are kinematically allowed in @xmath0 decays , where no collinear particles are present in the initial state .
a puzzling aspect of our analysis is the finding of a relevant momentum region corresponding to virtualities that are parametrically below the qcd scale @xmath51 .
however , our discussion in section [ sec : sudakov ] illustrates that the same phenomenon arises in the case of the sudakov form factor , if the external massless particles have off - shell momenta scaling like @xmath52 and @xmath53 with @xmath9 . then perturbatively the ultra - soft region of loop momenta @xmath54 gives rise to leading - order contributions and acts as a `` messenger '' between the two collinear particles , just like the soft - collinear modes connecting soft and collinear particles in our case .
the fact that modes with @xmath55 represent very long - range fluctuations compared to the scale of hadronic systems is not necessarily an argument against their existence .
it is to some extent a consequence of dimensional regularization and analyticity that the relevant momentum region is @xmath56 and not @xmath57 , in which case the virtuality would be of order @xmath58 .
ultimately , it is not the virtuality that is important but the fact that the plus and minus components of soft - collinear momenta are commensurate with certain components of collinear or soft momenta . as an analogy ,
let us recall the case of the hydrogen lamb shift , in which `` ultra - soft '' modes with time and space components of momentum scaling like @xmath59 are important .
these modes have virtuality @xmath60 , where @xmath61 is the typical momentum of the hydrogen atom .
also in this case it is not the off - shellness that matters , but rather the fact that one of the components ( the energy ) is comparable to that of the bound - state system . finally , let us mention that the introduction of a mass term @xmath62 in the propagator labeled with @xmath63 in figure [ fig : triangle ] would change the analysis of regions . in this case
there would be no contribution from the soft - collinear region , since the loop momentum @xmath64 in the soft - collinear propagator @xmath65 could be taylor - expanded and no pinch singularities would occur .
however , this `` simplification '' comes at a price : after the introduction of the mass term , the loop integrals occurring in the soft and collinear region are no longer regulated dimensionally .
in order to obtain the expansion of the massive integral using the strategy of regions , additional analytic regulators have to be introduced on the soft and collinear propagators @xcite . the procedure is straightforward as long as one is only interested in the expansion of a given loop integral , but would complicate the construction of the effective theory since the analytic regulators destroy longitudinal boost and gauge invariance .
we believe it is fair to say that to date it is not known how to consistently incorporate masses of @xmath66 into soft - collinear effective theory .
a related question one might worry about is whether the scaling laws for the propagator functions @xmath67 describing the soft - collinear exchange , which follow from the power - counting rules of the effective theory , could be invalidated by some non - perturbative effects not seen at the level of feynman diagrams , such as a generation of mass terms in exact propagators . in our opinion
such effects would not only upset power counting but threaten the usefulness of the effective theory as a whole .
for now we take the conservative point of view that in order to be useful the effective theory should at least reproduce correctly the analytic properties of perturbation theory . as we have shown , the inclusion of soft - collinear modes is then an indispensable part in the construction of the effective lagrangian .
one of us ( m.n . ) is grateful to the aspen center of physics , where part of this work was carried out .
we would like to thank b. lange , c. bauer and i. stewart for useful discussions .
the work of t.b . and r.j.h .
is supported by the department of energy under grant de - ac03 - 76sf00515 .
the research of m.n . is supported by the national science foundation under grant phy-0098631 . c. w. bauer , s. fleming , d. pirjol and i. w. stewart , phys .
d * 63 * , 114020 ( 2001 ) [ hep - ph/0011336 ] . c. w. bauer , d. pirjol and i. w. stewart , phys .
d * 65 * , 054022 ( 2002 ) [ hep - ph/0109045 ] .
j. chay and c. kim , phys .
d * 65 * , 114016 ( 2002 ) [ hep - ph/0201197 ] .
m. beneke , a. p. chapovsky , m. diehl and t. feldmann , nucl .
b * 643 * , 431 ( 2002 ) [ hep - ph/0206152 ] .
r. j. hill and m. neubert , nucl .
b * 657 * , 229 ( 2003 ) [ hep - ph/0211018 ] . c. w. bauer , d. pirjol and i. w. stewart , phys .
lett . * 87 * , 201806 ( 2001 ) [ hep - ph/0107002 ] .
e. lunghi , d. pirjol and d. wyler , nucl .
b * 649 * , 349 ( 2003 ) [ hep - ph/0210091 ] . c. w. bauer , d. pirjol and i. w. stewart , phys .
d * 67 * , 071502 ( 2003 ) [ hep - ph/0211069 ] .
s. w. bosch , r. j. hill , b. o. lange and m. neubert , phys .
d * 67 * , 094014 ( 2003 ) [ hep - ph/0301123 ] . j .-
chay and c. kim , preprints hep - ph/0301262 , phys .
d * 68 * , 034013 ( 2003 ) [ hep - ph/0305033 ] .
m. beneke , g. buchalla , m. neubert and c. t. sachrajda , phys .
lett . * 83 * , 1914 ( 1999 ) [ hep - ph/9905312 ] ; nucl .
b * 591 * , 313 ( 2000 ) [ hep - ph/0006124 ] ; nucl .
b * 606 * , 245 ( 2001 ) [ hep - ph/0104110 ] .
s. coleman and r. norton , nuov .
cim . * 38 * , 438 ( 1965 ) .
m. beneke and v. a. smirnov , nucl .
b * 522 * , 321 ( 1998 ) [ hep - ph/9711391 ] . for a review , see : v. a. smirnov , _ applied asymptotic expansions in momenta and masses _ , springer tracts in modern physics 177 ( 2002 ) . v. a. smirnov and e. r. rakhmetov , theor .
* 120 * , 870 ( 1999 ) [ teor
. mat .
fiz . * 120 * , 64 ( 1999 ) ] [ hep - ph/9812529 ] . j. h. khn , a. a. penin and v. a. smirnov , eur .
j. c * 17 * , 97 ( 2000 ) [ hep - ph/9912503 ] .
m. beneke and t. feldmann , phys .
b * 553 * , 267 ( 2003 ) [ hep - ph/0211358 ] .
s. mantry , d. pirjol and i. w. stewart , preprint hep - ph/0306254 . | it is argued that soft - collinear effective theory for processes involving both soft and collinear partons , such as exclusive @xmath0-meson decays , should include a new mode in addition to soft and collinear ones .
these `` soft - collinear messengers '' can interact with both soft and collinear particles without taking them far off - shell .
they thus can communicate between the soft and collinear sectors of the theory .
the relevance of the new mode is demonstrated with an explicit example , and the formalism incorporating the corresponding quark and gluon fields into the effective lagrangian is developed .
clns 03/1836 + slac
pub10098 + hep - ph/0308122 + * soft - collinear messengers : a new mode in soft - collinear effective theory * thomas becher@xmath1 , richard j. hill@xmath1 and matthias neubert@xmath2 + @xmath1_stanford linear accelerator center , stanford university + stanford , ca 94309 , u.s.a . _
+ @xmath2_newman laboratory for elementary - particle physics , cornell university + ithaca , ny 14853 , u.s.a . |
systems with large dielectric and piezoelectric responses are of interest both from a fundamental and a technological point of view@xcite . understanding the microsopic origin of a high sensitivity of the polarization of a ferroelectric oxide to applied field , in the case of the dielectric constant , or stress - induced changes in strain , in the case of piezoelectricity , can lead to the optimization of these properties through the appropriate choice of materials . empirically , large responses are observed in stoichiometric compounds such as pbtio@xmath0 and batio@xmath0 for temperatures near the ferroelectric phase transition @xcite . in solid solutions , such as pzt ,
enhanced responses are observed over a wide temperature range@xcite . in this paper
, we discuss the modeling of the piezoelectric and dielectric response of ferroelectric perovskites , specifically pbtio@xmath0 and batio@xmath0 , though the use of first - principles effective hamiltonians previously constructed in refs . .
the calculations give reasonable agreement with experiment , considering that this modeling gives only the contribution of the soft modes to the responses , and not other contributions such as thermal expansion and the response of other polar modes .
however , for temperatures near the ferroelectric phase transition , the soft modes are expected to dominate and thus this approach should capture the essential physics of the large observed responses .
in addition , it should correctly describe the trends between different materials .
finally , with this microscopic approach , we are able directly to relate the calculated response functions to the characteristic correlations of local unit cell polarizations , which should yield further insight into the nature of the large responses of interest .
the construction of first - principles effective hamiltonians for pbtio@xmath0 and batio@xmath0 has previously been described in detail @xcite . briefly , the model consists of one vector per five - atom unit cell which represents the local polarization associated with the ferroelectric distortion .
the potential @xmath1 for these vector degrees of freedom is expanded for the high - symmetry cubic perovskite reference structure , including local anharmonic terms , quadratic intersite interactions which are assumed to be dipolar beyond third neighbors , and lowest order coupling to homogeneous strain .
the extension of this model to include the effects of electric field is accomplished by writing the dependence of the polarization on the model degrees of freedom to lowest order : @xmath2 and adding to the potential the coupling term @xmath3 where @xmath4 is the macroscopic electric field@xcite .
the dielectric response is then simply obtained from @xmath5 , while the piezoelectric response can be expressed as @xmath6 , where the brackets are used to denote the thermal expectation value .
the effects of macroscopic stress @xmath7 can be included by adding to the potential the coupling term @xmath8 with these couplings , the thermodynamic identity @xmath9 = @xmath10 can be obtained ( the well - known equality of the direct and converse piezoelectric effects@xcite ) . in the rest of this paper
, we work at zero external stress and evaluate the piezoelectric response using @xmath11 calculations of the temperature dependent properties of the system are carried out using classical monte carlo calculations , as previously described@xcite .
the calculations of the response functions presented here involved 7x7x7 simulation cells , 10,000 monte carlo sweeps ( mcs ) for thermalization , 200,000 mcs for the computation of thermal expectation values , and several runs at each temperature with different random number seeds to estimate statistical error .
the calculations of the real - space correlation functions presented here involved 10x10x10 simulation cells , 10,000 monte carlo sweeps ( mcs ) for thermalization , and 100,000 mcs for the computation of thermal expectation values . in order to compute the temperature - dependence of the order parameter and thus identify the cubic - tetragonal transition , a small symmetry - breaking field @xmath12 of magnitude @xmath13 kv / cm was applied .
the response function results include this nonzero field .
in the monte carlo simulations described above , the response functions are computed using the fact that they can be expressed as correlation functions .
this allows the appropriate derivatives to be calculated in one monte carlo run .
specifically , we consider the dielectric tensor : @xmath14 where @xmath15 here , @xmath16 is the volume of the unit cell and @xmath17 is the polarization per unit cell , specifically @xmath18 or @xmath19 where we have defined @xmath20 using the approximation @xmath21 ( see footnote 7 in ref . ) and differentiating equation [ poltherm ] with respect to @xmath22 , we readily find the correlation function expression : @xmath23 where @xmath24 the piezoelectric tensor can be expressed in terms of correlation functions in an completely analogous way : @xmath25 where @xmath26 differentiating equation [ strainthermexp ] with respect to @xmath27 , we find the correlation function expression : @xmath28 when the homogeneous strain appears only up to quadratic order in the effective hamiltonian , as is the case for the effective hamiltonians available for pbtio@xmath0@xcite , batio@xmath0@xcite and knbo@xmath0@xcite , the piezoelectric tensor can be reexpressed in terms of correlations of local polar distortions . because the integral over strain is gaussian , completely equivalent expressions for the thermal expectation values involving @xmath29 can be obtained by making the following substitutions ( and cyclic permutations thereof ) : @xmath30 @xmath31 for the example of @xmath32 , this yields @xmath33 while equation [ piezocorrfcn ] proves more convenient for the actual computation of @xmath34 , it is equation [ piezocorrfcna ] that naturally leads to a microscopic interpretation of the soft - mode response , as we will discuss further below .
the results of the monte carlo calculations for @xmath37 are shown in figure [ fig1 ] . at room temperature ,
the calculated value is 66 , which should be compared to the single - crystal measurement @xmath37 = 80@xcite .
the agreement is surprisingly good , considering that the calculation includes only the soft - mode contribution .
the room temperature value is already considerably larger than the zero - temperature calculated value of 42 .
the overall temperature dependence is very similar to that observed experimentally@xcite , with a slow increase with temperature below the transition , a near - divergence both above and below t@xmath38 which is cut off by the first - order character of the cubic - tetragonal transition , and an enhanced value of the dielectric response in the high - temperature cubic phase .
the results of the monte carlo calculations for @xmath32 are shown in figure [ fig2 ] . at room temperature ,
the calculated value is 76 , which should be compared to the single - crystal measurement @xmath32 = 83.7@xcite . as for the dielectric response , the agreement is surprisingly good , considering that the calculation includes only the soft - mode contribution .
the room temperature value is already considerably larger than the zero - temperature calculated value of 48 .
the overall temperature dependence is very similar to that observed experimentally@xcite , with a slow increase with temperature below the transition , a near - divergence below t@xmath38 which is cut off by the first - order character of the cubic - tetragonal transition , and a drop to zero in the high - temperature cubic phase , as required by symmetry .
these results can be compared with the closely analogous calculation of the piezoelectric coefficients of batio@xmath0 ( figure 1 in @xcite ) , noting that for this system as well , remarkably good quantitative agreement with the experimentally measured values is obtained .
the near - divergences observed in the results described above can be already understood within a simple landau theory which describes coupling of a scalar polar mode @xmath39 with a one - component strain @xmath40 and electric field @xmath41 : @xmath42 where the temperature dependence enters through @xmath43 .
if we first eliminate @xmath40 , which is accomplished by the substitution @xmath44 , the following effective fourth - order free energy results : @xmath45 minimizing with respect to @xmath39 yields the following expressions for the dielectric susceptibility @xmath46 and piezoelectric coefficient @xmath47 as observed in the full simulations , @xmath48 diverges as t@xmath38 is approached from below , while @xmath49 diverges on both the low and high temperature sides , and is larger above t@xmath38 . in pbtio@xmath0
, these divergences are cut off due to the first - order character of the transition , so that , for example , a curie - weiss fit to @xmath49 above t@xmath38 yields a value of t@xmath50 about 80 k below t@xmath38 .
the correlation function expressions for the response functions allow the interpretation of the responses in terms of the characteristic intersite correlations between the local polar distortions .
some of the correlation functions of interest have been previously calculated in effective hamiltonian studies of knbo@xmath0@xcite and batio@xmath0@xcite , allowing us to connect to and build upon the insights derived in those investigations to develop a microscopic understanding of the dielectric and piezoelectric responses . for the dielectric response , the first term inside the parentheses of equation [ dicorrfcn ]
can be reorganized : @xmath51 if correlations become infinite ranged , as at a second order transition , this will scale like n , leading to a divergence in the thermodynamic limit . for the piezoelectric tensor , we can similarly reorganize the correlation functions which appear in equation [ piezocorrfcna ] , for example : @xmath52 the real - space correlation functions appearing in the expressions above are computed using fast fourier transforms of @xmath53 and @xmath54 . for the dielectric susceptibility , for example , the real space expressions can be written in terms of @xmath55 as follows ( taking @xmath56 ) : @xmath57 @xmath58 it can be checked that this gives the correct relation to @xmath59 when @xmath60 : @xmath61 in figure [ fig3 ] , we show the real - space decomposition in the @xmath62 plane of the high - t dielectric susceptibility correlation function @xmath63 for pbtio@xmath0 at two temperatures above t@xmath38 .
the correlation is rather isotropic , and the range of the correlations increases as expected as the temperature decreases towards t@xmath38 . for comparison , the same correlation function for batio@xmath0 at two temperatures above the cubic - tetragonal transition temperature is also shown in figure [ fig3 ] .
the highly anisotropic `` chain - like '' correlations discussed in previous work @xcite are clearly evident , and the correlation length transverse to the chains increases as the temperature decreases towards t@xmath38 .
similar behavior is expected for knbo@xmath0 .
the qualitative difference in the nature of the correlations can be directly attributed to the difference in the dispersion relation of the soft mode along r. the implications of the different character of these correlations for the dielectric behavior of pbtio@xmath0 and batio@xmath0 are currently under investigation . as can be seen from equation [ piezocorrfcna ] , the piezoelectric response depends on third - order two - site correlations of the form @xmath64 .
these have not to our knowledge been considered in previous studies .
calculations of these real - space correlations for pbtio@xmath0 and batio@xmath0 above and below the transition temperature are currently in progress .
as already discussed , the use of an effective hamiltonian for calculating the dielectric and piezoelectric response involves a number of approximations .
the effects of thermal expansion and the neglect of other polar modes can not be incorporated without adding additional degrees of freedom to the model ( unless one is willing to accept semi - empirical input ) .
however , the low - order expansion around the cubic reference structure can be improved , which in particular will yield better model results for the properties of the pbtio@xmath0 tetragonal ground state @xcite .
this could involve both @xmath65 and @xmath66 . in the case of the polarization @xmath66 , the next order terms involve including strain dependence and @xmath39 dependence of the effective charges .
first - principles calculations@xcite suggest that the former corrections are rather small compared to the latter .
calculations for these refinements for pbtio@xmath0 are currently in progress .
since the effective hamiltonian approach is limited to the description of soft - mode contributions to the responses , we focus on the near - divergent responses where these contributions strongly dominate . fortunately , this regime is , in any case , the one of greatest interest for understanding the origin of and engineering large responses in a variety of systems characterized by proximity to a lattice instability .
further insight can be gained from the real - space decomposition of the divergent responses based on their expressions in terms of correlation functions . finally , these results for pure systems suggest an interpretation for the origin of large responses in solid solutions . for such systems , there is a distribution of local environments , and for a given temperature some subset of lattice degrees of freedom will be marginally stable .
this idea has been investigated using first principles results in pb@xmath67ge@xmath68te , and more details are presented in refs . .
we thank r. e. cohen , a. garcia , r. m. martin , d. vanderbilt , and u. v. waghmare for helpful discussions . this work was supported by onr grant n00014 - 97 - 1 - 0047 and
the alfred p. sloan foundation , and was in part carried out at the aspen center for physics . | a method for the calculation of the temperature dependence of dielectric and piezoelectric responses , based on the use of a first - principles effective hamiltonian , is described .
results are presented for the ferroelectric perovskite pbtio@xmath0 . while the method includes only the soft - mode contributions to the responses ,
it is argued to give a good description of the divergences or near - divergences of the response functions near the cubic - tetragonal transition .
the expression of the response functions in terms of correlation functions is used to provide a real - space interpretation of the responses which clearly distinguishes between pbtio@xmath0 and the related materials batio@xmath0 and knbo@xmath0 . |
spacetime quantization was proposed earlier than renormalization theory as a possible way of regularizing quantum field theory @xcite . recently
, a deeper motivation was given @xcite : the concurrence of the principles of quantum mechanics and of classical general relativity leads to spacetime uncertainty relations ; the natural geometric background that implements those relations is a noncommutative model of spacetime .
more precisely , in order to give localization in spacetime an operational meaning , _ the energy transfer associated to the localization of an event by the heisenberg uncertainty principle should be limited so that the generated gravitational field does not trap the event itself inside an horizon ; otherwise the observation would be prevented .
_ this principle implies spacetime uncertainty relations which in a weaker form can be written as @xmath5 where @xmath6 is the planck length @xmath7 .
it is possible to implement exactly these relations by appropriate commutation relations between the components of the spacetime coordinates @xmath8 @xcite @xmath9=i\lambda_p^2q^{\mu\nu},\\ \label{q2 } [ q^\mu , q^{\nu\rho}]=0,\\ \label{q3 } q_{\mu\nu}q^{\mu\nu}=0,\\ \label{q4 } \left(\frac{1}{2}q_{\mu\nu}(*q)^{\mu\nu}\right)^2=i,\end{gathered}\ ] ] where @xmath10 is the hodge dual of @xmath11 .
these relations are covariant under the full poincar group .
the irreducible representations of the spacetime commutation relations ( 1 ) take the familiar form ( in absolute units , where @xmath12 ) @xmath13=i\sigma^{\mu\nu}i,\ ] ] where @xmath14 is a real antisymmetric matrix in the manifold @xmath15 defined by the conditions ( [ q3],[q4 ] ) with @xmath16 .
they evidently break lorentz covariance .
interest in the relations ( [ familiar ] ) was more recently raised by the occurrence of closely related forms of noncommutativity also in string theory @xcite .
there exists , however , an essentially unique , fully covariant representation where the pairwise commuting , selfadjoint operators @xmath17 have the full manifold @xmath15 as their joint spectrum .
the generalized weyl correspondence @xmath18 extends to any symbol @xmath19 , @xmath20 , where @xmath21 is the inverse fourier transform of @xmath22 , for @xmath14 fixed . in the above equation
, @xmath23 is to be understood in the sense of the joint functional calculus of the @xmath17s , and @xmath24 where @xmath25 and @xmath26 , @xmath27 .
the above correspondence induces a generalized twisted product @xmath28 on the symbols , by @xmath29 .
moreover , @xmath30 .
we denote by @xmath31 the enveloping c*-algebra of the resulting algebra ; it is isomorphic to @xmath32 , the c*-algebra of the continuous functions taking values in the algebra @xmath33 of the compact operators on the separable , infinite dimensional hilbert space , and vanishing at infinity .
the poincar group acts on the symbols in @xmath31 by @xmath34 the manifold @xmath15 is the orbit of the standard symplectic matrix @xmath35 under the action @xmath36 of the full lorentz group . by definition , the states of @xmath31 with _ optimal localization _ ( both in space and in time )
minimize @xmath37 .
this characterization is evidently invariant under rotations and translations , but not under lorentz boosts .
it can be shown ( see @xcite ) that the optimally localized states are of the form @xmath38 where @xmath39 is any probability measure on the distinguished subset @xmath40 of @xmath15 , the orbit of @xmath41 under the action of @xmath42 , and @xmath43 is the localization map with localization centre @xmath44 , @xmath45 in what follows , we will need only the localization map with localization centre @xmath46 , and in order to simplify the notation we will denote it by @xmath47 .
however , the results below also hold for a general @xmath48 .
it is convenient to introduce the enveloping c*-algebra @xmath49 generated by the restrictions @xmath50 of the symbols to @xmath40 .
then the localization map @xmath47 is the composition @xmath51 of the restriction map @xmath52 , with a positive map @xmath53 from @xmath49 to @xmath54 , which is a conditional expectation in the sense that is not a subalgebra of @xmath55 , it is a subalgebra ( actually , the centre ) of the multiplier algebra @xmath56 . ]
@xmath57 , @xmath58 , @xmath59 .
@xmath47 will also denote the normal extension to the multiplier algebra @xmath60 .
then , @xmath61 as a constant function of @xmath62 . by analogy with the definition of @xmath63 , the evaluation of an ordinary quantum field @xmath64 on the quantum spacetime is given by @xmath65 and is to be interpreted as a map from states on @xmath31 to smeared field operators , @xmath66 where the r.h.s
. is a quantum field on the ordinary spacetime , smeared with the test function @xmath67 defined by @xmath68 . if products of fields are evaluated in a state , the r.h.s .
will in general involve nonlocal expressions . as in ordinary quantum field theory , due to
the singular properties of fields , products of fields are not _ a priori _ well - defined . on the ordinary minkowski spacetime ,
well - defined products of fields are given by the so - called wick products .
they may be defined by bringing the positive and negative frequency parts of the fields in the product into `` normal order '' , which in momentum space corresponds to putting all destruction operators to the right .
another definition , which , contrary to normal - ordering may also be applied on curved spacetimes , is given in terms of the formal evaluation on the diagonal of a suitably subtracted product ; one has , for instance , at second order @xmath69 .
the two constructions , while equivalent on the ordinary minkowski spacetime , lead to inequivalent generalizations on the quantum spacetime . in @xcite , for instance
, an interaction lagrangean was given in terms of the _ usual _ normal ordering of positive and negative frequency parts of @xmath70 , @xmath71 another possibility will be investigated in @xcite .
there , we consider products of fields at different points as they arise in the context of the yang - feldman equation , @xmath72 , @xmath73 .
we then define the so - called quasiplanar wick products by allowing only terms which are local in a certain sense to be subtracted , and show that they are well - defined on the diagonal , i.e. in the limit of coinciding points where @xmath74 . in this paper
we consider yet another approach .
the evaluation on the diagonal is replaced by a suitable generalization compatible with the uncertainty relations , leading to a regularized nonlocal effective interaction .
the idea is that a product of fields at different points , @xmath75 , may be defined by interpreting @xmath76 as _ mutually independent quantum coordinates _ , that is , by defining @xmath77 and @xmath78 now
, the different spacetime components of each variable @xmath79 , @xmath80 , no longer commute with one another , hence the limit @xmath81 loses its natural meaning .
we can , however , identify the central elements , i.e. take @xmath82=iq^{\mu\nu}$ ] for all @xmath83 .
this amounts to taking the tensor products in ( [ ind_coord ] ) not over the complex numbers , but over the centre @xmath84 of ( the multiplier algebra of ) @xmath85 .
the limit @xmath81 will then be replaced by a _ quantum diagonal map _ which on each function of @xmath79 evaluates a state minimizing the square euclidean length , while leaving the mean coordinates invariant ( cf .
@xcite , @xcite ) . as a consequence of taking the tensor product in ( [ ind_coord ] ) over @xmath86
, the mean coordinates @xmath87 commute with the relative coordinates @xmath79 ( in the strong sense ) , e.g. for @xmath88 , @xmath89= [ q^\mu , q^\nu]\otimes i- i\otimes[q^\mu , q^\nu]= 0.\ ] ] this fact turns out to be crucial for the construction of the quantum diagonal map and provides an additional motivation for taking the tensor product over @xmath86 , actually , the tensor product @xmath90 of @xmath86-moduli can be defined as the completion relative to the maximal c*-seminorm of the quotient of the algebraic tensor product over @xmath91 modulo the two sided ideal generated by the multiples in @xmath92 of @xmath93 , where @xmath94 varies in @xmath86 .
it may be equivalently described as the fibrewise tensor product of bundles of c*-algebras ( continuous fields of c*-algebras , in the terminology of @xcite ) over @xmath15 .
another motivation to use a @xmath86-module tensor product is to view the components @xmath17 of the coordinates commutator as universal data which are the same for the different variables corresponding to independent events .
the @xmath17s are thus treated as a point independent geometric background , which , however , is translation invariant and lorentz covariant .
since the c*-algebra @xmath31 describes the regular representations of ( 1 ) , i.e. integrable to a representation of the weyl relations @xmath95 the uniqueness theorem of von neumann @xcite , applied to each fibre at @xmath96 , ensures that commutativity implies tensor factorization over @xmath86 .
this fact will allow us to obtain the desired map as a conditional expectation .
furthermore , we will use the tensor product of @xmath97 copies of the basic algebra as an auxiliary algebra , where the mean coordinates are ( affiliated to the algebra ) in the first factor , and where the algebra to which the difference variables @xmath79 are affiliated is identified with a subalgebra of the auxiliary algebra , associated to the factors from slot @xmath98 to @xmath97 .
the desired quantum diagonal map @xmath99 is then obtained by evaluating @xmath100 on such tensor products , where @xmath47 is the localization map ( [ loc_map ] ) with localization centre @xmath46 , and where @xmath52 is the restriction map .
it turns out that the application of @xmath101 to functions whose symbol do not depend on @xmath11 explicitly , yields expressions which in turn are independent of @xmath11 .
the quantum diagonal map replaces the ordinary evaluation at coinciding points .
contrary to the ordinary case , it yields a well - defined expression when applied to a product of fields , @xmath102 since a nonlocal regularizing kernel @xmath103 appears . we conclude that contrary to the ordinary case , no infinite counterterms have to be subtracted and @xmath104 may be used directly to define the interaction in the quantum theory . regarding the combinatorics ,
it is , however , convenient to additionally apply ordinary normal ordering , and to define a _ quantum wick power _ as @xmath105 the quantum wick power @xmath106 as well as @xmath104 may be understood as functions of @xmath107 , not explicitly depending on @xmath11 , taking values in the field operators . in other words ,
@xmath106 and @xmath104 formally are elements of @xmath108 , where @xmath109 is the field algebra .
once products of fields are given a precise meaning , one may apply an appropriate perturbative setup . since sharp localization in time is compatible with the spacetime uncertainty relations ( at the cost of complete delocalization in space ) , _ one _ possibility is , for instance , to follow the standard approach to perturbation theory in the interaction representation , involving integrations at sharp fixed times @xcite .
if the lagrangean is symmetric , the resulting @xmath0-matrix is formally unitary by construction ( at least before renormalization ) . in @xcite ,
such an approach was proposed , based on the interaction lagrangean ( [ old_wick ] ) .
unfortunately , the resulting perturbation theory is not free of ultraviolet divergences .
this fact was first observed in @xcite where , however , instead of the interaction picture used in @xcite , the theory was defined in terms of modified feynman rules which may be formally derived from a path integral formulation . as
first observed in @xcite the resulting theory violates unitarity , a defect which may be traced back to the problem of time ordering on a ( space / time)-noncommutative theory , as discussed in @xcite : the time ordering naturally defined in the interaction picture formulation ( cf ( * ? ? ?
* eq . ( 6.15 ) ) and subsequent comments , as well as section [ diagramm ] of the present paper ) does not violate unitarity . as a consequence ,
the formulation of the theory in terms of modified feynman rules is not equivalent to the one discussed here .
another inequivalent approach , which , however , yields a unitary perturbation theory was proposed in @xcite .
this approach is based on the yang - feldman equation and will be discussed elsewhere @xcite . instead
, we will again apply the standard approach to perturbation theory in the interaction representation , this time employing the quantum wick products .
the interaction hamiltonian on the quantum spacetime is then given by @xmath110 as a constant operator valued function of @xmath40 ( i.e. @xmath111 is formally in @xmath112 ) .
while in @xcite one still had to handle the dependence of the hamiltonian on @xmath14 , in the approach adopted here , @xmath111 is a _ constant _ function of @xmath113 . as a consequence
, our procedure leads to a unique prescription for the interaction hamiltonian on quantum spacetime .
the resulting effective non local hamiltonian is @xmath114 where @xmath115 is the effective nonlocal interaction lagrangean @xmath116 it will be shown in section [ finite ] that the corresponding perturbation theory is free of ultraviolet divergences .
the ultraviolet regularization arises as a point
split regularization by convolution with gaussian kernels , and we will show that , by insertion of an adiabatic switch , the perturbation series is order by order finite , and each term is a well defined , closed operator with a common core
. the only remaining source of divergences is then given by possible infinite volume effects arising in the adiabatic limit , which will be discussed elsewhere .
the ultraviolet finiteness of the theory presented in this paper is in accordance with the expectation that noncommutativity of spacetime may regularize the theory .
other examples for ultraviolet finite theories on noncommutative spaces were discussed in @xcite , for instance compact spacetimes , corresponding to finite dimensional algebras .
it is noteworthy that the transition matrix elements will vanish as gaussian functions of the energies and momentum transfers expressed in planck units .
while in the high energy limit the transition amplitudes vanish rapidly as a result of the quantum delocalization of the interaction , in the low energy limit one would expect that the corrections to the ordinary theory on minkowski space vanish .
this is clearly possible only after a finite renormalisation ; the structure of the needed counterterms and the dependence upon the planck length of the renormalisation constants will be studied elsewhere .
note that in the limit where the planck length can be neglected , the renormalized theory on quantum spacetime should coincide with the ordinary renormalized theory on minkowski space . at the physical values of the planck length ,
the effect of the quantum nature of spacetime should manifest itself as quadratic or higher order corrections , since gravitation is not explicitly taken into account , but manifests itself only through the commutator of the coordinates .
a weak point of the approach to quantum field theory on quantum spacetime presented here is that , while , as was first shown in @xcite , the prescription leading to ( [ old_wick ] ) does not alter the _ free _
hamiltonian , the prescription discussed here would indeed change it , replacing it by a deformed operator which would no longer be the zero component of a lorentz vector .
we therefore treat , in this paper , the interaction on a different footing than the unperturbed hamiltonian which we identify with that of the usual free theory . as a consequence
, lorentz invariance is violated in an essential way , since optimally localized states are defined relative to a particular lorentz frame .
however , spacetime translation and space rotation invariance are preserved .
moreover , the evaluation of optimally localized states on the difference variables @xmath79 automatically restricts the joint eigenvalues @xmath117 of @xmath17 to @xmath40 , the basis of @xmath15 , where the electric and magnetic parts of @xmath14 are equal or opposite .
this gives an a posteriori motivation for a similar choice , made in @xcite , which was motivated by simplicity and by the need of preserving space rotation invariance .
according to the previous discussion , @xmath118 fulfill the relations ( for any @xmath119 ) @xmath120=i\lambda_p^2\delta_{ij } { \mathcal q}^{\mu\nu},\\ \label{qn2 } [ { q_j}^\mu,{\mathcal q}^{\nu\rho}]=0,\\ \label{qn3 } { \mathcal q}_{\mu\nu}{\mathcal q}^{\mu\nu}=0,\\ \label{qn4 } \left(\frac{1}{2}{\mathcal q}_{\mu\nu}(*\mathcal q)^{\mu\nu}\right)^2=i.\end{gathered}\ ] ] the correspondence @xmath121 extends to the generalized symbols @xmath122 as usually , where @xmath123 it induces a product and an involution on the generalized @xmath124-symbols , and the enveloping c*-algebra of the resulting algebra is precisely @xmath125 .
[ ideals ] note that , since @xmath126 as c*-algebras , @xmath127 . closed 2-sided ideals @xmath128 in @xmath129 are then in a 1 - 1 correspondence with closed ideals in @xmath86 ( the kernel of the restriction to @xmath86 of the canonical extension to @xmath60 of the projection map mod @xmath128 ) ,
hence are in a 1 - 1 correspondence with the closed subsets of @xmath15 .
let us now introduce the coordinates of the mean event , denoted mean coordinates for short , @xmath130 as well as the separations @xmath131 then @xmath132 since the commutator @xmath133=i{\mathcal q}^{\mu\nu}$ ] does not depend on @xmath83 , the following strong commutation relations hold : @xmath134 we have the following factorization .
let @xmath135 be coordinates with characteristic length @xmath136 , i.e. @xmath137=\,\frac{i}{n}\,q^{\mu\nu}$ ] .
define @xmath138 and @xmath139 we immediately check that the above elements also fulfill the relations ( [ qn1][qn4 ] ) in the regular form , where @xmath140=iq^{\mu\nu}$ ] . by von neumann uniqueness
( at each fixed @xmath14 ; see @xcite ) , there exists a faithful * -homomorphism @xmath141 such that @xmath142 this follows from the fact that regularity implies that the map @xmath143 determines a * -homomorphism @xmath144 ( whose canonical extension to @xmath60 will still be denoted by @xmath145 ) ; the ranges of @xmath145 and @xmath146 commute for @xmath147 and @xmath148 is an isomorphism independent of @xmath149 . by the universal properties of the tensor product and its uniqueness for nuclear c*-algebras ( as @xmath31 ) ,
there is a * -homomorphism @xmath150 of @xmath151 to @xmath152 , s.t .
@xmath153 , @xmath154 . by assumption
, @xmath150 is faithful on @xmath86 , hence , by remark [ ideals ] on page , @xmath150 is faithful .
explicitly , @xmath155 where , of course , @xmath156 the quantum diagonal map @xmath157 is defined as @xmath158 where @xmath159 are the localization map and the restriction to @xmath40 ( projection of @xmath31 onto @xmath49 ) , respectively .
note that the generators @xmath160 of the algebra in which @xmath101 takes values have characteristic length @xmath136 . to motivate this choice ,
let us recall that the difference variables @xmath161 fulfill the commutation relations ( [ q1][q4 ] ) , and a short computation yields @xmath162 ( as constant functions of @xmath62 ; compare with ( [ loc_car ] ) ) .
in other words , @xmath163 minimizes the euclidean separation @xmath164 .
@xmath101 will also denote the normal extension of the above map to the multiplier algebra @xmath165 .
note also that , had we used @xmath166 instead of @xmath167 , we would have defined the same map @xmath101 , since the separations @xmath168 are invariant under translations .
let @xmath169 , @xmath170 . the explicit form of the quantum diagonal map on @xmath171
is given by @xmath172 where @xmath173 equivalently , @xmath174 where @xmath175 with @xmath176 and , with @xmath177 , @xmath178 in particular , @xmath179 is a constant function of @xmath62 .
_ a simple computation yields , by the the definition ( [ bqi ] ) of @xmath180 , @xmath181 by the above and ( [ qbar_cr]),([qbar_cr2 ] ) , we have @xmath182 as a constant function of @xmath14 ; ( [ rn ] ) then follows by a straightforward computation .
standard computations provide the configuration space kernel.@xmath183 the quantum diagonal map takes a particular simple form if evaluated in optimally localized states .
indeed , let @xmath184 denote the localization map @xmath185 applied to the mean position coordinates @xmath186 .
then a simple calculation yields the formula @xmath187 since the function @xmath188 may be understood as the best commutative analogue of an element @xmath171 of the noncommutative algebra , this formula provides an additional justification of the present approach .
also the discussion in @xcite and @xcite . )
the uncertainty relations ( 2 ) are compatible with sharp localization in time , at the cost of total delocalization in space .
consistently , the centre valued map @xmath189 extends to a a positive partial trace @xmath190 ( see @xcite for details ) , which commutes with the restriction @xmath191 to @xmath40 . for a fixed choice of a frame of reference , we formulate a traditional perturbative setup in the spirit of @xcite .
consider for simplicity the @xmath192 interaction ; then the formal interaction hamiltonian will be defined as @xmath193 where the interaction lagrangean @xmath194 may be either the @xmath195 _ quantum wick power _ , defined by evaluating the quantum diagonal map on a normally ordered product of fields in the creation an destruction operators ( @xmath196 ) is called normally ordered or wick ordered and denoted @xmath197 , if all creation operators stand left of the destruction operators . ] , @xmath198 or the _ regularized product _ , defined by evaluating the quantum diagonal map on a product of fields as it stands , without application of normal order , @xmath199 clearly , the first definition yields a well - defined expression , but , contrary to the ordinary case , normal ordering is not necessary due to the regulating kernel @xmath103 which renders the second product well - defined as well , where @xmath200 denotes the ordinary convolution , is well - defined . ] .
in fact , both the quantum wick power @xmath106 as well as @xmath104 may be understood as functions of @xmath107 , not explicitly depending on @xmath11 , taking values in the field operators . in other words ,
@xmath106 and @xmath104 formally are elements of @xmath201 , where @xmath109 is the field algebra and @xmath104 are affine maps from states on @xmath31 to quadratic forms . ] . in the following ,
however , we will base our investigation on an interaction given by a quantum wick power .
for one thing , it simplifies the combinatorics , and in view of the adiabatic limit , which we hope to study in a later paper , normal ordering may even be necessary .
the resulting hamiltonian @xmath111 is formally affiliated to @xmath112 , where @xmath109 is the free bose field algebra on the ordinary spacetime .
roughly speaking , @xmath111 is a function from @xmath40 to ( formal ) field operators , i.e. to quadratic forms . in order to retrieve the hamiltonian of the equivalent theory on the ordinary spacetime , one has to integrate over some probability measure @xmath39 over @xmath40 , defining @xmath202 .
since , however , @xmath111 is a constant function of @xmath14 , the choice of @xmath39 is irrelevant .
this fact should be contrasted with the case considered in @xcite , where the non - irrelevant choice of a particular measure though the most reasonable in that context was to some extent arbitrary , as well as the special rle played by @xmath40 .
the resulting hamiltonian @xmath203 for the equivalent theory on the ordinary spacetime can then be put in the form ( [ h_i ] ) , where the effective nonlocal lagrangean is given by @xmath204 note that in the perturbation series ( ( [ bogodyson ] ) here below ) the time ordering of products @xmath205 will refer to the variables @xmath206 rather than to the integration variables in ( [ l_eff ] ) .
the fundamental result of this section is that the finite 4-volume theory yields a finite perturbation series .
more precisely , we turn the coupling constant @xmath207 into a smooth function of @xmath208 vanishing at infinity sufficiently fast , of the form @xmath209 , and we show that the corresponding dyson series is well defined at all orders . well - known methods from ordinary quantum field theory are employed . for any schwartz function @xmath207 of the form @xmath209 , @xmath210 , @xmath211 ,
the formal series @xmath212= t\exp\left\{-i\int d^4x\;\lambda(x){\mathcal l_{\text{eff}}}(x)\right\}= i+\sum_{n=1}^\infty(-i)^n s^{(n)}[\lambda]\ ] ] is finite at all orders .
more precisely , it is possible by wick reduction to put the @xmath213 order contribution in the form of a finite sum of closable operators with common core @xmath214 ( the subspace of the fock space consisting of the vectors with finitely many particles and with schwartz @xmath124-particle components for each @xmath124 ) . by construction ,
$ ] is unitary . while the existence of the adiabatic limit @xmath216 is questionable due to the breakdown of lorentz covariance , the infinite volume limit @xmath217 ( with @xmath218 fixed ) of the gell - mann low formula for @xmath219/\langle s[\lambda'\otimes\lambda'']\rangle_0 $ ] exists as a quadratic form .
indeed , the only terms in the perturbation expansion of @xmath220 $ ] which are divergent in the limit @xmath217 ( with @xmath218 fixed ) are precisely those containing vacuum vacuum parts .
the actual behaviour of the full adiabatic limit will be investigated elsewhere .
_ we shall follow standard conventions ( see e.g. @xcite ) : in particular , @xmath221 are the destruction and creation operators on the symmetric fock space @xmath222 , @xmath223 , and @xmath224 as quadratic forms on @xmath225 , where @xmath226 , and @xmath227 .
the cutoff lagrangian @xmath228 is given by @xmath229 we introduce the following compact notations @xmath230 standard computations yield @xmath231= & \int_{{\mathbb r}^{4nn}}d{\underline{a}}_1\dotsm d{\underline{a}}_n\ ; \bigg(\prod_{1\leq m < n}\theta\big(\kappa^0({\underline{a}}_{m+1}^0-{\underline{a}}_m^0)\big)\bigg)\\ & g_\lambda({\underline{a}}_1)\dotsm g_\lambda({\underline{a}}_n)\ ; { \protect{{\boldsymbol{:}\!\phi^{(n)}({\underline{a}}_1)\!\boldsymbol{:}}}}\dotsm{\protect{{\boldsymbol{:}\!\phi^{(n)}({\underline{a}}_n)\!\boldsymbol{:}}}},\end{aligned}\ ] ] where @xmath232 is the time component of the mean point @xmath233 of @xmath234 , and @xmath235 the set of all subsets of @xmath236 ( including the trivial subsets ) , we shall write @xmath237 for any choice of @xmath238 elements of @xmath239 , namely @xmath240 with @xmath241 , @xmath242 . then , by ( [ freefield ] ) , we get @xmath243=&\sum_{j_\bullet } \int_{\mathbb r^{3nn}}d{\underline{\boldsymbol{k}}}_1\dotsm d{\underline{\boldsymbol{k}}}_n \frac{k_\lambda^{j_\bullet}({\underline{\boldsymbol{k}}}_1,\dotsc,{\underline{\boldsymbol{k}}}_n ) } { \sqrt{\omega^{(n)}({\underline{\boldsymbol{k}}}_1)\dotsm \omega^{(n)}({\underline{\boldsymbol{k}}}_n)}}\\ \nonumber & \bigg(\prod_{u_1\in j_1}a^\dagger({\boldsymbol{k}}_{u_1})\bigg ) \bigg(\prod_{v_1\in i_n\backslash j_1}a({\boldsymbol{k}}_{v_1})\bigg)\dotsm\\ \label{sk } & \dotsm\bigg(\prod_{u_n\in j_n}a^\dagger({\boldsymbol{k}}_{u_n})\bigg ) \bigg(\prod_{v_n\in i_n\backslash j_n}a({\boldsymbol{k}}_{v_n})\bigg)\end{aligned}\ ] ] where @xmath244 here @xmath245 is a diagonal @xmath246 matrix , with diagonal entries @xmath247 and @xmath248 by the second wick theorem , each term in the above sum may be written as a sum of terms with contractions , where the surviving creation and destruction operators appear in normal order ( creation on the left ) , and the contracted pairs are replaced by @xmath249 distributions . since each integration variable @xmath250 appears as the argument of precisely one operator @xmath251 , ( @xmath196 ) , then the arising product of @xmath252s
is well defined by the following elementary the map [ good ] @xmath253 on the schwartz functions , with @xmath254 a real linear functional on @xmath255 , @xmath256 , is a well defined distribution if and only if the functionals @xmath257 are linearly independent . by performing @xmath258 integrations , the above distribution always takes the form @xmath259 where the linear maps @xmath260 , @xmath261 , provide a linear injection of @xmath262 into @xmath255 .
after complete wick reduction , the generic term will be of the form @xmath263 where @xmath264 for a suitable set of linear maps ( see remark [ good ] ) @xmath265 with @xmath266 .
corresponds to the term with no contractions ; in that case , the above mentioned linear maps reduce to renaming the integration variables . ]
the proof is complete by ( * ? ? ?
* theorem x.44 ) , if we show that the function @xmath267 is in @xmath268 . to this end
, it is enough to prove that @xmath269 is bounded by a schwartz function .
due to the form of @xmath207 , the integrations over the space and time variables in ( [ k ] ) factorize , and @xmath270 is the product of a function with gaussian decay with the fourier transform evaluated at some point continuously depending on the @xmath271 s of an @xmath272 function ; the latter is then a bounded , continuous function .
indeed , we will show in appendix [ lambdasecondoproof ] that @xmath273 is a schwartz function , by carrying out explicitly the above mentioned computations.@xmath183 note that the kernels @xmath274 appearing in the above proof are transition amplitudes of scattering processes with @xmath258 incoming and @xmath275 outgoing particles .
they decay as gaussian functions of the energies and momentum transfers expressed in planck units .
we complement our discussion with some formal remarks aiming to clarify the relation between the approach followed here and the usual formulation of the perturbation theory in terms of feynman diagrams .
we have already shown in @xcite that the perturbative setup dealt with in @xcite ( as well as the approach based on the yang - feldman equation ) is inequivalent to the by now standard setup in terms of the modified feynman rules @xcite . in this section
we will make explicit that in the framework of the regularized interaction proposed in this paper , feynman propagators are no more available at all . in this discussion
we ignore all problems which may arise in the adiabatic limit , and use the formal ( i.e. defined with constant coupling constant @xmath207 ) time dependent interaction hamiltonian @xmath203 throughout , @xmath276 where the integral kernel @xmath277 is of the form @xmath278 and the gaussian kernel @xmath279 does not depend on @xmath280 .
introducing the time ordered kernels @xmath281 formal computations yield the formal dyson expansion ( see @xcite ) @xmath282 where we use the notations ( [ smart1][smart8 ] ) . by integrating over the time variables @xmath283 , we obtain @xmath284 here @xmath285 is the ordering of the wick monomials with respect to the average time , i.e. the time of the mean position of each @xmath286 : more precisely , with @xmath287 the time component of the four vector @xmath288 , we have @xmath289 where @xmath290 can be thought of as a `` fat vertex '' , i.e. a multi - vertex actually consisting of @xmath124 minkowski vectors .
the time ordering thus defined , as well as the one discussed in @xcite , arises from ordering the variables @xmath206 in products @xmath205 .
however , in both approaches causality is violated at the planck length scale ( cf .
also @xcite ) .
a typical manifestation of the violation of causality is the following : suppose that @xmath291 i.e. @xmath292 , then it is not forbidden that @xmath293 for some @xmath294 . in other words , a single field @xmath295 belonging to the first wick monomial can be subsequent in time to some field @xmath296 .
note however , that the above picture is consistent with the request that the large scale limit should reproduce the ordinary ( non regularized ) theory with @xmath297 interaction ; in particular , as @xmath298 , the multi - vertices shrink to their mean positions , and the above average time - ordering reduces to the usual one .
we conclude our discussion by showing that it is not possible to absorb the time ordering in feynman propagators , @xmath299\omega)$ ] , as one usually does in the framework of ordinary local theories .
in fact , in the approach followed here , we can not construct feynman propagators _ at all _ , while in the original hamiltonian approach based on ( [ old_wick ] ) as well as in the approach based on the yang - feldman equation it is still possible to rewrite the time ordering in terms of feynman propagators _ together with _
advanced and retarded propagators @xcite . to see what happens in the approach adopted here ,
consider as an example the second order contribution to the dyson series with two internal contractions in quantum @xmath300-interaction , as depicted in figure [ qfish ] . here ,
the two multi - vertices are sets of four distinguishable points , @xmath301 and @xmath302 , and at most one line originates from each of these points .
a kernel @xmath279 belongs to each multi - vertex .
the internal contractions lead to positive and negative frequency parts of the commutator function , @xmath303 , where @xmath304 . ,
while the time ordering is done with respect to the time components of the mean coordinates of each vertex , leading to the heaviside functions @xmath305 .
we may thus conclude that the distributions arising in the fish graph will not yield feynman propagators at all .
this is consistent with the fact that the usual interpretation of feynman graphs as pictorial representations of scattering processes ( `` first a is annihilated , then b is created '' ) should break down on quantum spacetime , as it needs sharply localized events , which are no longer available .
in this appendix , we explicitly compute the kernel @xmath273 , for future reference .
recall that , for any @xmath306 , we write @xmath307 , and @xmath308 .
let @xmath309 be any schwartz function , and @xmath209 . by equation ( [ k ] ) , @xmath310 where @xmath311 is given by @xmath312 the orthogonal matrix @xmath245 is given by ( [ u ] ) .
we now set @xmath313 then , if @xmath314 is the orthogonal projection of @xmath315 onto @xmath316 , we set @xmath317 as a projection onto @xmath318 , @xmath319 ; in the same spirit , @xmath320 .
finally , @xmath321 , and @xmath322 is the identity .
we finally need the maps @xmath323 , @xmath324 from @xmath315 to @xmath325 , with the corresponding maps @xmath326 , @xmath327 from @xmath318 to @xmath328 .
we may now take a @xmath246 real orthogonal matrix @xmath333 , such that @xmath334 then , @xmath335 is the orthogonal projection of @xmath336 onto @xmath337 , and @xmath338 , as usually .
furthermore , @xmath339 .
with the change of integration variables @xmath340 , we obtain @xmath341 the integration over the variables @xmath342 , @xmath343 , is completely separated from the integration over the variables @xmath344 .
hence @xmath345 the first factor is a fourier transform of a gaussian function .
renaming the integration variables @xmath346 , @xmath347 where we used @xmath348 the integral over @xmath353 in the above expression is the fourier transform of an @xmath272 function , hence it vanishes at infinity .
moreover , it is the product of a schwartz function in @xmath354 , times a @xmath355 function of @xmath356 , vanishing at infinity . as a consequence ,
@xmath357 is a schwartz function .
h. bozkaya , p. fischer , h. grosse , m. pitschmann , v. putz , m. schweda and r. wulkenhaar , _ space / time noncommutative field theories and causality _ , tuw preprint 02 - 18 [ arxiv : hep - th/0209253 ] . m. chaichian , a. demichev and p. presnajder , _ quantum field theory on noncommutative spaces and the persistence of ultraviolet divergences _ , nucl . phys .
* 567 * , 360390 ( 2000 ) [ arxiv : hep - th/0003270 ] .
s. cho , r. hinterding , j. madore and h. steinacker , _ finite field theory on noncommutative geometries _ ,
int . j. mod .
d * 9 * ( 2000 ) 161 [ arxiv : hep - th/9903239 ] .
a. connes , m. r. douglas , and a. schwarz,_noncommutative geometry and matrix theory : compactification on thori _ , jhep*9802 * , 003 ( 1998 ) [ arxiv : hep - th/9711162 ] .
9711162 ; dyson , _
the @xmath0-matrix in quantum electrodynamics _ , phys
. rev . * 75 * , 1736 ( 1949 ) .
t. filk , _ divergencies in a field theory on quantum space _ , phys .
b * 376 * , 53 ( 1996 ) .
j. gomis and t. mehen , _ space - time noncommutative field theories and unitarity _ nucl .
b * 591 * , 265276 ( 2000 ) [ arxiv : hep - th/0005129 ] .
k. fredenhagen , _ quantum fields and noncommutative space - time _ , in f. scheck , w. werner , h. upmeier ( eds . ) , _ noncommutative geometry and the standard model of elementary particle physics .
proceedings , hesselberg march 1999 _ , lecture notes in physics * 596 * , springer ( 2002 ) | we discuss a formulation of quantum field theory on quantum space time where the perturbation expansion of the @xmath0-matrix is term by term ultraviolet finite .
the characteristic feature of our approach is a quantum version of the wick product at coinciding points : the differences of coordinates @xmath1 are not set equal to zero , which would violate the commutation relation between their components .
we show that the optimal degree of approximate coincidence can be defined by the evaluation of a conditional expectation which replaces each function of @xmath2 by its expectation value in optimally localized states , while leaving the mean coordinates @xmath3 invariant .
the resulting procedure is to a large extent unique , and is invariant under translations and rotations , but violates lorentz invariance . indeed
, optimal localization refers to a specific lorentz frame , where the electric and magnetic parts of the commutator of the coordinates have to coincide @xcite . employing an adiabatic switching
, we show that the s - matrix is term by term finite .
the matrix elements of the transfer matrix are determined , at each order in the perturbative expansion , by kernels with gaussian decay in the planck scale . the adiabatic limit and
the large scale limit of this theory will be studied elsewhere . _
dedicated to rudolf haag on the occasion of his 80@xmath4 birthday . _ |
two novel phenomena have been observed at intermediate transverse momentum ( @xmath1gev/@xmath2 ) in heavy - ion collisions at rhic .
first , the azimuthal di - hadron correlation with a high @xmath0 ( @xmath3 gev/@xmath2 ) trigger particle is significantly broadened on the away - side ( double - peak structure ) in central au+au collisions relative to p+p collisions @xcite .
second , the bayon to meson ratio is significantly enhanced @xcite@xcite . in order to understand the physics mechanism(s ) for these phenomena , we study two - particle azimuthal correlation of net - charge and compare to that of total - charge . since the net - charge is dominated by net - protons , and the total - charge is the sum of mesons and bayons , the comparison between net - charge and total - charge azimuthal correlations may help further our understanding of baryon - meson effects at rhic .
we obtain the raw azimuthal correlations from positively ( @xmath4 ) and negatively ( @xmath5 ) charged hadrons with a charged high @xmath0 trigger particle ( @xmath6 ) . the trigger and associated particle @xmath0 ranges are @xmath7 gev/@xmath2 and @xmath8 gev/@xmath2 , respectively . both the trigger and associate particles
are restricted to @xmath9 .
tracking efficiency is corrected for associated particles .
the two - particle acceptance effect is corrected for by the event - mixing technique .
the correlations are normalized per trigger particle . by taking the difference and the sum
, we obtain the net - charge ( @xmath10 ) and total - charge ( @xmath11 ) correlations : @xmath12 @xmath13 we obtain the corresponding backgrounds by using the same technique from mixed - events and adding in @xmath14 modulation , where @xmath14 is presently taken from the charged hadron measurements @xcite@xcite .
the final correlation signal is given by @xmath15 where @xmath16 is the normalization factor from zero yield at 1 radian ( zya1 ) method .
the systematic uncertainty estimate is done according to ref .
@xcite .
background subtracted azimuthal correlations of net - charge ( data points ) and total - charge ( solid histogram ) in 12% central au+au collisions at 200 gev .
the trigger and associated @xmath17 ranges are 3 - 4 gev/@xmath2 and 1 - 3 gev/@xmath2 , respectively .
errors shown for the data points are statistical only .
the total - charge result is scaled by 1/15 to compare with the net - charge result .
the dashed histograms show the systematic errors on the total - charge result , obtained from background subtraction using the modified reaction - plane @xmath14 @xcite and the 4-particle cumulant @xmath14 @xcite.,scaledwidth=70.0% ]
figure [ fig : corr ] shows the background subtracted results .
we find the correlation shapes are similar between total - charge and net - charge .
the net - charge correlation shows a significant away - side broadening ( double - peak structure ) , similar to previously observed in total - charge @xcite .
we study the @xmath0 dependence of the correlated yields by dividing the data into four associated @xmath0 bins . in different @xmath0 bins ,
the correlation shapes are also found to be similar between total - charge and net - charge .
@xmath17 spectra of net - charge ( triangles ) and total - charge ( squares ) correlated with a high @xmath17 trigger particle ( @xmath18 gev/@xmath2 ) on both near - side ( filled symbols ) and away - side ( open symbols ) in auau central collisions at 200 gev .
errors shown are systematic errors .
the inclusive ( round points ) @xmath17 spectrum is also shown for comparison .
, scaledwidth=70.0% ] we obtain associated particle @xmath0 spectra by integrating over the near - side ( @xmath19 ) and the away - side ( @xmath20 ) .
the obtained spectra are shown in fig .
[ fig : spec ] . for comparison
, we also show the inclusive charged hadron @xmath0 spectrum ( i.e. without requirement of a high @xmath0 trigger particle ) .
we fit the spectra by an exponential function : @xmath21 , where @xmath22 is the inverse slope parameter .
the fit results are listed in table [ table : t ] .
.inverse slope parameter @xmath22 in gev/@xmath2 from exponential fit to the @xmath0 spectra .
errors are statistical only . [ cols="^,^,^",options="header " , ] as seen from the extracted values of the inverse slope parameter , the shapes of net - charge and total - charge correlated spectra on near - side are similar , and they are both harder than the inclusive spectrum .
the away - side correlated net - charge spectrum seems to be harder than the total - charge , and neither is much harder than the inclusive one .
ratios of correlated yields of net - charge over total - charge on away - side ( squares ) and near - side ( triangles ) in 12% central au+au collisions at 200 gev .
the trigger @xmath17 range is from 3 - 4 gev/@xmath2 .
vertical bars are statistical errors .
horizontal bars are the @xmath17 bin size . the curve shows the inclusive ratio of @xmath23 , where the shaded area is statitical uncertainty ( the data at @xmath242.5 gev/@xmath2 are from tof measurement , and the @xmath252.5 gev/@xmath2 data are from tpc measurement ) @xcite . , scaledwidth=70.0% ] the inclusive bayon to meson ratio is strongly enhanced at intermediate @xmath0 as mentioned earlier .
we want to study the bayon / meson ratio of the correlated hadrons with high @xmath0 trigger particles .
we take the ratio of the @xmath0 spectra on near- and away - side separately .
the obtained results are shown in fig .
[ fig : ratio ] , and compared with the inclusive ratio .
the ratio of the correlated net - charge to total - charge yields on away - side is larger than that on the near - side .
the away - side ratio is similar to the inclusive ratio , with an average difference of @xmath26 .
the near - side ratio seems smaller than the inclusive ratio with an average difference of @xmath27 .
figure [ fig : ratio ] shows that the near - side ratio is independent of @xmath0 .
a fit of the near - side ratio to a constant yields @xmath28 , with a @xmath29 .
this is to be compared to the inclusive ratio of @xmath30 . on the other hand
, the away - side ratio appears to increase with @xmath0 , similar to the inclusives . a constant fit yields @xmath31 , with a significantly larger @xmath32 .
we have presented results on azimuthal correlation of net - charge with high @xmath0 trigger particles . the net - charge correlation shape is found to be similar to that of total - charge .
we studied the associate @xmath0 spectra of correlated net - charge , and the net - charge to total - charge ratio . on the near - side , net - charge and total - charge spectra
have similar shape ; the net - charge / total - charge ratio is constant over @xmath0 .
however , they are both harder than the inclusives , suggesting jet nature for the correlated particles . on the away - side ,
the correlated spectra are not much harder than the inclusives , suggesting partial equilibration of the correlated hadrons with the medium .
the net - charge / total - charge ratio increases with @xmath0 and is similar to inclusives . | 1.0 cm 1.0 cm we presented results on azimuthal correlation of net - charge with high @xmath0 trigger particles .
it is found that the net - charge correlation shape is similar to that of total - charge . on the near - side ,
the net - charge and total - charge @xmath0 spectra have similar shape and both are harder than the inclusives . on the away - side ,
the correlated spectra are not much harder than the inclusives , and the net - charge / total - charge ratio increases with @xmath0 and is similar to the inclusive ratio . |
the warping degree and a monotone diagram is defined by kawauchi for an oriented diagram of a knot , a link @xcite or a spatial graph @xcite .
the warping degree represents such a complexity of a diagram , and depends on the orientation of the diagram . for an oriented link diagram @xmath0
, we say that @xmath0 is monotone if we meet every crossing point as an over - crossing first when we travel along all components of the oriented diagram with an order by starting from each base point .
this notion is earlier used by hoste @xcite and by lickorish - millett @xcite in computing polynomial invariants of knots and links .
the warping degree @xmath1 of an oriented link diagram @xmath0 is the smallest number of crossing changes which are needed to obtain a monotone diagram from @xmath0 in the usual way .
we give the precise definitions of the warping degree and a monotone diagram in section 2 .
let @xmath3 be the diagram @xmath0 with orientations reversed for all components , and we call @xmath3 the inverse of @xmath0 .
let @xmath5 be the crossing number of @xmath0 .
we have the following theorem in @xcite which is for a knot diagram : @xcite let @xmath0 be an oriented knot diagram which has at least one crossing point . then we have @xmath6 further , the equality holds if and only if @xmath0 is an alternating diagram .
[ dk ] let @xmath0 be a diagram of an @xmath7-component link ( @xmath8 ) . let @xmath9 be a diagram on a knot component @xmath10 of @xmath11 , and we call @xmath9 a component of @xmath0 .
we define a property of a link diagram as follows : a link diagram @xmath0 has _ property @xmath12 _ if every component @xmath9 of @xmath0 is alternating , and the number of over - crossings of @xmath9 is equal to the number of under - crossings of @xmath9 in every subdiagram @xmath13 for each @xmath14 .
note that a diagram @xmath0 has property @xmath12 if @xmath0 is an alternating diagram in the case that @xmath15 .
we generalize theorem [ dk ] to a link diagram : let @xmath0 be an oriented link diagram , and @xmath4 the number of components @xmath9 such that @xmath9 has at least one self - crossing
. then we have @xmath16 further , the equality holds if and only if @xmath0 has property @xmath12 .
[ mainthm ] for example , the link diagram @xmath0 in figure [ 8 - 2_5 ] has @xmath17 .
let @xmath0 be a diagram of a link .
let @xmath18 be the unlinking number of @xmath0 .
as a lower bound for the value @xmath2 , we have the following inequality : we have @xmath19 the rest of this paper is organized as follows . in section 2 , we define the warping degree @xmath1 of an oriented link diagram @xmath0 . in section 3 , we define the linking warping degree @xmath20 , and consider the value @xmath21 to prove theorem [ mainthm ] . in section 4
, we show relations of the linking warping degree and the linking number . in section 5
, we apply the warping degree to a link itself . in section 6
, we study relations to unknotting number and crossing number . in section 7
, we define the splitting number and consider relations between the warping degree and the splitting number . in section 8 , we show methods for calculating the warping degree and the linking warping degree .
let @xmath11 be an @xmath7-component link , and @xmath0 a diagram of @xmath11 . we take a sequence @xmath22 of base points @xmath23 ( @xmath24 ) , where every component has just one base point except at crossing points .
then @xmath25 , the pair of @xmath0 and @xmath22 , is represented by @xmath26 with the order of @xmath22 .
a self - crossing point @xmath27 of @xmath28 is a _
warping crossing point of _ @xmath28 if we meet the point first at the under - crossing when we go along the oriented diagram @xmath28 by starting from @xmath23 ( @xmath24 ) .
a crossing point @xmath27 of @xmath28 and @xmath29 is a _
warping crossing point between _ @xmath28 _ and _ @xmath29 if @xmath27 is the under - crossing of @xmath28 ( @xmath30 ) .
a crossing point @xmath27 of @xmath25 is a _
warping crossing point of _ @xmath25 if @xmath27 is a warping crossing point of @xmath28 or a warping crossing point between @xmath28 and @xmath29 @xcite .
for example in figure [ wcp ] , @xmath27 is a warping crossing point of @xmath31 , and @xmath32 is a warping crossing point between @xmath31 and @xmath33 .
we define the warping degree for an oriented link diagram @xcite .
the _ warping degree _ of @xmath25 , denoted by @xmath34 , is the number of warping crossing points of @xmath25 .
the _ warping degree _ of @xmath0 , denoted by @xmath1 , is the minimal warping degree @xmath34 for all base point sequences @xmath22 of @xmath0 .
ozawa showed that a non - trivial link which has a diagram @xmath0 with @xmath35 is a split union of a twist knot or the hopf link and @xmath7 trivial knots ( @xmath36 ) @xcite .
fung also showed that a non - trivial knot which has a diagram @xmath0 with @xmath35 is a twist knot @xcite .
for an oriented link diagram and its base point sequence @xmath26 , we denote by @xmath37 the number of warping crossing points of @xmath28 .
we denote by @xmath38 the number of warping crossing points between @xmath28 and @xmath29 . by definition
, we have that @xmath39 thus , the set of the warping crossing points of @xmath25 is divided into two types in the sense that the warping crossing point is self - crossing or not . the pair @xmath25 is _ monotone _ if @xmath40 .
for example , @xmath25 depicted in figure [ monotone ] is monotone .
note that a monotone diagram is a diagram of a trivial link .
hence we have @xmath41 , where @xmath18 is the unlinking number of @xmath0 ( @xcite , @xcite ) .
in this section , we prove theorem [ mainthm ] . we first define the linking warping degree , which is like a restricted warping degree and which has relations to the crossing number and the linking number ( see also section 4 ) .
the number of non - self warping crossing points does not depend on the orientation .
we define the _ linking warping degree _ of @xmath25 , denoted by @xmath42 , by the following formula : @xmath43 where @xmath44 are components of @xmath25 @xmath45 . the _ linking warping degree _ of @xmath0 , denoted by @xmath20 , is the minimal @xmath42 for all base point sequences @xmath22 .
it does not depend on any choices of orientations of components .
for example , the diagram @xmath0 in figure [ stacked ] has @xmath46 .
a pair @xmath25 is _ stacked _ if @xmath47 . a diagram @xmath0 is _ stacked _ if @xmath48 .
for example , the diagram @xmath49 in figure [ stacked ] is a stacked diagram .
we remark that a similar notion is mentioned in @xcite .
note that a monotone diagram is a stacked diagram .
a link @xmath11 is _ completely splittable _ if @xmath11 has a diagram @xmath0 without non - self crossings .
notice that a completely splittable link has some stacked diagrams .
the _ linking crossing number _ of @xmath0 , denoted by @xmath50 , is the number of non - self crossing points of @xmath0 .
remark that @xmath50 is always even . for an unordered diagram @xmath0
, we assume that @xmath9 and @xmath13 denote subdiagrams of @xmath0 with an order .
we have the following relation of linking warping degree and linking crossing number .
we have @xmath51 further , the equality holds if and only if the number of over - crossings of @xmath9 is equal to the number of under - crossings of @xmath9 in every subdiagram @xmath13 for every @xmath14 .
let @xmath22 be a base point sequence of @xmath0 , and @xmath52 the base point sequence @xmath22 with the order reversed .
we call @xmath52 the reverse of @xmath22
. since we have that @xmath53 , we have the inequality @xmath54 .
let @xmath0 be a link diagram such that the number of over - crossings of @xmath9 is equal to the number of under - crossings of @xmath9 in every subdiagram @xmath13 for each @xmath14 .
then we have @xmath55 for every base point sequence @xmath22 .
hence we have @xmath56 . on the other hand
, we consider the case the equality @xmath57 holds .
for an arbitrary base point sequence @xmath22 of @xmath0 and its reverse @xmath52 , we have @xmath58 then we have @xmath59 . hence we have @xmath60 for every base point sequence @xmath22 .
let @xmath61 be the base point sequence which is obtained from @xmath62 by exchanging @xmath63 and @xmath64 ( @xmath65 ) .
then , the number of over - crossings of @xmath66 is equal to the number of under - crossings of @xmath66 in the subdiagram @xmath67 of @xmath25 because we have @xmath68 .
this completes the proof .
[ 2dc ] we next consider the value @xmath21 for an oriented link diagram @xmath0 and the inverse @xmath3 .
we have the following proposition : let @xmath0 be an oriented link diagram .
the value @xmath21 does not depend on the orientation of @xmath0 .
let @xmath69 be @xmath0 with the same order and another orientation .
since we have @xmath70 or @xmath71 , we have @xmath72 for each @xmath9 and @xmath73
. then we have @xmath74 a link diagram is a _ self - crossing diagram _ if every component of @xmath0 has at least one self - crossing . in other words , a diagram @xmath0 of an @xmath7-component link @xmath11
is a self - crossing diagram if @xmath75 .
we have the following lemma : let @xmath0 be a self - crossing diagram of an @xmath7-component link
. then we have @xmath76 further , the equality holds if and only if @xmath0 has property @xmath12 .
we have @xmath77 where the first inequality is obtained by theorem [ dk ] , and the second inequality is obtained by lemma [ 2dc ] .
hence we have the inequality .
the equality holds if and only if @xmath0 has property @xmath12 which is obtained by theorem [ dk ] and lemma [ 2dc ] .
[ self - crossing ] we give an example of lemma [ self - crossing ] . in figure
[ parallel ] , there are three diagrams with 12 crossings .
the diagram @xmath0 is a diagram such that any component is alternating and has 3 over - non - self crossings and 3 under - non - self crossings .
then we have @xmath78 .
the diagram @xmath69 is a diagram which has a non - alternating component diagram .
then we have @xmath79 .
the diagram @xmath80 is a diagram such that a component has 2 over - non - self crossings and 4 under - non - self crossings .
then we have @xmath81 .
lemma [ self - crossing ] is only for self - crossing link diagrams .
we prove theorem [ mainthm ] which is for every link diagram .
_ proof of theorem [ mainthm ] .
_ for every component @xmath9 such that @xmath9 has no self - crossings , we apply a reidemeister move of type i as shown in figure [ lr-1 ] .
then we obtain the diagram @xmath73 from @xmath9 , and @xmath73 satisfies @xmath82 and @xmath83 .
for example the base points @xmath23 , @xmath84 in figure [ lr-1 ] satisfy @xmath85 , @xmath86 .
we remark that every @xmath9 and @xmath73 are alternating .
we denote by @xmath69 the diagram obtained from @xmath0 by this procedure . since every component has at least one self - crossing , we apply lemma [ self - crossing ] to @xmath69
. then we have @xmath87 and we obtain @xmath88 hence we have @xmath16 the equality holds if and only if @xmath0 has property @xmath12 . @xmath89
in this section , we consider the relation of the linking warping degree and the linking number . for a crossing point @xmath27 of an oriented diagram
, @xmath90 denotes the sign of @xmath27 , namely @xmath91 if @xmath27 is a positive crossing , and @xmath92 if @xmath27 is a negative crossing . for an oriented subdiagram @xmath13 ,
the _ linking number _ of @xmath9 with @xmath93 is defined to be @xmath94 the linking number of @xmath9 with @xmath93 is independent of the diagram ( cf .
@xcite , @xcite ) .
we have a relation of the linking warping degree and the linking number of a link diagram in the following proposition : for a link diagram @xmath0 , we have the following ( i ) and ( ii ) .
( i ) : : we have @xmath95 further , the equality holds if and only if under - crossings of @xmath9 in @xmath13 are all positive or all negative with an orientation for every subdiagram @xmath13 ( @xmath96 ) .
( ii ) : : we have @xmath97 ( i ) : : for a subdiagram @xmath13 ( @xmath96 ) with @xmath98 , we show that @xmath99 let @xmath100 be the warping crossing points between @xmath9 and @xmath93 , and @xmath101,@xmath102,@xmath103 the signs of them .
since a stacked diagram is a diagram of a completely splittable link , we have @xmath104 by applying crossing changes at @xmath100 for @xmath13 .
then we have @xmath105 hence we obtain @xmath95 the equality holds if and only if under - crossings of @xmath9 in @xmath13 are all positive or all negative with an orientation for every subdiagram @xmath13 ( @xmath96 ) .
+ ( ii ) : : by the above equality ( [ sign - o - u ] ) , we observe that @xmath106 , where @xmath107 ( resp .
@xmath108 ) is an under - crossing ( resp .
an over - crossing ) of @xmath9 in @xmath13 , @xmath109 and @xmath110 .
a similar fact is also mentioned in @xcite .
we have @xmath111 hence we have the modular equality @xmath112 [ lld ] in figure [ ex3components ] , @xmath0 has @xmath113 , @xmath49 has @xmath114 , and @xmath115 has @xmath116 , where @xmath117 of @xmath0 denotes that @xmath118 , @xmath119 , and @xmath120 .
the _ total linking number _ of an oriented link @xmath11 is defined to be @xmath121 with a diagram and an order .
we have the following corollary : we have @xmath122 where @xmath22 is a base point sequence of @xmath0 . [ total ]
corollary [ total ] is useful in calculating the total linking number of a diagram .
for example in figure [ linkcor ] , the diagram @xmath0 with @xmath123 components and @xmath124 crossing points has @xmath125 .
we have that the total linking number of @xmath0 is @xmath126 by summing the signs of only @xmath123 crossing points .
in this section , we consider the minimal @xmath21 for minimal crossing diagrams @xmath0 of @xmath11 in the following formula : @xmath127 where @xmath128 denotes the crossing number of @xmath11 . in the case where @xmath129 is a non - trivial knot , we have @xmath130 further , the equality holds if and only if @xmath129 is a prime alternating knot @xcite . note that the condition for the equality of ( [ ek ] ) requires that @xmath0 is a minimal crossing diagram in the definition of @xmath131 .
we next define @xmath132 and @xmath133 as follows : @xmath134 @xmath135 as a generalization of the above inequality ( [ ek ] ) , we have the following theorem : for an @xmath7-component link @xmath11 , we have @xmath136 further , the equality holds if and only if every self - crossing diagram @xmath0 of @xmath11 with @xmath137 has property @xmath12 .
let @xmath0 be a self - crossing diagram of @xmath11 with @xmath138 .
we assume that @xmath0 satisfies the equality @xmath139
. then we have @xmath140 where the first inequality is obtained by theorem [ dk ] , and the second inequality is obtained by lemma [ 2dc ] .
if @xmath0 has a non - alternating component @xmath9 , or @xmath0 has a diagram @xmath13 such that the number of over - crossings of @xmath9 is not equal to the number of under - crossings of @xmath9 , then we have @xmath141 . on the other hand , the equality holds if @xmath0 has property @xmath12 .
we have the following example : for non - trivial prime alternating knots @xmath142 ( @xmath143 ) , we have a non - splittable link @xmath11 by performing @xmath144-full twists for every @xmath10 and @xmath145 ( @xmath24 ) with @xmath146 as shown in figure [ full ] , where we assume that @xmath147 and @xmath148 have the same sign .
note that we do not change the type of knot components @xmath10 .
let @xmath0 be a diagram of @xmath11 with @xmath149 .
then we notice that @xmath0 is a self - crossing diagram with @xmath137 .
we also notice that @xmath0 has property @xmath12 because @xmath150 and @xmath151 , and @xmath152 and @xmath153 in the case where @xmath154 .
hence we have @xmath155 in this case .
we have the following corollary : let @xmath11 be an @xmath7-component link whose all components are non - trivial .
then we have @xmath156 further , the equality holds if and only if every diagram @xmath0 of @xmath11 with @xmath149 has property @xmath12 .
since every diagram @xmath0 of @xmath11 is a self - crossing diagram , we have @xmath157 and @xmath158 .
we also consider the minimal @xmath2 and the minimal @xmath4 for diagrams @xmath0 of @xmath11 in the following formulas : @xmath159 @xmath160 note that the value @xmath161 and @xmath162 also do not depend on the orientation of @xmath11 .
jin and lee mentioned in @xcite that every link has a diagram which restricts to a minimal crossing diagram for each component .
then we have the following proposition : the value @xmath162 is equal to the number of non - trivial knot components of @xmath11 .
the following corollary is directly obtained from theorem [ mainthm ] .
we have @xmath163 for a diagram @xmath0 with @xmath149 , we have @xmath164 where the second inequality is obtained by theorem [ mainthm ] . [ fcl ]
we have the following question : when does the equality @xmath165 hold ? in figure [ ex - fc ] , there are two link diagrams @xmath0 and @xmath49 .
we assume that @xmath0 ( resp .
@xmath49 ) is a diagram of a link @xmath11 ( resp .
@xmath166 ) .
we have @xmath167 because we have @xmath168 and we know @xmath169 , @xmath170 , and @xmath171 , where @xmath9 is any diagram of @xmath172 . on the other hand
, we have that @xmath173 because @xmath174 .
in this section , we enumerate several relations of the warping degree , the unknotting number or unlinking number , and the crossing number .
let @xmath175 be @xmath0 with orientation forgotten .
we define the minimal warping degree of @xmath0 for all orientations as follows : @xmath176 note that the minimal @xmath177 for all diagrams @xmath0 of @xmath11 is equal to the ascending number @xmath178 @xcite : @xmath179 let @xmath49 be a knot diagram , and @xmath0 a diagram of an @xmath7-component link .
we review the relation of the unknotting number @xmath180 ( resp .
the unlinking number @xmath18 ) and the crossing number @xmath181 ( resp .
@xmath5 ) of @xmath49 ( resp .
@xmath0 ) .
the following inequalities are well - known @xcite : @xmath182 @xmath183 moreover , taniyama mentioned the following conditions
@xcite : the necessary condition for the equality of ( [ ucdk ] ) is that @xmath49 is a reduced alternating diagram of some @xmath184-torus knot , or @xmath49 is a diagram with @xmath185 .
the necessary condition for the equality of ( [ ucdl ] ) is that every @xmath9 is a simple closed curve on @xmath186 and every subdiagram @xmath13 is an alternating diagram .
hanaki and kanadome characterized the link diagrams @xmath0 which satisfy @xmath187 as follows @xcite : let @xmath188 be a diagram of an @xmath7-component link .
then we have @xmath189 if and only if exactly one of @xmath190 is a reduced alternating diagram of a @xmath184-torus knot , the other components are simple closed curves on @xmath186 , and the non - self crossings of the subdiagram @xmath13 are all positive , all negative , or empty for each @xmath14 .
in addition , they showed that any minimal crossing diagram @xmath0 of a link @xmath11 with @xmath191 satisfies @xmath187 .
abe and higa study the knot diagrams @xmath0 which satisfy @xmath192 let @xmath0 be a knot diagram with @xmath193 .
they showed in @xcite that for any crossing point @xmath27 of @xmath0 , one of the components of @xmath194 is a reduced alternating diagram of a @xmath184-torus knot and the other component of @xmath194 has no self - crossings , where @xmath194 is the diagram obtained from @xmath0 by smoothing at @xmath27 .
in addition , they showed that any minimal crossing diagram @xmath0 of a knot @xmath129 with @xmath195 satisfies the above condition . by adding to ( [ ucdk ] ) , we have the following corollary : for a knot diagram @xmath49 , we have @xmath196 further ,
if we have @xmath197 then @xmath49 is a reduced alternating diagram of some @xmath184-torus knot , or @xmath49 is a diagram with @xmath185 . by adding to ( [ ucdl ] ) , we have the following corollary .
( i ) : : for an @xmath7-component link diagram @xmath0 , we have @xmath198 ( ii ) : : we have @xmath199 if and only if every @xmath9 is a simple closed curve on @xmath186 and the number of over - crossings of @xmath9 is equal to the number of under - crossings of @xmath9 in every subdiagram @xmath13 for each @xmath14 .
+ ( iii ) : : if we have @xmath200 then every @xmath9 is a simple closed curve on @xmath186 and for each pair @xmath201 , @xmath202 , the subdiagram @xmath13 is an alternating diagram .
( i ) : : the equality @xmath203 holds because @xmath41 holds for every oriented diagram .
we show that @xmath204 .
let @xmath0 be an oriented diagram which satisfies @xmath205 then @xmath0 also satisfies @xmath206 for every component @xmath9 because of the orientation of @xmath0 . by lemma [ 2dc ]
, we have @xmath207 then we have @xmath208 by ( [ 101 ] ) and ( [ 102 ] ) . hence we obtain the inequality @xmath209 + ( ii ) : : suppose that the equality @xmath210 holds .
then the equalities @xmath211 and @xmath212 hold by ( [ 101 ] ) and ( [ 102 ] ) , where @xmath0 has an orientation such that @xmath213 .
the equality ( [ 103 ] ) is equivalent to that @xmath214 for every @xmath9 .
we prove this by an indirect proof .
we assume that @xmath215 for a component @xmath9 . in this case
, we have the inequality @xmath216 by theorem [ dk ] since @xmath9 has a self - crossing .
we also have @xmath217 because @xmath218 and ( [ 103 ] ) . by substituting ( [ 106 ] ) for ( [ 105 ] ) , we have @xmath219 this implies that the assumption @xmath215 is contradiction . therefore every @xmath9 is a simple closed curve .
the inequality ( [ 104 ] ) is equivalent to that the number of over - crossings of @xmath9 is equal to the number of under - crossings of @xmath9 in every subdiagram @xmath13 for each @xmath14 by lemma [ 2dc ] . on the other hand ,
suppose that every @xmath9 is a simple closed curve , and the number of over - crossings of @xmath9 is equal to the number of under - crossings of @xmath9 in every subdiagram @xmath13 for each @xmath14 , then we have @xmath220 + ( iii ) : : this holds by corollary [ 3cor](i ) and above taniyama s condition .
[ 3cor ] let @xmath129 be a knot , and @xmath11 an @xmath7-component link .
let @xmath221 be the unknotting number of @xmath129 , and @xmath222 be the unlinking number of @xmath11 .
the following inequalities are also well - known @xcite : @xmath223 @xmath224 the following conditions are mentioned by taniyama @xcite : the necessary condition for the equality of ( [ uck ] ) is that @xmath129 is a ( @xmath225)-torus knot ( @xmath27:odd,@xmath226 ) .
the necessary condition for the equality of ( [ ucl ] ) is that @xmath11 has a diagram @xmath0 such that every @xmath9 is a simple closed curve on @xmath186 and every subdiagram @xmath13 is an alternating diagram . by adding to ( [ uck ] ) and ( [ ek ] )
, we have the following corollary : ( i ) : : we have @xmath227 ( ii ) : : we have @xmath228 if and only if @xmath129 is a prime alternating knot .
+ ( iii ) : : if we have @xmath229 then @xmath129 is a ( @xmath225)-torus knot ( @xmath27:odd,@xmath226 ) . by adding to ( [ ucl ] ) , we have the following corollary : for a diagram of an unoriented @xmath7-component link , we have @xmath230 further ,
if the equality @xmath231 holds , then @xmath11 has a diagram @xmath188 such that every @xmath9 is a simple closed curve on @xmath186 and for each pair @xmath232 , the subdiagram @xmath13 is an alternating diagram .
we prove the inequality @xmath233 .
let @xmath0 be a minimal crossing diagram of @xmath11 which satisfies @xmath234 .
then we obtain @xmath235 the condition for the equality is due to above taniyama s condition .
in this section , we define the splitting number and enumerate relations of the warping degree and the complete splitting number . the _ splitting number _ ( resp . _
complete splitting number _ ) of @xmath0 , denoted by @xmath236 ( resp .
@xmath237 ) , is the smallest number of crossing changes which are needed to obtain a diagram of a splittable ( resp .
completely splittable ) link from @xmath0 .
the splitting number of a link which is the minimal @xmath236 for all diagrams @xmath0 is defined by adams @xcite . the _ linking splitting number _ ( resp . _ linking complete splitting number _ ) of @xmath0 , denoted by @xmath238 ( resp .
@xmath239 ) , is the smallest number of non - self - crossing changes which are needed to obtain a diagram of a splittable ( resp .
completely splittable ) link from @xmath0 .
we have the following propositions : ( i ) : : we have @xmath240 + ( ii ) : : we have @xmath241 [ lsplit ] we give examples of proposition [ lsplit ] .
the diagram @xmath0 in figure [ ex - sd ] has @xmath242 .
the diagram @xmath49 in figure [ ex - sd ] has @xmath243 .
the diagram @xmath0 in figure [ ex - splitl ] has @xmath244 .
the diagram @xmath49 in figure [ ex - splitl ] has @xmath245 .
the diagram @xmath0 in figure [ ex - lsplitld ] has @xmath246 .
the diagram @xmath49 in figure [ ex - lsplitld ] has @xmath247 .
we raise the following question : when does the equality @xmath248 @xmath249 or @xmath250 hold ?
in this section , we show methods for calculating the warping degree and linking warping degree by using matrices .
first , we give a method for calculating the warping degree @xmath1 of an oriented knot diagram @xmath0 .
let @xmath251 be a base point of @xmath0 . we can obtain the warping degree @xmath252 of @xmath253 by counting the warping crossing points easily .
let @xmath254 $ ] be a sequence of some `` @xmath255 '' and `` @xmath256 '' , which is obtained as follows .
when we go along the oriented diagram @xmath0 from @xmath251 , we write down `` @xmath255 '' ( resp .
`` @xmath256 '' ) if we reach a crossing point as an over - crossing ( resp . under - crossing ) in numerical order .
we next perform normalization to @xmath254 $ ] , by deleting the subsequence `` @xmath257 '' repeatedly , to obtain the normalized sequence @xmath258
. then we have @xmath259 where @xmath260 denotes the number of entries in @xmath258 .
thus , we obtain the warping degree @xmath1 of @xmath0 . in the following example , we find the warping degree of a knot diagram by using the above algorithm . for the oriented knot diagram @xmath0 and the base point @xmath251 in figure [ 948 ] , we have @xmath261 and @xmath254=[oouuouuouuouoouoou]$ ] . by normalizing @xmath254
$ ] , we obtain @xmath262 $ ] . hence we find the warping degree of @xmath0 as follows : @xmath263 for some types of knot diagram , this algorithm is useful in formulating the warping degree or looking into its properties .
we enumerate the properties of an oriented diagram of a pretzel knot of odd type in the following example : let @xmath264 be an oriented pretzel knot diagram of odd type ( @xmath265 : odd@xmath266 ) , where the orientation is given as shown in figure [ pretzel ] .
we take base points @xmath251 , @xmath267 in figure [ pretzel ] .
then we have @xmath268 and @xmath269 hence we have @xmath270 in this case .
in particular , if @xmath0 is alternating i.e. @xmath271 , then we have that @xmath272 we next explore how to calculate the linking warping degree @xmath20 by using matrices . for a link diagram @xmath0 and a base point sequence @xmath22 of @xmath0
, we define an @xmath7-square matrix @xmath273 by the following rule : we note that @xmath42 is obtained by summing the upper triangular entries of @xmath279 , that is @xmath280 and we notice that @xmath281 where @xmath274 is an entry of @xmath279 ( @xmath282 ) . for the base point sequence @xmath283 which is obtained from a base point sequence @xmath22 by exchanging @xmath63 and @xmath64 ( @xmath65 ) , the matrix @xmath284 is obtained as follows : @xmath285 where @xmath286 with respect to the linking warping degree , we have @xmath287 where @xmath288 are entries of @xmath279 .
to enumerate the permutation of the order of @xmath289 , we consider a matrix @xmath290 , where @xmath291 denotes @xmath292 ( @xmath293 ) or the identity matrix @xmath294 . since @xmath295 depends on the choices of @xmath296 ( @xmath297 ) , we also denote @xmath295 by @xmath298 , where @xmath299 ( @xmath300 ) and we regard @xmath301 in the case @xmath302 . hence we obtain the following formula : @xmath303 thus , we obtain the warping degree of an oriented link diagram by summing the warping degrees @xmath304 ( @xmath24 ) and the linking warping degree @xmath20 .
99 t. abe and r. higa : _ an upper bound for the band - unknotting number of a knot _ , in preparation . c. adams : _ splitting versus unlinking _ , j. knot theory ramifications .
* 5 * ( 1996 ) , 295299 .
cromwell : _ knots and links _ , cambridge university press , 2004 .
r. hanaki and j. kanadome : _ on an inequality between unknotting number and crossing number of links _ , to appear in j. knot theory ramifications .
j. hoste : _ a polynomial invariant of knots and links _ , pacific j. math .
* 124 * ( 1986 ) , 295320 .
g. jin and j. lee : _ link diagrams realizing prescribed subdiagram partitions _
, kobe j. math .
* 18 * ( 2001 ) , 199202 .
a. kawauchi : _ lectures on knot theory _
( in japanese ) , kyoritsu shuppan , 2007 .
a. kawauchi : _ on a complexity of a spatial graph _ , in : knots and soft - matter physics , topology of polymers and related topics in physics , mathematics and biology , bussei kenkyu 92 - 1 ( 2009 - 4 ) , 16 - 19 . w. b. r. lickorish and k. c. millett : _ a polynomial invariant of oriented links _
, topology .
* 26 * ( 1987 ) , 107141 .
y. nakanishi : _ a note on unknotting number _ , math .
notes kobe univ .
* 9 * ( 1981 ) , 99108 .
m. ozawa : _ ascending number of knots and links _ , to appear in j. knot theory ramifications .
d. rolfsen : _ knots and links _ , publish or perish , inc , 1976 .
a. shimizu : _ the warping degree of a knot diagram _ , to appear in j. knot theory ramifications .
stoimenow : _
gau diagram sums on almost positive knots _ , compositio mathematica .
* 140 * ( 2004 ) , 228254 .
k. taniyama : _ unknotting numbers of diagrams of a given nontrivial knot are unbounded _ , j. knot theory ramifications * 18 * ( 2009 ) , 10491063 . | for an oriented link diagram @xmath0 , the warping degree @xmath1 is the smallest number of crossing changes which are needed to obtain a monotone diagram from @xmath0 .
we show that @xmath2 is less than or equal to the crossing number of @xmath0 , where @xmath3 denotes the inverse of @xmath0 and @xmath4 denotes the number of components which have at least one self - crossing . moreover
, we give a necessary and sufficient condition for the equality .
we also consider the minimal @xmath2 for all diagrams @xmath0 . for the warping degree and linking warping degree
, we show some relations to the linking number , unknotting number , and the splitting number . |
doubly charged higgs bosons ( ) appear in theories beyond the standard model @xcite .
many authors have pointed out that these doubly charged scalar particles occur naturally in the left - right symmetric models which allow small neutrino masses @xcite , and stress the importance of looking for their existence @xcite .
supersymmetric left - right models in which the @xmath62 gauge symmetry is broken by triplet higgs fields do not conserve baryon and lepton numbers .
they also yield a natural embedding of a large mass for the right - handed majorana neutrino needed for the implementation of the see - saw mechanism for small neutrino masses .
recently , it has pointed out that such models can lead to `` light '' doubly charged higgs particles , in the 100 gev/@xmath63 mass range @xcite .
an important feature of the doubly charged higgs bosons is that they couple , at tree level , only to charged leptons and to other higgs and gauge bosons , in order to conserve electric charge .
thus doubly charged higgs bosons could be copiously produced in pairs via @xmath64 at lep ii with a differential cross section at tree level depending on the doubly charged higgs boson mass and the centre - of - mass energy of the colliding beams . in left - right symmetric models ,
two higgs triplets arise , conventionnaly labelled left and right handed .
the cross section of @xmath65 is not equal to that for @xmath65 .
the events are characterized by the @xmath66 dependence , where @xmath67 is the angle of the with respect to the electron beam axis , anticipated for the production of a pair of spin 0 particles via the s - channel z or @xmath68 exchange .
the partial widths of the dominant decay modes , @xmath69 and @xmath70 , have been calculated by huitu and collaborators @xcite and by swartz @xcite .
for instance for h decaying into @xmath71 it is , at tree level : @xmath72 where @xmath73 is the mass of the doubly charged higgs boson , @xmath74 is the mass of the tau lepton and @xmath75 is the unknown h@xmath71 yukawa coupling constant .
if @xmath76 , values of @xmath77 can ensure the fast decay of the higgs particle ( @xmath78 s ) .
observation of a four lepton final state would provide a clean signature for the identification of the event . on the other hand for values of @xmath79 smaller than about @xmath80
long lived doubly charged higgs bosons could be detected as events with large impact parameter tracks , kinked tracks , or stable doubly charged massive particles .
opal has searched for charged massive particles at centre - of - mass energies up to 183 gev @xcite .
previous searches for pair production have been performed with the markii @xcite and the opal @xcite detectors using data collected near the z peak and have excluded the existence of the with mass less than 45.6 gev / c@xmath81 and for values of @xmath75 that extend down to zero , apart from a small band around @xmath82 .
significant couplings to electrons or muons are not likely as there are indirect constraints @xcite on the decays of @xmath83 from high energy bhabha scattering , of @xmath84 from the absence of muonium anti - muonium transitions , and of @xmath85 from limits on the flavour changing decay @xmath86 . in this paper , however , we will present our results based on general searches for @xmath65 four leptons with missing energy ( a ) , and four leptons without missing energy ( b ) .
we have considered the following channels : ( a ) @xmath87 , @xmath88 and @xmath89 , and ( b ) @xmath90 , @xmath91 and @xmath92 .
the data set is from collisions recorded with the opal detector at lep , and is composed of events at centre - of - mass energies between @xmath93 189 gev and 209 gev with a total integrated luminosity of about 614 pb@xmath0 .
a complete description of the opal detector can be found in @xcite .
monte carlo samples are used to model the pair production of doubly charged higgs bosons as well as to estimate the expected background due to standard model processes .
the simulation of the signal events with one non - zero @xmath94 coupling at a time and with zero lifetime has been done with the monte carlo generator pythia @xcite modified according to @xcite .
events have been generated for various mass points ranging from 45 gev/@xmath1 to half the centre - of - mass energy , @xmath95 . at each point on the ( @xmath73,@xmath95 ) plane ,
10000 events for each of the six @xmath96 decays have been generated .
the main sources of background arise from standard model four - fermion final states ( @xmath97 ) and a small contribution from two - photon ( @xmath98 hadrons , @xmath99 ) and two - fermion ( @xmath100 ff ) processes . for two - photon processes , the phojet @xcite ,
pythia @xcite , and herwig @xcite generators have been used to simulate hadronic final states .
the vermaseren @xcite generator has been used to estimate the background contribution from all two - photon @xmath101 final states .
all other four - fermion final states have been simulated with grc4f @xcite .
for the two - fermion final states , @xcite was used for the ee@xmath102 final state , and koralz @xcite and kk2f @xcite for the @xmath103 and @xmath104 states .
the multi - hadronic events , @xmath105 , were simulated using pythia and kk2f .
all events were processed through the full simulation of the opal detector @xcite , with the same analysis chain being applied to simulated events and to data .
the final states resulting from @xmath64 processes followed by @xmath96 decays consist of four leptons .
two different analyses are applied depending on the decay assumed : for @xmath106 , @xmath107 or @xmath108 , a selection for four lepton candidates plus missing energy ( resulting from the decays of the tau leptons ) is applied , while for @xmath109 , @xmath110 or @xmath111 , a selection for four leptons ( electrons or muons ) without missing energy is used . although lepton identification is used in the event selection , it is not used for splitting the selected events into the different channels within the two analyses .
the analyses are similar to the r - parity violation search @xcite that considers four leptons with or without missing energy topologies .
these event selection procedures are described in detail in @xcite and are only briefly outlined below .
the cut values used in the present analysis are optimized for the searches for @xmath112 ( in the missing energy analysis ) and for @xmath113 ( in the no missing energy analysis ) .
the selections are then applied as such for the other channels considered , and the efficiencies are calculated separately for each channel . the visible energy and the momentum of the event are calculated using the method described in @xcite .
a preselection is applied , consisting of * data quality requirements as described in @xcite ; * vetoes on the energy deposited in the forward detectors to reduce background due to two - photon events and interactions of beam particles with the beam pipe or residual gas : the energy deposited in each silicon - tungsten forward calorimeter and in each forward detector has to be less than 5 gev ( these detectors are located in the forward region , with polar angle -axis points to the centre of the lep ring , and positive @xmath114 is along the electron beam direction .
the angles @xmath67 and @xmath115 are the polar and azimuthal angles , respectively . ]
@xmath116 , surrounding the beam pipe ) ; there should be no signal in the mip plug scintillators . *
a requirement that at least four and no more than 10 tracks are reconstructed . after the preselection
, a series of cuts is applied as follows . for the missing energy selection : ( a1 ) : : in order to reduce the background from two - photon processes and from radiative return events ( @xmath117 ) with the @xmath68 escaping into the beam pipe : + * the event s transverse momentum , , is required to be larger than @xmath118 . *
the polar angle , @xmath119 , of the missing momentum direction is required to satisfy @xmath120 . *
the event s longitudinal momentum , , is required to be smaller than @xmath121 .
( a2 ) : : the visible energy , @xmath122 , of the event is required to be between @xmath123 and @xmath124 .
( a3 ) : : there have to be at least three tracks with a transverse momentum with respect to the beam axis larger than 1.5 gev/@xmath125 .
( a4 ) : : there have to be at least three well identified isolated leptons @xcite ( e , @xmath126 , @xmath127 ) , each with a transverse momentum larger than 1.5 gev/@xmath125 . ( a5 ) : : the total leptonic momentum , @xmath128 , defined as the scalar sum of the momenta of all identified leptons , is required to be greater than 0.40 @xmath129 . for the no missing energy selection : ( b1 ) : : the polar angle , @xmath119 , of the missing momentum direction has to satisfy @xmath120 , in order to reject events in which a possible missing momentum points towards the beam pipe .
( b2 ) : : @xmath122 is required to be larger than @xmath130 .
( b3 ) : : there have to be at least three isolated tracks each with a transverse momentum larger than 1.5 gev/@xmath125 .
( b4 ) : : there have to be at least three well identified isolated leptons @xcite ( e , @xmath126 , @xmath127 ) , each with a transverse momentum larger than 1.5 gev/@xmath125 . ( b5 ) : : the total leptonic momentum is required to be greater than 0.60 @xmath129 . ( b6 ) : : in order to select events with a pair of particles with identical masses , the following procedure is applied .
pairs are formed using the four most energetic tracks , and the invariant mass is computed for each pair .
events are selected if one of the three possible pairings satisfies @xmath131 , where @xmath132 is the invariant mass of the pair @xmath133 . only pairs with invariant masses
@xmath132 greater than 20 gev/@xmath1 are used in the computation .
the numbers of events remaining after each cut are listed in tables [ tab : cut_flow_miss ] and [ tab : cut_flow_nomiss ] .
the poor agreement between data and monte carlo expectation in the early stages of the selections is due to the beam - related backgrounds and incomplete modelling of two - photon processes .
when the background from these processes has been effectively reduced , the agreement between data and monte carlo is satisfactory .
figure [ fig : distributions ] shows the distributions of some of the variables used in the selections for the data and the background simulation ( the contributions from all centre - of - mass energies are added ) . for the missing energy selection ,
figure [ fig : distributions]a ) shows the distributions of the number of tracks with a transverse momentum with respect to the beam axis larger than 1.5 gev/@xmath125 after cut ( a2 ) has been applied .
the distributions for a signal sample for @xmath134 gev/@xmath1 and @xmath112 at @xmath135 gev are also shown with arbitrary normalisation .
similarly , figure [ fig : distributions]b ) shows the distributions of the total leptonic momentum divided by the visible energy of the event after all but cut ( a5 ) have been applied .
for the no missing energy selection , figure [ fig : distributions]c ) shows the distributions of the number of well identified isolated leptons after cut ( b3 ) and figure [ fig : distributions]d ) the distributions of the total leptonic momentum divided by the visible energy of the event after cut ( b4 ) .
the distributions for a signal sample for @xmath134 gev/@xmath1 and @xmath113 at @xmath135 gev are also shown with arbitrary normalisation .
the signal detection efficiencies are given in table [ tab : results ] .
for values of @xmath136 smaller than approximately @xmath80 , the efficiency to detect @xmath96 decreases due to the long decay length of the @xmath137 . at very small @xmath136
( @xmath138 to @xmath139 ) the @xmath137 could traverse the tracking chambers of the detector without decaying .
we assume zero detection efficiency for values of @xmath136 smaller than @xmath80 .
the following systematic errors on the signal detection efficiencies have been considered : the statistical uncertainty on the determination of the efficiency from the monte carlo simulation ( less than 1% absolute uncertainty ) ; the systematic uncertainty on the measurement of the integrated luminosity ( 0.20% to 0.24% ) ; and the errors due to the modelling of the physics variables ( between 0.5% and 4.5% absolute uncertainty ) .
these last errors are determined by shifting each cut by the maximum amount such that distributions from high statistics monte carlo samples remain in agreement with the data . for the expected number of background events , an uncertainty due to monte carlo statistics ( 4% to 8% ) and a systematic error due to the modelling of the physics variables ( 5% to 7% ) have been taken into account .
the total systematic uncertainty was calculated by summing in quadrature the individual errors .
correlations between systematic uncertainties at different centre - of - mass energies have been taken into account .
in addition to effects included in the detector simulation , a relative efficiency loss of 1.5% to 2.9% arises from beam - related background in the silicon - tungsten forward calorimeter and in the forward detector .
this is estimated using random beam crossing events .
as shown in table [ tab : results ] the observed numbers of events selected by the missing energy and the no missing energy selections ( 17 and 10 ) are consistent with the numbers expected from standard model processes ( 25.19 @xmath140 0.92 @xmath140 1.06 and 6.57 @xmath140 0.50 @xmath140 0.31 ) .
no evidence for production has therefore been observed and limits on the doubly charged higgs boson production cross - section and mass have been derived . for each selected event ,
a reconstructed mass is computed .
figure [ fig : recmass ] shows data and background as well as typical signal reconstructed mass distributions for the e@xmath141e@xmath142 and @xmath143 states .
the momenta of the four lepton candidates are constrained so that the total energy is the centre - of - mass energy , and the total momentum adds to zero . for events with unambiguous charge assignments ( two positive lepton candidates and two negative lepton candidates ) , the charge is used to assign leptons to the hypothesized doubly - charged higgs bosons .
the reconstructed mass is then the average of the invariant masses reconstructed for the two boson hypotheses in the event . for events for which the charge assignment is ambiguous ( three or more lepton candidates of the same sign ) , the assignment of lepton candidates to hypothesized higgs bosons
is chosen to minimize the difference between the two reconstructed masses of the higgs bosons , and the average reconstructed mass of the two bosons is used as the event reconstructed mass . in the channel without missing energy ,
the reconstructed mass resolution is characterized by two components : a core of width 400 mev/@xmath63 with approximately 96% of the signal , and a tail of width 3 gev/@xmath63 , with approximately 4% of the signal . in the channel with missing energy , for the @xmath143 final state , for example
, approximately 40% of the signal is concentrated within a peak of width 2.5 gev/@xmath63 , while the remainder has a long tail .
the first bin of the histograms collects events which fail to have a reconstructed mass ( e.g. missing a lepton candidate ) , and the last bin contains all overflows ( events with a reconstructed mass larger than 130 gev/@xmath63 .
the distributions of the reconstructed masses of signal monte carlo events for various test masses are formed , with bin widths of 500 mev/@xmath63 . to obtain the expected signal distribution for an arbitrary test mass ,
the available signal histograms are interpolated @xcite .
histograms of the reconstructed masses of standard model monte carlo events , averaging over 22 gev/@xmath63 wide mass intervals , are used as the background .
a likelihood ratio method @xcite has been used to determine an upper limit for the production cross - section .
this method combines the results obtained at different centre - of - mass energies , taking into account the number of candidates and their reconstructed masses , the signal detection efficiencies , the expected number of background events and the distributions of the reconstructed masses for background and for signal events .
the confidence levels are computed by binning the data in the reconstructed mass separately for the different centre - of - mass energies are estimated by weighting by @xmath144 , where @xmath145 is @xmath146 , and combining them .
systematic uncertainties on the efficiencies and on the number of expected background events have been taken into account in the limit calculation according to @xcite . within supersymmetric left - right symmetric models , this results in limits on @xmath147 given as a function of the mass , where @xmath148 is the branching ratio of the decay @xmath149 .
figure [ fig : cross_limit_all ] shows the 95% confidence level upper limit on @xmath147 for each of the six possible leptonic decays assuming conservatively that the efficiency for all other decays is negligible . as an example
, figure [ fig : cross_limit_comb ] shows the 95% confidence level upper limit on the cross - section for pair production of followed by a decay into @xmath108 at @xmath135 gev , obtained from the data collected at centre - of - mass energies between 189 gev and 209 gev . comparing this limit with the expected cross - sections for left - handed and right - handed pair production @xcite , @xmath150 and @xmath151 , yields 95% c.l .
lower mass limits of 99.0 gev/@xmath1 and 98.5 gev/@xmath1 , respectively , assuming a 100% decay into @xmath108 .
as listed in table [ tab : mass_limits ] the left - handed mass limits for all channels taken independently lie between 99.0 gev/@xmath1 ( for 100% decay into @xmath108 ) and 100.5 gev/@xmath1 ( for @xmath110 ) . for right - handed
the mass limits lie between 98.5 gev/@xmath1 and 100.1 gev/@xmath1 .
the opal data sets at centre - of - mass energies between 189 gev and 209 gev , corresponding to a total luminosity of about 614 pb@xmath0 , have been searched for evidence of the reaction @xmath64 followed by the decays @xmath152 in r - parity conserving supersymmetric left - right symmetric models .
no significant excess of events has been observed in the data .
production cross - section limits have been derived for values of the yukawa coupling constant @xmath153 larger than @xmath80 .
a lower mass limit of 98.5 gev/@xmath1 at the 95% confidence level has been obtained for higgs bosons decaying via a single channel @xmath152 with 100% branching ratio .
we particularly wish to thank the sl division for the efficient operation of the lep accelerator at all energies and for their close cooperation with our experimental group .
we thank our colleagues from cea , dapnia / spp , ce - saclay for their efforts over the years on the time - of - flight and trigger systems which we continue to use .
we also gratefully thank r.n .
mohapatra for valuable support on theoretical issues .
in addition to the support staff at our own institutions we are pleased to acknowledge the + department of energy , usa , + national science foundation , usa , + particle physics and astronomy research council , uk , + natural sciences and engineering research council , canada , + israel science foundation , administered by the israel academy of science and humanities , + minerva gesellschaft , + benoziyo center for high energy physics , + japanese ministry of education , science and culture ( the monbusho ) and a grant under the monbusho international science research program , + japanese society for the promotion of science ( jsps ) , + german israeli bi - national science foundation ( gif ) , + bundesministerium fr bildung und forschung , germany , + national research council of canada , + research corporation , usa , + hungarian foundation for scientific research , otka t-029328 , t023793 and otka f-023259 , + fund for scientific research , flanders , f.w.o .- vlaanderen , belgium . + g.b .
gelmini and m. roncadedli , phys .
* b99 * ( 1981 ) 411 ; + r.n .
mohapatra and j.d .
vergados , phys .
lett . * 47 * ( 1981 ) 1713 ; + v. barger , h. baer , w.y .
keung and r.j.n .
phillips , phys .
* d26 * ( 1982 ) 218 .
a. g. frodesen , o. skeggestad , and h. tofte , _ `` probability and statistics in particle physics '' _ , universitetsforlaget , 1979 , isbn 82 - 00 - 01 - 01906 - 3 ; + s. l. meyer , _ `` data analysis for scientists and engineers '' _ , john wiley and sons , 1975 , isbn 0 - 471 - 59995 - 6 ; + a. l. read , _ `` modified frequentist analysis of search results ( the cl(s ) method ) '' _ , cern - open-2000 - 205 , in 1st workshop on confidence limits ( 2000 ) 81 ; + t. r. junk , nucl . instr . meth .
, * a434 * ( 1999 ) 435 . | a search for pair - produced doubly charged higgs bosons has been performed using data samples corresponding to an integrated luminosity of about 614 pb@xmath0 collected with the opal detector at lep at centre - of - mass energies between 189 gev and 209 gev .
no evidence for a signal has been observed .
a mass limit of 98.5 gev/@xmath1 at the 95% confidence level has been set for the doubly charged higgs particle in left - right symmetric models .
this is the first search for doubly charged higgs bosons at centre - of - mass energies larger than 91 gev .
european organization for nuclear research cern - ep/2001 - 082 + november 16th , 2001 * * search for doubly charged higgs bosons with the opal detector at lep the opal collaboration _ to be submitted to phys .
lett .
b _ the opal collaboration g.abbiendi@xmath2 , c.ainsley@xmath3 , p.f.kesson@xmath4 , g.alexander@xmath5 , j.allison@xmath6 , g.anagnostou@xmath7 , k.j.anderson@xmath8 , s.arcelli@xmath9 , s.asai@xmath10 , d.axen@xmath11 , g.azuelos@xmath12 , i.bailey@xmath13 , e.barberio@xmath14 , r.j.barlow@xmath6 , r.j.batley@xmath3 , p.bechtle@xmath15 , t.behnke@xmath15 , k.w.bell@xmath16 , p.j.bell@xmath7 , g.bella@xmath5 , a.bellerive@xmath17 , g.benelli@xmath18 , s.bethke@xmath19 , o.biebel@xmath19 , i.j.bloodworth@xmath7 , o.boeriu@xmath20 , p.bock@xmath21 , j.bhme@xmath15 , d.bonacorsi@xmath2 , m.boutemeur@xmath22 , s.braibant@xmath14 , l.brigliadori@xmath2 , r.m.brown@xmath16 , h.j.burckhart@xmath14 , j.cammin@xmath4 , s.campana@xmath18 , r.k.carnegie@xmath17 , b.caron@xmath23 , a.a.carter@xmath24 , j.r.carter@xmath3 , c.y.chang@xmath9 , d.g.charlton@xmath25 , p.e.l.clarke@xmath26 , e.clay@xmath26 , i.cohen@xmath5 , j.couchman@xmath26 , a.csilling@xmath27 , m.cuffiani@xmath2 , s.dado@xmath28 , g.m.dallavalle@xmath2 , s.dallison@xmath6 , a.de roeck@xmath14 , e.a.de wolf@xmath14 , p.dervan@xmath26 , k.desch@xmath15 , b.dienes@xmath29 , m.donkers@xmath17 , j.dubbert@xmath22 , e.duchovni@xmath30 , g.duckeck@xmath22 , i.p.duerdoth@xmath6 , e.etzion@xmath5 , f.fabbri@xmath2 , l.feld@xmath20 , p.ferrari@xmath31 , f.fiedler@xmath14 , i.fleck@xmath20 , m.ford@xmath3 , a.frey@xmath14 , a.frtjes@xmath14 , d.i.futyan@xmath6 , p.gagnon@xmath31 , j.w.gary@xmath18 , g.gaycken@xmath15 , c.geich-gimbel@xmath4 , g.giacomelli@xmath2 , p.giacomelli@xmath2 , m.giunta@xmath18 , j.goldberg@xmath28 , k.graham@xmath13 , e.gross@xmath30 , j.grunhaus@xmath5 , m.gruw@xmath14 , p.o.gnther@xmath4 , a.gupta@xmath8 , c.hajdu@xmath32 , m.hamann@xmath15 , g.g.hanson@xmath31 , k.harder@xmath15 , a.harel@xmath28 , m.harin-dirac@xmath18 , m.hauschild@xmath14 , j.hauschildt@xmath15 , c.m.hawkes@xmath7 , r.hawkings@xmath14 , r.j.hemingway@xmath17 , c.hensel@xmath15 , g.herten@xmath20 , r.d.heuer@xmath15 , j.c.hill@xmath3 , k.hoffman@xmath8 , r.j.homer@xmath7 , d.horvth@xmath33 , k.r.hossain@xmath23 , r.howard@xmath11 , p.hntemeyer@xmath15 , p.igo-kemenes@xmath21 , k.ishii@xmath10 , a.jawahery@xmath9 , h.jeremie@xmath34 , c.r.jones@xmath3 , p.jovanovic@xmath7 , t.r.junk@xmath17 , n.kanaya@xmath13 , j.kanzaki@xmath10 , g.karapetian@xmath34 , d.karlen@xmath17 , v.kartvelishvili@xmath6 , k.kawagoe@xmath10 , t.kawamoto@xmath10 , r.k.keeler@xmath13 , r.g.kellogg@xmath9 , b.w.kennedy@xmath16 , d.h.kim@xmath35 , k.klein@xmath21 , a.klier@xmath30 , s.kluth@xmath19 , t.kobayashi@xmath10 , m.kobel@xmath4 , t.p.kokott@xmath4 , s.komamiya@xmath10 , r.v.kowalewski@xmath13 , t.krmer@xmath15 , t.kress@xmath18 , p.krieger@xmath36 , j.von krogh@xmath21 , d.krop@xmath31 , t.kuhl@xmath15 , m.kupper@xmath30 , p.kyberd@xmath24 , g.d.lafferty@xmath6 , h.landsman@xmath28 , d.lanske@xmath37 , i.lawson@xmath13 , j.g.layter@xmath18 , a.leins@xmath22 , d.lellouch@xmath30 , j.letts@xmath31 , l.levinson@xmath30 , j.lillich@xmath20 , c.littlewood@xmath3 , s.l.lloyd@xmath24 , f.k.loebinger@xmath6 , j.lu@xmath11 , j.ludwig@xmath20 , a.macchiolo@xmath34 , a.macpherson@xmath38 , w.mader@xmath4 , s.marcellini@xmath2 , t.e.marchant@xmath6 , a.j.martin@xmath24 , j.p.martin@xmath34 , g.martinez@xmath9 , g.masetti@xmath2 , t.mashimo@xmath10 , p.mttig@xmath30 , w.j.mcdonald@xmath23 , j.mckenna@xmath11 , t.j.mcmahon@xmath7 , r.a.mcpherson@xmath13 , f.meijers@xmath14 , p.mendez-lorenzo@xmath22 , w.menges@xmath15 , f.s.merritt@xmath8 , h.mes@xmath39 , a.michelini@xmath2 , s.mihara@xmath10 , g.mikenberg@xmath30 , d.j.miller@xmath26 , s.moed@xmath28 , w.mohr@xmath20 , t.mori@xmath10 , a.mutter@xmath20 , k.nagai@xmath24 , i.nakamura@xmath10 , h.a.neal@xmath40 , r.nisius@xmath14 , s.w.oneale@xmath7 , a.oh@xmath14 , a.okpara@xmath21 , m.j.oreglia@xmath8 , s.orito@xmath10 , c.pahl@xmath19 , g.psztor@xmath41 , j.r.pater@xmath6 , g.n.patrick@xmath16 , j.e.pilcher@xmath8 , j.pinfold@xmath23 , d.e.plane@xmath14 , b.poli@xmath2 , j.polok@xmath14 , o.pooth@xmath14 , a.quadt@xmath4 , k.rabbertz@xmath14 , c.rembser@xmath14 , p.renkel@xmath30 , h.rick@xmath18 , n.rodning@xmath23 , j.m.roney@xmath13 , s.rosati@xmath4 , k.roscoe@xmath6 , y.rozen@xmath28 , k.runge@xmath20 , d.r.rust@xmath31 , k.sachs@xmath17 , t.saeki@xmath10 , o.sahr@xmath22 , e.k.g.sarkisyan@xmath42 , a.d.schaile@xmath22 , o.schaile@xmath22 , p.scharff-hansen@xmath14 , m.schrder@xmath14 , m.schumacher@xmath15 , c.schwick@xmath14 , w.g.scott@xmath16 , r.seuster@xmath43 , t.g.shears@xmath44 , b.c.shen@xmath18 , c.h.shepherd-themistocleous@xmath3 , p.sherwood@xmath26 , a.skuja@xmath9 , a.m.smith@xmath14 , g.a.snow@xmath9 , r.sobie@xmath13 , s.sldner-rembold@xmath45 , s.spagnolo@xmath16 , f.spano@xmath8 , m.sproston@xmath16 , a.stahl@xmath4 , k.stephens@xmath6 , d.strom@xmath35 , r.strhmer@xmath22 , l.stumpf@xmath13 , b.surrow@xmath15 , s.tarem@xmath28 , m.tasevsky@xmath14 , r.j.taylor@xmath26 , r.teuscher@xmath8 , j.thomas@xmath26 , m.a.thomson@xmath3 , e.torrence@xmath35 , d.toya@xmath10 , t.trefzger@xmath22 , a.tricoli@xmath2 , i.trigger@xmath14 , z.trcsnyi@xmath46 , e.tsur@xmath5 , m.f.turner-watson@xmath7 , i.ueda@xmath10 , b.ujvri@xmath46 , b.vachon@xmath13 , c.f.vollmer@xmath22 , p.vannerem@xmath20 , m.verzocchi@xmath9 , h.voss@xmath14 , j.vossebeld@xmath14 , d.waller@xmath17 , c.p.ward@xmath3 , d.r.ward@xmath3 , p.m.watkins@xmath7 , a.t.watson@xmath7 , n.k.watson@xmath7 , p.s.wells@xmath14 , t.wengler@xmath14 , n.wermes@xmath4 , d.wetterling@xmath21 g.w.wilson@xmath47 , j.a.wilson@xmath7 , t.r.wyatt@xmath6 , s.yamashita@xmath10 , v.zacek@xmath34 , d.zer-zion@xmath48 @xmath7school of physics and astronomy , university of birmingham , birmingham b15 2tt , uk @xmath2dipartimento di fisica dell universit di bologna and infn , i-40126 bologna , italy @xmath4physikalisches institut , universitt bonn , d-53115 bonn , germany @xmath18department of physics , university of california , riverside ca 92521 , usa @xmath3cavendish laboratory , cambridge cb3 0he , uk @xmath17ottawa - carleton institute for physics , department of physics , carleton university , ottawa , ontario k1s 5b6 , canada @xmath14cern , european organisation for nuclear research , ch-1211 geneva 23 , switzerland @xmath8enrico fermi institute and department of physics , university of chicago , chicago il 60637 , usa @xmath20fakultt fr physik , albert ludwigs universitt , d-79104 freiburg , germany @xmath21physikalisches
institut , universitt heidelberg , d-69120 heidelberg , germany @xmath31indiana university , department of physics , swain hall west 117 , bloomington in 47405 , usa @xmath24queen mary and westfield college , university of london , london e1 4ns , uk @xmath37technische hochschule aachen , iii physikalisches institut , sommerfeldstrasse 26 - 28 , d-52056 aachen , germany @xmath26university college london , london wc1e 6bt , uk @xmath6department of physics , schuster laboratory , the university , manchester m13 9pl , uk @xmath9department of physics , university of maryland , college park , md 20742 , usa @xmath34laboratoire de physique nuclaire , universit de montral , montral , quebec h3c 3j7 , canada @xmath35university of oregon , department of physics , eugene or 97403 , usa @xmath16clrc rutherford appleton laboratory , chilton , didcot , oxfordshire ox11 0qx , uk @xmath28department of physics , technion - israel institute of technology , haifa 32000 , israel @xmath5department of physics and astronomy , tel aviv university , tel aviv 69978 , israel @xmath10international centre for elementary particle physics and department of physics , university of tokyo , tokyo 113 - 0033 , and kobe university , kobe 657 - 8501 , japan @xmath30particle physics department , weizmann institute of science , rehovot 76100 , israel @xmath15universitt hamburg / desy , ii institut fr experimental physik , notkestrasse 85 , d-22607 hamburg , germany @xmath13university of victoria , department of physics , p o box 3055 , victoria bc v8w 3p6 , canada @xmath11university of british columbia , department of physics , vancouver bc v6 t 1z1 , canada @xmath23university of alberta , department of physics , edmonton ab t6 g 2j1 , canada @xmath32research institute for particle and nuclear physics , h-1525 budapest , p o box 49 , hungary @xmath29institute of nuclear research , h-4001 debrecen , p o box 51 , hungary @xmath22ludwigs - maximilians - universitt mnchen , sektion physik , am coulombwall 1 , d-85748 garching , germany @xmath19max - planck - institute fr physik , fhring ring 6 , 80805 mnchen , germany @xmath40yale university , department of physics , new haven , ct 06520 , usa @xmath49 and at triumf , vancouver , canada v6 t 2a3 @xmath50 and royal society university research fellow @xmath51 and institute of nuclear research , debrecen , hungary @xmath52 and heisenberg fellow @xmath53 and department of experimental physics , lajos kossuth university , debrecen , hungary @xmath54 and mpi mnchen @xmath55 and research institute for particle and nuclear physics , budapest ,
hungary @xmath56 now at university of liverpool , dept of physics , liverpool l69 3bx , uk @xmath57 and university of california , riverside , high energy physics group , ca 92521 , usa @xmath58 and cern , ep div , 1211 geneva 23 @xmath59 and universitaire instelling antwerpen , physics department , b-2610 antwerpen , belgium @xmath60 now at university of kansas , dept of physics and astronomy , lawrence , ks 66045 , usa @xmath61 now at university of toronto , dept of physics , toronto , canada |
the dynamics of a flexible polymer in a moving fluid depends strongly on the properties of the velocity gradients .
a remarkable example is given by the coil
stretch transition in a planar extensional flow @xmath0 @xcite .
this flow is characterised by a direction of pure compression and a direction of pure stretching with constant strain rate @xmath1 .
the elongation of the polymer is controlled by the weissenberg number @xmath2 , where @xmath3 is the longest relaxation time of the polymer .
the value @xmath4 marks the coil
stretch transition : for @xmath5 the polymer remains coiled under the action of the entropic elastic force , whereas for @xmath6 the drag force exerted by the flow overcomes the entropic force and the polymer unravels almost completely . as a consequence , the probability density of the extension , @xmath7 , has a single pronounced peak , whose position depends on @xmath8 . for @xmath5 ,
the peak is in the neighbourhood of the equilibrium size of the polymer , @xmath9 ; for @xmath6 , the peak approaches the maximum length @xmath10 .
the ability of a ( non - uniform ) laminar flow to deform an isolated polymer has now been demonstrated for various flow configurations ( see , e.g. , the reviews by @xcite and @xcite ) . the corresponding problem for random flows
was first studied by @xcite , who observed that the strain tensor of an incompressible random flow always has a direction of stretching , although such direction fluctuates in time and in space .
lumley further remarked that the vorticity penalises polymer stretching , for it prevents the polymer from remaining aligned with the principal axes of the strain tensor .
he then concluded that random flows can stretch polymers if the product of the lagrangian correlation time of the strain tensor and the modulus of the maximum eigenvalue of the deformation tensor exceeds a critical value proportional to @xmath11 .
@xcite confirmed these conclusions by showing evidences of a significant amount of polymer stretching in a low - reynolds - number random flow generated by viscoelastic instabilities . in random flows , the time scale describing the stretching of line elements is the reciprocal of the maximum lyapunov exponent @xmath12 .
hence the appropriate definition of the weissenberg number is @xmath13 . @xcite and
@xcite related the deformation of a polymer in a random flow to the statistics of the stretching rate or , more precisely , to the entropy function associated with its probability density function ( e.g. , * ? ? ?
based on this analysis , @xcite predicted the existence of the coil
stretch transition for any random flow with positive @xmath12 , and proposed the following explanation . for an infinitely extensible polymer ( @xmath14 ) ,
the stationary probability density function ( pdf ) of the extension has a power - law tail : @xmath15 for @xmath16 , where @xmath17 depends on the form of the entropy function .
the exponent @xmath17 is positive for @xmath5 , it decreases with increasing @xmath8 , and becomes negative as @xmath8 exceeds @xmath18 .
consequently , @xmath7 is normalizable only for @xmath5 , whereas for @xmath19 there is unbounded growth of the extension and the assumption of infinite extensibility becomes inadequate .
this abrupt change in the statistics of the extension is interpreted as indicating the coil
stretch transition in random flows .
the threshold @xmath4 reproduces ( to within a numerical factor ) lumley s criterion so that the coil stretch transition occurs at @xmath20 . ] .
if the random flow is isotropic and has zero correlation time , @xmath7 can be written explicitly and the behaviour predicted by @xcite can be obtained by direct computation @xcite . in this particular case , @xmath21 , where @xmath22 is the spatial dimension of the flow .
a polymer having finite maximum extension ( @xmath23 ) reaches a stationary configuration at any @xmath8 since the entropic force forbids extensions greater than @xmath10 .
according to the analysis of the infinitely extensible case , @xmath7 is now expected to display a power - law behaviour @xmath24 for @xmath25 . in a short - correlated flow , the coil
stretch transition results from the combination of this power law and the cutoff at @xmath26 : as @xmath8 increases , the intermediate power law raises and the peak of @xmath7 moves from @xmath9 to extensions near to @xmath10 @xcite . an important qualitative difference between extensional and random flows must nevertheless be emphasised . in the former case
, @xmath7 has a pronounced peak at an extension near either @xmath9 or @xmath10 depending on whether @xmath8 is less or greater than @xmath18 . in the latter case
, @xmath7 has broad tails signaling the coexistence of coiled and stretched configurations with predominance of either configuration according to the value of @xmath8 . in random flows , the coil
stretch transition is not as sharp as in the extensional case because of the random nature of the velocity gradient and the presence of vorticity .
the coil stretch transition and the relation between the exponent @xmath17 and the maximum lyapunov exponent @xmath12 were investigated experimentally by @xcite and @xcite . in these experiments ,
the statistics of the extension was measured by following a fluorescently labelled polymer in a random velocity field generated by elastic turbulence .
the pdf of polymer extension was also investigated by means of direct numerical simulations of the continuum equation for the polymer conformation tensor @xcite and by means of brownian dynamics simulations for elastic dumbbells @xcite or multi - bead chains @xcite .
the results of the simulations support the picture of polymer dynamics presented above .
the weissenberg number suffices to determine the elongation of a polymer in laminar flows .
by contrast , in random flows an additional dimensionless number may influence the deformation of the polymer , namely , the kubo number @xmath27 , where @xmath28 is the lagrangian correlation time of the velocity gradient .
the case @xmath29 has been briefly reviewed above . in this paper , we examine the effect of a nonzero @xmath30 and show that the dynamics at @xmath31 differs significantly from the one predicted for short - correlated flows . to emphasise the basic physical mechanisms , we consider a simplified situation .
the polymer molecule is modelled as an elastic dumbbell @xcite .
this approximation is appropriate when attention is restricted to the extension of the polymer @xcite . as for the random advection
, we consider a two - dimensional linear renewing flow . in renewing flows ,
time is split into intervals of length @xmath28 and the velocities in different intervals are independent and identically distributed ( e.g. , * ? ? ?
linear renewing flows were used by @xcite to study the kinematic dynamo effect .
the magnetic field is actually stretched by the velocity gradient in the same way as an infinitely extensible polymer .
the use of a linear renewing flow enables us to obtain semi - analytical results and to accurately compute the statistics of polymer extension with moderate numerical effort .
the remainder of the paper is divided as follows . in section [ sec : dumbbell ] , we briefly review the elastic dumbbell model . in section [ sec : renewing ] , we introduce the renewing random flow . the results are presented in section [ sec : results ] .
finally , some conclusions are drawn in section [ sec : concl ] .
an elastic dumbbell is composed of two beads joined by a spring .
the beads represent the two ends of the polymer ; the spring models the entropic force . for the sake of simplicity , we assume that the flow transporting the dumbbell is two - dimensional .
the vector separating the ends of the polymer , @xmath32 , satisfies the stochastic ordinary differential equation ( e.g. , * ? ? ?
* ) : @xmath33 where @xmath34 and @xmath35 is white noise , i.e. , a gaussian process with zero mean and two - time correlation @xmath36 .
the three terms on the right - hand - side of equation result from the drag force , the entropic elastic force , and thermal noise , respectively .
the @xmath37 matrix @xmath38 is the velocity gradient evaluated along the trajectory of the centre of mass of the dumbbell : @xmath39 .
the function @xmath40 is identically equal to @xmath41 for an infinitely extensible polymer ( hookean model ) and has the form @xmath42 for a finitely extensible polymer with nonlinear elasticity ( fene model ) . in the former case
, @xmath32 is defined on @xmath43 ; in the latter case , @xmath32 belongs to @xmath44 .
equation holds under some assumptions on the dynamics of the beads .
the velocity field is assumed to be linear at the size of the dumbbell .
the drag on a bead is given by the stokes law .
furthermore , hydrodynamic interactions between the beads and inertial effects are disregarded .
according to the assumptions of the dumbbell model , in the reference frame of the centre of mass of the polymer the velocity field is of the form : @xmath45 .
the velocity gradient @xmath38 is constant in each of the time intervals @xmath46 , @xmath47 .
we denote by @xmath48 the ( random ) value taken by @xmath38 in @xmath49 : @xmath50 for all @xmath51 .
the random matrices @xmath48 are identically distributed and statistically independent .
we assume that , for a fixed @xmath52 , @xmath48 is gaussian , zero - mean , statistically isotropic , and traceless ( so that the velocity field is incompressible ) . as a result
, @xmath38 takes the form : @xmath53 with @xmath54 and @xmath55 positive constants .
the elements @xmath56 satisfy : @xmath57 for all @xmath51 , where the @xmath58 are gaussian random variables such that @xmath59 for all @xmath60 and @xmath52 and @xmath61 for all @xmath62 and @xmath63 .
the mean - squared strain and rotation rates are @xmath64 and @xmath65 , respectively .
we set @xmath66 to reproduce the relation holding for the solution of the navier stokes equation ( e.g. , * ? ? ?
* ) . strictly speaking ,
the flow considered is not statistically stationary in time .
nevertheless , over time ranges longer than @xmath28 , it can be considered as a good approximation to a stationary flow with correlation time @xmath28 owing to the invariance with respect to the transformation @xmath67 and thanks to the independence of the matrices @xmath38 in different time intervals @xmath49 @xcite . the @xmath68-correlated flow ( i.e. white - in - time noise )
is recovered by letting @xmath28 tend to zero while holding @xmath69 constant . as mentioned in the introduction
, the elongation of a polymer is related to the statistics of the stretching rate of the flow . for a review on the entropy function and the generalized lyapunov exponents in statistical physics ,
we refer the reader to the book by @xcite . here
, we briefly recall some basic elements of the theory .
if @xmath70 denotes a fluid - line element , the stretching rate at time @xmath71 is defined as @xmath72 .
$ ] the maximum lyapunov exponent is the long - time limit of the average stretching rate : @xmath73 , where the average is taken over the realizations of @xmath38 .
the pdf of @xmath74 for long @xmath71 takes the large - deviation form @xcite : @xmath75 where @xmath76 is the entropy function .
@xmath76 is non - negative and attains its minimum value at the point @xmath12 .
equivalently , the stretching properties of the flow can be characterised by the generalized lyapunov exponents , defined as the rate of growth of the moments of @xmath77 : @xmath78 the function @xmath79 is the legendre transform of @xmath76 , @xmath80 $ ] , and satisfies @xmath81 . for small @xmath82 , @xmath79 can be obtained by using the quadratic approximation @xmath83 so that @xmath84 . for the linear renewing flow , we can compute @xmath79 by adapting to the present problem the method described by @xcite ( see also * ? ? ?
a line element satisfies the equation : @xmath85 since @xmath38 is constant in @xmath49 , the solution of equation for @xmath51 is : @xmath86 hence @xmath87 where @xmath88 is written @xmath89 ( s\zeta_{2,n}-\sqrt{2}\varomega\zeta_{3,n})c_n^- & 2\omega_n c_n^+ -s\zeta_{1,n}c_n^- \end{pmatrix}\ ] ] with @xmath90 , @xmath91 , and @xmath92 ( we remind the reader that @xmath58 is the constant value taken by the random function @xmath56 in @xmath49 ) . if @xmath93 is real , then @xmath94 can be written in terms of hyperbolic functions ; for a non - real complex @xmath93 , @xmath94 involves trigonometric functions . by using the statistical isotropy of the flow and the fact that the matrices @xmath48 are identically distributed and independent for different @xmath52 , it is possible to show the following relation @xcite : @xmath95 where @xmath96 is any constant unit vector and the average @xmath97 is taken over the statistics of the gradient in the interval @xmath49 only .
iterating equation yields : @xmath98 with @xmath99 .
it follows that @xmath79 and @xmath12 can be written as @xmath100 given that the @xmath48 are identically distributed , the average can be taken over any interval @xmath49 .
equations provide a simple way to compute the function @xmath79 and the lyapunov exponent .
the behaviour of @xmath12 as a function of @xmath28 is reported in figure [ fig : lyap ] .
( circles ) . dashed and dotted lines represent the asymptotic behaviours for @xmath101 and @xmath102 , respectively .
inset : generalized lyapunov exponents @xmath79 for three values of @xmath103 .
note that @xmath104 in agreement with the properties of the generalized lyapunov exponents proved by @xcite.,width=377 ] the following asymptotic behaviours hold @xcite : @xmath105 for @xmath66 , one obtains @xmath106 as @xmath102 .
thus the lyapunov exponent becomes independent of @xmath28 for large @xmath28 , and the convergence to the asymptotic value is exponential .
by contrast , the variance @xmath107 monotonically increases like @xmath108 both for small and large @xmath28 @xcite , and therefore the generalized lyapunov exponents do not saturate to a constant value ( figure [ fig : lyap ] ) .
we stress the fact that the above asymptotic behaviours hold for any two - dimensional random flow that is incompressible and statistically isotropic and not only for the renewing flow .
in order to investigate the influence of temporal correlations of the velocity gradients on the dynamics of polymers , we numerically integrated equation for the elastic dumbbell model , where the velocity gradients were determined by equation .
we computed the pdf of the polymer elongation @xmath109 for various values of @xmath30 and @xmath110 .
the numerical integration has been performed using the predictor
corrector scheme proposed by @xcite for the fene model and the stochastic runge
kutta algorithm introduced by @xcite for the hookean model .
the statistics of polymer elongation has been computed by following the dynamics of a single dumbbell for @xmath111 time intervals @xmath49 . , for the hookean model at @xmath112 .
note the power - law behaviours @xmath113 for @xmath114 and @xmath15 for @xmath115 .
inset : exponent @xmath17 as a function of @xmath116 . here
dash - dotted and dashed lines are the asymptotic behaviours and for @xmath118 and @xmath119 , respectively.,width=377 ] within the hookean model , the pdf of the extension behaves like @xmath15 for @xmath16 @xcite . the coil
stretch transition is identified by the change of sign of @xmath17 , which occurs at @xmath120 .
the effects of the temporal correlation of the velocity gradients can be quantified through the dependence of @xmath17 on the kubo number .
the numerical simulations of the hookean model show that for small values of @xmath116 the tail of @xmath7 is almost independent of @xmath116 ; conversely , for @xmath121 , the tail raises as @xmath116 increases and approaches the slope @xmath122 ( figure [ fig : linear ] ) . because of long temporal correlations , even polymers with a short relaxation time occasionally experience significant stretching events , and this effect produces a power - law tail close to the coil stretch transition also for small values of @xmath110 .
this intuitive idea can be rationalized by the following argument .
the exponent @xmath17 determines the lowest order such that @xmath123 diverges and satisfies @xmath124 @xcite . for @xmath110
near to 1/2 , @xmath17 is not far from zero and it is appropriate to use the quadratic approximation for @xmath79 .
the equation @xmath124 then yields @xcite : @xmath125 while the dependence of @xmath17 on the weissenberg number is contained entirely in the second factor , the kubo number enters the expression for @xmath17 through the ratio @xmath126 . for the linear renewing flow
, the asymptotic behaviours discussed in section [ sec : renewing ] give : @xmath127 one obtains that for small values of @xmath116 the tail of @xmath7 is independent of @xmath116 , whereas @xmath128 as @xmath116 is increased , in agreement with our numerical findings .
for the fene model at various values of @xmath116 for @xmath112 ( panel @xmath129 ) and @xmath130 ( panel @xmath131 ) .
the maximum extension is set to @xmath132.,title="fig:",width=253 ] for the fene model at various values of @xmath116 for @xmath112 ( panel @xmath129 ) and @xmath130 ( panel @xmath131 ) .
the maximum extension is set to @xmath132.,title="fig:",width=253 ] although for an infinitely extensible polymer @xmath7 is normalizable only for @xmath133 ( equivalently for @xmath134 ) , the calculation leading to the power - law behaviour remains formally valid also if @xmath135 ( @xmath134 ) . in this case
, the stationary pdf of the extension exists only if the nonlinearity of the elastic force is taken into account , and the power - law prediction is expected to hold for intermediate extensions @xmath136 . to allow the development of the intermediate power law , we performed numerical simulations of the fene model at artificially high @xmath137 . for @xmath138
the results agree with those obtained for the hookean model ( figure [ fig : r1000 ] , panel ( a ) ) . for @xmath139
an increase of @xmath116 produces a decrease of the intermediate slope of @xmath7 , which can be negative even for very large @xmath110 ( figure [ fig : r1000 ] , panel ( b ) ) . according to equation , near @xmath140 the range of variation of @xmath17 as a function of @xmath110
can be made arbitrarily small by increasing @xmath116 , to the extent that the intermediate slope of @xmath7 becomes almost independent of @xmath110 ( figure [ fig : r1000_ku10 ] ) . in a long - correlated flow ( @xmath141 ) .
the maximum extension is set to @xmath132.,width=321 ] our findings show that the picture of a coil
stretch transition associated with the increase of the intermediate slope of @xmath7 does not hold in a long - correlated flow . at large @xmath116
, the appearance of the stretched state occurs through the drop of the maximum at @xmath142 and the simultaneous rise of a second maximum near @xmath10 ( figure [ fig : r1000_ku10 ] ) . for realistic values of @xmath137 ,
it is difficult to detect a clean power law at intermediate extensions and the above scenario becomes even more relevant .
the numerical simulations of the fene model for @xmath143 show that , for short - correlated flows ( @xmath144 ) , the coil
stretch transition manifests through the gradual widening of @xmath7 and the displacement of its maximum from values close to @xmath9 towards values close to @xmath10 ( see figure [ fig : ku - r20 ] , panel ( a ) ) .
this behaviour is in agreement with the theoretical prediction for the @xmath68-correlated isotropic flow @xcite and with recent experimental measurements of polymer elongation in a random shear flow @xcite . for long - correlated flows ( @xmath121 ) ,
the dependence of @xmath7 on @xmath110 is very different . in this case
, increasing @xmath110 at fixed @xmath116 rather produces the drop of the peak at @xmath9 and the simultaneous formation of a second peak near @xmath10 ( see figure [ fig : ku - r20 ] , panel ( b ) ) . the pdf of the extension is bimodal and the two maxima are clearly distinct : intermediate extensions are much less likely than in the small-@xmath116 case .
the shape of @xmath7 for large @xmath116 can be explained intuitively as follows .
if the lagrangian correlation time of the velocity gradient is long , a typical trajectory of a polymer is composed of long portions where the gradient can be thought as frozen with a fixed direction of stretching
. some of those portions will be characterised by a weak stretching rate , some by a strong one .
the response of the polymer to the velocity gradient along one of those portions of the trajectory will be similar to the response that a polymer would have in an extensional flow with strain rate comparable to the stretching rate . as mentioned earlier
, the pdf of @xmath109 in an extensional flow has a sharp peak either at small extensions or at long extensions depending on the strain rate , and intermediate extensions are very unlikely .
therefore , the evolution of an isolated polymer at large @xmath116 consists of a sequence of long deformation events , whose intensity can produce either a coiled or a highly stretched extension , whence the bimodal shape of @xmath7 .
, for the fene model , at various @xmath110 numbers , in a short - correlated flow ( @xmath145 , panel @xmath129 ) and in a long - correlated flow ( @xmath141 , panel @xmath131 ) .
the maximum extension is fixed to @xmath146.,title="fig:",width=321 ] , for the fene model , at various @xmath110 numbers , in a short - correlated flow ( @xmath145 , panel @xmath129 ) and in a long - correlated flow ( @xmath141 , panel @xmath131 ) .
the maximum extension is fixed to @xmath146.,title="fig:",width=321 ]
we have studied the influence of long temporal correlations on the dynamics of a hookean dumbbell and of a fene dumbbell in a two dimensional random renewing flow .
the time correlation of this simple model flow can be changed arbitrarily , and this property enabled us to obtain semi - analytic predictions for the kubo - number dependence of the statistics of polymer elongations .
it is known that the pdf of the elongation of hookean dumbbells is characterised by a power - law tail and that this power law also describes the behavior of the pdf of fene dumbbells at intermediate elongations .
we have shown that in the long - correlated limit ( @xmath119 ) the power - law tail becomes almost insensitive to @xmath110 and its slope is not an effective indicator for the coil stretch transition . in the case of the fene model , the long temporal correlations strongly affect the shape of the pdf , which shows two distinct maxima corresponding to the coiled and stretched configurations . for @xmath147 we have found a new scenario for the coil stretch transition , which occurs through the drop of the peak at @xmath9 and the simultaneous rise of the second peak near @xmath10 .
our findings can be understood in terms of basic properties of long - correlated flows , and the underlying mechanisms does not rely on the peculiar choice of the model flow used in our study .
it is therefore arguable that the phenomena discussed here could be observed also in realistic flows . in particular
, it would be interesting to investigate by means of numerical simulations or experimental measurements whether the presence of long - lived structures in high - reynolds - number flows could result in the appearence of the two - peak coil
stretch transition scenario depicted here . | the effects induced by long temporal correlations of the velocity gradients on the dynamics of a flexible polymer are investigated by means of theoretical and numerical analysis of the hookean and fene dumbbell models in a random renewing flow . for hookean dumbbells ,
we show that long temporal correlations strongly suppress the weissenberg - number dependence of the power - law tail characterising the probability density function ( pdf ) of the elongation . for the fene model ,
the pdf becomes bimodal , and the coil
stretch transition occurs through the simultaneous drop and rise of the two peaks associated with the coiled and stretched configurations , respectively . |
paraffin films and other surface coatings play an integral part in several emerging technologies that employ vapors of alkali - metal atoms , including atomic magnetometers , clocks , and quantum and nonlinear optical devices .
paraffin was first shown to preserve the spin polarization of alkali atoms in 1958 by robinson , ensberg , and dehmelt @xcite and was first studied extensively by bouchiat and brossel.@xcite it has been investigated by several others in the decades since,@xcite but the details of its operation as an anti - relaxation coating remain poorly understood .
paraffin coatings enable narrow zeeman resonance linewidths , and they have recently been the subject of renewed interest due to advances in the technology of alkali - metal magnetometers @xcite that have led to the development of detectors with sensitivity comparable to or better than superconducting quantum interference devices ( squids ) .
modern magnetometers have enabled significant advances in low - magnetic - field nuclear magnetic resonance ( nmr),@xcite magnetic resonance imaging ( mri),@xcite and medical imaging,@xcite as well as paleomagnetism,@xcite explosives detection,@xcite and ultra - sensitive tests of fundamental physics.@xcite paraffin - coated cells also feature narrow hyperfine resonance linewidths and have been explored in the context of secondary frequency standards;@xcite they have also been employed in experiments involving spin squeezing,@xcite quantum memory,@xcite and `` slow light.''@xcite while cells with diameters from a few to tens of centimeters are typically employed , miniature millimeter - sized cells with paraffin coating have also been explored.@xcite in addition , similar coatings with silicon head groups have been used in magneto - optical traps,@xcite hollow photonic fibers,@xcite and noble - gas - atom optical pumping cells.@xcite alkali atoms in the vapor phase depolarize upon contact with the bare surface of a glass container , limiting the coherence lifetime of the spin ensemble . in order to prevent such depolarization
, vapor cells typically include either a buffer gas or an anti - relaxation surface coating .
the inclusion of up to several atmospheres of a chemically inert buffer gas slows diffusion of alkali atoms to the cell walls , but there are several advantages to the use of a surface coating , including lower laser power requirements , larger optical signals , reduced influence of magnetic - field gradients , and a smaller collision rate with other atoms and molecules in the cell .
anti - relaxation coatings can allow an alkali atom to experience thousands of collisions with the walls of the cell without depolarizing , and paraffin in particular has been demonstrated to allow up to 10,000 bounces.@xcite as the size of the vapor cell decreases , the surface - area - to - volume ratio increases , requiring improvement in the quality of the surface coating to compensate for the resulting increase in the rate of collisions with the surface , in order to maintain the same spin - coherence lifetime . in miniature alkali vapor cells with volume 1 - 10 mm@xmath0 or less , the use of surface coatings enables more sensitive magnetometers and clocks than the use of buffer gas , assuming that a coating with appropriate spin - preservation properties can be employed.@xcite a paraffin coating significantly reduces the probability of spin destruction during a collision with the surface because it contains no free electron spins and it features a lower adsorption energy than the bare glass , thus reducing the residence time of an adsorbed alkali atom on the surface of the paraffin relative to the glass surface .
the residence time of an atom at a surface site can be expressed as @xmath1 , where @xmath2 is the adsorption energy , @xmath3 is the boltzmann constant , and @xmath4 is the temperature.@xcite @xmath5 is the time constant for a perfectly elastic collision and therefore gives the high - temperature limit where the thermal energy @xmath6 .
minimal @xmath7 is desirable because an adsorbed alkali atom dephases from the ensemble of atoms in the bulk of the cell , as a result of experiencing both a different magnetic field than in the cell interior and a fluctuating magnetic field generated by the hydrogen nuclei of the paraffin material.@xcite the adsorption energy for alkali atoms on a paraffin surface is small , roughly 0.06 - 0.1 ev,@xcite and assuming @xmath5@xmath810@xmath9 s ( the period of a typical molecular vibration ) gives a residence time of approximately 0.1 ns at room temperature .
the performance of paraffin [ c@xmath10h@xmath11 coatings quickly degrades at temperatures above 60 - 80@xmath12c,@xcite but operation at higher temperature is beneficial for some devices because it increases the saturated vapor pressure of the alkali atoms , and thus the atomic density . for most types of paraffin , such as tetracontane ( @xmath13=40 )
, the critical temperature corresponds to the melting point,@xcite but for longer - chain polyethylene coatings,@xcite which have a much higher fusion temperature of 130@xmath12c , the mechanism for the decreased performance above 60@xmath12c is not fully understood .
recent magnetometers have achieved ultra - high sensitivity better than 1 ft/@xmath14 to near - dc @xcite and radio - frequency @xcite magnetic fields by operating at very high vapor density , but the high operating temperatures of these magnetometers ( @xmath4@xmath15100@xmath12c for cesium vapor and @xmath4@xmath15150@xmath12c for potassium vapor ) prevent the use of paraffin coatings .
in addition , paraffin does not survive the elevated temperatures required by the anodic bonding process used in the production of microfabricated vapor cells .
surface coatings with superior temperature stability are therefore required for use with high - density or microfabricated alkali vapor cells .
high - temperature coatings also allow experimentation with potassium and sodium vapor , which have lower vapor pressures compared to rubidium and cesium at a given temperature .
recent efforts at developing alternatives to paraffin have mainly focused on certain silane coatings that resemble paraffin , containing a long chain of hydrocarbons but also a silicon head group that chemically binds to the glass surface.@xcite such materials do not melt and remain attached to the glass surface at relatively high temperatures , enabling them to function as anti - relaxation coatings at much higher temperatures than paraffin . in particular , a multilayer coating of octadecyltrichlorosilane [ ots , ch@xmath16(ch@xmath17)@xmath18sicl@xmath16 ] has been observed to allow from hundreds up to 2100 bounces with the cell walls @xcite and can operate in the presence of potassium and rubidium vapor up to about 170@xmath12c.@xcite however , the quality of such coatings with respect to preserving alkali polarization is highly variable , even between cells coated in the same batch , and remains significantly worse than that achievable with paraffin .
in addition , it was shown recently that an alkene coating , which resembles paraffin except with an unsaturated c = c double bond , can allow up to two orders of magnitude more bounces than paraffin , but only to temperatures of 33@xmath12c,@xcite and properties such as stability of the coating remain to be studied .
paraffin thus remains the most widely used anti - relaxation coating for alkali spins . in order to facilitate the design and development of new coating materials for alkali - metal cells and hollow fibers , it is therefore necessary to develop a more detailed understanding of the interactions between alkali atoms and the paraffin surface , many aspects of which are not yet fully understood .
indeed , the production of high - quality paraffin cells remains more of an art than a science , with little understanding of why only certain coating procedures work or the reasons that variations in those procedures affect the anti - relaxation quality of the coating .
as an example , paramagnetic impurities could couple to and depolarize the alkali spins , so some researchers observe that purification of the paraffin by distillation is necessary to produce a high - quality coating;@xcite however , others have been able to use paraffins as received.@xcite in addition , a so - called `` ripening '' phase is required , which involves annealing of the paraffin - coated cell at 50 - 60@xmath12c in the presence of the alkali metal for an extended period of time ( typically hours to days ) before the cell may be used , although the specific processes that occur during ripening remain unknown . much of the behavior of the paraffin coating during operation remains equally mysterious .
for example , the paraffin coating introduces a small shift in the hyperfine frequency of the atomic ground state , which varies between cells and must be accounted for in the use of alkali vapor cells as frequency standards.@xcite in addition , the measured vapor pressure in a paraffin - coated cell is smaller than the expected saturated vapor pressure at the temperature of the cell , implying that alkali atoms can be absorbed into , and thus can diffuse within , the bulk volume of the paraffin.@xcite it has been observed @xcite that alkali vapor density increases significantly when a paraffin - coated cell is exposed to light ( particularly uv and near - uv light ) due to the light - induced atomic desorption ( liad ) effect,@xcite which causes absorbed atoms to be ejected from the paraffin coating .
recently , liad effects on alkali spin relaxation have been investigated,@xcite with a focus on non - thermal control of alkali vapor density.@xcite the exact mechanisms of liad in paraffin are not yet understood ; however , it is clear that a combination of surface processes and light - enhanced diffusion of alkali atoms within the bulk of the coating are important .
liad has been used with silane coatings to load photonic fibers,@xcite and it has been used with bare pyrex surfaces to load a chip - scale bose - einstein condensate ( bec),@xcite although in the latter case the desorption efficiency might be increased by orders of magnitude with a paraffin or silane coating . any coating used for liad loading of an ultra - cold atom chip would require compatibility with ultra - high vacuum . similarly , enhanced alkali - atom density is also observed upon the application of large electric fields ( 1 - 8 kv / cm ) across paraffin - coated cells,@xcite although again the exact mechanism remains unknown . in this work we use modern surface science methods as the basis for an investigation of the interaction of alkali - metal atoms with various paraffin materials . unlike most previous studies , which primarily measured properties of the alkali vapor such as density and relaxation time
, we instead observe the properties of the coating itself .
fourier transform infrared spectroscopy ( ftir ) and differential scanning calorimetry ( dsc ) were used to understand bulk properties of the coating materials , while atomic force microscopy ( afm ) , near edge x - ray absorption fine - structure ( nexafs ) , and x - ray photoelectron spectroscopy ( xps ) were used to obtain information about the coated surface and its interaction with the alkali atoms .
these and similar techniques have been employed previously to study paraffins,@xcite but not in the context of their use with alkali atoms . in addition , we compared the liad yields of several different paraffin materials .
an array of techniques enables a more complete characterization of the atom - surface interactions than can otherwise be achieved .
the work described here is intended to demonstrate the power of these methods to thoroughly characterize and help understand the behavior of paraffin and other coating materials , in order to guide the creation of new coatings to enhance the performance of rapidly developing technologies such as microfabricated magnetometers and clocks , nonlinear and quantum optical devices , and portable liad - loaded devices .
we selected paraffin waxes that have been successfully implemented as anti - relaxation coatings in alkali vapor cells , including the @xmath13-alkanes eicosane [ ch@xmath16(ch@xmath17)@xmath20ch@xmath16 ] , dotriacontane [ ch@xmath16(ch@xmath17)@xmath21ch@xmath16 ] , and tetracontane [ ch@xmath16(ch@xmath17)@xmath22ch@xmath16 ] , as well as long - chain polyethylene [ ch@xmath16(ch@xmath17)@xmath10ch@xmath16 where @xmath13 is large and varies between molecules ] .
we also considered a proprietary paraffin from dr .
mikhail balabas that is used in the manufacture of magnetometer cells , which we refer to here as pwmb ; this wax is obtained by fractionation of polyethelyene wax at 220@xmath12c.@xcite finally , we considered the commercially available waxes fr-130 parowax ( from the luxco wax company ) and paraflint , which are both expected to contain a mixture of various @xmath13-alkanes . unless otherwise noted , samples were coated following the typical procedures used in the manufacture of alkali vapor cells , which involve evaporating the material at high temperature in vacuum and allowing it to condense on the inner surface of the cell .
fourier transform infrared spectroscopy ( ftir ) analysis was performed on several types of paraffin in order to identify the general functional groups present in the waxes .
ftir is used extensively for structure determination and chemical analysis because it gives bond - specific information.@xcite transitions due to specific bending and stretching modes of bonds absorb infrared light at specific frequencies allowing for identification of the bonds present .
spectra were obtained on a varian 640-ir system using a room - temperature dtgs ( deuterated triglycine sulfate ) detector .
the aperture of the instrument was completely open , and a nominal 4 @xmath23 spectral resolution was used .
the strong optical absorption of the si - o - si structure makes glass substrates opaque to infrared analysis below 2000 @xmath23 . to widen the range of ftir characterization , native oxide terminated , 500 @xmath24 m thick , ( 111)-oriented
, low - doped ( phosphorus , 74 - 86 @xmath25 cm resistivity ) silicon wafers ( silicon quest international ) were instead used as substrates for paraffin coating .
the silicon phonon absorption occurs below 600 @xmath23 and therefore allows for a more thorough characterization of the various waxes .
in addition , the surface of the silicon wafer , as received , contains a very thin layer of oxidized silicon , which is assumed to be similar enough to the surface of glass cells that it does not disturb the behavior of the wax .
a si thickness of 500 @xmath24 m or greater is necessary to achieve spectra with 4 @xmath23 resolution without complication due to interference from ghost zero - path difference ( etaloning effect ) peaks .
high - resistivity si improves signal to noise by minimizing free - carrier absorption of the ir light .
uncoated si samples were first situated in the beam path with the surface plane perpendicular to the light propagation direction , and 256 background spectra of the uncoated silicon wafers were obtained in transmission mode before the waxes were rub - coated onto the warmed surface , giving a layer with thickness of hundreds of nm to several @xmath24 m . to minimize baseline drift ,
256 scans were collected in transmission mode under the exact same sample placement as for background spectra collection .
absorption and/or si phonon absorption at 600 @xmath23.,width=302 ] the transmisison spectra for tetracontane , fr-130 , and pwmb are shown in fig .
[ fig_ftir ] .
the spectrum for tetracontane is in agreement with literature reports : ch@xmath17 and ch@xmath16 stretching modes at 2850 , 2920 , and 2954 @xmath23 , an hch scissor at 1474 @xmath23 , a ch@xmath16 asymmetric bending mode at 1464 @xmath23 , a u+w ( methyl symmetric and methylene wagging ) mode at 1370 @xmath23 , and a methylene rocking doublet at 719 and 729 @xmath23.@xcite similar ftir spectra were obtained for the fr-130 and pwmb paraffins .
the peaks contained within broken - line boxes are due to environmental changes that occurred between background and sample collection .
the peaks for co@xmath17 at 2300 @xmath23 and 676 @xmath23 may be positive or negative depending on the extent of bench purging with n@xmath17 .
the broad peak at 600 @xmath23 is due to the si bulk crystal which can be either positive or negative due to slight changes in sample temperature or angle of incidence within the bench .
the periodic undulations in the pwmb spectrum are due to the thickness of the film creating an etalon effect .
these results show no observable carbon - carbon double bonds within the sensitivity of the experiment ; specifically , there is no detectable c = c double bond stretching in the range of 1600 - 1700 @xmath23 , and there is no detectable = c - h stretching mode within the range of 3050 - 3100 @xmath23 .
differential scanning calorimetry ( dsc ) was used to assess phase transitions associated with crystallinity of the bulk paraffins .
dsc measures differences in heat flow into a sample and a reference as a function of temperature as the two are subjected to controlled temperature changes .
it can be used to characterize crystallization behavior and assess sample purity by analysis of the heat flow behavior near phase transitions.@xcite the dsc scans were obtained on a ta q200 - 1305 advantage instrument .
wax samples of 1 - 4 mg were pressed into aluminum pans and the reference was an empty aluminum pan .
measurements were made in a nitrogen atmosphere .
the typical scanning conditions were temperature scanning at a rate of 10 @xmath26c / min over a range of 10@xmath26c to 90 - 140@xmath26c , depending on the melting point of the wax .
figure [ fig_dsc1 ] shows the observations for several waxes expected to be crystalline at typical cell operating temperatures below 60@xmath12c .
pure linear - chain alkanes ( including eicosane , dotriacontane , and tetracontane ) all display relatively sharp peaks indicative of the crystalline phase transitions expected for pure @xmath13-alkanes .
purified tetracontane displayed melting and fusion peaks at 79@xmath12c , as well as a phase transition between orthorhombic and hexagonal crystal structure at 62 - 63@xmath12c.@xcite dotriacontane similarly showed distinct melting / fusion and phase transition peaks .
the long - chain polyethylene , while not completely monodisperse , showed a relatively sharp melting point extrapolated to 122@xmath12c and a fusion peak at 118@xmath12c , as well as a weak endothermic process during heating at 100@xmath12c with a corresponding exothermic process during cooling at 80@xmath12c .
-alkanes ; these waxes are not expected to be crystalline at typical operating temperatures.,width=302 ] for comparison , fig .
[ fig_dsc2 ] displays the dsc scans for several waxes that apparently do not exist in a crystalline state at typical operating temperatures .
the fr-130 parowax displayed a relatively sharp melting point at 48 - 51@xmath12c and phase transition between 25 - 35@xmath12c , implying that this wax is fairly monodisperse.@xcite in contrast , the pwmb wax and paraflint do not display sharp melting or fusion peaks , indicating a lack of homogeneous bulk crystallinity .
instead , they feature drawn - out melting and fusion profiles that are consistent with films containing a mixture of various saturated @xmath13-alkanes;@xcite it is possible that these waxes contain branched alkanes as well . in particular , maximum operational temperatures of pwmb wax around 60@xmath12c are well within the material s phase transition .
these results indicate that crystallinity in the bulk is not necessary for an effective anti - relaxation coating , consistent with observations that some working coatings are often partially melted at standard operating temperatures .
atomic force microscopy ( afm ) @xcite of the surface of coated operational cells and coated silicon surfaces was performed to investigate the surface topography of the paraffins .
afm has been employed in the study of organic systems , especially polymeric films on solid inorganic substrates.@xcite measurements were taken using a di dimension 3100 scanning probe microscope using silicon probes with a nominal tip radius of 7 nm .
images taken of polyethylene are shown in fig .
[ fig_afm1 ] .
we observe that surfaces of polyethylene and tetracontane melted onto the silicon substrate showed periodic ridges indicative of crystalline surface structure , consistent with the dsc results . in contrast , we show an image in fig .
[ fig_afm2 ] of a pwmb - coated glass surface after exposure to rubidium atoms ; this surface was part of an operational magnetometer cell , coated in the standard manner described above , that was broken open for this experiment .
this coating does not display any crystalline structure , again consistent with the dsc observations , and it features structures that may indicate either the presence of rubidium clusters on the surface or artifacts of such clusters having reacted upon exposure to air .
similar clusters have been observed on silane coatings @xcite and are speculated to represent regions of the film with an increased probability of causing alkali depolarization . near edge x - ray absorption fine structure ( nexafs )
spectroscopy was used to characterize the molecular bonds present in the paraffin .
the nexafs experiments were carried out on the u7a nist / dow materials characterization end station at the national synchrotron light source at brookhaven national laboratory .
the x - ray beam was elliptically polarized ( polarization factor of 0.85 ) , with the electric field vector dominantly in the plane of the storage ring .
the photon flux was 10@xmath27 s@xmath28 at a typical storage ring current of 500 ma .
a toroidal spherical grating monochromator was used to obtain monochromatic soft x - rays at an energy resolution of 0.2 ev .
nexafs spectra were acquired for incident photon energy in the range 225 - 330 ev , which includes the carbon k edge .
each measurement was taken on a fresh spot of the sample in order to minimize possible beam damage effects .
the partial - electron - yield ( pey ) signal was collected using a channeltron electron multiplier with an adjustable entrance grid bias ( egb ) .
all the data reported here are for a grid bias of -150 v. the channeltron pey detector was positioned at an angle of 45@xmath12 with respect to the incoming x - ray beam and off the equatorial plane of the sample chamber . to eliminate the effect of incident beam intensity fluctuations and monochromator absorption features ,
the pey signals were normalized by the incident beam intensity obtained from the photo yield of a clean gold grid located along the path of the x - ray beam .
a linear pre - edge baseline was subtracted from the normalized spectra , and the edge jump was arbitrarily set to unity at 320 ev , far above the carbon k - edge , a procedure that enabled comparison of different nexafs spectra for the same number of carbon atoms .
energy calibration was performed using a highly oriented pyrolytic graphite ( hopg ) reference sample .
the hopg 1s to @xmath29 transition was assigned an energy of 285.3 ev according to the literature value .
the simultaneous measurement of a carbon grid ( with a 1s to @xmath29 transition of 285 ev ) allowed the calibration of the photon energy with respect to the hopg sample .
charge compensation was carried out by directing low - energy electrons from an electron gun onto the sample surface .
spectra are shown in fig .
[ fig_nexafs1](a ) for tetracontane melted onto silicon , with one sample having been slowly heated to 220@xmath12c over the course of about an hour prior to the measurement , featuring characteristic peaks due to a c - h@xmath30 bond near 288 ev and a @xmath31 c c bond near 292.3 ev , in agreement with previous observations.@xcite the spectrum for polyethylene deposited on glass , shown in fig .
[ fig_nexafs1](b ) , displays these features as well as a small @xmath29 peak near 285.3 ev that is characteristic of a c = c double bond.@xcite another experiment consisted of measuring the spectra from pwmb - coated glass slides , which were initially contained within glass cells ; the standard coating procedure described above deposited pwmb on the slides in addition to the inside surface of the cells . in order to avoid exposure to air
, the cells were broken open inside a glovebag containing an argon atmosphere , and the slides were then inserted into the load lock of the end station .
after initial measurements were taken , the samples were exposed to air for several hours before conducting additional measurements .
figure [ fig_nexafs2](a ) shows spectra of the surface of a pwmb sample that had not been exposed to alkali vapor , with the spectra appearing qualitatively similar both before and after exposure to air .
in addition to the c - h@xmath30 and c - c peaks , there is a peak at 285.3 ev due to a c = c double bond .
these same features are evident in the spectra of a pwmb sample that had been exposed to cesium vapor for extended periods of time , shown in fig .
[ fig_nexafs2](b ) .
the c = c peak is notably larger for the cesium - exposed sample before exposure to air , and is indeed larger than for the sample that had not been exposed to cesium ; the reason for this is unknown but may be related to the ripening process , with the decrease in height after exposure to air likely due to oxidation , and additional measurements will be necessary to determine if the effect is repeatable .
the peaks at approximately 244 and 249 ev are assigned as the third harmonics of the cesium m5 and m4 edges , respectively.@xcite the c = c peak is also seen in the spectrum of the fractionated pwmb material melted directly onto glass ( not shown ) . finally , x - ray photoelectron spectroscopy ( xps ) was employed to determine the elemental composition of paraffin samples as well as to identify the chemical states of the elements present .
xps is well - suited to a study of atomic vapor cells as it may be used to examine the nature of the alkali - paraffin interactions present in the cell .
in addition , angle - resolved xps ( arxps ) offers a means to investigate the distribution of alkali metal in the coating as a function of depth.@xcite xp spectra were acquired on a vg scientific escalab2 spectrometer with al k@xmath32 radiation ( @xmath33 = 1486.6 ev ) and a base operating pressure of approximately 10@xmath34 torr .
survey scans were collected with a 1-ev step size , 100-ms dwell time , and 100-ev pass energy .
higher - resolution scans collected with a 0.05-ev step size , 100-ms dwell time , and 20-ev pass energy were obtained for the c 1@xmath35 , rb 3@xmath36 and 3@xmath37/3@xmath38 regions .
curve fitting of the core - level xps lines was carried out with casaxps software using a nonlinear shirley background subtraction and gaussian - lorentzian product function .
tetracontane ( sigma - aldrich ; @xmath1598.5@xmath39 purity ) was coated onto piranha - cleaned ( 1:3 = 30@xmath39 h@xmath17o@xmath17 : h@xmath17so@xmath40 ; ca .
80@xmath12c ; 1 h ) si(100 ) substrates through immersion of 1-@xmath41 wafers into melted paraffin wax at approximately 80@xmath12c to give a visibly thick coating ( ca .
100 @xmath24 m ) .
coated samples were placed in a cylindrical pyrex cell with a rb source at one end ( residual pressure @xmath85x10@xmath42 torr ) .
exposure to rb vapor was accomplished by heating the sealed pyrex cell to about 60@xmath12c for 48 hours . after trapping excess rb vapor onto a room - temperature cold spot on the cell for 12 hours
, the cell was broken open in air , and the samples were immediately transferred into the xps antechamber .
light exposure was minimized throughout the process in order to prevent light - induced desorption of the rb atoms from the paraffin film . and
3@xmath37/3@xmath38 signals.,width=226 ] xp spectra of both unexposed tetracontane and rb - exposed tetracontane were collected .
comparison of the survey scans ( fig .
[ fig_xps1 ] ) clearly shows the appearance of rb 3@xmath36 ( 110 ev ) and 3@xmath37/3@xmath43 ( 247/238 ev ) signals in the rb - exposed sample .
notably , curve fitting of the c 1@xmath35 ( 284.5 ev ) signal ( fig .
[ fig_xps2 ] ) indicates the presence of a single carbon species and suggests that there are no rb - c bonds in this sample .
rb on the surface is merely physisorbed and not chemisorbed . /3@xmath44
signals.,width=226 ] analogous to the rb - exposed tetracontane , samples of cs - exposed pwmb clearly show the appearance of cs signals in the xp spectrum , as shown in fig .
[ fig_xps3 ] . however , in contrast to the tetracontane samples , the cs - exposed pwmb carbon peak exhibits an asymmetry that suggests the presence of more than one carbon species . indeed , spectral deconvolution of the carbon region ( fig
. [ fig_xps4 ] ) indicates two components , with one at lower binding energy ( ca .
281.5 ev ) than the expected c - c / c - h signal at 284.5 ev .
this signal is reproducible at different positions on the sample and does not appear in the unexposed pwmb sample , so it is probably not a result of surface charging effects .
the lower binding energy component may be assigned to cs - bound carbon @xcite and represents an alkali - carbon interaction not present in the rb - exposed tetracontane .
we speculate that this bound state arises from a reaction between cs and c = c double bonds present in pwmb ; such a pathway not only agrees with the observed c = c nexafs signal but also points to the importance of c = c double bonds in understanding the alkali - paraffin interactions .
in addition to the standard analytical techniques for studying surfaces and bulk materials , such as those described above , alkali vapor cell coatings have in the past typically been studied using more specialized physical methods , such as observations of zeeman and hyperfine lifetimes and frequency shifts .
such studies permit direct comparison of the suitability of different coating materials for use in clocks and magnetometers .
along these lines , we describe here an experiment that allows comparison of the effect of desorbing light on alkali atoms absorbed into different types of paraffin coatings .
the results of such an experiment , combined with the information provided by the other tests described above , can guide the design of coating materials for liad - loaded atomic devices . for this experiment ,
we use two different cell geometries .
the first geometry features lockable stems to prevent ejected atoms from leaving the main cell body .
the lock is implemented using a sliding glass `` bullet '' and permits large vapor density changes to be maintained after repeated exposures to desorbing light ; for more details on the experimental setup and photographs of cells with this lockable stem design , see ref . .
all such cells are spherical with diameter of @xmath45 cm , and the stems remain locked at all times .
the second geometry features cylindrical cells , approximately 9 cm long and 2 cm in diameter , without stem locks .
all cells of either geometry contain rubidium in natural abundance .
for the first geometry , the atomic density is determined by monitoring the absorption of a weak ( 1 @xmath24w ) probe laser beam tuned to the @xmath46 transitions of the @xmath47rb @xmath48 line . for the second geometry ,
the density is determined by quickly sweeping the probe beam across all @xmath49 transitions of the @xmath47rb and @xmath50rb @xmath48 lines and fitting the measured transmission spectrum at each point in time . in order to induce atomic desorption , the cells are fully illuminated by off - resonant light produced by a 405 nm ( blue ) laser diode .
cells are characterized by the liad yield @xmath51 , where @xmath52 is the maximum rb vapor density measured after exposure to desorbing light , and @xmath53 is the initial density prior to illumination . [
cols="<,^,^,^",options="header " , ] results of the liad experiment are summarized in table [ table_liad ] , with an error in the determination of @xmath54 of approximately 15% for each cell .
the preparation temperature is given at which each material was coated on the inside surface of its cell .
for the data shown , desorbing light intensity of 5 mw/@xmath41 was used with the cells of the first geometry , and intensity of 2 mw/@xmath41 was used with cells of the second geometry .
in addition to paraffin materials , several alkene materials are also considered .
the materials labeled `` alkene 80 '' and `` alkene 110 '' are produced by fractionation of alpha olefin fraction c20 - 24 from chevron philips chemical company , which contains a mixture of straight - chain alkenes , with primarily between 18 and 26 carbon atoms ; the number in the material label refers to the temperature of distillation .
alkene 80 is the highly effective anti - relaxation coating material described in ref . .
enet4160 from gelest , inc .
contains 1-triacontene and small amounts of other alkenes with between 26 and 54 carbon atoms .
the liad yield is highly dependent on cell geometry and can not be directly compared between samples of different geometries , even when accounting for differing intensity of the desorbing light .
samples of pwmb and deuterated polyethylene show appreciable increase in rubidium density after exposure to desorbing light for both cell geometries , as does alkene 110 for the second geometry .
it is interesting to note that the nexafs spectra of both pwmb and polyethylene show evidence of unsaturated c = c bonds .
although alkene 80 and enet4160 do not display any measurable liad effect , these results nevertheless suggest that covalently bound alkali atoms could act in part as a reservoir for the liad effect .
tetracontane also shows a large liad yield , but only for the second experiment .
for both cell geometries , deuterated polyethylene gives a larger liad yield when the material is deposited at higher temperature , and it also presents very different dynamical behavior depending on the preparation temperature , as shown in fig .
[ fig_liad ] ; after the desorbing light turns off , the vapor density for the sample prepared at 360@xmath12c is observed to be less than the initial density due to depletion of atoms from the coating.@xcite these results are preliminary and motivate a more comprehensive study of cell - to - cell variation and the effects of preparation conditions ( including particularly temperature ) , material deuteration , unsaturated bonds , and other coating parameters .
= 0 s and turned off at @xmath55=45 s.,width=302 ]
these measurements provide important details about the properties of effective anti - relaxation coatings and demonstrate the utility of applying surface - science methods .
for instance , paraffin does not need to be in crystalline form , as evidenced by the observations of the pwmb wax using dsc and afm .
particularly interesting is the detection of c = c double bonds in the pwmb wax using nexafs , a surface technique , while transmission ftir , a bulk measurement technique , does not show any c = c bonds above the detection limit of approximately 5% ; the discrepancy may be due to lack of sensitivity in the ftir measurement , but it is also possible that the molecules with unsaturated bonds represent impurities expelled to the surface of the material and not extant in its bulk .
it is unknown if the double bonds are present in the raw material or if they form during fractionation when heated to 220@xmath12c .
alkanes are known to decompose at high temperatures,@xcite potentially leading to the presence of unsaturated alkenes in the deposited pwmb surface coating , although the nexafs spectrum of tetracontane shown in fig .
[ fig_nexafs1 ] does not display evidence of double - bond formation after heating to 220@xmath12c .
decomposition could explain the observed increase in liad yield for polyethylene materials deposited at higher temperatures .
the alkene coating described in ref . also contains double bonds and allows approximately 10@xmath56 bounces with the surface , significantly more than any other known material .
the presence of c = c double bonds in effective anti - relaxation coatings is unexpected because unsaturated bonds increase the polarizability of the surface , and effective coatings have long been assumed to require low polarizability to enable short alkali atom residence time , although these recent results indicate that this assumption may be mistaken .
the unsaturated double - bond sites may also react with alkali atoms to form alkali - carbon bonds within the coating material , as detected in the pwmb samples using xps .
in fact , it is likely that passivation of the double - bond sites near the surface of the material is a necessary part of the ripening process .
we may also compare the dsc observations of the thermal properties of the materials with prior measurements of the temperature dependence of the wall shift in paraffin - coated cells of the rubidium 0 - 0 hyperfine frequency , which is used for frequency reference in atomic clocks .
the hyperfine frequency in a cell coated with tetracontane exhibits hysteresis:@xcite the frequency changes little until the solid tetracontane is heated to about 80@xmath12c , at which point it decreases significantly as the wax melts , and the frequency does not increase again upon cooling until the liquid tetracontane reaches a temperature 1@xmath12c colder than this and resolidifies .
this behavior agrees well with the measured temperatures of the melting and fusion peaks of tetracontane .
in addition , the temperature dependence of the hyperfine frequency in cells coated with paraflint is observed to change sign at 72@xmath12c,@xcite which correlates with the endothermic peak measured with dsc .
our observations suggest several avenues for further research .
specific surface characteristics such as roughness and fractional coverage can be studied in order to optimize the composition of coating materials and the deposition procedure .
the absence or presence of alkali - carbon bonds has significant implications with regard to the mechanism of the ripening and liad processes , and so future work will focus on the effect of unsaturated bonds on both the anti - relaxation effectiveness and liad efficiency of materials .
for example , angle - resolved xps studies can reveal depth - dependent changes in the distribution of alkali and alkali - bound species in paraffins before and after treatment with desorbing light .
this work can also be extended to include additional techniques for the study of coated surfaces , such as sum frequency generation ( sfg ) spectroscopy @xcite and raman spectroscopy;@xcite the latter may be particularly useful for its ability to observe the surfaces of intact , operational vapor cells .
in conclusion , using modern analytic techniques we show a systematic study of paraffin waxes which can be extended to a broader set of waxes and other anti - relaxation surface coatings to understand the fundamental properties of these coatings .
combining a number of different surface science methods gives information regarding the chemical nature of the bulk coating materials as well as the thin films at the surface .
this knowledge will inform the design of coatings used with atomic devices . as this research continues
, it will hopefully lead to the development of more effective and robust anti - relaxation coatings for use in a variety of alkali - vapor - based technologies under a wide range of operational conditions .
the authors thank daniel fischer , kristin schmidt , and ed kramer for assistance with the nexafs measurements , and joel ager , joshua wnuk , david trease , and gwendal kervern for helpful discussions and other assistance .
sjs , djm , mhd , ap , and db , the advanced light source , and the dsc , ftir , and afm studies were supported by the director , office of science , office of basic energy sciences , materials sciences division and nuclear science division , of the u.s .
department of energy under contract no .
de - ac02 - 05ch11231 at lawrence berkeley national laboratory .
other parts of this work were funded by nsf / dst grant no .
phy-0425916 for u.s .- india cooperative research , by an office of naval research ( onr ) muri grant , and by onr grant n0001409wx21049 . certain commercial equipment , instruments , or materials are identified in this document
. such identification does not imply recommendation or endorsement by the u.s .
department of energy , lawrence berkeley national laboratory , the advanced light source , or the national institute of standards and technology , nor does it imply that the products identified are necessarily the best available for the purpose .
t. karaulanov , m. t. graf , d. english , s. m. rochester , y. rosen , k. tsigutkin , d. budker , e. b. alexandrov , m. v. balabas , d. f. j. kimball , f. a. narducci , s. pustelny , and v. v. yashchuk , phys .
a * 79 * , 012902 ( 2009 ) . | many technologies based on cells containing alkali - metal atomic vapor benefit from the use of anti - relaxation surface coatings in order to preserve atomic spin polarization . in particular , paraffin has been used for this purpose for several decades and has been demonstrated to allow an atom to experience up to 10,000 collisions with the walls of its container without depolarizing , but the details of its operation remain poorly understood .
we apply modern surface and bulk techniques to the study of paraffin coatings , in order to characterize the properties that enable the effective preservation of alkali spin polarization .
these methods include fourier transform infrared spectroscopy , differential scanning calorimetry , atomic force microscopy , near - edge x - ray absorption fine structure spectroscopy , and x - ray photoelectron spectroscopy .
we also compare the light - induced atomic desorption yields of several different paraffin materials .
experimental results include the determination that crystallinity of the coating material is unnecessary , and the detection of c = c double bonds present within a particular class of effective paraffin coatings .
further study should lead to the development of more robust paraffin anti - relaxation coatings , as well as the design and synthesis of new classes of coating materials . |
in spite of the numerous efforts to achieve a satisfactory understanding of entanglement for systems of identical particles , there is no general agreement on the appropriate generalization of concepts valid for non - identical constituents @xcite .
that is because many concepts are usually only discussed in the context of quantum systems for which the hilbert space @xmath0 is a simple tensor product with no additional structure .
an example is the hilbert space @xmath1 of two non identical particles . in this case
the partial trace @xmath2 for @xmath3 to obtain the reduced density matrix has a good physical meaning : it corresponds to observations only on the subsystem @xmath4 .
in contrast , the hilbert space of a system of @xmath5 identical bosons ( fermions ) is given by the symmetric ( antisymmetric ) @xmath5-fold tensor product of the single - particle spaces .
the consequence is that any multi - particle state contains _ intrinsic _ correlations between subsystems due to quantum indistinguishability .
this , in turn , forces a departure from the straightforward application of entanglement - related concepts like singular value decomposition ( svd ) , schmidt rank , or entanglement entropy . in this context ,
schliemann et al .
@xcite have introduced an analogue of the schmidt rank , the ` slater rank ' to study entanglement in two - fermion systems , using a new version of the svd adapted to deal with antisymmetric matrices .
the extension of these ideas to the boson case was worked out in @xcite and @xcite .
these approaches also have not found general acceptance .
the problems arising in the interpretation of the slater rank and the von neumann entropy of the reduced density matrix ( obtained by partial tracing ) for these systems have been analyzed in @xcite . numerous other proposals for the treatment of identical particles
have recently been put forward .
but , as a closer look at the literature on the subject @xcite reveals , it is apparent that there is no consensus yet as to what the proper formalism should be . in this paper , we propose an approach to the study of entanglement based on the theory of operator algebras .
the foundational results of gelfand , naimark and segal on the representation theory of @xmath6-algebras , abbreviated as the _ _ gns - construction__@xcite are used in order to obtain a generalized notion of entanglement . in particular , the notion of _ partial trace _ is replaced by the much more general notion of _ _ restriction of a state to a subalgebra__@xcite .
this will allow us to treat entanglement of identical and non - identical particles on an equal footing , without the need to resort to different separability criteria according to the case under study . in order to display the usefulness of our approach
, several explicit examples will be worked out . in particular
we show that the gns - construction gives zero for the von neumann entropy of a fermionic or a bosonic state containing the least possible amount of correlations .
we believe that this settles an issue that has caused a lot of confusion regarding the use of von neumann entropy as a measure of entanglement for identical particles @xcite .
a vector state of a quantum system is usually described by a vector @xmath7 in a hilbert space @xmath0 ( pure case ) . more generally , a state is a density matrix @xmath8 , a linear map satisfying @xmath9 ( normalization ) , @xmath10 ( self - adjointness ) and positivity @xmath11 . for _
pure _ states the additional condition @xmath12 is required , which amounts to the assertion that @xmath13 is of the form @xmath14 for some normalized vector in @xmath0 .
now , given that the expectation value of an observable @xmath15 is defined by @xmath16 , we can equivalently regard @xmath13 as a _ linear functional _
@xmath17 on a unital ( @xmath6- ) algebra @xmath18 of observables with unity @xmath19 ( we consider only unital algebras ) .
that is , @xmath20 , for @xmath21 .
the normalization and positivity conditions above then take the form @xmath22 and @xmath23 ( for any @xmath21 ) .
such a positive linear functional of unit norm is called a _ state _ on the algebra @xmath18 .
as already mentioned , in the bipartite case @xmath24 , the definition of @xmath25 involves a partial trace operation .
it is therefore natural to ask for a characterization of this operation in terms of the notion of states on an algebra . for this purpose
consider the subalgebra @xmath26 consisting of all operators of the form @xmath27 , for @xmath28 an observable on @xmath29 . defining a new state @xmath30 as the _ restriction _ of @xmath31 to the subalgebra @xmath26 , one easily checks that for any observable @xmath28 of @xmath29 the equality @xmath32 holds . as we show below
, an algebraic description of the quantum system where the basic objects are a @xmath6-algebra @xmath18 and a state @xmath33 on the algebra provides the solution to some of the problems that appear when @xmath0 does not have the form of a ` simple tensor product ' .
the basic idea of the gns - construction is that given an algebra @xmath18 of observables and a state @xmath33 on this algebra , we can construct the hilbert space on which the algebra of observables acts on .
the key steps are : ( a ) using @xmath33 , we can endow @xmath18 itself with an inner product .
so it becomes an ` inner product ' space @xmath34 .
( b ) this inner product may be degenerate in the sense that the norm of some non - null elements of @xmath34 may be zero .
( c ) if we remove these null vectors by taking the quotient of @xmath34 by the null space @xmath35 of zero norm vectors to get @xmath36 , then we have a well - defined positive definite inner product on @xmath36 .
hence we get a well - defined hilbert space ( after completion ) .
( d ) the algebra of observables @xmath18 acts naturally on this hilbert space in a simple manner .
we now make this set of ideas more precise . from the mathematical point of view
, the algebra of observables is a @xmath6-algebra .
this guarantees that one has enough structure to perform all the tasks in the list above .
a @xmath6-algebra is a ( complete normed ) algebra @xmath37 , together with an antilinear _ involution _
@xmath38 , such that the basic property @xmath39 is satisfied for all @xmath40 in @xmath18 .
the prototypical example of a @xmath6-algebra is the algebra @xmath41 of all bounded operators on a hilbert space @xmath0 , with the involution given by the adjoint : @xmath42 . here
we will only be interested in _ unital _ algebras , that is , we assume the existence of a unit @xmath43 for the algebra . given a state @xmath33 on a @xmath6-algebra @xmath18 , we can obtain a representation @xmath44 of @xmath18 on a hilbert space @xmath45 as follows . since @xmath18 is an algebra ,
it is in particular a vector space .
when elements @xmath46 are regarded as elements of a vector space @xmath34 we write them as @xmath47 .
we then set @xmath48 .
this is almost a scalar product , @xmath49 , but there could be a null space @xmath50 of zero norm vectors : @xmath51 schwarz inequality shows that @xmath50 is a left ideal : @xmath52 it also shows that @xmath53 we denote the elements of the quotient space @xmath54 by @xmath55\rangle$ ] , where @xmath56~=~\alpha + \mathcal n_\omega$ ] @xmath57 .
it has a well - defined scalar product @xmath58 |[\beta]\rangle = \omega ( \alpha^ * \beta ) \ ] ] ( it is independent of the choice of @xmath40 from @xmath56 $ ] because of ( [ leftideal ] ) ) and no nontrivial null vectors .
moreover we have a representation @xmath44 of @xmath18 on @xmath59\rangle= |[\alpha\beta]\rangle$ ] .
( to show this , use @xmath60 ) .
this representation @xmath44 is in general reducible .
we decompose @xmath45 into a direct sum of irreducible spaces : @xmath61 .
let @xmath62 be the corresponding orthogonal projections and define @xmath63\rangle\|\;\;(\mu_i>0\ ; \mbox{always})\ ; \;\;\;\;\;\mbox { and}\;\;\;\;\;\ ; this is a crucial fact .
since @xmath71\rangle \langle [ \chi_i]|$ ] is a pure state , it shows that @xmath33 is _ pure _ if and only if the representation @xmath72 is _ irreducible _ , a well - known result . in particular
, the von neumann entropy of @xmath33 , @xmath73 , is zero if and only if @xmath45 is irreducible . _
the latter is a property that depends on _ both _ the algebra @xmath18 and the state @xmath33_. consider now a ( unital ) subalgebra @xmath74 of @xmath18 and let @xmath75 denote the _ restriction _ to @xmath26 of a pure state @xmath33 on @xmath18 .
we can apply the gns - construction to the pair @xmath76 and use the von neumann entropy of @xmath75 to study the entropy which arises from the restriction . in order to illustrate the above gns - construction , consider the algebra @xmath78 of @xmath79 complex matrices . denoting by @xmath80 the @xmath81 matrix with one on its @xmath82 entry and zero elsewhere , we can write any @xmath83 as @xmath84 .
it is readily checked that the map @xmath85 given by @xmath86 defines a _ state _ on the algebra , as long as @xmath87 .
the vector space @xmath88 is generated by the four vectors @xmath89 ( @xmath90 ) and the null space @xmath91 is generated by those @xmath40 such that @xmath92 .
since @xmath93 the explicit form of the null vector condition is : @xmath94 _ case 1 : @xmath95 _ for @xmath96 the only solution to this equation is @xmath97 , implying that there are no null vectors : @xmath98 .
therefore , in this case the gns - space is given by @xmath99 .
the matrices @xmath80 act on this space as : @xmath100\rangle= \delta_{jk}|[e_{il}]\rangle.\ ] ] the matrix of @xmath101 for instance is : @xmath102 when we order the basis as @xmath103 .
this representation is clearly reducible , the subspaces @xmath104 spanned by @xmath105\rangle\}_{k= 1,2}$ ] being invariant .
hence @xmath44 is the direct sum of two isomorphic irreducible representations , the corresponding decomposition of @xmath106 being @xmath107 next we decompose @xmath108 into pure states .
the unity @xmath43 of @xmath18 is just @xmath109 so that @xmath110 \rangle~=|[e_{11}]\rangle + |[e_{22}]\rangle\ ] ] gives the decomposition of @xmath111 into irreducible subspaces .
the norms of the two components are @xmath112 and @xmath113 by ( [ eq : inner ] ) .
so @xmath110\rangle~=~\sqrt{\lambda}|[\chi_{1}]\rangle + \sqrt{1-\lambda}|[\chi_{2}]\rangle , \;\;\;\;\langle[\chi_{i}]|\chi_{j}]\rangle = \delta_{ij } , \ ] ] with @xmath71\rangle$ ] as in ( [ eq : mu - chi ] ) .
it follows that @xmath114\rangle \langle [ \chi_{1}]| + ( 1-\lambda ) |[\chi_{2}]\rangle \langle [ \chi_{2}]|,\ ] ] so that @xmath108 is not pure .
it has von neumann entropy @xmath115 _ case 2 : @xmath116 _ if we choose @xmath117 , from ( [ eq : condition ] ) we see that @xmath118 , since it is spanned by elements of the form @xmath119 that is , by linear combinations of @xmath120 and @xmath121 .
accordingly , the gns - space @xmath122 is generated by @xmath123\rangle$ ] and @xmath124\rangle$ ] . in this case the representation of @xmath18 is irreducible and given by @xmath81 matrices @xmath125 : @xmath126\rangle= \delta_{jk}|[e_{i2}]\rangle.\ ] ] the state @xmath108 is pure with zero entropy .
a similar situation is found for @xmath127 .
let @xmath128 and consider the state vector @xmath129 then @xmath130 can be thought of as a state @xmath33 on the algebra @xmath18 of linear operators on @xmath0 .
this algebra is isomorphic to the @xmath131 matrix algebra @xmath132 and is generated by elements of the form @xmath133 ( @xmath134 ) , with @xmath135 and @xmath136 the pauli matrices . in this context , the _ entanglement _ of @xmath7 is understood in terms of correlations between local " measurements performed separately on subsystems @xmath4 and @xmath137 .
measurements performed on @xmath4 correspond to the restriction of @xmath33 to the subalgebra @xmath138 generated by elements of the form @xmath139 .
we now study this case .
we set @xmath140 . in order to construct the gns - space @xmath141
, we first notice that @xmath142 , so that in this case there are no nontrivial null states , that is , @xmath143 .
we obtain @xmath144 , with basis vectors @xmath145\rangle \equiv |[\sigma_\nu\otimes \mathds 1_2]\rangle$ ] and inner product @xmath146|[\sigma_\nu]\rangle= \delta_{\mu\nu}$ ] .
the action of @xmath147 on @xmath141 is given , as explained above , by linear operators @xmath148 : @xmath149\rangle = |[\alpha \beta]\rangle.\ ] ] the rhs can be explicitly computed using the identity @xmath150 .
one then finds that the gns space splits into the sum of two invariant subspaces : @xmath151 they are spanned by @xmath152\big\rangle , ~~\big|[(1/2)(1-\sigma_3)\otimes \mathds 1_2]\big\rangle\big\}\ ] ] and @xmath153\big\rangle , ~~\big|[\sigma_-\otimes \mathds 1_2]\big\rangle\big\},\ ] ] where @xmath154 the corresponding projections are @xmath155 we obtain @xmath156\rangle\|^2= \frac{1}{2}\omega_a(\mathds 1_{\mathcal a}+(-1)^i\sigma_3\otimes \mathds 1_2)=\frac{1}{2},\ ] ] @xmath157\rangle=\frac{1}{\sqrt 2 } \big(|[\sigma_0]\rangle+(-1)^i|[\sigma_3]\rangle\big),\;\;\ ; \mbox{with } \,\,\,\langle[\chi_i]|[\chi_j]\rangle = \delta_{ij}.\ ] ] thus , the representation of @xmath158 as a density matrix on the gns - space @xmath141 is : @xmath159\rangle\langle [ \chi_1]|+\frac{1}{2}|[\chi_2]\rangle\langle [ \chi_2]|.\ ] ] the von neumann entropy computed via the gns - construction is therefore @xmath160 reproducing the standard result , as expected .
let @xmath161 be the hilbert space of a one - particle system .
the group @xmath162 acts on @xmath163 by the representation @xmath164 and the algebra of observables is given by a @xmath165-representation of the _ group algebra _
@xmath166 on @xmath167 .
its elements are of the form @xmath168 where @xmath40 is a complex function on @xmath169 and @xmath170 the haar measure @xcite .
the elements @xmath171 span the matrix algebra @xmath172 .
we can understand ( [ matrixalgebra ] ) in terms of matrix elements @xmath173 which are just the integrals of the functions @xmath174 .
consider now a fermionic system with single - particle space @xmath167 .
the hilbert space of this system is the fock space @xmath175 , where @xmath176 is the space of antisymmetric @xmath177-tensors in @xmath167 .
let @xmath178 denote an orthonormal basis for @xmath167 .
then , the set @xmath179 provides an orthonormal basis for @xmath180 .
we can alternatively consider the canonical anticommutation relations ( car ) algebra @xmath181 , and obtain all basis vectors by repeated application of creation operators to the vacuum vector @xmath182 : @xmath183 a self - adjoint operator @xmath4 on @xmath167 can be made to act on @xmath180 in a way that preserves the antisymmetric character of the vectors by considering combinations of the form @xmath184 _ the map @xmath185 is a lie algebra homomorphism_. we further comment on this important point below . at the group level , we may consider exponentials of such operators , of the form @xmath186 we then see that the operators of the form @xmath187 act properly on @xmath180 . _
the map @xmath188 is an isomorphism from @xmath189 into @xmath190_. this is also important as discussed below .
these constructions are most conveniently expressed in terms of a _ coproduct _ @xmath191@xcite .
in fact , an approach based on hopf algebras ( as explained in @xcite ) , has the great advantage that para- and braid - statistics can be _ automatically _ included . in our present case , the construction of the observable algebra corresponds to the following simple choice for the coproduct : @xmath192 , ( @xmath193 ) , linearly extended to all of @xmath194 .
this choice fixes the form of ( [ eq : alpha^k ] ) . at the lie algebra level
, it reduces to ( [ liealgebra ] ) .
physically , the existence of such a coproduct is very important , since it allows us to homomorphically represent the one - particle observable algebra on the @xmath177-particle sector . in a many particle system , the choice to perform observations of only one - particle observables corresponds to restricting the full - algebra of observables to the homomorphic image of the one - particle observable algebra obtained by using the coproduct .
now , if for some reason we perform only partial one - particle observations ( for instance , if at the one - particle level , we decide to measure -or have access to- only the spin degrees of freedom , or only the position ) , the one - particle observable algebra will have to be restricted accordingly and hence its homomorphic image at the @xmath177-particle level will be a subalgebra of the original algebra . as we will see in what follows
, the gns approach covers all such cases whether or not particles are identical .
in particular it merits to reemphasize that it covers observations of particles obeying para- and braid- statistics , including anyons . keeping the same notation as above ,
put @xmath196 .
we focus our attention on the two - fermion space @xmath197 , with basis @xmath198 @xmath199 being an orthonormal basis for @xmath167 . the algebra @xmath18 of observables for the two - fermion system is the matrix algebra generated by @xmath200 ( @xmath201 ) .
it is isomorphic to @xmath202 .
now , @xmath203 acts on @xmath167 through the defining representation ( @xmath204 ) , so that one particle observables are given - at the two fermion level - by the action of @xmath205 on @xmath206 .
this action is given by the restriction of the operators @xmath207 to the space of antisymmetric vectors .
let 3 be the defining ( or fundamental ) @xmath208 representation on @xmath167 .
then the restriction can be obtained from the decomposition @xmath209 of the @xmath208 representation . the @xmath210 span this @xmath211 representation . _ choice 1 for @xmath26 : _
let @xmath212 be any two - fermion state vector .
if we take @xmath26 to be the full algebra of one - particle observables acting on @xmath206 , then @xmath213 and the gns - representation corresponding to the pair @xmath214 is irreducible , the state remaining unchanged upon restriction .
this is just the fact that the @xmath215 representation of @xmath208 is irreducible and corresponds to the fact that , for @xmath196 , all two - fermion vector states @xmath7 have slater rank 1 .
we hence get zero for the von neumann entropy .
notice , however , that the _ von neumann entropy computed by partial trace is equal to 1 for all choices of @xmath7 ( cf .
@xcite ) , in disagreement with the gns - approach_. _ choice 2 for @xmath26 : _ the situation changes drastically if we make a different choice for the subalgebra @xmath26 of @xmath18 .
let us , for the sake of concreteness , choose @xmath26 to be given by those one - particle observables pertaining _ only _ to the one - particle states @xmath216 and @xmath217 .
in this case , @xmath26 will be the five dimensional algebra generated by @xmath218 @xmath219 and @xmath43 , the unit matrix . as a physical illustration for
the meaning of @xmath220 , let us think of @xmath221 as @xmath222 quarks .
observables acting just on @xmath223 and @xmath224 amounts to the isospin group @xmath225 acting just on @xmath226 and @xmath227 in the @xmath215 representation . and the group algebra @xmath228 is also generated by @xmath220 .
we now explicitly perform the gns - construction for the particular choice @xmath229 let @xmath230 denote the corresponding state : @xmath231 and put @xmath232 now notice that @xmath233 independently of @xmath234 .
the same holds true for @xmath235 so that both are null vectors : @xmath236\rangle = |[m^{22}]\rangle= 0 $ ] .
_ case 1 : @xmath237 _ for @xmath238 , there are no more linearly independent null vectors .
we can see this from @xmath239 therefore , the null space @xmath240 is two - dimensional and the gns - space @xmath241 is the three - dimensional space with basis @xmath242\rangle , |[m^{21}]\rangle,|[e^3]\rangle\}$ ] , where @xmath243 . since @xmath244 if @xmath245 , we immediately recognize that , in terms of irreducibles , @xmath246 .
call @xmath247 and @xmath248 the corresponding projections . after noting that @xmath249 ~=~ [ \mathds 1_2]$ ] , we obtain @xmath250\rangle=| [ m^{11 } ] \rangle , ~~ p_2 | [ \mathds 1_{\mathcal a } ] \rangle=| [ e^3 ] \rangle\ ] ] the corresponding ` weights ' @xmath251\rangle\|^2 $ ] are computed using the inner product of @xmath252 .
we obtain @xmath253 hence , @xmath254\rangle \langle[m^{11}]|\right ) + \sin^2\theta\left(\frac{1}{\sin^2\theta } |[e^3]\rangle \langle[e^3]|\right).\ ] ] the result for the entropy as a function of @xmath234 is therefore @xmath255 _ case 2 : @xmath256 _ it is readily checked that at @xmath257 additional null states appear as compared to case 1 .
thus , for @xmath256 the null vectors are spanned by @xmath258 the gns space @xmath259 is two - dimensional and irreducible .
it is spanned by @xmath260\rangle , \;\;\;|[m^{21}]\rangle.\ ] ] since @xmath261 acts nontrivially on this space , and the smallest nontrivial representation of @xmath26 is its two - dimensional irr , this representation is irreducible .
hence @xmath262 is pure with zero entropy . for completeness
we note that the projector to @xmath263 is @xmath264 .
_ case 3 : @xmath265 _ for @xmath265 instead , _ all _ of @xmath266 are null vectors .
so @xmath267 is one - dimensional and spanned by @xmath268\rangle $ ] .
clearly @xmath269 is pure with zero entropy .
the decomposition of @xmath270 as a direct sum of irreducible subspaces as we change the value of @xmath234 is , therefore , as follows : @xmath271 this result should be contrasted against the fact that the @xmath211 representation , when regarded as a representation space for @xmath272 acting on @xmath273\rangle$ ] , @xmath274\rangle$ ] and @xmath275\rangle$ ] splits as @xmath276 .
consider , in the spirit of @xcite , a one - particle space describing fermions with two degrees of freedom which we call _ external _
( e.g. _ ` left ' _ and _ ` right ' _ ) and two degrees of freedom which we call _ internal _
e.g. _ ` spin 1/2 ' _ ) .
here it is convenient to use a description in terms of fermionic creation / annihilation operators @xmath278 , with @xmath279 standing for _ ` left ' _ , @xmath280 for _ ` right ' _ and @xmath281 for spin up and down , respectively . a basis for @xmath206
is then given by the vectors @xmath282 , @xmath283 and @xmath284 , with @xmath285 .
again we consider a @xmath234-dependent state vector , this time given by @xmath286 at the two - particle level , the full observable algebra @xmath18 is the matrix algebra @xmath287 . from this algebra
we pick the subalgebra @xmath26 of _ one - particle observables corresponding to measurements at the left location_. this is the six - dimensional algebra generated by @xmath288 _ case 1 : @xmath237 _ for @xmath289 we readily find that a basis of null vectors of @xmath290 are @xmath291 and @xmath292 .
the gns hilbert space @xmath293 is hence four - dimensional and spanned by the vectors @xmath294\rangle$ ] and @xmath295\rangle\}_{i=1,2,3}$ ] . let @xmath296 be the gns representation of @xmath26 on @xmath270 .
we can evidently find the decomposition of @xmath270 into irreducible subspaces under @xmath296 by computing the casimir operator and the highest weight vectors of the lie algebra @xmath297 given by the representation @xmath298 .
we find @xmath299 , with @xmath300 spanned by @xmath301\rangle=|[a_1^\dagger a_2]\rangle$ ] and @xmath302\rangle$ ] and @xmath303 spanned by @xmath304\rangle$ ] and @xmath305\rangle=|[a_2^\dagger a_1]\rangle$ ] .
the two representations are isomorphic .
\langle\chi_i|\chi_j\rangle = \delta_{ij}.\ ] ] this gives the following result for entropy : @xmath313 _ case 2 : @xmath314 _ consider first @xmath256 .
in this case , @xmath315 the null vectors @xmath316 are obtained by solving @xmath317 for @xmath318 .
that shows that @xmath319 \rangle , \label{null}\ ] ] the quotient space @xmath320 is @xmath321 and is isomorphic to @xmath303 above : @xmath322 \rangle , |[a_2^\dagger a_1]\rangle\right\ } = \mathds c^2 .
\label{quotient}\ ] ] for @xmath265 , when @xmath323 , we find instead a @xmath321 isomorphic to @xmath300 above : @xmath324\rangle , |[\mathds 1_{\mathcal a}- a_2^\dagger a_2]\rangle , |[a_1^\dagger a_1]\rangle , |[a_2^\dagger a_1]\rangle \right\ } , \\ { \hat{\mathcal a_0}}/{\mathcal n_{\frac{\pi}{2},0 } } & = & \mbox{span } \left\ { |[a_1^\dagger a_2]\rangle , |[a_2^\dagger a_2 ] \rangle\right\ } = \mathds c^2\end{aligned}\ ] ] the gns representations on both these @xmath321 s are irreducible
. hence @xmath262 and @xmath325 are pure states with zero entropy .
the decomposition of @xmath270 into irreducible subspaces , as a function of @xmath234 , is the following : @xmath326 the significant aspect of this example is the fact that for the values of @xmath327 for which the slater rank of @xmath328 is one , namely @xmath256 and @xmath329 , we obtain exactly zero for the entropy . in previous treatments of entanglement for identical particles ,
the minimum value for the von neumann entropy of the reduced density matrix ( obtained by partial trace ) has been found to be @xmath330 ( cf .
@xcite and references therein ) .
this has been a source of embarrassment : it seems to suggest that different entanglement criteria have to be adopted , depending on whether one is dealing with non - identical particles , or with bosons , or fermions .
we have shown here that , by replacing the notion of _ partial trace _ by the more general one of _ restriction to a subalgebra _ , all cases can be treated on an equal footing .
here we consider the bosonic analogue of example 3 .
consider the one - particle space @xmath195 with an orthonormal basis @xmath331 .
the two - boson space @xmath206 is the space of symmetrized vectors in @xmath332 .
it corresponds to the six - dimensional space obtained from the decomposition @xmath209 of @xmath208 .
an orthonormal basis for @xmath206 is given by vectors @xmath333 , where @xmath334 the algebra @xmath18 of observables for the two boson system is thus isomorphic to @xmath335 .
consider the state @xmath336 corresponding to the vector @xmath337 to the subalgebra @xmath26 of one - particle observables pertaining _ only _ to the one - particle vectors @xmath216 and @xmath217 . proceeding in the same way as in the previous examples , we recognize that the 6 representation , when regarded as a representation space for @xmath272 acting nontrivially on @xmath216 and @xmath217 , splits as @xmath338 .
the basis vectors for these three invariant subspaces are given below : @xmath339 the one - particle observables on @xmath206 are obtained from the operators @xmath340 ( with @xmath90 ) , as well as from the unit operator on @xmath167 , by means of the coproduct .
thus , the subalgebra @xmath26 is generated by operators of the form @xmath341 , with both @xmath342 and @xmath343 belonging to the _ same _ irreducible component of @xmath206 .
( note that the image of unity on @xmath167 under the coproduct @xmath191 is @xmath43 .
hence by taking combinations of images of the above @xmath167-observables under @xmath191 , we see that @xmath26 contains @xmath344 ) . in other words
, @xmath26 is given by block - diagonal matrices , with each block corresponding to one of the irreducible components in the decomposition @xmath338 .
the dimension of @xmath26 is therefore @xmath345 .
the construction of the gns - representation corresponding to each particular value of the parameters @xmath234 and @xmath346 is performed following the same procedure as in example 3 .
let us introduce the notation @xmath347 , for any pair @xmath348 in ( [ invariant ] ) .
then , from ( [ eq : bosons ] ) we see that as long as the @xmath349-coefficients are all different from zero , those elements of @xmath26 of the form @xmath350 ( @xmath351 ) and @xmath352 ( @xmath353 ) generate the null vectors . that these generate all the null vectors follows from the fact that ( [ eq : bosons ] ) contains one basis element for every irreducible component , so that no further linear relation can arise that lead to null vectors .
so in this case we have @xmath354 in terms of irreducible subspaces , one can readily see that @xmath355 decomposes according to @xmath356 . in general , we can read off the decomposition of @xmath357 into irreducible subspaces from ( [ eq : bosons ] ) , depending on which of its coefficents vanish .
for example , if only the first one vanishes , @xmath358 it is interesting to consider the entropy as a function of @xmath349 . for the case
in which all @xmath349-coefficients are non - zero , we have : @xmath110\rangle= |[b_{1,1}]\rangle+ |[b_{1/2,1/2}]\rangle + |[b_{\tilde 0 , \tilde 0}]\rangle,\ ] ] from which the entropy is readily computed as before .
the result is : @xmath359 - \cos^2\theta\log ( \cos\theta)^2.\ ] ] the analytic formulae for entropy when one or more of the coefficients in ( [ invariant ] ) vanish can be obtained from ( [ eq : boson - entropy ] ) by taking suitable limits on @xmath234 and @xmath346 .
we can see that the entropy vanishes whenever @xmath360 lies in a single irreducible component .
this happens precisely at those points of the two - sphere generated by the parameters @xmath349 that correspond to the coordinate axes .
there are therefore six points where the entropy vanishes exactly .
this is depicted in figure [ fig ] , where the @xmath349-sphere has been mapped to the @xmath361-@xmath362 plane through a stereographic projection .
the figure shows the entropy as a function of the coordinates of that plane . the entropy equation ( [ eq : boson - entropy ] ) as a function of @xmath361 and @xmath362 , the coordinates of a plane representing the @xmath349-sphere through stereographic projection .
darker regions correspond to lower values of the entropy .
five of the six vanishing points of the entropy can be seen on the picture ( black spots ) .
the sixth one , corresponding to the north - pole of the sphere , lies ` at infinity ' in this representation . ]
we have presented a new approach to the study of quantum entanglement based on restriction of states to subalgebras .
the gns - construction allows us to obtain a representation space for the subalgebra such that its decomposition into irreducible subspaces can be used to study quantum correlations .
we showed that , when applied to bipartite systems for which the hilbert space is a ` simple ' tensor product , our method reproduces the standard results on entanglement .
we furthermore showed , with explicit examples , how the formalism can be applied to systems of identical particles .
our results demonstrate in a clear fashion that the von neumann entropy indeed remains a suitable entanglement measure , when understood in terms of states on algebras of observables .
the formalism used for the treatment of identical particles , using coproducts to identify algebras of subsytems can be easily generalized to more sophisticated situations such as those of particles obeying para- and braid statistics .
our results can hence be extended to the study of entanglement of such particles .
it can even be extended to study for instance a @xmath177-particle subsystem in an @xmath5 particle hilbert space .
the authors would like to thank alonso botero for discussions that led to this work .
arq and afrl acknowledge the warm hospitality of t.r .
govindarajan at the institute of mathematical sciences , chennai , where the main part of this work was done .
we also thank m. asorey , s. ghosh , k. gupta , a. ibort , g. marmo and v. p. nair for fruitful discussions .
apb is supported by doe under grant number de - fg02 - 85er40231 and by the institute of mathematical sciences , chennai .
arq is supported by cnpq under process number 307760/2009 - 0 .
afrl is supported by universidad de los andes .
26ifxundefined [ 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/0953-4075/44/i=19/a=192001 [ * * , ( ) ] link:\doibase 10.1103/physreva.64.022303 [ * * , ( ) ] link:\doibase 10.1103/physreva.64.042310 [ * * , ( ) ] link:\doibase 10.1103/physreva.64.054302 [ * * , ( ) ] http://dx.doi.org/10.1023/a:1015439502289 [ * * , ( ) ] , link:\doibase 10.1103/physreva.70.012109 [ * * , ( ) ] link:\doibase 10.1103/physreva.79.042109 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.80.517 [ * * , ( ) ] link:\doibase 10.1088/1742 - 6596/171/1/012032 [ * * , ( ) ] link:\doibase 10.1016/j.aop.2012.02.002 [ * * , ( ) ] link:\doibase 10.1103/physrevb.76.113304 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/revmodphys.81.865 [ * * , ( ) ] link:\doibase 10.1103/physreva.72.022302 [ * * , ( ) ] link:\doibase 10.1103/physreva.83.012113 [ * * , ( ) ] link:\doibase 10.1103/physreva.67.024301 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.91.097902 [ * * , ( ) ] link:\doibase 10.1103/physreva.65.042101 [ * * , ( ) ] http://dx.doi.org/10.1140/epjd/e2011-10654-x [ * * , ( ) ] , http://stacks.iop.org/1751-8121/44/i=17/a=175302 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1090/s0002 - 9904 - 1947 - 08742 - 5 [ * * , ( ) ] @noop _ _ , ed .
, texts and monographs in physics ( , ) link:\doibase 10.1103/physrevlett.92.107902 [ * * , ( ) ] link:\doibase 10.1209/0295 - 5075/86/20005 [ * * , ( ) ] @noop _ _ ( , ) | a novel approach to entanglement , based on the gelfand - naimark - segal ( gns ) construction , is introduced .
it considers states as well as algebras of observables on an equal footing .
the conventional approach to the emergence of mixed from pure ones based on taking partial traces is replaced by the more general notion of the restriction of a state to a subalgebra . for bipartite systems of nonidentical particles ,
this approach reproduces the standard results .
but it also very naturally overcomes the limitations of the usual treatment of systems of identical particles .
this gns approach seems very general and can be applied for example to systems obeying para- and braid- statistics including anyons . |
there is currently considerable experimental activity in the area of harmonically confined fermion vapors @xcite .
this has been the motivation for the present letter , in which equations are set up involving the particle density @xmath0 and the kinetic energy density @xmath1 .
we have written both quantities solely as functions of the radial distance @xmath2 , as we shall be dealing with an arbitrary number , say @xmath3 closed shells in an isotropic external potential .
two equations form the basis for what constitutes a density functional theory of independent harmonically confined fermions in three dimensions .
the first is a form of differential virial equation , specific to an isotropic oscillator potential @xmath4 where @xmath5 , with @xmath6 the force constant and @xmath7 the fermion mass .
the second equation relates the turning points of the particle density @xmath0 and kinetic energy density @xmath1 , which we define from the wave function form @xmath8 two other forms of the kinetic energy density will be used below , the first from the gradient of the wave function @xmath9 and the other derived as the average of @xmath1 and @xmath10 , @xmath11
as to the differential virial equation mentioned above , the early work of march and young @xcite , which however was restricted to one - dimensional motion along the @xmath12 axis , say , gave by expansion of the equation of motion of the dirac density matrix around its diagonal : @xmath13 this was termed a differential virial equation since , by multiplying by @xmath12 and integrating over all @xmath12 the usual integral virial theorem was recovered .
we note from the definitions ( [ ti])-([tibar ] ) above , with the consequence that @xmath14 that eq .
( [ vir1d ] ) can be rewritten @xmath15 while , for one dimension eq .
( [ virmod1d ] ) is valid for a general confining potential @xmath16 , its generalization of present interest refers to a three - dimensional isotropic harmonic oscillator with closed - shells occupancy .
it is not our purpose in this letter to attempt any general proof , but we shall mention later a number of convincing confirmations of our result . this is @xmath17 but with @xmath18 now restricted to the harmonic form ( [ potential ] ) .
the factor 3/2 , by comparison with eq .
( [ virmod1d ] ) , is readily understood as due to dimensionality , as is confirmed by multiplying eq .
( [ vir3d ] ) throughout by @xmath2 to form the virial of the force @xmath19 .
a volume integration , followed by integration by parts on the lhs of the resulting equation yields @xmath20 where @xmath21 is the total kinetic energy .
it is a very simple matter to verify eq .
( [ vir3d ] ) directly for the lowest shell only occupied , when @xmath22 and @xmath23 with @xmath24 being the harmonic oscillator length and @xmath25 a normalization constant . more generally , numerical tests of eq .
( [ vir3d ] ) are shown in fig .
[ fig1 ] for three values of the number of filled shells . in these calculations
we have used the explicit expressions for the particle and kinetic energy densities reported by brack and van zyl @xcite in terms of laguerre polynomials . as a last remark we note that , although the focus of the present work is on spherical harmonic confinement , any additional term that may enter the rhs of eq .
( [ vir3d ] ) for more general central @xmath18 potentials does not contribute to the integral virial . indeed
( [ virint ] ) is the correct form of the virial theorem for such confinements .
having given these arguments for the form ( [ vir3d ] ) of the differential virial equation for spherical harmonic confinement , we pass to the second central result of the present study , namely a relation between the turning points of kinetic energy and particle densities . again
, we can make a useful analogy with the one - dimensional harmonic oscillator . with @xmath26
singly occupied shells we can write , following lawes and march @xcite : @xmath27 the single lowest occupancy corresponding to @xmath28 .
we now assert that , for the three - dimensional oscillator potential energy ( [ potential ] ) the appropriate generalization of eq .
( [ turning1d ] ) reads @xmath29 again , no general proof will be attempted .
but for the lowest shell @xmath30 only occupied , we can again use the explicit gaussian wave function @xmath31 given above to calculate @xmath1 in eq .
( [ turning ] ) and @xmath32 is also readily obtained .
equation ( [ turning ] ) is confirmed for @xmath30 after a short calculation . more generally , fig .
[ fig2 ] reports numerical tests of eq .
( [ turning ] ) for three values of the number of filled shells .
again the specific expressions of brack and van zyl @xcite for @xmath1 and @xmath0 have been used .
further confirmation , now in integral form but for general @xmath33 , comes by multiplying eq .
( [ turning ] ) throughout by @xmath34 and doing a volume integration .
integration by parts readily yields , invoking the equipartition theorem for the harmonic oscillator , @xmath35 but the integral on the rhs is related to the total number @xmath26 of fermions occupying @xmath3 closed shells , the simple result @xmath36 following as a consequence of eq .
( [ turning ] ) .
this is readily confirmed by calculating the eigenvalue sum @xmath37 from the oscillator levels plus the known degeneracies .
again , for @xmath28 or equivalently @xmath30 , @xmath38 follows immediately from the virial result @xmath39 since @xmath40 for the lowest shell . before considering the consequences of eq .
( [ turning ] ) , when combined with the differential virial theorem , we want to stress that , for @xmath41 , the pauli exclusion principle plays a role in determining both particle and kinetic energy densities . therefore , as a further confirmation of eq .
( [ turning ] ) we have used explicit harmonic oscillator wave functions ( see _ e.g. _ morse and feshbach @xcite ) to construct the dirac density matrix @xmath42 for two shells occupied , _
i.e. _ for @xmath43 .
since the degeneracy of the levels for a given total quantum number @xmath44 is @xmath45 , one is then dealing for singly occupied levels with 4 particles , and the dirac matrix is readily constructed as @xmath46 \exp[-(r^2+{r'}^2)/2a_{ho}^2]\;.\ ] ] forming the kinetic energy density from @xmath47 and using the explicit form of @xmath48 following from differentiating with respect to @xmath2 the diagonal density @xmath49 , once again confirms the correctness of eq .
( [ turning ] ) for @xmath50 .
given therefore eqs .
( [ vir3d ] ) and ( [ turning ] ) as the two pillars of the present study we go on to construct the density functional theory of this problem of three - dimensional harmonic confinement .
first of all , let us use eqs .
( [ tibar ] ) and ( [ tigivsti ] ) to remove @xmath51 in eq .
( [ vir3d ] ) in favour of @xmath1 and @xmath52 . then invoking eq .
( [ turning ] ) the derivative @xmath53 can be eliminated , the result being a differential equation for the particle density @xmath0 .
this reads @xmath54 + [ ( m+2)\hbar \omega -v(r)]\rho'(r ) + \frac{3}{2}\frac{\partial v(r)}{\partial r } \rho(r)=0\ ; , \label{eqdiff}\ ] ] which we must again emphasize is valid for the three - dimensional potential energy ( [ potential ] ) .
equation ( [ eqdiff ] ) is evidently a third - order , linear , homogeneous equation for the particle density @xmath0 of @xmath3 closed shells . since the pioneering work of von weizscker @xcite
, it has been clear that a major objective of density functional theory is to construct directly the fermion particle density from a given one - body potential .
equation ( [ eqdiff ] ) achieves that aim quite explicitly for closed shells in the harmonic potential ( [ potential ] ) .
it is the three - dimensional generalization of the early result of lawes and march @xcite for one - dimensional harmonic confinement .
but there is a further consequence of the two basic equations ( [ vir3d ] ) and ( [ turning ] ) . for one can alternatively eliminate the potential , in order to directly relate kinetic energy and particle densities .
one finds almost immediately @xmath55\;.\ ] ] as emphasized earlier , @xmath51 and @xmath1 differ only by a constant times the laplacian of the particle density .
the resulting relation can be integrated , using an appropriate integrating factor , to give @xmath1 in terms of integrals of @xmath0 and its low - order derivatives .
explicitly we find @xmath56=\frac{\hbar^2}{12 m } \frac{1}{\rho^{5/3}(r)}\frac{\partial}{\partial r } \nabla^2 \rho(r)\;. \label{leili}\ ] ] one can reduce the order of the differential equation ( [ leili ] ) by explicit integration and by using the `` boundary condition '' @xmath57_{r=0}=(m+2)\hbar \omega$ ] from eq .
( [ turning ] ) to obtain @xmath58 a second integration leads to the final result @xmath59 ^ 2}{\rho^{8/3}(s ) } \left[\frac{1}{2s}+\frac{\rho'(s)}{3\rho(s)}\right]\right\}\rho^{5/3}(r ) \label{func}\ ] ] where @xmath60 ^ 2/\rho(r)$ ] is the von weizscker `` surface '' contribution to the kinetic energy density @xcite and the constant @xmath61 is given by @xmath62 .
we have assumed @xmath63 .
the form @xmath64 in eq .
( [ func ] ) suggests , with of course the exact density @xmath0 , the introduction of the thomas - fermi ( tf ) kinetic energy density defined by @xmath65 the constant @xmath66 is a multiplicative factor which we do not need to specify for the present purposes .
thus , as in earlier work in this journal @xcite on the one - dimensional case of harmonic confinement , we obtain that @xmath67 and @xmath68 enter as building blocks the non - local kinetic energy density functional ( [ func ] ) .
in summary , we have derived two relatively simple equations , namely eqs .
( [ vir3d ] ) and ( [ turning ] ) , the first being a differential form of the virial theorem and the second relating turning points of kinetic energy density @xmath1 and particle density @xmath0 .
these allow _
( i ) _ the differential equation ( [ eqdiff ] ) for @xmath0 and _ ( ii ) _ the non - local kinetic energy density @xmath1 in eq .
( [ func ] ) to be obtained .
this work was partially supported by murst through prin2000 .
n.h.m . wishes to acknowledge generous support from sns for a stay in pisa during which much of his contribution to the present work was carried out . n.h.m . also acknowledges valuable discussions with dr .
t. gl and with professors l. c. balbs , a. holas and .
nagy on the general area embraced by the present investigation .
m. o. mewes , g. ferrari , f. schreck , a. sinatra , and c. salomon , phys .
rev . a * 61 * ( 2000 ) 011403(r ) ; f. schreck , g. ferrari , k. l. corwin , g. cubizolles , l. khyakovich , m. o. mewes , and c. salomon , cond - mat/0011291 . | two equations are constructed which reflect , for fermions moving independently in a spherical harmonic potential , a differential virial theorem and a relation between the turning points of kinetic energy and particle densities .
these equations are used to derive a differential equation for the particle density and a non - local kinetic energy functional . and
fermi gases , density functional theory |
we deal in this paper with the following question .
question : : given a unitary random matrix ensemble @xmath8 defined on positive definite hermitian matrices @xmath1 of size @xmath9 , where the real analytic potential @xmath3 is such that the equilibrium measure associated with @xmath3 has a density on @xmath10 that vanishes like a square root at @xmath4 .
what are the scaling limits around @xmath4 of the eigenvalue correlation kernel as @xmath5 , @xmath11 ?
one expects a universality type result for this situation where the limiting kernels do not depend on the exact form of @xmath3 .
this is in line with known universality results in the usual unitary random matrix ensembles of the form @xmath12 defined on all hermitian matrices @xmath1 of size @xmath9 , @xcite . in ( [ ensemble1 ] ) and ( [ ensemble2 ] ) and throughout the paper we assume that the confining potentials @xmath3 and @xmath13 are real analytic and satisfy @xmath14 the random matrix ensemble ( [ ensemble1 ] ) is restricted to positive definite matrices , thereby creating a hard edge at @xmath4 .
this has an influence on the local eigenvalue behavior near @xmath4 since eigenvalues are always positive .
it is well - known , see e.g. @xcite , that the eigenvalue correlation kernel for the ensemble ( [ ensemble1 ] ) has the form @xmath15 where @xmath16 is the @xmath17-th degree orthonormal polynomial with respect to the weight @xmath18 on @xmath10 .
the limiting mean eigenvalue density @xmath19 is the minimizer of the weighted energy @xmath20 taken over probability density functions @xmath21 supported on @xmath10 , see @xcite .
it does not depend on @xmath22 . as @xmath3 is real analytic ,
the support of @xmath23 consists of a finite union of intervals @xcite . for @xmath24 with @xmath25 equal to @xmath26 ( left figure ) , @xmath27 ( middle figure ) and @xmath28 ( right figure ) .
[ figure1],title="fig : " ] for @xmath24 with @xmath25 equal to @xmath26 ( left figure ) , @xmath27 ( middle figure ) and @xmath28 ( right figure ) .
[ figure1],title="fig : " ] for @xmath24 with @xmath25 equal to @xmath26 ( left figure ) , @xmath27 ( middle figure ) and @xmath28 ( right figure ) .
[ figure1],title="fig : " ] we can then distinguish two situations .
the first is that @xmath4 does not belong to the support @xmath29 of @xmath23 .
then the hard edge at @xmath4 does not lead to a new phenomenon .
generically one finds that the eigenvalue correlation kernel ( [ kernel1 ] ) has the sine kernel @xmath30 as scaling limit in the bulk and the airy kernel @xmath31 as scaling limit at the ( soft ) edge points of @xmath29 .
these are also the generic scaling limits for the correlation kernel associated with the matrix ensemble ( [ ensemble2 ] ) , see @xcite .
the second situation is that @xmath4 belongs to @xmath29 . in that case ,
one typically finds that @xmath23 is unbounded at @xmath4 with a square root type singularity at @xmath4 .
then instead of the airy kernel we are led to a bessel kernel @xmath32 that describes the local eigenvalue correlations near @xmath4 , see @xcite .
changing parameters in @xmath3 we may create a transition from the airy kernel to the bessel kernel .
this happens for example if @xmath33 when considered on @xmath34 this is a shifted and scaled gue ensemble .
for @xmath35 we have @xmath36.\ ] ] these are all semi - circle laws which tend as @xmath37 to the semi - circle law on @xmath38 $ ]
. then the left soft edge meets with the hard edge at @xmath4 . for @xmath39
we have @xmath40,\ ] ] with @xmath41 which has a square - root singularity at @xmath4 .
see figure [ figure1 ] for the density @xmath42 for the values @xmath43 , @xmath44 and @xmath45 .
thus our question asks about the transitional case .
what happens when the soft edge meets the hard edge ?
note that this transition from soft to hard edge is different from the one considered in @xcite where the authors start from a hard edge situation and let the parameter @xmath22 increase to infinity .
other types of singular edge behaviors in unitary random matrix ensembles have been studied in @xcite , where a soft edge coincides with a spectral singularity , and in @xcite , where the limiting mean eigenvalue density vanishes faster than in the regular case .
our question is related to the calculation of the so - called janossy densities for unitary random matrix ensembles @xmath46 in an asymptotic limit .
indeed , it was shown in @xcite that the janossy densities for an interval @xmath47 are expressed in terms of the orthogonal polynomials for the weight @xmath48 restricted to the complement of @xmath49 . if @xmath23 vanishes like a square root at a left edge point @xmath50 , and if @xmath51 $ ] , then after linear scaling we can reduce this to the case @xmath52 , and then the janossy densities are expressed in terms of the orthogonal polynomial kernel ( [ kernel1 ] ) with @xmath53 .
our first result is that there is indeed a one - parameter family of limiting kernels that depends on @xmath22 .
[ theorem11 ] for every @xmath2 , there is a one - parameter family of kernels @xmath54 depending on @xmath55 , such that the following holds .
let @xmath3 be real analytic on @xmath10 such that @xmath23 vanishes like a square root at @xmath4 and such that there are no other singular points in the external field @xmath3 .
then there are positive constants @xmath56 and @xmath57 such that @xmath58 whenever @xmath5 such that @xmath59 and @xmath60 . the limit ( [ limitkernel ] ) holds uniformly for @xmath61 and @xmath62 in compact subsets of @xmath63 .
see equation ( [ defc1vc2v ] ) below for explicit formulas for the constants @xmath64 and @xmath57 .
we also refer to @xcite and section 3 below for the notion and the classification of singular points in an external field .
in contrast to other works on universality , see e.g. @xcite , we do not have to go through an explicit riemann - hilbert steepest descent analysis in order to prove theorem [ theorem11 ] .
instead we prove theorem [ theorem11 ] by relating the random matrix ensemble ( [ ensemble1 ] ) to the random matrix ensemble ( [ ensemble2 ] ) defined on all @xmath9 hermitian matrices where @xmath65 and with @xmath22 replaced by @xmath66 .
the equilibrium measure @xmath67 for ( [ ensemble2 ] ) vanishes quadratically at @xmath4 .
the universal limiting kernels for this situation were found in @xcite .
they are constructed out of @xmath21-functions associated with the hastings - mcleod solution of the painlev ii equation . using this connection we find the following representation for the limiting kernels @xmath54 .
[ theorem12 ] let @xmath2 .
then the kernels @xmath54 have the form @xmath68 where @xmath69 and @xmath70 are solutions of the system of differential equations @xmath71 g_{\alpha}(x;s ) \end{pmatrix } = \begin{pmatrix } 2q + \alpha/(2x ) & 2 + ( q^2 + r + \frac{s}{2})/x \\[5pt ] -2 x - q^2 + r - \frac{s}{2 } & -2q - \alpha/(2x ) \end{pmatrix } \begin{pmatrix } f_{\alpha}(x;s ) \\[5pt ] g_{\alpha}(x;s ) \end{pmatrix}\ ] ] where @xmath72 is the hastings - mcleod solution of the painlev ii equation with parameter @xmath73 @xmath74 and @xmath75 .
the solutions of ( [ diffsystem ] ) are characterized by the asymptotic behavior @xmath76 as @xmath77 .
[ remark1 ] the hastings - mcleod solution of ( [ painleve2 ] ) with @xmath78 was introduced in @xcite .
it is chararacterized by the asymptotic behavior @xmath79 as @xmath80 where @xmath81 is the usual airy function . for general @xmath2 , @xmath82 , the hastings - mcleod solution of ( [ painleve2 ] )
is characterized by the asymptotic behavior @xmath83 as @xmath80 and @xmath84 as @xmath85 , see ( * ? ? ?
* chapter 11.7 ) .
it was shown in @xcite that the hastings - mcleod solution has no poles on the real line and so the system ( [ diffsystem ] ) is well - defined for every @xmath55 .
note that @xmath86 satisfies @xmath87 and @xmath88 satisfies @xmath89 which are versions of the painlev xxxiv equation , see @xcite . the functions @xmath69 and @xmath70
also satisfy the following linear differential equation with respect to @xmath90 @xmath91 g_{\alpha}(x;s ) \end{pmatrix } = \begin{pmatrix } q & 1 \\[5pt ] -x & -q \end{pmatrix } \begin{pmatrix } f_{\alpha}(x;s ) \\[5pt ] g_{\alpha}(x;s ) \end{pmatrix}\ ] ] which together with ( [ diffsystem ] ) constitutes the lax pair for painlev xxxiv , see @xcite
. we will not use ( [ diffsystem2 ] ) in what follows . in @xcite ,
a limiting kernel is obtained which is built out of solutions to the lax pair for painlev xxxiv as well . however , the relevant solution to painlev xxxiv in this paper is different from ours , and not related to the hastings - mcleod solution to painlev ii .
[ remark2 ] the system ( [ diffsystem ] ) can also be considered for complex values of @xmath61 .
the functions @xmath92 and @xmath93 have asymptotics in the complex @xmath61-plane given by @xmath94 uniformly as @xmath95 in the sector @xmath96 for any @xmath97 .
the positive real line @xmath98 is an anti - stokes line for the system ( [ diffsystem ] ) , where the asymptotic behaviors ( [ falphaasymptotics ] ) and ( [ galphaasymptotics ] ) are not valid .
as usual one has a two - term asymptotic approximation along the anti - stokes line which leads to ( [ realfalphaasymptotics ] ) and ( [ realgalphaasymptotics ] ) . besides the eigenvalue correlation kernel , one might be interested in the limiting distribution of the ( re - scaled ) smallest eigenvalue of the ensemble ( [ ensemble1 ] ) .
these distributions are given by fredholm determinants of the limiting kernels ( [ integrablekernel ] ) .
for general @xmath22 we do not know if these kernels allow painlev expressions but for @xmath53 we can relate them to the well - known tracy - widom distribution @xcite @xmath99 where @xmath100 is the hastings - mcleod solution of painlev ii .
indeed , since @xmath3 is real analytic on @xmath10 we can also consider @xmath3 on @xmath101 for some @xmath102 .
for small enough @xmath103 the equilibrium measure associated with @xmath3 on @xmath101 will not depend on @xmath103 and it will vanish like a square root at @xmath4 .
then for the ensemble ( [ ensemble1 ] ) with @xmath53 and defined on hermitian matrices with eigenvalues @xmath104 we have that the distribution of the smallest eigenvalue tends to the tracy - widom distribution ( [ twdist ] ) as @xmath105 .
here we assume for convenience that @xmath106 ; otherwise under the assumption ( [ eq - limits ] ) we find a shifted version of ( [ twdist ] ) .
moreover , denoting the smallest eigenvalue by @xmath107 we have for @xmath108 , @xmath109 here @xmath110 denotes the probability in the ensemble ( [ ensemble1 ] ) on positive - definite matrices , while @xmath111 is the probability in the ensemble ( [ ensemble1 ] ) on hermitian matrices with eigenvalues @xmath104 .
hence for an appropriate constant @xmath112 we have if @xmath53 and @xmath106 , @xmath113 for @xmath114 , the right - hand side of ( [ cutofftw ] ) will be different , but the scaling with @xmath115 in the left - hand side will be the same .
in this section we relate the correlation kernels ( [ kernel1 ] ) for the random matrix ensemble ( [ ensemble1 ] ) to the eigenvalue correlation kernel for the random matrix ensemble ( [ ensemble2 ] ) with @xmath116 and parameter @xmath66 instead of @xmath22 .
we use @xmath117 to denote the correlation kernel .
thus @xmath118 where @xmath119 is the @xmath17-th degree orthonormal polynomial with respect to the weight @xmath120 on @xmath34 . to distinguish this kernel from the kernel ( [ kernel1 ] ) that is related to @xmath3 and parameter @xmath22 we write in this section @xmath121 the following result
is not new , see e.g. ( * ? ? ?
* appendix b ) and ( * ? ? ?
* exercises 4.4 ) . for the convenience of the reader
we have included a proof .
[ kernel - relation ] let @xmath3 be defined on @xmath10 and let @xmath13 be defined by ( [ defw ] ) . assume that ( [ asympvw ] ) is satisfied . then for @xmath122
, we have @xmath123 while for @xmath124 , we have in addition @xmath125 in the proof we use the abbreviations @xmath126 we also write @xmath127 for the orthonormal polynomial of degree @xmath128 with respect to the weight @xmath129 on @xmath10 , and @xmath130 and @xmath131 for the orthonormal polynomials with respect to the weights @xmath132 and @xmath133 , respectively .
note that @xmath127 and @xmath130 are well - defined for @xmath122 , while @xmath131 is only defined for @xmath124 . since @xmath134 is an even weight on @xmath34 ,
the even degree orthogonal polynomials are even and the odd degree orthogonal polynomials are odd .
therefore there exists a polynomial @xmath135 of degree @xmath128 such that @xmath136 for @xmath137 , we then have @xmath138 where we put @xmath139 .
a similar calculation shows that @xmath140 therefore @xmath141 has the orthogonality conditions that characterize the orthonormal polynomial @xmath127 and it follows that @xmath142 if @xmath143 , we find in a similar way that @xmath144 from ( [ kernel2 ] ) , we now get for @xmath145 , @xmath146 since @xmath147 is even for even @xmath17 and odd for odd @xmath17 , we see that the odd @xmath17-terms in the two sums of ( [ 5-knn1 ] ) cancel out , while the even @xmath17-terms are equal .
we then obtain by ( [ p2n ] ) and ( [ kernel1 ] ) that @xmath148 this proves ( [ kernel - relation1 ] ) .
if @xmath143 and if we use ( [ q2n+1 ] ) we obtain ( [ kernel - relation2 ] ) in a similar way .
we will also connect the density @xmath23 of the equilibrium measure for @xmath3 with the density @xmath67 of the equilibrium measure for @xmath13 . the equilibrium density
@xmath23 minimizes the energy ( [ energy1 ] ) .
it is characterized as the unique probability density function on @xmath149 with the property that for some constant @xmath150 , @xmath151 the density @xmath67 is the limiting mean eigenvalue density for the matrix ensemble ( [ ensemble2 ] ) as @xmath152 in such a way that @xmath153 .
then @xmath67 minimizes the weighted energy @xmath154 among all probability density functions @xmath21 on @xmath34 .
the density @xmath67 is uniquely characterized by the following variational conditions for some constant @xmath155 : @xmath156 [ lemma - eqdensity ] we have that @xmath157 we first show that @xmath158 is indeed a probability density .
this follows from the positivity of @xmath23 and the fact that @xmath159 to prove part ( i ) of the lemma , it is now sufficient to show that @xmath158 satisfies the variational conditions ( [ varcond2a ] ) and ( [ varcond2b ] ) .
we have @xmath160 this relation implies ( [ varcond2a ] ) and ( [ varcond2b ] ) by virtue of ( [ varcond1a ] ) and ( [ varcond1b ] ) . since these variational conditions uniquely characterize @xmath67 the lemma follows . in exactly the same way we find the following relation between the equilibrium densities of the sets @xmath29 and @xmath161 .
thus @xmath162 is the unique probability density on @xmath29 that minimizes the ( unweighted ) energy @xmath163 [ lemma - omega ] let @xmath162 be the equilibrium density of the set @xmath29 and let @xmath164 be the equilibrium density of @xmath165
. then @xmath166
the proof of theorem [ theorem11 ] is based on the following result for the matrix ensemble ( [ ensemble2 ] ) for the case that @xmath67 vanishes quadratically at @xmath4 .
there is a one - parameter family of limiting kernels that are connected with the hastings - mcleod solution of the painlev ii equation @xmath167 .
we denote these kernels by @xmath168 and we refer to ( [ kcritalpha1 ] ) below for the precise form of @xmath169 .
the superscript ii refers to the classification of singular points for the matrix ensemble ( [ ensemble2 ] ) see @xcite according to which there are three cases singular case i : : : equality holds in ( [ varcond2b ] ) for some @xmath170 .
any such point @xmath61 is then a singular exterior point in the external field @xmath13 .
singular case ii : : : @xmath67 vanishes at an interior point of its support . such a point is a singular interior point in the external field @xmath13 .
singular case iii : : : @xmath67 vanishes at an edge point to higher order than a square root .
such a point is a singular edge point in the external field @xmath13 . *
( ( * ? ? ?
* theorem 1.2 ) ) * [ theoremckv ] let @xmath13 be real analytic on @xmath34 such that ( [ asympvw ] ) holds .
suppose that @xmath171 and @xmath172 , and assume that there no other singular points besides @xmath4 .
define constants @xmath173 where @xmath174 is the density of the equilibrium measure of the support of @xmath67 .
let @xmath5 such that @xmath175 exists and put @xmath176 and @xmath177 .
then @xmath178 uniformly for @xmath61 , @xmath62 in compact subsets of @xmath179 .
now we are ready for the proof of theorem [ theorem11 ] with constants @xmath180 \qquad \mbox { and } \qquad c_{2,v } = \frac{2\pi}{c_{1,v}^{1/3 } } \lim_{x \to 0 + } \left[\sqrt x \
\omega_{v}(x ) \right],\ ] ] where @xmath181 is the density of the equilibrium measure of the support @xmath29 of @xmath23 .
we define @xmath182 then by lemma [ lemma - eqdensity ] we have that @xmath183 , which implies by the assumptions on @xmath23 near @xmath4 , that @xmath184 > 0.\ ] ] it follows from lemma [ lemma - eqdensity ] that there are no other singular points besides @xmath4 in the external field @xmath13 ( in fact , the absence of type i singular points follows from the proof of the lemma ) .
the conditions of theorem [ theoremckv ] are therefore satisfied . from ( [ defc1vc2v ] ) ,
( [ defc1wc2w ] ) and lemmas [ lemma - eqdensity ] and [ lemma - omega ] it follows that @xmath185 then if @xmath5 such that ( [ eq - limits ] ) and if @xmath56 and @xmath186 , we get from from theorem [ theoremckv ] that @xmath187 and if @xmath143 we also have @xmath188 then straightforward calculations based on ( [ kernel - relation1 ] ) and ( [ scalinglimitw3 ] ) show that the limit ( [ limitkernel ] ) exists with @xmath189 while if @xmath143 , then from ( [ kernel - relation2 ] ) and ( [ scalinglimitw4 ] ) we also get @xmath190 given @xmath194 and @xmath55 , we abbreviate @xmath195 and @xmath196 , and we let @xmath197 be the unique solution of the equation @xmath198 with asymptotics @xmath199 uniformly as @xmath200 in the sector @xmath201 for any @xmath202
. since @xmath203 has no real poles for @xmath204 , @xmath205 and @xmath206 are well - defined for real @xmath90 .
the functions @xmath205 and @xmath206 extend to analytic functions on @xmath207 $ ] , with branch points in @xmath4 .
we denote these extensions also by @xmath205 and @xmath206 . in the kernel
@xmath208 we need their values on the real line .
indeed , we have for real @xmath61 , @xmath62 , and @xmath90 , @xmath209 } \frac{\phi_{\alpha,1}(x;s ) \phi_{\alpha,2}(y;s ) - \phi_{\alpha,1}(y;s)\phi_{\alpha,2}(x;s)}{2\pi i(x - y)}.\ ] ] we first rewrite @xmath208 in a ` real ' form , using the functions ( see also @xcite ) , @xmath210 then we get from ( [ kcritalpha1 ] ) that @xmath211 and from ( [ diffeqphi ] ) we get the differential equation @xmath212 [ lemma - even - odd ] both @xmath213 and @xmath214 are real for real @xmath61 and @xmath90 .
in addition , @xmath215 is an even function and @xmath216 is an odd function , and they have the following asymptotic behaviors @xmath217 as @xmath77 .
let @xmath219 be the contour consisting of four straight rays oriented to infinity , where @xmath220 see figure [ contourgamma ] .
the contour @xmath221 divides the complex plane into four sectors @xmath222 , @xmath223 as shown in figure [ contourgamma ] .
* @xmath234 is analytic in @xmath235 .
* @xmath236 satisfies the following jump relations on @xmath237 , @xmath238 \label{rhp psi : b2 } \psi_{\alpha,+}(z;s ) & = \psi_{\alpha,-}(z;s ) \begin{pmatrix } 1 & 0 \\ -e^{\pi i \alpha } & 1 \end{pmatrix } , \qquad \mbox{for $ z\in\gamma_2 $ , } \\[1ex ] \label{rhp psi : b3 } \psi_{\alpha,+}(z;s ) & = \psi_{\alpha,-}(z;s ) \begin{pmatrix } 1 & e^{-\pi i \alpha } \\ 0 & 1 \end{pmatrix } , \qquad \mbox{for $ z\in\gamma_3 $ , } \\[1ex ] \label{rhp psi : b4 } \psi_{\alpha,+}(z;s ) & = \psi_{\alpha,-}(z;s ) \begin{pmatrix } 1 & -e^{\pi i \alpha } \\ 0 & 1 \end{pmatrix } , \qquad \mbox{for $ z\in\gamma_4$. } \end{aligned}\ ] ] * @xmath236 has the following behavior at infinity , @xmath239 here @xmath240 denotes the third pauli matrix . * @xmath236 has the following behavior near the origin .
if @xmath241 , @xmath242 and if @xmath243 , @xmath244 \operatorname{{\mathcal o}}\begin{pmatrix } |z|^{\alpha } & |z|^{-\alpha } \\ |z|^{\alpha } & |z|^{-\alpha } \end{pmatrix } , & \mbox{as $ z\to 0,\ , z \in s_2 $ , } \\[3ex ] \operatorname{{\mathcal o}}\begin{pmatrix } |z|^{-\alpha } & |z|^{\alpha } \\ |z|^{-\alpha } & |z|^{\alpha } \end{pmatrix } , & \mbox{as $ z\to 0,\ , z \in s_4$. } \end{cases}\ ] ] it was shown in @xcite that the riemann - hilbert problem for @xmath236 has a unique solution and that @xmath245 \psi_{\alpha}(z;s ) \begin{pmatrix } 1 \\ 0 \end{pmatrix } , & \mbox{for $ z\in s_2 $ , } \\[2ex ] \psi_{\alpha}(z;s ) \begin{pmatrix } 1 \\
e^{\pi i \alpha } \end{pmatrix } , & \mbox{for $ z\in s_3 $ , } \\[3ex ] \psi_{\alpha}(z;s ) \begin{pmatrix } 0 \\ e^{-\pi i \alpha } \end{pmatrix } , & \mbox{for $ z \in s_4 $ , $ { { \operatorname{re}}}z > 0 $ , } \\[3ex ] \psi_{\alpha}(z;s ) \begin{pmatrix } 0 \\
e^{\pi i \alpha } \end{pmatrix } , & \mbox{for $ z \in s_4 $ , $ { { \operatorname{re}}}z < 0$. } \end{cases}\ ] ] it is easy to check that @xmath246 and @xmath247 with @xmath248 also satisfy the conditions of the riemann - hilbert problem for @xmath236 .
therefore by the uniqueness of the solution we have @xmath249 from the first equality in ( [ symmetries2 ] ) it follows ( as noted in remark 2 of @xcite ) that @xmath250 which shows that @xmath213 and @xmath214 are real , since we obtain from ( [ deffalpha ] ) and ( [ symmetries ] ) that @xmath251 , \\
f_{\alpha,2}(x;s ) & = { { \operatorname{im}}}\left[e^{\frac{1}{2 } \pi i \alpha \operatorname{sgn}(x ) } \phi_{\alpha,1}(x;s ) \right ] .
\end{aligned}\ ] ] from ( [ rhp psi : c ] ) , ( [ definition : phi_alpha ] ) , and ( [ symmetries3 ] ) we obtain the asymptotic behaviors ( [ f1asymptotics ] ) and ( [ f2asymptotics ] ) . from ( [ scalinglimitv1 ] ) and ( [ kcritalpha2 ] ) , we get @xmath254 putting the fractions onto a common denominator , and using the even / odd properties of lemma [ lemma - even - odd ] , we find @xmath255 which leads to ( [ integrablekernel ] ) with @xmath256 and @xmath257 the system of differential equations ( [ diffsystem ] ) follows from ( [ diffeqf ] ) , ( [ deffalpha ] ) , and ( [ defgalpha ] ) and the asymptotic behavior ( [ realfalphaasymptotics])-([realgalphaasymptotics ] ) follows from ( [ f1asymptotics])-([f2asymptotics ] ) and ( [ deffalpha])-([defgalpha ] ) .
99 g. akemann , d. dalmazi , p.h .
damgaard , and j.j.m .
verbaarschot , qcd3 and the replica method , nucl.phys .
b 601 ( 2001 ) , 77124 .
p. bleher and a. its , semiclassical asymptotics of orthogonal polynomials , riemann - hilbert problem , and universality in the matrix model , ann .
( 1999 ) , 185266 .
, double scaling limit in the random matrix model : the riemann - hilbert approach , comm .
pure appl .
56 ( 2003 ) , 433516 .
a. borodin and p. forrester , increasing subsequences and the hard - to - soft - edge transition in matrix ensembles , j. phys .
a 36 ( 2003 ) , 29632981 .
a. borodin and a. soshnikov , janossy densities i. determinantal ensembles , j. statist .
( 2003 ) , 595610 .
t. claeys and a.b.j .
kuijlaars , universality of the double scaling limit in random matrix models , comm .
pure appl . math .
59 ( 2006 ) , 15731603 .
t. claeys , a.b.j .
kuijlaars , and m. vanlessen , multi - critical unitary random matrix ensembles and the general painlev ii equation , arxiv math - ph/0508062 , to appear in ann . math .
t. claeys and m. vanlessen , universality of a double scaling limit near singular edge points in random matrix models , comm .
phys . 273 ( 2007 ) , 499532 .
p. deift , orthogonal polynomials and random matrices : a riemann - hilbert approach , courand lecture notes in mathematics vol . 3 , new york university , 1999 .
p. deift and d. gioev , universality at the edge of the spectrum for unitary , orthogonal and symplectic ensembles of random matrices , comm .
pure appl . math .
60 ( 2007 ) , 867910 .
p. deift , t. kriecherbauer , and k.t - r mclaughlin , new results on the equilibrium measure for logarithmic potentials in the presence of an external field , j. approx .
theory 95 ( 1998 ) , 388475 .
p. deift , t. kriecherbauer , k.t - r mclaughlin , s. venakides , and x. zhou , uniform asymptotics for polynomials orthogonal with respect to varying exponential weights and applications to universality questions in random matrix theory , comm .
pure appl .
( 1999 ) , 13351425 .
fokas , a.r .
its , a.a .
kapaev , and v.yu .
novokshenov , painlev transcendents : the riemann - hilbert approach , mathematical surveys and monographs vol .
128 , amer .
soc . , providence , r.i . , 2006 .
forrester , log - gases and random matrices , ` www.ms.unimelb.edu.au/~matpjf/matpjf.html ` s.p .
hastings and j.b .
mcleod , a boundary value problem associated with the second painlev transcendent and the korteweg - de vries equation , arch .
rational mech .
73 ( 1980 ) , 3151 .
its , a.b.j .
kuijlaars , and j. ostensson , critical edge behavior in unitary random matrix ensembles and the thirty fourth painlev transcendent , arxiv 0704.1972 .
kuijlaars and k.t - r mclaughlin , generic behavior of the density of states in random matrix theory and equilibrium problems in the presence of real analytic external fields , comm .
pure appl .
53 ( 2000 ) , 736785 .
a.b.j . kuijlaars and m. vanlessen , universality for eigenvalue correlations from the modified jacobi unitary ensemble , internat .
notices 2002 ( 2002 ) , 15751600 .
mehta , random matrices , 3rd edition , elsevier , amsterdam , 2004 .
saff and v. totik , logarithmic potentials with external fields , springer - verlag , new york , 1997 .
tracy and h. widom , level spacing distributions and the airy kernel , comm .
159 ( 1994 ) , 151174 .
, level spacing distributions and the bessel kernel , comm .
( 1994 ) , 289309 .
m. vanlessen , strong asymptotics of laguerre - type orthogonal polynomials and applications in random matrix theory , constr .
approx . 25 ( 2007 ) , 125175 . | unitary random matrix ensembles @xmath0 defined on positive definite matrices @xmath1 , where @xmath2 and @xmath3 is real analytic , have a hard edge at @xmath4 . the equilibrium measure associated with @xmath3 typically vanishes like a square root at soft edges of the spectrum . for the case that the equilibrium measure vanishes like a square root at @xmath4 ,
we determine the scaling limits of the eigenvalue correlation kernel near @xmath4 in the limit when @xmath5 such that @xmath6 . for each value of @xmath2
we find a one - parameter family of limiting kernels that we describe in terms of the hastings - mcleod solution of the painlev ii equation with parameter @xmath7 . |
the electronic properties of most materials are determined by their crystal structure or lack thereof . for purely crystalline materials , well established methods of condensed matter
physics @xcite allow the nearly complete experimental characterization , as well as theoretical description , of the resulting electronic bands and associated density of states ( dos ) , all the way to transport and thermal properties .
in fact , our present understanding of the underlying mechanisms allows the manufacture of tailored artificial crystal structures such as photonic,@xcite phononic,@xcite polaritonic@xcite or plasmonic@xcite lattices .
while in the former two systems classical counterparts of electronic bands and transport properties are manifestly observed , in the latter two , electronic quasi - particle scattering is shown to lead to the formation of controllably engineered excitation bands and , in particular , the _ gaps _ between these as is required for a multitude of applications.@xcite in a strongly disordered system , anderson localization@xcite suppresses transport even in regions with a finite dos.@xcite particularly in low - dimensional systems , the so - called scaling hypothesis@xcite establishes the expectation that all states remain localized for non - interacting quasi - particles , and hence there seems little room for a similarly controlled `` engineering '' of _ bands _ of extended states .
nevertheless , such a complementary approach has already enjoyed some successes . local positional correlation in a disordered material has been shown to lead to resonant scattering events generating extended states at isolated energies in the spectrum.@xcite with much longer - ranged correlated disorder , even in one - dimensional ( 1d ) systems , effective metal - insulator transitions can be induced .
@xcite similarly , certain disordered configurations can lead to an optimization of the quantum interference process mediating excitonic transport in molecular networks .
@xcite in fact , a controlled disorder can induce highly selective transport properties,@xcite giving rise to materials with interesting new functionalities , as has been recently explored with photonic crystals .
@xcite in this paper we show how to open a channel of perfect transmission in an otherwise disordered system .
our approach uses a simple product structure for the random potential which , except for certain resonance conditions , leads to anisotropically localized states .
we believe our results to be of particular relevance for ultracold atoms or bose - einstein condensates in optical lattices as these are ideal for studying disorder effects.@xcite in fact , very recently the first direct observations of localization of matter waves in 3d disordered optical potentials have been reported .
@xcite here , by engineering the underlying disorder , our study shows how extended matter waves emerge from a background of localized states ( cp .
[ fig - states ] ) .
furthermore , our disorder structure allows for independent tailoring of the transport properties of the system in each direction , and gives rise to energy coexistence of extended and localized states , which can be manipulated with a high degree of control .
we prove our findings analytically and we corroborate them by extensive numerical studies . in section [ sec - locspec ] , we describe the chosen @xmath1-disorder , restricting ourselves to 2d for presentational simplicity .
we then deduce the form of the resulting dos and show that the eigenstates exhibit localization . in section [ sec - deloca ] , we discuss the conditions for the existence of extended states and show the appearance of a perfectly transmitting channel .
section [ sec - engineering ] presents a discussion on the engineering of the transport properties and spectral regions with mixed localized and extended states . in section [ sec - conclusions ]
we draw our conclusions .
details on numerical techniques and some lengthy derivations are provided in several appendices .
we work with a 2d tight - binding model described by the hamiltonian @xmath2 on a lattice with @xmath3 sites . here , @xmath4 denotes the @xmath5 hamiltonian matrix acting in the @xmath6 direction for each vertical arm at position @xmath7 such that @xmath8 and @xmath9 , with @xmath10 ( @xmath11 ) the usual annihilation ( creation ) operators at the site with coordinates @xmath12 . also , @xmath13 is the hopping along the @xmath7 direction .
the set @xmath14 gives the on - site energies and @xmath15 is the hopping term in the @xmath6-direction between sites @xmath16 and @xmath17 .
a pictorial representation of the lattice is shown in fig .
[ fig - lattice](a ) .
the energies and states of @xmath18 are the solutions of the schrdinger equation @xmath19 where @xmath20 contains the wavefunction amplitudes in the @xmath7-th vertical arm of the system and @xmath21 is the energy . here and in the following , we shall assume hard wall ( fixed ) boundary conditions in @xmath7 and @xmath6 direction , although the formalism does of course work with periodic boundaries as well .
we now introduce disorder into the system in the following way .
let @xmath22 and @xmath23 be random , uncorrelated sequences with corresponding probability distributions @xmath24 and @xmath25 which for simplicity will be taken as box - distributions of widths @xmath26 , @xmath27 , and mean values @xmath28
, @xmath29 .
the on - site random energies will be constructed from the _ product _ of these two sequences , i.e.@xmath30 additionally , we also choose the hopping strengths in @xmath6 direction to be random and determined by @xmath31 where @xmath32 is another _ independent _ random sequence . for simplicity , we choose the elements @xmath33 to have dimensions of energy measured in units of the hopping @xmath34 , thus @xmath35 and @xmath36 will be dimensionless .
this choice of @xmath37 and @xmath15 above gives rise to patterns for the disordered energy landscape and the random ( vertical ) @xmath6 couplings with a characteristic correlation in the @xmath7 and @xmath6 directions , as can be seen in figs .
[ fig - lattice](b ) and [ fig - lattice](c ) , hence the name @xmath1-disorder. let us remark that these choices , particularly for @xmath38 , resemble and generalise quasi-1d ladder " models currently discussed in the literature to describe electronic transport in dna @xcite and mesoscopic devices.@xcite however we believe that a faithful realization of our model can be implemented using optical potentials and matter waves , where single - site resolution and control has already been demonstrated.@xcite equations and allow us to factorize the @xmath4 matrix as @xmath39 where @xmath40 does not depend on the @xmath7 coordinate .
the matrix @xmath40 can be diagonalised via @xmath41 , where @xmath42 is the matrix whose columns are the orthonormal eigenvectors of @xmath40 and @xmath43 contains the eigenvalues @xmath44 in its diagonal . performing the change to a new basis @xmath45 where @xmath46 , we can reduce eq . to @xmath47
this is simply a set of @xmath48 decoupled schrdinger equations , @xmath49 each of which corresponds to a 1d disordered channel with random on - site energies @xmath50 for the @xmath51-th _ channel_. the disorder is uncorrelated within each channel and the distribution is @xmath52 , which is again a box - distribution with mean @xmath53 and width @xmath54 , for the @xmath51-th case . from eqs .
, it follows that the energy spectrum of the 2d @xmath55 system can be obtained from the union of the different spectra for the corresponding decoupled 1d channels .
the spectrum will be determined by the properties of the random distributions @xmath33 , @xmath35 , and @xmath36 . for finite @xmath48
the dos per site of the 2d system can be written as @xmath56 where @xmath57 is the dos per site of the @xmath51-th decoupled channel , characterized by a disorder distribution of mean @xmath58 and width @xmath59 . in the limit @xmath60 ,
. becomes @xmath61 where @xmath62 corresponds to the dos per site for the eigenvalues of @xmath40 . due to the nature of @xmath1-disorder
, the dos of the 2d system follows from a convolution of 1d distributions .
eqs . and can be used to obtain the dos numerically .
furthermore , since they only require calculations of 1-d distributions , they can be very efficiently combined with the functional equation formalism ( fef),@xcite to obtain the dos of the system for @xmath63 and either finite @xmath48 or @xmath64 . a brief overview of the fef is given in appendix [ ap - fef ] .
some examples of the dos for @xmath1-disorder are shown in fig .
[ fig - dosgeneric ] , where we see that the global distribution of states is a superposition of the dos from the individual channels . and maximum localization lengths @xmath65 in the @xmath7 and @xmath6 directions for @xmath1-systems as computed from the fef . for panel ( a )
@xmath63 and @xmath66 for @xmath67 , @xmath68 .
the dashed lines show @xmath69 for the @xmath48 decoupled infinite 1d channels that get added according to eq . to produce @xmath70 . for comparison
, the shaded histogram has been obtained from exact diagonalization of a @xmath71-site system , averaging over 25 realizations of the @xmath72 sequence .
the inset shows the fixed finite sequences @xmath35 and @xmath36 .
( b ) @xmath73 with parameters @xmath74 , @xmath75 , @xmath76 and @xmath77 ( solid ) , @xmath78 ( dashed ) . in this case @xmath79 .
the dos was calculated using eq . where the integral over @xmath80 $ ] was discretized using @xmath81 points .
notice the different scale ( right ) for the localization lengths , obtained according to eqs . and
.,title="fig : " ] + and maximum localization lengths @xmath65 in the @xmath7 and @xmath6 directions for @xmath1-systems as computed from the fef . for panel
( a ) @xmath63 and @xmath66 for @xmath67 , @xmath68 .
the dashed lines show @xmath69 for the @xmath48 decoupled infinite 1d channels that get added according to eq . to produce @xmath70 . for comparison , the shaded histogram has been obtained from exact diagonalization of a @xmath71-site system , averaging over 25 realizations of the @xmath72 sequence .
the inset shows the fixed finite sequences @xmath35 and @xmath36 .
( b ) @xmath73 with parameters @xmath74 , @xmath75 , @xmath76 and @xmath77 ( solid ) , @xmath78 ( dashed ) . in this case @xmath79 .
the dos was calculated using eq . where the integral over @xmath80 $ ] was discretized using @xmath81 points .
notice the different scale ( right ) for the localization lengths , obtained according to eqs . and
.,title="fig : " ] the states of the 2d system can be obtained from the eigenstates of the 1d decoupled channels by undoing the change of basis , @xmath82 .
if we consider the @xmath83-th eigenenergy of the @xmath51-th decoupled channel with corresponding eigenstate @xmath84 , the vector @xmath85 will only have one non - zero component at the @xmath51-th position .
it then follows that @xmath86 where @xmath87 means the @xmath51-th column of @xmath42 , which corresponds to the @xmath51-th eigenvector of the matrix @xmath40 which we denote by @xmath88 .
the 2d eigenstates of are thus obtained as @xmath89 where we must consider all 1d eigenstates @xmath90 for all the decoupled channels @xmath91 , giving the complete basis of @xmath55 states in the original coordinates .
notice that all eigenstates of a certain decoupled channel @xmath51 get multiplied by the same eigenvector of @xmath40 . from eqs .
and , and making use of known results on anderson localization in 1d,@xcite we conclude that , for _ generic _ cases , the eigenstates will be _ anisotropically localized _ for all energies , with different localization lengths in the @xmath7 and @xmath6 directions .
localization persists although our model only contains @xmath92 independent random on - site energies @xmath93 , as compared to the standard anderson model with @xmath3 uncorrelated values for @xmath37 .
an example of a localized eigenstate is shown in fig .
[ fig - states](b ) . for a given energy @xmath21 , the transport properties of the system in the @xmath7 direction , as @xmath94 , will be determined by the largest possible value of the localization length in this direction , @xmath95 .
it then follows that @xmath96 where the localization length for each channel @xmath51 at energy @xmath21 is defined as the inverse of the corresponding lyapunov exponent , @xmath97.@xcite additionally , localization in the @xmath6 direction will be determined by the lyapunov exponent of the 1d system defined by the @xmath40 matrix only , @xmath98 , for large enough @xmath48 .
the localization length in @xmath6 will change with the eigenvalue @xmath99 , and thus all wavefunctions obtained from the same decoupled channel [ same @xmath100 in eqs .
] will exhibit the same @xmath6-spreading .
the relevant localization length in @xmath6 as a function of the energy can be defined as @xmath101 where @xmath102 \label{eq - channelspectrum}\ ] ] denotes the boundaries of the energy spectrum of the @xmath51-th channel as @xmath94 , and @xmath103 .
that is , at energy @xmath21 , the relevant localization length in @xmath6 will be the maximum of @xmath104 over all channels whose spectrum includes @xmath21 .
the lyapunov exponents for the decoupled 1d channels , in the thermodynamic limit , can be obtained numerically using the fef ( see appendix [ ap - fef ] ) , which in combination with and permits the calculation of the relevant localization lengths for a 2d system with generic @xmath1-disorder , like those shown in fig .
[ fig - dosgeneric ] .
we emphasize that , irrespective of the detailed energy dependence of these localization lengths , all @xmath105 values are finite and hence correspond to localized states . in order to confirm the validity of the calculation of the localization lengths
, we carried out transport simulations in the @xmath7 direction using the transfer - matrix method ( tmm)@xcite ( see appendix [ ap - tmm ] ) on eq . without performing any decoupling .
this technique is capable of giving the whole spectrum of characteristic lyapunov exponents and thus automatically provides the largest decay ( localization ) length for an initial excitation travelling through the system . in fig .
[ fig - tmmvsfef ] we show the whole lyapunov spectrum of an @xmath1-disordered system ( @xmath94 and finite @xmath48 ) obtained from tmm , and compare it with the results from the fef on the decoupled channels .
the excellent agreement observed confirms the correctness of the decoupling transformation and of eq . .
from tmm ( dashed lines ) and fef ( solid lines ) for a system with @xmath1-disorder ( @xmath63 , @xmath106 ) and parameters @xmath78 , @xmath68 , and @xmath79 .
the inset shows the 10-element sequence considered for @xmath107 .
the thick black line highlights the trajectory of the minimum lyapunov exponent for each energy , i.e. the inverse of the maximum localization length , as computed by the fef .
the blue circles show every 50th data point of the corresponding tmm result ( errors are within symbol size ) . ]
despite the localized nature of eigenstates in generic @xmath1-disorder , a _ genuine delocalization _ in the @xmath7-direction can be achieved as well .
this is demonstrated in fig .
[ fig - trans - e - alll ] , where we show results from tmm calculations of the localization length @xmath95 for a system with @xmath106 and different values of @xmath108 ( here we assume a constant @xmath109 for presentational simplicity only ) .
we see that the enhancement of the localization length is always found inside the region @xmath110 .
in fact , as shown in fig . [ fig - trans - e - alll ] , for certain @xmath109 the localization length becomes comparable to @xmath111 independently of the longitudinal disorder @xmath112 , signalling the appearance of extended states .
this is in contrast to the @xmath113 dependence expected for quasi-1d systems,@xcite which is found at most other @xmath109 values .
direction obtained from tmm for a system of @xmath106 , @xmath114 and @xmath115 for different values of @xmath108 . for clarity only every 50th data point has been indicated by a symbol .
errors are within symbol size . for the chosen realisation of @xmath116 values ( shown in inset ) , @xmath117 corresponds to @xmath118 for one of the @xmath119 channels .
note that the tmm calculations were terminated at a fixed value of @xmath120 and hence the @xmath121 values for @xmath117 in the energy range @xmath110 do not diverge .
the right frame shows the dependence of @xmath122 versus @xmath112 at energy @xmath123 .
the green lines represent @xmath113 power laws . ] the reason for the existence of this band of extended states can be understood by returning to eq . .
we see that if at least one of the eigenvalues @xmath124 of @xmath40 is zero , then the corresponding 1d equation in no longer describes localization but rather an extended channel . as each @xmath100 value depends on the distributions @xmath35 and @xmath36 , the condition @xmath118
can be controlled by fine - tuning these parameters in our 2d disordered system , therefore inducing _ a perfectly transmitting channel_. indeed , the relation between the @xmath36 and @xmath35 values can be derived straightforwardly .
the characteristic polynomial of @xmath40 is @xmath125 and if @xmath126 then this will correspond to at least one eigenvalue of @xmath40 being zero . assuming that the @xmath107 disorders have been fixed , the condition can be satisfied by choosing the remaining @xmath127 values appropriately . for example , for @xmath128 , @xmath129 , @xmath130 , the condition @xmath118 reads @xmath131 respectively . in the most general case
one could select arbitrarily @xmath132 values for @xmath127 and then choose the remaining one such that @xmath133 .
from here on , we shall restrict ourselves to the case where @xmath134 constant for all @xmath6 .
this simplification makes the hoppings in the @xmath6 direction constant within each vertical arm , i.e. @xmath135 .
it should be clear , however , that this condition is in general not necessary for the existence of extended states .
because of the tridiagonal nature of @xmath40 , its determinant involves only even powers of @xmath109 , and it is a polynomial of order @xmath48 [ @xmath136 for even [ odd ] @xmath48 .
we can then consider only @xmath137 without loss of generality .
therefore , the spectrum of @xmath109 satisfying @xmath138 contains at most @xmath139 or @xmath140 independent positive real values .
@xcite for any of these resonant @xmath109 values @xcite ( @xmath118 ) ,
the spectrum of the otherwise disordered system has at least one perfectly conducting channel populated by @xmath111 extended states in the @xmath7 direction , in the energy interval @xmath141 $ ] .
the eigenstates of the resonant channel [ with @xmath118 in eq . ]
are bloch states , @xcite @xmath142 with energy @xmath143 $ ] , @xmath144 , for hard - wall boundary conditions @xmath145
. however , the 2d eigenstates given by eq .
, are in general still localized in the @xmath6 direction .
an example of such states is shown in fig .
[ fig - states](a ) . the presence of these extended states changes also the properties of the dos of the system . according to eq .
, the dos now contains the contribution from the spectrum of a periodic chain , @xmath146 for @xmath141 $ ] , and its characteristic van - hove singularities at the band edges , as shown in fig .
[ fig - dosresonant](a ) . direction for @xmath1-disorder with resonant channel , obtained from fef .
( a ) @xmath63 and @xmath147 for @xmath148 , @xmath76 and @xmath149 .
the dashed lines show @xmath69 for the decoupled infinite 1d channels , that get added according to eq . to produce @xmath70 .
the resonant channel contribution is highlighted in gray .
the inset shows the fixed finite sequence @xmath35 .
( b ) @xmath150 with parameters @xmath151 , @xmath75 , @xmath76 and different @xmath152 . in this case
dos was calculated using eq . where the integral over @xmath154 $ ] was discretized using 1000 points .
notice the different scale ( right ) for the inverse localization lengths.,title="fig : " ] + direction for @xmath1-disorder with resonant channel , obtained from fef .
( a ) @xmath63 and @xmath147 for @xmath148 , @xmath76 and @xmath149 .
the dashed lines show @xmath69 for the decoupled infinite 1d channels , that get added according to eq . to produce @xmath70 .
the resonant channel contribution is highlighted in gray .
the inset shows the fixed finite sequence @xmath35 .
( b ) @xmath150 with parameters @xmath151 , @xmath75 , @xmath76 and different @xmath152 . in this case
dos was calculated using eq . where the integral over @xmath154 $ ] was discretized using 1000 points .
notice the different scale ( right ) for the inverse localization lengths.,title="fig : " ] in general , there is a clear correlation between the dependences of @xmath155 and @xmath121 on @xmath109 , as shown in fig .
[ fig - stability ] .
therefore very large localization lengths occur in the vicinity of the resonant @xmath109 values , which will lead to effectively extended channels in finite systems , even if the resonant condition is not precisely met . for the case considered in fig .
[ fig - stability ] , corresponding to a quasi-1d system with @xmath106 , we see that deviations from the resonant @xmath109 value up to @xmath156 still ensure localization lengths in the @xmath7 direction of at least several thousands of lattice sites .
more interestingly , we have also checked that @xmath157 of random spatial deviations in each @xmath15 value i.e. when the factorization for the vertical hoppings is weakly broken ,
do have a similar effect only .
this tolerance of the delocalization channel is very significant , and makes our results relevant for potential experimental realizations , where deviations from the theoretically obtained parameters are to be expected .
furthermore , the situation becomes even more robust when the width of the system is increased , as we discuss below .
( right scale ) and localization length @xmath158 ( left scale , black solid and dashed lines ) as a function of @xmath109 for the @xmath1-system used in fig .
[ fig - trans - e - alll ] .
four resonant values appear in the interval shown @xmath117 , @xmath159 , @xmath160 , and @xmath161 .
the shaded regions highlight the intervals of @xmath109 corresponding to @xmath162 and @xmath163 deviation around the resonances .
the dashed line corresponds to calculations of @xmath158 where each @xmath15 value was allowed to deviate randomly up to @xmath162 from the target @xmath164 value . ]
the resonant condition requires @xmath165 to be an eigenvalue of the matrix @xmath40 , defined in eq . .
for short @xmath48 this implies tuning the vertical hopping elements via @xmath109 , for which several isolated values lead to the appearance of a transmitting channel . for large @xmath48
, however , the resonant condition translates into whether the dos for the system described by @xmath40 satisfies @xmath166 .
if the latter is true , then @xmath165 is an eigenvalue as @xmath60 , and at least one of the decoupled channels of the system will not correspond to a disordered chain .
therefore the infinite system ( @xmath64 , @xmath63 ) will always exhibit a perfect transmission band in the @xmath7 direction for @xmath141 $ ] as long as @xmath167 .
the matrix @xmath40 has the structure of a 1d anderson model with disordered onsite potentials , @xmath107 , constant hopping strength , @xmath108 , and hardwall boundaries . by gershgorin
s circle theorem,@xcite the spectrum of eigenvalues @xmath99 as @xmath60 fills the interval @xmath168 $ ] .
the condition for the existence of a perfect conducting channel is then @xmath169 the above inequality can be satisfied either by changing @xmath109 or the @xmath107 distribution . as an example , we see that when @xmath170 , the resonant channel will emerge ( for large @xmath48 ) for any value of @xmath109 .
numerical results for dos and localization length for @xmath1-systems in the thermodynamic limit ( @xmath73 ) when the above inequality is fulfilled are shown in fig .
[ fig - dosresonant](b ) .
we emphasize that for large @xmath48 , due to the nature of the dos in the 1d anderson model ( smooth and differentiable ) , the condition @xmath167 implies that the fraction of values of @xmath99 arbitrarily close to @xmath171 is finite , and therefore the number of channels with very large localization lengths grows with @xmath48 .
these will then lead to a dense continuum of effectively extended states for @xmath141 $ ] in systems with large but finite @xmath111 and @xmath48 whenever the above condition is satisfied .
we can estimate the relation between @xmath121 and @xmath48 for nearly resonant channels , i.e. when holds .
if we assume that @xmath172 is roughly constant around @xmath165 , then for finite @xmath48 we will find a value of @xmath99 as small as @xmath173 .
the localization in the corresponding decoupled channel which provides the maximum localization length
, characterized by a disorder distribution of width @xmath174 and mean @xmath175 , at @xmath123 is given by ( cp .
[ fig - trans - e - alll ] ) @xmath176,\ ] ] since @xmath174 is small .
@xcite this leads to @xmath177 therefore the minimum localization length at the band centre of the nearly resonant channels , which will emerge when eq .
is fulfilled , scales as @xmath178 .
we can roughly estimate the order of magnitude of @xmath179 by assuming a constant dos which leads to @xmath180 .
for the typical parameters used in our simulations , we then obtain @xmath181 as approximate lower bound .
we note that this estimate for large @xmath48 already works quite well for the @xmath106 case of fig .
[ fig - stability ] . in general , for large enough @xmath48 , localization of the wavefunctions given in is to be expected in the @xmath6-direction , with localization lengths determined by the eigenvectors of @xmath40 ( cp . fig .
[ fig - states ] ) .
the properties of the 1d disordered system described by the latter matrix depend on @xmath109 and @xmath107 , corresponding , respectively , to hopping and on - site energies .
hence we can write the localization length in the @xmath6-direction as @xmath182 , \label{eq : lly}\ ] ] in the weak disorder approximation , @xmath183.@xcite as explained in section [ sec - localization ] , the transverse localization is determined by the value of @xmath99 , and thus all states of a certain decoupled channel characterized by @xmath100 [ see eqs . ] will exhibit the same @xmath6-spreading .
localization in @xmath6 can then also be tailored by changing @xmath109 and/or the disorder distribution @xmath107 .
for example , for fixed disorder , @xmath184 can be increased by choosing larger @xmath109 values .
expression can be used to estimate the value of @xmath109 that we need if we want the extended states that we have generated in the @xmath7-direction to be also _ effectively _ delocalized ( @xmath185 ) in the vertical direction . by using eq . for @xmath165
we estimate that these states will be effectively delocalized in the @xmath6 direction when @xmath186^{1/2}. \label{eq : delocay}\ ] ] for the case @xmath170 we see that for systems a few hundred sites wide a value of @xmath187 will roughly ensure effective delocalization over the whole system of the resonant channel states , as shown in fig . [ fig - delocay](a ) .
alternatively , we can also find effective delocalization in @xmath6 , while keeping the localized character in the @xmath7 direction , as displayed in fig .
[ fig - delocay](b ) .
therefore , the localization properties of @xmath1-disorder can be engineered almost at will , and independently in the @xmath7 and @xmath6 directions , giving rise to the 4 basic configurations shown in figs .
[ fig - states ] and [ fig - delocay ] .
in fig . [ fig - tpd1 ] we show the diagram of transport regimes @xmath188 versus @xmath21 for a 2d system of size @xmath189 . by comparing the magnitude of the localization length against the system size , we can distinguish regions of efficient transport in direction @xmath7 , in @xmath6 or in both .
the lines separating the different phases are calculated in appendix [ ap - transport ] .
we confirm the validity of the diagram by numerical calculations of localization lengths estimated from transmission probabilities in @xmath7 and @xmath6 with open boundary conditions averaged over 100 realizations of the system , displayed by the color density plots in fig .
[ fig - tpd1 ] .
transmission in the @xmath7 direction emerges from the existence of resonant or quasi - resonant channels , whereas in @xmath6 it is due to effectively delocalized states only .
thus it must be clear that in the thermodynamic limit , @xmath190 , only the efficient transport along @xmath7 remains .
the diagram is system - size dependent with respect to the transport properties in the @xmath6-direction .
nevertheless , for finite systems , there exists the interesting possibility of realizing switching devices whose transmitting behaviour in @xmath7 and @xmath6 is highly controllable . versus @xmath21 in a 2d system with @xmath1-disorder of size @xmath189 and parameters @xmath191 , @xmath77 , @xmath192 , @xmath193 .
the dashed lines mark the limits of the efficient transport regimes obtained analytically and given by eq .
( straight horizontal dashed line ) , eq .
( vertical dashed lines ) and eqs . and ( bent dashed line ) .
the density plots in the background ( color scales are indicated above the main panel ) show numerically calculated localization lengths @xmath121 ( white - to - blue semitransparent ) and @xmath184 ( white - to - red ) .
the @xmath194 space was discretized in 675 points ( @xmath195 ) for the calculation , and the results averaged over 100 disorder realizations at every point .
the density plot uses a linear interpolation to smooth the visualization .
labels @xmath7 and @xmath6 highlight regions where @xmath196 and @xmath185 respectively .
the gray wedges at the lower corners correspond to regions outside the spectrum of the system . ]
different configurations of the diagram can be obtained by changing the disorder parameters @xmath26 , @xmath152 , @xmath27 and @xmath197 .
for example , @xmath197 controls the appearance of the resonant channels in @xmath7 , as well as the width of the plateau region in the border of effective delocalization in @xmath6 . on the other hand
, @xmath198 breaks the symmetry of the diagram , the region of efficient transport in @xmath6 shifts laterally preserving the plateau region and decaying asymmetrically on the sides . in fig .
[ fig - tpd2 ] , we show an alternative diagram of @xmath197 vs @xmath21 , where different transport regimes can be triggered simply by changing the mean value of the @xmath107 distribution , even for a fixed disorder realization . versus @xmath21 in a 2d system with @xmath1-disorder of size @xmath189 and parameters @xmath191 , @xmath77 , @xmath192 , @xmath199 .
the dashed lines mark the limits of the efficient transport regimes and can be obtained from the equations in appendix [ ap - transport ] .
the density plots in the background show numerical calculations of the localization lengths as in fig .
[ fig - tpd1 ] . ] in a finite @xmath1-system , the fulfilment of the resonance condition ( as discussed in section [ sec - zero ] ) induces @xmath111 bloch states in the @xmath7 direction for @xmath141 $ ] .
nevertheless , the other @xmath200 states coming from the remaining decoupled channels will be localized in @xmath7 , and can in principle have energies inside the range @xmath201 $ ] .
therefore , in general , we should expect coexistence of extended and localized states in @xmath201 $ ] .
the overlapping of the spectrum of resonant and localized channels can be seen in fig .
[ fig - dosresonant](a ) .
we have checked that the level spacing distribution in this region includes that for bloch states on a background of poissonians corresponding to localized states , as also found in ref . for a ladder model supporting coexistence . as @xmath111 and @xmath48
grow , the range @xmath201 $ ] will potentially become densely populated by extended and localized states .
the filling of that energy interval by bloch states in a continuous manner will render the presence of localized states irrelevant , as far as transport in the @xmath7 direction is concerned .
the @xmath51-th decoupled channel is characterized by an on - site energy distribution with mean @xmath202 and width @xmath203 , and whose spectral boundaries are given by eq . .
thus @xmath152 controls the position of the spectrum of the different channels .
the avoidance of coexistence is only possible if the @xmath72 s have all the same sign , i.e. @xmath204 .
otherwise the spectrum of all decoupled channels always includes @xmath123 , as the lower and upper bounds have opposite signs , thus giving rise to overlapping with the range @xmath201 $ ] . for simplicity
we consider @xmath205 .
then we can work with @xmath206 and consider positive energies only without loss of generality . in order to avoid coexistence
, we have to ensure that the lowest energy of all regions of localized states is greater than @xmath207 . in view of eq .
this condition is written as @xmath208 where @xmath209 stands for the minimum non - vanishing @xmath206 .
let us recall that the resonance condition is satisfied and thus zero is an eigenvalue of @xmath40 [ eq . ] .
once the disorder realization of the system is generated , i.e. fixed sequences @xmath33 , @xmath35 and @xmath109 , we can obtain the value of @xmath209 and then change @xmath152 to satisfy the inequality .
we note that is formally independent of the system size , but only reasonable for moderate @xmath48 .
since we are in the region of resonance , for large @xmath48 the @xmath209 will get arbitrarily small ( section [ sec - largesize ] ) and the required @xmath152 to avoid overlapping will be very large .
it is also possible to obtain some general bounds for @xmath152 without having to apply the above inequality to each particular realisation of the disorder . for a fixed value of the width of the system @xmath48
, we can estimate @xmath209 from the distribution @xmath210 around @xmath211 .
a rough estimation is obtained by assuming a constant distribution , @xmath180 , from which it follows that @xmath212 .
condition translates into @xmath213 which gives an approximation , usually overestimated , of the minimum @xmath152 value to avoid coexistence . in fig .
[ fig - over ] we show examples of spectrum engineering of the overlapping between extended and localized states .
-system with resonant channel and parameters @xmath106 , @xmath68 , @xmath214 for different @xmath152 .
black lines indicate dos for @xmath63 as obtained from fef , while blue ( red ) points on the @xmath7 axis mark eigenenergies of extended ( localized ) states for a finite realization with @xmath215 .
the shaded gray region highlights the contribution of the resonant channel to the total dos .
the inset shows the @xmath107 sequence considered characterized by @xmath216 .
according to eq .
blue - red coexistence should be avoided for @xmath217 . ]
our results show how a controlled choice of quite a simple , structured disorder as given by eqs . and can lead to extended states . by varying the parameters of the individual disorder distributions , we can engineer the localization and transport properties of the system independently in each direction .
we can furthermore control the energy coexistence of extended and localized states in the spectrum .
although our approach is based on the anderson model , we expect our results to hold in the many realms of anderson localization physics in general.@xcite for an experimental realisation of our model , the robustness of our results to small deviations away from the exact resonance conditions for quasi-1d systems , as shown in section [ sec - stability ] , and the broadening of the resonance condition for 2d systems ( section [ sec - largesize ] ) is of course reassuring
. nevertheless , the disorder does require single - site control in order to adjust the @xmath37 and @xmath15 values as needed .
the key feature expected of any experimental endeavour to measure our model is henceforth the attempt to achieve full control by implementing single site resolution for quantum state manipulation as well as quantum state analysis .
fortunately , important progress in this direction has already been achieved , e.g. in optical lattices.@xcite a scheme combining a disordered optical potential and an additional harmonic trap has been recently proposed @xcite to observe experimentally the coexistence of extended and localized wavefunctions .
we emphasize that our @xmath1-disorder can provide a genuine and controllable coexistence of extended and localized wavefunctions that does not occur in usual disordered systems exhibiting delocalization transitions .
while we have analysed 2d models for simplicity , the product structure for the disorder presented here can be straightforwardly extended to 3d systems , where similar features are to be expected , in particular , the possibility to engineer the transport properties in the three spatial directions independently .
we thank kai bongs for stimulating discussions and encouragement , and andreas buchleitner for a careful reading of the manuscript .
we gratefully acknowledge epsrc ( ep / f32323/1 , ep / c007042/1 , ep / d065135/1 ) for financial support .
acknowledges financial support from the german dfg ( bu 1337/5 - 1 , bu 1337/8 - 1 ) and the spanish micinn ( fis2009 - 07880 ) , and the hospitality of departamento de fsica fundamental at the university of salamanca . a.c . and r.a.r .
are thankful to the royal society for financial assistance ( india - uk science networks ) .
a.c . gratefully acknowledges the hospitality of the centre for scientific computing in the department of physics , university of warwick .
the dos and the localization length of an infinite 1d disordered chain , can be calculated numerically using the fef.@xcite this technique has been applied to obtain the spectral properties in the thermodynamic limit of different quantum - wire models with correlated and uncorrelated disorder.@xcite here we give a brief overview of the formalism applied to 1d tight - binding models with diagonal disorder , described by the equation @xmath218 , where the on - site energies are obtained randomly from a continuous distribution @xmath219 .
for such a system , the functional equation reads , @xmath220 where @xmath221 , with the conditions @xmath222 and @xmath223 for integer @xmath83 .
the angular variable , @xmath224 , follows from the representation of the wavefunction amplitude @xmath225 in polar coordinates .
the function @xmath226 corresponds to the cumulative distribution function of @xmath224 in the disordered chain in the thermodynamic limit .
the dos @xmath227 and the lyapunov exponent @xmath228 as functions of the energy are written in terms of @xmath226 as , @xmath229 where @xmath230 , and the prime indicates differentiation with respect to @xmath224 . by solving numerically eq . for different values of the energy , the spectral and localization properties of the infinite system can be obtained using the latter expressions .
the transfer - matrix method ( tmm ) allows for a very memory efficient way to iteratively calculate the decay length of electronic states in a quasi-1d system of width @xmath48 for lengths @xmath231.@xcite equation ( [ eq : canonical ] ) has to be rearranged into a form where the @xmath232 amplitudes of sites in layer @xmath233 when @xmath7 is chosen as the direction of transfer is calculated solely from parameters of sites in previous layers @xmath7 and @xmath234 , @xmath235 equation can be expressed in standard transfer - matrix form as @xmath236 = \underbrace { \left [ \begin{array}{cc } t^{-1}(e\openone - \boldsymbol{\epsilon}_x ) & -\openone \\ \openone & \mathbf{0 } \end{array } \right]}_{\mathbb{t}_n } \left [ \begin{array}{c } \boldsymbol{\psi}_x \\ \boldsymbol{\psi}_{x-1 } \end{array } \right ] \label{eq - tmm},\ ] ] where @xmath237 and @xmath238 are the @xmath239 zero and unit matrices , respectively .
formally , the transfer matrix @xmath240 is used to ` transfer ' electronic amplitudes @xmath241 from one slice to the next and repeated multiplication of this gives the global transfer matrix @xmath242 .
the limiting matrix @xmath243 exists@xcite and has eigenvalues @xmath244 , @xmath245 .
the inverse of these lyapunov exponents @xmath246 are estimates of decay / localization lengths and the physically most relevant largest decay length in the @xmath7 direction is @xmath247 .
the transport properties of the system along @xmath7 and @xmath6 can be engineered by tuning the parameters of the system : @xmath152 , @xmath197 , @xmath26 , @xmath27 , @xmath109 and @xmath21 . a qualitative diagram of the transport regimes , for sufficiently large but finite @xmath111 and @xmath48 , can be obtained by using the weak disorder expressions for the localization lengths @xmath121 and @xmath184 . for simplicity
we assume here that @xmath248 and we only focus on relations @xmath188 versus @xmath21 .
the basic condition to have resonant channels for large @xmath48 is given by eq . .
let us approximate that @xmath172 is constant for all @xmath99 and thus @xmath180 .
then for finite @xmath48 , when the above condition is satisfied , we will find a value of @xmath99 as small as @xmath249 .
the maximum localization length in @xmath7 ( provided by a channel with a disorder distribution of width @xmath250 and mean @xmath251 ) , then satisfies @xmath252.\ ] ] we can ensure efficient transport in @xmath7 by enforcing that the minimum bound for @xmath121 is larger than @xmath111 .
this implies @xmath253 this condition gives a continuum region of @xmath109 values providing transmitting behaviour in @xmath7 .
note , however , that @xmath196 will also be achieved outside this region , around resonant values of @xmath109 ( @xmath254 ) .
the localization length in @xmath6 is determined by the localization properties of the eigenstates of the matrix @xmath40 . the localization length in @xmath99
is given by @xmath182 , \label{eq - llyap}\ ] ] for @xmath255 $ ] and @xmath183 .
in this weak disorder approximation the localization length has a parabolic dependence in @xmath99 with its maximum value at @xmath256 . since the localization length in @xmath6
is determined by the value of @xmath99 , all states of a certain decoupled channel will exhibit the same @xmath6-spreading [ see eq . ] .
the spectrum of a channel characterized by @xmath100 is @xmath257,\ ] ] where @xmath103 . for a given energy @xmath21
, the maximum localization length in @xmath6 , and thus the relevant for transport processes , is given by @xmath104 such that @xmath100 is the closest value to @xmath197 whose spectrum includes @xmath21 .
for @xmath258 we get the maximum attainable value of @xmath184 in the system , and thus @xmath259 for @xmath260 , \label{eq - plateau}\ ] ] since any other @xmath100 whose spectrum overlaps with this range will provide smaller values of @xmath184 .
the localization length in @xmath6 thus displays a characteristic plateau structure [ see fig .
[ fig - dosgeneric](b ) ] whose width is proportional to the parameter @xmath197 . for energies larger than the interval above , there is overlapping of the spectra of channels with @xmath261 ( recall that we assume @xmath262 ) . for a given energy @xmath21 ,
the maximum @xmath184 is attained for @xmath100 such that @xmath21 coincides with the upper edge of the spectrum of the @xmath100-channel , since this will be the value of @xmath99 closest to @xmath197 .
therefore we can identify @xmath263 and substitute in eq . to obtain @xmath184 as function of the energy . for energies smaller than the interval
, it can be seen that the maximum @xmath184 is attained for @xmath100 such that @xmath21 coincides with the lower edge of the spectrum of the @xmath100-channel .
thus @xmath264 is obtained from eq . for @xmath265 .
the expression and its energy - range of validity depends on whether the contributing @xmath100 is positive ( @xmath266 ) or negative ( @xmath267 ) , as well on the sign of ( @xmath268 ) .
therefore , from the plateau structure in the interval , the localization length in @xmath6 decays with @xmath21 . for @xmath77
the dependence of @xmath264 is symmetric around @xmath123 .
a non - vanishing @xmath152 induces a shift in the position of the plateau and can lead to an asymmetric decay of @xmath264 from its maximum value , left and right from the plateau .
these features can be observed in the numerical results shown in fig .
[ fig - dosgeneric](b ) . in order to have efficient transport along @xmath6 we require @xmath269 .
following the reasoning given above it is possible to write general relations for @xmath188 to ensure the latter condition . in order to avoid cumbersome expressions ,
we show them here only for the case @xmath77 : @xmath270 , \label{eq - ty1}\ ] ] and the latter relations are symmetric around @xmath123 .
as discussed above , a non - vanishing @xmath152 will shift the plateau structure , and it can break the symmetric behaviour around it .
the formulas above provide a qualitative understanding of the influence of the different parameters on the transport regimes in the @xmath6 direction . while not all solutions of @xmath138 correspond to real @xmath109 values
, this does not represent a problem .
if @xmath48 is very small , the values of @xmath107 may be tuned to ensure real solutions , but in our experience for @xmath272 some real solutions are found without further restrictions on the values of @xmath107 .
nevertheless , the @xmath273 distribution can be shifted by changing @xmath197 to ensure @xmath274 , in which case all resonant @xmath109 values are real . | we describe how to engineer wavefunction delocalization in disordered systems modelled by tight - binding hamiltonians in @xmath0 dimensions .
we show analytically that a simple product structure for the random onsite potential energies , together with suitably chosen hopping strengths , allows a resonant scattering process leading to ballistic transport along one direction , and a controlled coexistence of extended bloch states and anisotropically localized states in the spectrum .
we demonstrate that these features persist in the thermodynamic limit for a continuous range of the system parameters .
numerical results support these findings and highlight the robustness of the extended regime with respect to deviations from the exact resonance condition for finite systems .
the localization and transport properties of the system can be engineered almost at will and independently in each direction .
this study gives rise to the possibility of designing disordered potentials that work as switching devices and band - pass filters for quantum waves , such as matter waves in optical lattices . |
in certain cases a commutative ring - spectrum @xmath1 can be equipped with a distinguished element @xmath2 called a complex orientation of @xmath1 ( see @xcite ) .
the pair @xmath3 is called a complex oriented ring spectrum .
given a complex orientation @xmath4 of @xmath1 , every smooth complex projective variety @xmath5 can be equipped with a homological class @xmath6 called the fundamental class of @xmath5 ( here @xmath7 stays for the complex dimension of @xmath5 ) .
this class has the property that the cap - product @xmath8 provides an isomorphism of cohomology and homology groups of @xmath5 .
the isomorphism is often called the poincar duality isomorphism . from the modern point of view
it looks pretty interesting to obtain an analogue of this result in the context of algebraic geometry .
it is reasonable in this case to choose and fix a field @xmath9 and consider a symmetric commutative ring @xmath0-spectrum @xmath10 in the sense of voevodsky @xcite ( for the concept of symmetric @xmath0-spectrum see jardine @xcite ) .
the @xmath0-spectrum @xmath10 determines bi - graded cohomology and homology theories ( @xmath11 and @xmath12 ) on the category of algebraic varieties @xcite .
( we also assume the spectrum @xmath13 to be a ring - spectrum _
i.e. _ be endowed with a multiplication @xmath14 , which induces product structures in ( co)homology . ) in some cases @xmath10 can be equipped with a distinguished element @xmath15 , which morel calls an orientation of @xmath10 . following him the pair @xmath16
is called an oriented symmetric commutative ring @xmath0-spectrum .
the orientation @xmath17 equips both cohomology @xmath11 and homology @xmath12 with trace structures @xcite .
the latter means that for every projective morphism @xmath18 of @xmath9-smooth irreducible varieties with @xmath19 there are two operators @xmath20 and @xmath21 satisfying a list of natural properties .
define now a fundamental class of a @xmath9-smooth projective equi - dimensional variety @xmath5 of dimension @xmath7 as @xmath22 , where @xmath23 is the structure morphism .
our main result claims that the map @xmath24 is a grade - preserving isomorphism ( poincar duality isomorphism ) .
there are at least two interesting examples of oriented symmetric commutative ring @xmath0-spectra .
the first one is a symmetric model @xmath25 of the algebraic cobordism @xmath0-spectrum @xmath26 of voevodsky @xcite .
this symmetric commutative ring @xmath0-spectrum @xmath25 together with an orientation @xmath27 is described in details in ( * ? ? ?
* sect.6.5 ) . so that , every smooth irreducible projective variety @xmath28 of dimension @xmath7 has the fundamental class @xmath29 \in \mathbb { mgl}_{2d , d}(x)$ ] and the cap - product with this class @xmath30 is an isomorphism .
the second example is the eilenberg mac lane @xmath0-spectrum @xmath31 ( it is intrinsically a symmetric @xmath0-spectrum representing the motivic cohomology ) .
this @xmath0-spectrum @xmath31 is constructed in @xcite and we briefly describe its orientation here .
recall that for a smooth variety @xmath28 the first chern class of a line bundle with value in the motivic cohomology defines a functorial isomorphism @xmath32 .
thus , @xmath33 and the class of the line bundle @xmath34 over @xmath35 is a free generator of @xmath36 .
this class provides the required orientation of @xmath31 .
similarly to the case of algebraic cobordism , one has the fundamental class @xmath29 \in \textrm h^\mathcal m_{2d , d}(x)$ ] in motivic homology and the isomorphism : @xmath37 to embellish the latter result , let us mention that unlike the topological context in algebro - geometrical case the canonical pairing @xmath38 is generally degenerated even with rational coefficients @xcite .
the paper is organized as follows .
section 1 is devoted to product structures in extraordinary cohomology and homology theories . in section 2
we formulate poincare duality theorem and derive it from two projection formulae , which are proven in sections 3 and 4 .
finally , in appendices a and b we display some useful properties of orientable theories
. * acknowledgements . *
the first author is in debt to a.merkurjev for inspiring discussions at the initial stage of the work .
he is especially grateful to the institute for advanced study ( princeton ) for excellent working conditions .
the main result of the paper was obtained during the stay of the second author at universitt essen and the current text was mostly written during his short - time visits to universitt bielefeld and ihs ( bures - sur - yvette ) .
the second author is very grateful to all of these institutes for shown hospitality and excellent working possibilities during these visits .
* @xmath39 is a category of smooth quasi - projective algebraic varieties over a field @xmath9 .
* @xmath40 always denotes a diagonal morphism ; * symbol @xmath41 denotes trivial one - dimensional bundle ; * for a vector bundle @xmath42 over @xmath5 we write @xmath43 for its section sheaf ; * for a vector bundle @xmath42 over @xmath5 we write @xmath44 for the dual to @xmath42 ; * @xmath45 is the projective bundle of lines in @xmath42 ; * @xmath46 \in { \mathbb p}^n $ ] * typically @xmath47 is regarded as a hyperplane in @xmath48 * @xmath35 is a space defined in @xcite * @xmath49 ; some products in ( co)homology consider a symmetric @xmath0-spectrum @xmath10 @xcite endowed with a multiplication @xmath52 making @xmath10 a symmetric commutative ring @xmath0-spectrum . then the spectrum @xmath10 determines bigraded cohomology and homology theories on the category of algebraic varieties @xcite . a ring structure in cohomology
is then given by the cup - product satisfying the following commutativity law .
for @xmath53 and @xmath54 , one has : @xmath55 where @xmath56 is the involution described in appendix b. then the pair @xmath61 is called _ an oriented symmetric commutative @xmath0-spectrum_. if @xmath10 can be endowed with an element @xmath15 satisfying the conditions @xmath62 and @xmath63 then @xmath64 is called _ an orientable symmetric commutative @xmath0-spectrum_. for such a @xmath0-spectrum @xmath65 by lemma [ chernandepsilon ] and the commutativity law is reduced to @xmath66 . in this case it is convenient to set @xmath67 , @xmath68 , @xmath69 , and @xmath70 , where @xmath11 ( resp .
@xmath12 ) are ( co)homology theories represented by the @xmath0-spectrum @xmath10 .
the functors @xmath71 and @xmath72 are ( co)homology theories taking values in the category of @xmath73-graded abelian groups .
although all our duality results hold for bigraded ( co)homology groups , we shall work , for simplicity , with the @xmath73-grading just introduced .
multiplicativity of the @xmath0-spectrum @xmath10 provides a canonical way @xcite to supply the functors @xmath74 and @xmath75 ( contravariant and covariant , respectively ) with a product structure consisting of two cross - products @xmath76 and two slant - products @xmath77 one also defines two inner products @xmath78 as @xmath79 and @xmath80 , correspondingly .
the cup - product makes the group @xmath81 an associative skew - commutative @xmath73-graded unitary ring and this structure is functorial .
( skew - commutativity is not obvious and implied by the orientability of @xmath10 as it is shown in appendix [ aux2 ] ) .
the cap - product makes the group @xmath82 a unital @xmath81-module ( @xmath83 for every @xmath84 ) and this structure is functorial in the sense that @xmath85 .
we shall need below the following associativity relations , which are completely analogous to ones existing in the topological context ( see , for example , @xcite ) . for @xmath86 , @xmath87 , @xmath88 , @xmath89 , and @xmath90
, we have : for the final object @xmath101 in @xmath39 one , clearly , has @xmath102 .
this provides us with a distinguished element @xmath103 ( fundamental class of the point ) such that for any smooth @xmath5 and arbitrary @xmath104 , one has : @xmath105 .
( here we assume the standard identification @xmath106 . )
one can easily verify that the canonical isomorphism @xmath107 may be written as @xmath108 . throughout the paper
we implicitly use this construction and usually denote @xmath109 by @xmath110 .
poincar duality theorem let @xmath10 be an orientable symmetric commutative ring @xmath0-spectrum .
then the involution @xmath111 from @xmath112 coincides with the identity as explained in appendix [ aux2 ] . so that , the commutativity law is reduced to @xmath113 . setting @xmath67 , @xmath68
, we see that the functor @xmath114 takes value in the category of skew - commutative @xmath73-graded rings . for what follows
it is convenient to give the following let @xmath10 be an orientable symmetric commutative ring @xmath0-spectrum .
a chern element is an element @xmath115 such that @xmath116 and the family @xmath117 is a free basis of the free rank two @xmath118-module @xmath119 .
another term , which can be used for a chern element , is a non - homogeneous orientation of @xmath10 .
a chern element @xmath17 lifts to a chern structure in the cohomology theory @xmath74 in the sense of ( * ? ? ? * def.3.2 ) and to a commutative chern structure in the homology theory @xmath75 ( resp .
* defs.2.1.1 , 2.2.12 ) ) .
in fact , for every line bundle @xmath120 over @xmath121 there exists a diagram of the form @xmath122 , where @xmath123 is a finite dimensional @xmath9-vector space , @xmath124 is a torsor under a vector bundle over @xmath5 , and the morphism @xmath125 is such that the line bundles @xmath126 and @xmath127 are isomorphic @xcite .
denote by @xmath128 the class in @xmath129 such that @xmath130 .
by @xcite the element @xmath128 is well - defined and the assignment @xmath131 is a chern structure in @xmath74 .
moreover , the family of operators @xmath132 forms a commutative chern structure in the homology theory @xmath75 . any chern structure in @xmath74 ( resp .
on @xmath75 ) lifts to a trace structure in the cohomology ( resp .
homology ) , see ( * ? ? ?
* thm.4.1.2 ) ( resp .
* thm.5.1.4 ) ) .
namely , to every projective morphism @xmath18 of smooth varieties over @xmath9 one assigns two grade - preserving operators @xmath133 and @xmath134 satisfying a list of natural properties .
precise definitions of trace structures in a ring ( co)homology theory is given in @xcite .
the operators @xmath135 and @xmath136 are called trace operators .
( by historical reasons they called _ integrations _ in @xcite . )
the trace structures @xmath137 and @xmath138 are explicit and unique up to the following normalization condition . for a smooth divisor
@xmath139 : @xmath140 @xmath141 and @xmath142 .
let @xmath10 be an orientable symmetric commutative ring @xmath0-spectrum equipped with a chern element @xmath115 .
for a smooth projective variety @xmath5 with the structure morphism @xmath23 we call @xmath147 the * fundamental class * of @xmath5 in @xmath75 and denote it by @xmath148 .
definitely , the class @xmath29 $ ] depends rather on the pair @xmath149 than on the @xmath0-spectrum @xmath10 .
however , we prefer to use this simplified notation , since we always keep in mind one chosen and fixed chern element @xmath17 throughout the paper .
[ poincareduality ] let @xmath10 be an orientable symmetric commutative ring @xmath0-spectrum equipped with a chern element @xmath152 .
then for every smooth projective variety @xmath5 the maps @xmath153 and @xmath154 are mutually inverse isomorphisms . if @xmath155 and @xmath5 is equi - dimensional of dimension @xmath7 then @xmath29 \in a_{2d , d}(x)$ ] . in this case
the isomorphism @xmath153 identifies @xmath156 with @xmath157 .
one can extract the following nice consequence of the poincar duality theorem , which enables us to interpret trace maps in a way topologists like to . to proof the first equality
, one should just check that @xmath164 .
taking into account that @xmath29=f^![y]$ ] , one immediately derives the desired relation from the first projection formula below ( theorem [ projectionformula ] ) .
the second statement can be proven in a similar way , but requires the `` dual '' projection formula that we do not consider here .
the proof of theorem [ poincareduality ] is based on two projection formulae for cap- and slant - products .
consider the cartesian square @xmath176\ar_{\operatorname{id}}[d]&x\times x\ar^\tau[d]\\ x\ar^-\delta[r]&x\times x.}\ ] ] since the map @xmath177 is flat , the square is transversal due to ( * ? ? ?
* b.7.4 . ) . by the base change property [ basechange ]
, one has : @xmath178 . by theorem [ projectionformula ] ,
one has : @xmath179 that implies [ m1a ] . to get [ m1b ] one uses cohomological projection formula ( [ coproj ] ) instead .
_ proof of theorem [ poincareduality ] .
_ let @xmath190 denote corresponding projections .
observe that for every @xmath191 one has the relation @xmath192 .
( in fact , the element @xmath193 is of degree zero , because the map @xmath193 is grade - preserving . )
thus , one has : @xmath194\overset{\ref{as2}}=&\alpha{\smallsmile}(\delta_!(1)/{\lbrack}x{\rbrack})=\alpha.\end{aligned}\ ] ] * zero - section morphisms of line bundles : @xmath199 ; * closed embeddings @xmath200 of smooth divisors ; * zero - sections of a finite sum of line bundles : @xmath201 * zero - sections of arbitrary vector bundles : @xmath202 ; * closed embeddings @xmath203 ; * projections @xmath204 ; @xmath218_{k_b } \\ { \mathbb p}({\mathbf{1}}\oplus{\mathcal n})\ar@{^(->}[r]_{k_0}\ar@/^0.7pc/[d]^p & b&x\ar@{_(->}[l]^{k_1}\\ y\ar@{^(->}[r]_{j_0}\ar@{^(->}[u]^{s } & y\times{\mathbb a}^1\ar@{^(->}[u]^{t } & y\ar@{_(->}[l]^{j_1}\ar@{^(->}[u]^{i}}\ ] ] first , we shall show that the morphism @xmath219 in diagram [ keydiagram ] belongs to the class @xmath196 .
let @xmath220 and @xmath221 .
using lemma [ useful ] we can rewrite @xmath222 as @xmath223 , where @xmath224 and @xmath225 .
from the gysin exact sequence , we have : @xmath226 therefore , @xmath227 and @xmath228 .
( the second relation yields from [ 2.5 ] : @xmath229 . ) thus , one has : @xmath230 applying lemma [ lifting ] to the left - hand - side square of diagram [ keydiagram ] and denoting @xmath231 by @xmath232 , one has : @xmath233 similarly : @xmath234 by the proposition assumption , we have the relation @xmath235 . combining this with equalities [ intermidiatefinal ] , [ intermidiateone ] , and [ intermidiatetwo ] ,
one gets : @xmath236 we now move the desired relation one more step further to the right in diagram [ keydiagram ] and show that @xmath237 .
observe that @xmath238 is a monomorphism .
therefore , it suffices to check that for every elements @xmath239 and @xmath240 we have : @xmath241 setting @xmath242 , @xmath243 , and applying lemma [ lifting ] to the right - hand - side square of diagram [ keydiagram ] , one has : @xmath244 . in the same way : @xmath245 . combining these two relations with [ intermidiaterelation ] , one sees that @xmath237 .
[ splbdl ] let @xmath246 be an @xmath247-dimensional vector bundle over a variety @xmath163 which splits in the sum of line bundles .
then the zero - section morphism @xmath248 belongs to the class @xmath196 . in order to proceed with the case of an arbitrary vector - bundle
, we need the homological analogue of the splitting principle .
consider a vector bundle @xmath252 of constant rank @xmath247 over a smooth irreducible variety @xmath163 .
let @xmath253 be the corresponding principal @xmath254-bundle over @xmath163 , @xmath255 be the diagonal tori , and @xmath256 be the orbit variety with the projection morphism @xmath257
. finally , we denote by @xmath258 the pull - back of the vector bundle @xmath259 .
[ splitting ] the bundle @xmath260 splits in a direct sum of line bundles and the map @xmath261 is a universal splitting epimorphism ( i.e. for any base - change @xmath262 the induced map @xmath263 is a splitting epimorphism ) .
the projection @xmath264 and the natural @xmath265-action on @xmath253 makes it a principal @xmath265-bundle over @xmath266 .
moreover , if @xmath267 is the pull - back of @xmath253 , there is a natural isomorphism of principal @xmath254-bundles @xmath268 over @xmath266 .
the bundle @xmath269 over @xmath266 corresponds exactly to the principal @xmath254-bundle @xmath270 .
thus , the mentioned isomorphism of principal @xmath254-bundles over @xmath266 shows that the bundle @xmath269 splits in a direct sum of line bundles ( say corresponding to the fundamental characters @xmath271 of the tori @xmath265 ) .
this proves the first assertion of the proposition . to prove the second one ,
consider a borel subgroup @xmath272 in @xmath254 ( say the subgroup of all upper triangle matrices ) and let @xmath273 be the maximal unipotent subgroup of @xmath272 ( the group of upper triangle matrices with 1 s on the diagonal ) .
let @xmath274 ( this is just the flag bundle over @xmath163 associated to @xmath259 ) .
the bundle @xmath275 comes equipped with projections @xmath276 and @xmath277 , where the projection @xmath278 is induced by the inclusion @xmath279 . using the natural @xmath273-action on @xmath253 , it is easy to check that there is a tower of morphisms : @xmath280 which has a principal @xmath281-bundle on each level ( each level is a torsor over the trivial rank one vector bundle ) . by the strong homotopy invariance property @xcite ,
the induced map on homology @xmath282 is an isomorphism .
as it was already mentioned , @xmath275 is a full flag bundle over @xmath163 associated to the bundle @xmath259 .
thus , there is a tower of morphisms @xmath283 in which each level is a projective bundle associated to a vector bundle . by the projective bundle theorem ( pbt ) [ pbt2 ]
, we have a split epimorphism in homology induced on each floor .
therefore , the map @xmath284 is a split epimorphism as well .
letting @xmath266 be as above , denote by @xmath289 the pull - back of the bundle @xmath287 with respect to the morphism @xmath290 . then by proposition [ splitting ]
the bundle @xmath289 splits in a direct sum of line bundles and the induced map @xmath291 is an epimorphism .
let @xmath292 and @xmath293 be morphisms induced by zero - sections of the corresponding vector bundles .
then the diagram @xmath294^{\bar p } & { \mathbb p}({\mathbf{1}}\oplus\mathcal v)\\ y^{\prime}\ar^{\bar s}[u]\ar^{p}[r]&y\ar^{s}[u]}\ ] ] is transversal .
let @xmath209 and @xmath295 .
choosing @xmath296 such that @xmath297 and applying lemma [ lifting ] , one gets : @xmath298 and @xmath299 two expressions on the right - hand - sides coincide by proposition [ splbdl ] .
applying proposition [ reduction ] we reduce the question to the case of the zero - section morphism @xmath213 of the normal bundle @xmath300 .
the morphism @xmath207 belongs to @xmath196 by proposition [ vectorbdls ] . 1 . [ diamond1](left distributivity )
let @xmath306 , and @xmath290 be projective morphisms .
if @xmath307 then @xmath308 , provided that both sides of the equality are well defined .
[ diamond2 ] given a transversal square with projective morphisms @xmath125 and @xmath309 @xmath310\ar_f[d]\ar@{ .. >}^h[dr]&y\ar^f[d]\\ x\ar_g[r]&z } \ ] ] one has the following equalities : @xmath311 .
[ diamond3 ] in the square above : @xmath312 .
[ diamond4 ] let @xmath313 be the standard embedding @xmath314 and @xmath315 be the projection map .
let @xmath316 be the same as in the projective bundle theorem ( see [ pbt2 ] )
. then @xmath317 .
part [ diamond1 ] immediately follows from the definition of the operation @xmath318 , [ diamond2 ] and [ diamond3 ] are trivial corollaries of the transversal base - change property , [ diamond4 ] easily follows from the pbt .
fix now a variety @xmath319 and take the @xmath247-dimensional projective space @xmath320 over @xmath5 .
( up to the end of this section all the schemes are considered over the base scheme @xmath5 and the product is implicitly taken over @xmath5 . ) due to the pbt , the element @xmath321 may be decomposed as @xmath322 where @xmath323 is the canonical generator of @xmath324 as an @xmath81-algebra and @xmath325 ( see ( * ? ? ?
* lemma 1.9.3 ) ) .
this equality together with the previous lemma gives us the following decomposition of the identity operator @xmath326 .
taking into account the relation @xmath327 , where @xmath328 is the standard embedding , we can rewrite the cap - product with @xmath193 operator in the form : @xmath329 applying lemma [ diamond ] to transversal squares : @xmath330\ar@{^(->}[r]\ar@{ .. >}_{p_{1,n}s_{ij}}[dr]&{\mathbb p}^n\times{\mathbb p}^{n - j}\ar^{p_{1,n - j}}[d]\\ { \mathbb p}^{n - i}\ar@{^(->}_{s_i}[r]&{\mathbb p}^n}\ ] ] ( where we denote by @xmath331 the projection map @xmath332 ) , one gets the following equality : @xmath333(here we set @xmath334 , unless @xmath335 . )
let us check the first statement .
for @xmath339 we , trivially , have @xmath340 . applying the map @xmath341 to both sides of [ dec ] and then using lemma [ diamond].c for
the transversal squares @xmath342\ar_{p_{1,n - j}}[d]&{\mathbb p}^{n - j}\ar^{p_{n - j}}[d]\\ { \mathbb p}^n\ar_{p_n}[r]&x,}\ ] ] one gets : @xmath343 taking into account that due to [ diamond].d @xmath344 , ( @xmath345 ) , one has : @xmath346 by the pbt , for any @xmath347 we can choose an element @xmath348 such that @xmath349 rewriting the proposition statement in our notation , we should verify the relation @xmath355 .
we proceed by induction on @xmath247 .
the case @xmath339 is trivial .
let the proposition hold for @xmath356 .
then for @xmath357 , by lemma [ diag ] , we have : @xmath358 and @xmath359 by the induction hypothesis the expressions on the right - hand - side coincide .
the induction runs .
given @xmath362 and @xmath182 one should verify that @xmath363 clearly , both sides of ( [ projectioncase ] ) are @xmath81-linear . by the pbt , @xmath364 is generated as an @xmath81-module by the elements @xmath365 .
thus , it suffices to check the proposition just for these elements . from (
* lemma 1.9.1 ) , we have a relation @xmath366 in @xmath364 , where @xmath367 is the standard embedding map and the element @xmath368 is considered here as lying in @xmath364 via the pull - back operator for the projection @xmath369 .
denote by @xmath370 the projection map @xmath371 . since @xmath372
, we have by proposition [ thecaseofclosedimbedding ] : @xmath373 using proposition [ upthendowninhomology ] , one has : @xmath374 proof of the second projection formula the strategy of the proof of theorem [ slantproductandpushforward ] is very similar to one used in the previous section .
it is again convenient to introduce a class @xmath375 consisting of projective morphisms @xmath18 such that for any @xmath180 , @xmath376 , and @xmath84 the relation @xmath377 holds in @xmath183 .
( here @xmath185 .
below we use similar notation rules . )
[ slantforthesectionofprojectivebundle ] let @xmath379 be a vector bundle over a smooth variety @xmath163 and let @xmath378 be the zero - section of the projection @xmath380
. then the morphism @xmath207 belongs to the class @xmath375 .
let @xmath181 and @xmath381 .
using functoriality of the slant - product , an associativity relation , and formula [ coproj ] , one gets : @xmath382 ( here the relation @xmath383 appears from the base - change property applied to the product with @xmath0 . ) as well as in the proof of theorem [ projectionformula ] our arguments are based on the deformation diagram which obtained from ( [ keydiagram ] ) by multiplication with a variety @xmath180 . for convenience , we reproduce this diagram here .
@xmath387_{k_b } \\
t \times { \mathbb p}({\mathbf{1}}\oplus{\mathcal n})\ar@{^(->}[r]_{k_0}&t \times b&t\times x\ar@{_(->}[l]^{k_1}\\ t \times y\ar@{^(->}[r]_{j_0}\ar@{^(->}[u]^{s}&t \times y\times{\mathbb a}^1\ar@{^(->}[u]^{i_t } & t \times y\ar@{_(->}[l]^{j_1}\ar@{^(->}[u]^{i}}\ ] ] since the map @xmath396 is an isomorphism , we can set @xmath397 and @xmath243 . applying lemma [ lifting ] again
, one gets : @xmath398 combining these equalities with relation ( [ slantintermidiaterelation ] ) proves the proposition .
consider the following commutative diagram with transversal square : @xmath404&x\times{\mathbb p}^{n - r}\ar_i[l]\ar^{p_r}[d]&t\times x\times { \mathbb p}^{n - r}\ar_{\bar q}[l]\ar^{p_r}[d]\\ & x&t\times x\ar_q[l ] } \ ] ] clearly , both sides of ( [ proj1relation ] ) are @xmath183-linear .
so , we may assume that @xmath405 .
since @xmath406 , one has : @xmath407 by proposition [ upthendowninhomology ] : @xmath408 applying the base - change property to the square in the diagram above , we get : @xmath409 the proposition is proven . some properties of a trace structure [ aux ] in this appendix we give a brief description of some useful properties of a trace structure , which are utilized in the paper .
although we make a deal with both cohomological and homological contexts , we present all the results for homology .
cohomological variant is `` dual '' in the obvious sense and may be found in @xcite .
all the proofs for the homological case not provided below can be found in @xcite .
it is not hard to check that this definition is accordant to ones given in @xcite or @xcite .
let us check , for example , that for a closed embedding @xmath125 condition [ trans2 ] implies the isomorphism : @xmath414 .
the short exact sequence above may be viewed as a total complex of the bicomplex : @xmath415&{\bar g}^*t_y\ar^{-df}[r]&{\bar g}^*f^*t_x\\ 0\ar[r]&t_{y^\prime}\ar_{d\bar f}[r]\ar^{d\bar g}[u]&{\bar f}^*t_{x^\prime}.\ar_{dg}[u]}\ ] ] since [ trans2 ] is exact , the bicomplex is acyclic . on the other hand , it is quasiisomorphic to the two - term complex @xmath416 .
[ lifting ] suppose , we are given a transversal square @xmath418 & x\\ y^\prime\ar_{\bar g}[r]\ar^{\bar f}[u]&y\ar[u]_f}\ ] ] with projective morphism @xmath125 .
let @xmath209 and @xmath419 .
then the following relations hold : [ useful ] consider the following diagram with transversal square @xmath428_{k^1}\\ v\ar[r]_{k^0}\ar@/^0.7pc/[d]^p & x\\ w\ar[r]_{j}\ar[u]^{i } & y,\ar@{^(->}[u]^{q}}\ ] ] where @xmath290 is projective , @xmath429 is a closed embedding , @xmath430 is the open complement of @xmath163 in @xmath5 , @xmath431 is the corresponding open embedding , @xmath432 , and the map @xmath433 induces an isomorphism in homology . then @xmath434 .
let @xmath347 .
since the map @xmath435 is an isomorphism and @xmath436 , we can , using the transversal base - change property , lift @xmath437 up to @xmath438 , such that @xmath439 .
then , the gysin exact sequence implies that @xmath440 .
chern element and homothety involution [ aux2 ] let @xmath448 and take the point @xmath449 as a distinguished point .
let @xmath10 be a commutative ring symmetric @xmath0-spectrum .
for @xmath450 consider a map @xmath451 sending @xmath452 $ ] to @xmath453 $ ] ( preserving the distinguished point ) .
it defines an involution @xmath454 on the sphere @xmath0-spectrum @xmath455 @xcite . for any space @xmath5 it defines an involution on the cohomology theory @xmath11 as follows : @xmath456 where @xmath457 is the @xmath0-suspension isomorphism and
@xmath458 is its inverse
. set @xmath459 the following lemma shows that @xmath65 for a certain class of @xmath0-spectra .
we show that for any @xmath450 one has @xmath460 .
let @xmath461 be a linear embedding .
for @xmath121 consider the projection @xmath462 .
the map @xmath463 is surjective since @xmath110 and @xmath464 is a free base of the bigraded @xmath465-module @xmath466 .
now the proof completes as in ( * ? ? ?
* lemma 1.6 ) | frank adams introduced the notion of a complex oriented cohomology theory represented by a commutative ring - spectrum and proved the poincar duality theorem for this general case . in the current paper we consider oriented cohomology theories on algebraic varieties represented by multiplicative symmetric @xmath0-spectra and
prove the duality theorem , which mimics the result of adams .
this result is held , in particular , for motivic cohomology and algebraic cobordism of voevodsky . |
physical properties of low - dimensional magnetic materials have attracted considerable attention at the border of condensed matter physics . in particular ,
the bond - alternating quantum magnets with both ferromagnetic ( f ) and antiferromagnetic ( af ) couplings have been extensively investigated , because they have exhibited quite fascinating properties owing to the competition of f and af interactions .
a simple af bond - alternating chain ( bac ) may be a f - af alternating quantum heisenberg chain , and the spin-1/2 copper - based compound @xmath2@xcite is a typical example , in which the haldane - like behavior has been observed .
this kind of alternating chains have been actively studied both theoretically and experimentally in the last decades .
another interesting antiferromagnetic bac is the f - f - af alternating heisenberg chain with period @xmath3 . according to oshikawa - yamanaka - affleck ( oya)@xcite ,
the f - f - af chain with spin @xmath4 may exhibit a magnetization plateau at @xmath5 , where @xmath6 is the magnetization per site . in a typical example of spin-1/2 f - f - af chain of @xmath7 with strong f coupling ( @xmath8 ) and weak af coupling ( @xmath9 ) ,
however , no plateau in the low - temperature magnetic curve has been observed@xcite .
such a breakdown of the magnetization plateau in small bond ratio of @xmath10@xcite suggests that the oya condition is only a necessary condition .
there is another intriguing bac with period @xmath3 , which consists of af - af - f alternations , and is a ferrimagnet .
the examples of af - af - f chains are @xmath11_{n}$ ] with spin @xmath12@xcite , @xmath13_{n}$ ] with @xmath14 @xmath15 and @xmath16 @xmath17@xcite , _ and @xmath18_{n}$ ] with @xmath12af - af - f-3_. these compounds have strong f and weak af couplings , and no magnetization plateaus have been seen experimentally .
if the ratio @xmath19 f - f - af - af , hagiwara , which possesses a relatively large ratio @xmath20 observed experimentally , in this paper , we shall invoke the transfer matrix renormalization group ( tmrg ) technique @xcite to study the thermodynamic properties of the af - af - f chain and f - f - af chain , respectively .
the rest of this paper is organized as follows . in sec .
ii , we shall construct the model hamiltonian for the trimerized j - j - j@xmath21 heisenberg spin chain . in sec .
iii , we shall present our tmrg results on the thermodynamic behaviors of the systems .
the comparisons between the tmrg results and the results of the modified spin - wave ( msw ) theory will be discussed in sec .
finally , a brief summary will be given .
let us consider a trimerized @xmath22 j - j - j@xmath21 heisenberg quantum spin chain .
the hamiltonian of the system reads @xmath23where @xmath24 and @xmath25 are exchange integrals with @xmath24 , @xmath26 denoting the antiferromagnetic coupling and @xmath24 , @xmath27 the ferromagnetic coupling , @xmath28 is the external magnetic field , and we take @xmath29 .
the schematic spin arrangements of the system are shown in figs .
[ chain ] .
when @xmath30 ( or @xmath31 ) , the system becomes a uniform @xmath22 heisenberg ferromagnetic ( or antiferromagnetic ) chain . when @xmath32 , as indicated in figs .
[ chain ] , if @xmath33 , @xmath34 , it is a ferrimagnetic spin chain ( i.e. the configuration is af - af - f ) ; if @xmath35 , @xmath36 , it becomes an antiferromagnetic system ( i.e. the configuration is f - f - af ) .
the competition of f and af couplings in these two trimerized chains would give rise to plenty of interesting characteristics .
the magnetic properties of this model with s=1/2 , 1 , 3/2 and 2 in a magnetic field at zero temperature have been investigated by use of the density matrix renormalization group methodgubo .
it is purpose of this paper to study the thermodynamic properties of these trimerized quantum heisenberg spin chains by means of the tmrg technique@xcite .
the tmrg method was detailed in two nice reviews@xcite , and we shall not repeat the technical skills here for concise . in our calculations ,
the number of kept optimal states is taken as @xmath37 for the susceptibility @xmath38 , and @xmath39 for the specific heat @xmath40 , where the width of the imaginary time slice is taken as @xmath41 .
we have used different @xmath42 and @xmath43 to verify the accuracy of calculations . at high temperature ,
the error caused by the trotter - suzuki decomposition is important and is of the order @xmath44 for a fixed @xmath6 . at low temperature ,
the error resulting from the basis truncation becomes important , which drops exponentially with increasing @xmath6 initially and reaches to a finite value .
the trotter - suzuki error is less than @xmath45 , and the truncation error is smaller than @xmath46 in our calculations .
the physical quantities presented below are calculated down to @xmath47 ( in units of @xmath8 ) .
[ mh - aaf](a ) shows the magnetization per site @xmath6 as a function of magnetic field @xmath28 at different temperatures for the af - af - f chain with @xmath48 . at zero temperature ,
a magnetization plateau at @xmath1 was observed@xcite .
as illustrated in fig .
[ chain](a ) , the ground state of the @xmath22 af - af - f chain in the absence of the magnetic field corresponds to @xmath1 .
when the external field is applied , the ground state with @xmath1 still persists for @xmath49 , which is consistent with the oya condition@xcite . for @xmath50 the plateau state
is destructed , and @xmath51 increases rapidly with increasing the field . when the field is increased to @xmath52 , @xmath51 becomes saturated , and the ground state becomes a fully polarized state .
when the temperature is increased , the magnetization plateau retains at low temperature , and is gradually smeared out at high temperature owing to the thermal fluctuations .
it is noting that the magnetization @xmath51 as a function of @xmath28 at finite temperature starts from zero , that is nothing but the result of mermin - wagner theorem .
[ mh - aaf](b ) gives the temperature dependence of the magnetization @xmath53 under different magnetic fields for the af - af - f chain with @xmath48 .
it can be seen that @xmath54 behaves distinctly for different ranges of the external field . at @xmath55 , @xmath54
first decreases rapidly , and then declines slowly with increasing temperature ; at @xmath56 , @xmath54 first increases , and then declines slowly with increasing temperature ; at @xmath57 , @xmath54 declines with increasing temperature . at @xmath58 and @xmath59 ,
@xmath53 appears to remain constant with zero curvature at low temperature , and declines slowly with temperature .
it is found that just below and above @xmath60 and @xmath59 , the curvature of @xmath54 changes abruptly with opposite signs , implying that @xmath58 and @xmath59 could be viewed as the crossover fields .
this observation is also consistent with fig .
[ mh - aaf](a ) where there are two crossing points in the curves for different temperatures at @xmath60 and @xmath59 .
[ ffa - m](a ) shows the field dependence of the magnetization per site @xmath51 at different temperatures for the f - f - af chain with @xmath61 .
a magnetization plateau at @xmath1 is also observed at zero temperature . as indicated in fig .
[ chain](b ) , the ground state of the @xmath22 af - af - f chain in the absence of a magnetic field corresponds to @xmath62 . with increasing the magnetic field ,
@xmath6 first increases till to @xmath1 at @xmath63 , then goes into the plateau state with @xmath1 for @xmath64 , beyond which the plateau state is destroyed , and @xmath51 then increases rapidly to a saturation at @xmath65 where the system becomes fully polarized .
the occurrence of the plateau state with @xmath1 is also consistent with the oya condition@xcite .
it is seen that at low temperature the magnetization plateau still persists , though the width of plateau is getting narrow . at high temperature ,
the plateau is gradually smeared out by thermal fluctuations , and m(h ) increases nonmonotonically till the saturation with increasing the magnetic field .
the temperature dependence of the magnetization per site @xmath54 under different fields for the f - f - af chain with @xmath61 is presented in fig .
[ ffa - m](b ) .
one may see that @xmath54 shows different behaviors at low temperature in different ranges of the magnetic field . at @xmath66 ,
@xmath54 increases first and then declines smoothly with increasing temperature ; at @xmath67 , @xmath54 decreases dramatically at low temperature and then slowly with increasing temperature ; at @xmath68 , @xmath54 increases at low temperature and then declines slowly with increasing temperature ; at @xmath69 , @xmath54 declines remarkably with increasing temperature . at @xmath70 and @xmath71 ,
@xmath54 remains almost unchanged at lower temperature , and then declines slowly with increasing @xmath72 .
again , the curvature of @xmath54 changes abruptly with opposite signs just below and above @xmath70 and @xmath73 .
the results presented in fig . [ ffa - m](b ) suggest that at low temperature the system may enter into different states in different ranges of the magnetic field , namely , for @xmath66 , the system is in an ordering state ; for @xmath67 the system could go into the spin - canting state ; for @xmath68 the system enters into the plateau state where the excitations are gapful ; and for @xmath69 the system goes into another ordering state before saturation .
it is interesting to note that the magnetization as a function of magnetic field and temperature exhibits different behaviors for the af - af - f chain and the f - f - af chain , for the former is a ferrimagnet , while the latter is an antiferromagnet .
however , both systems shows a magnetization plateau with @xmath74 at low temperature , being in agreement with the oya condition .
let us now present the tmrg results of the susceptibility ( @xmath75 ) as a function of temperature ( @xmath72 ) for the af - af - f chain . in the absence of the magnetic field ,
the susceptibility diverges when the temperature tends to zero , and with increasing temperature , @xmath75 decreases rapidly , as shown in fig . [
xt](a ) . for different @xmath76 ,
@xmath77 exhibits similar behaviors at low @xmath72 , but distinct behaviors at moderate @xmath72 , as manifested in the inset of fig .
[ xt](a ) , where @xmath78 versus @xmath72 is plotted for various @xmath76 , and a low-@xmath72 divergence and a round minimum are observed for @xmath79 , indicating that when the f interaction @xmath8 predominates , the system shows a ferrimagnetic behavior ; when the af interaction @xmath31 is predominant , the system falls into a trimerized bond state . in the presence of the magnetic field
, the susceptibility @xmath75 shows quite different behaviors , as presented in figs . [
xt](b ) and ( c ) at @xmath61 .
unlike the case of @xmath80 , @xmath38 under nonzero @xmath28 exhibits a peak at low temperature except @xmath58 and @xmath59 where with increasing @xmath72 , @xmath75 first increases and then drops slowly for the former , and goes to divergence at low temperature for the latter . with the increase of @xmath28 , the peak of @xmath75 appears to become rounded , and
the peak height of @xmath75 gets suppressed for @xmath55 , then increased for @xmath56 , and then becomes suppressed again for @xmath81 .
these observations show again that the system can enter into different states within different ranges of the magnetic field . for the f - f - af chain ,
the temperature dependence of the susceptibility @xmath82 for different @xmath76 and @xmath28 is shown in figs . [ xt](d)-(f ) , respectively . in the absence of the magnetic field
, the susceptibility reveals peaks at low temperature for different @xmath76 , and @xmath83 at @xmath84 .
when @xmath76 is either smaller or larger , e.g. , @xmath85 , @xmath86 , @xmath87 , @xmath75 shows one peak ; while @xmath88 and @xmath89 , @xmath75 shows a double - peak structure , as manifested in fig .
[ xt](d ) .
the calculated results indicate that the system shows an af behavior .
it seems that the double - peak structure comes from the two different kinds of excitations induced by the competition between f and af interactions . in the presence of the magnetic field , the temperature dependence of the susceptibility @xmath75 at @xmath61 for different external fields
are presented in figs .
[ xt](e ) and ( f ) .
it is found that @xmath77 behaves variously in different ranges of the magnetic field .
when @xmath28 is small , @xmath77 shows a double - peak structure at low temperature with the first peak sharp and the second peak round ; with increasing @xmath28 , the height of the first peak grows dramatically , while that of the second round peak leaves almost unchanged ; when @xmath90 , the first peak is remarkably suppressed at @xmath91 , and vanishes at @xmath92 , while the second round peak becomes slightly higher , as shown in fig .
[ xt](e ) .
when @xmath93 , the second peak of @xmath75 moves to the low - temperature side , and at @xmath94 , the two peaks merge into a single ; when @xmath28 is increased further , the peak of @xmath75 is suppressed again .
the behaviors of @xmath77 show that the f - f - af chain can enter into different states at low temperatures under different magnetic fields , consistent with the results presented in fig .
[ ffa - m](b ) . to verify our tmrg results ,
we have included the experimental data of an af - af - f chain compound @xmath95 $ ] and a f - f - af chain compound @xmath96 in figs .
[ theory - tmrg ] for a comparison .
[ theory - tmrg](a ) shows the experimental data of @xmath78 versus @xmath72 for @xmath95 $ ] , whose magnetic structure is an af - af - f chain with spin-@xmath97 @xcite .
owing to our computing capacity , we can not at present calculate directly the magnetic properties of the spin-@xmath98 af - af - f chain by using the tmrg method . as a qualitative understanding , however , in the inset of fig .
[ theory - tmrg](a ) , we have presented our tmrg result of a spin-@xmath0 af - af - f chain with @xmath99 and @xmath100 .
it can be seen that the experimental and calculated curves share qualitatively similar characteristics .
[ theory - tmrg](b ) gives a comparison of the experimental susceptibility of the compound @xmath7 , which is a f - f - af chain with spin-@xmath0 @xcite , with the tmrg result .
it is found that our tmrg result fits well with the experimental data , with the coupling parameters @xmath100 , @xmath101 @xmath102 , and @xmath103 .
our calculated result is in agreement with the previous theoretical result@xcite . in this subsection
, we shall discuss the specific heat of the spin-1/2 af - af - f and f - f - af chains , as shown in figs .
[ ct ] for different @xmath104 and @xmath28 . for the af - af - f chain , in absence of the magnetic field , the specific heat @xmath40 as a function of temperature shows clearly a double - peak structure , as indicated in fig .
[ ct](a ) . with increasing @xmath104 ,
the low - temperature peak of @xmath40 leaves almost unchanged , while the second peak moves to the high - temperature side , and gets rounded .
this feature of the specific heat shows that the system might exist two different kinds of excitations due to the competition of f and af interactions , as will be discussed in next section .
it seems that the first peak corresponds to the f excitations , while the second peak corresponds to the af excitations . in presence of the magnetic field ,
the temperature dependence of the specific heat c for the af - af - f chain with @xmath61 is shown in figs .
[ ct](b ) and ( c ) under different fields .
when @xmath28 is small , the double - peak structure of @xmath40 is observed ; with increasing @xmath28 , the first peak tends to vanish , while the second peak remains almost unchanged , as shown in fig .
[ ct](b ) .
when @xmath105 , the two peaks become a single .
when @xmath106 , as presented in fig .
[ ct](c ) , the first peak recovers anew , and with the increase of @xmath28 , the first peak moves to the lower temperature side till @xmath107 , then shifts to the high temperature side for @xmath108 , and tends to vanish for @xmath109 , while the second peak shifts slightly to the higher temperature side with increasing @xmath28 .
the results show that the system is in a few different states under different magnetic fields . for the f - f - af chain ,
the temperature dependence of the specific heat @xmath40 is shown in figs .
[ ct](d)-(f ) for different @xmath76 and @xmath28 . in absence of the magnetic field
, @xmath40 also shows a double - peak structure ; with increasing @xmath76 , the height of the first peak of @xmath40 is suppressed but its position leaves unchanged , while the second peak tends to shift to the high - temperature side , as illustrated in fig .
[ ct](d ) .
it is seen that the qualitative behavior is similar to that of the af - af - f chain presented in fig .
[ ct](a ) .
the results suggest that the f - f - af chain has also two kinds of excitations due to the competition of f and af interactions .
[ ct](e ) and ( f ) gives the temperature dependence of the specific heat @xmath40 for the f - f - af chain with @xmath61 under different magnetic fields .
similar to the af - af - f chain , with increasing @xmath28 , the first peak of c is gradually suppressed , while the second peak retains nearly intact till @xmath110 .
when @xmath111 , the first peak is completely suppressed ; when @xmath112 , with the increase of @xmath28 , the first peak recovers again with the peak height decreased and the peak position shifted to the lower temperature side , while the second peak moves slightly to the high temperature side with the peak height a little bit enhanced ; when @xmath113 , the situation changes , namely , with increasing @xmath28 , the first peak of @xmath40 is enhanced and moves to the high temperature side , and the second peak is also promoted .
the present observation displays that the f - f - af chain falls into different thermodynamic states under different external fields , and has some behaviors similar to the af - af - f chain , although both chains have quite different ground states , as the latter is a ferrimagnet , while the former is an antiferromagnet .
the double - peak structure of the temperature dependence of the specific heat can be viewed as a minimum at low temperature and a schottky - like maximum at high temperature .
the schottky - like anomaly can be fitted with a so - called two - level system , where the thermal population of the levels is governed by the maxwell - bolzman statistics , @xmath114 ( @xmath115 ) with the distribution function @xmath116 , and the total energy @xmath117 .
the specific heat can be obtained by @xmath118 .
define @xmath119 , @xmath120 .
the schottky - like specific heat can be written as @xmath121similarly , it is found that the magnetic susceptibility can be fitted bymohn , @xmath122where @xmath123 is the excitation gap .
however , our tmrg results can not be well fitted by these above two formulae . by noting that the present systems such as the af - af - f and f - f - af chains are quantum trimerized systems with a period @xmath3 , and the double - peak structure may be regarded as the consequence of the competition between f and af excitations , the thermodynamic properties of these systems
could be mimicked by a superposition of double two - level systems . as a result
, the susceptibility for the af - af - f and f - f - af chains may be expected to fit by @xmath124and the double - peak structure of the specific heat can be fitted by the following form : @xmath125where @xmath126 , @xmath127 are fitting parameters , and @xmath128 , @xmath129 are the excitation gaps for the double two - level systems . in these above equations , @xmath130 and @xmath131
are defined by eqs .
( x - twolevel ) and ( [ c - twolevel ] ) , respectively . for the af - af - f chain , in absence of the magnetic field
, we have found that @xmath132 can not be fitted by eq .
( [ x - double ] ) , but can be well fitted by the equation @xmath133 with @xmath134 , @xmath135 , and @xmath136 the fitting parameters , as shown in fig .
[ 3l - xt](a ) for @xmath137 as an example .
we note that the susceptibility of a spin-@xmath0 ferromagnetic system was expanded as @xmath138 with @xmath139 @xmath140 at low temperatures@xcite .
thus , it is not surprising that the present system shows a similar behavior , as the af - af - f chain is a ferrimagnet .
however , in presence of the magnetic field , the susceptibility @xmath77 of the af - af - f chain can be nicely fitted by eq .
( [ x - double ] ) , as compared in fig .
[ 3l - xt](b ) for @xmath61 and @xmath141 as an example .
[ 3l - xt](c ) and ( d ) give the tmrg fitting results in terms of eq .
( [ x - double ] ) for the susceptibility @xmath132 of the f - f - af chain in absence and presence of the magnetic field , respectively .
one may see that apart from the case where @xmath77 is divergent at lower temperature and can not be fitted by eq .
( [ x - double ] ) , the susceptibility as a function of temperature for both trimerized chains can be well fitted by a superposition of the double two - level systems . figs .
[ 3l - ct ] present the tmrg results of the specific heat as a function of temperature fitting to a superposition of double two - level systems characterized by eq .
( [ c - double ] ) for the af - af - f and f - f - af chains .
it can be found that the specific heat with double - peak structure can be well fitted by eq .
( [ c - double ] ) at low temperature , with only a slightly quantitative deviation at high temperature , showing that the main features of the specific heat of both trimerized chains can be reproduced by a superposition of double two - level systems .
the fitting results could give the two excitation gaps .
the reasons for such nice fittings come from the fact that these two trimerized chains have three branches of excitations ( see next section ) , one gapless f excitations , and two gapful excitations , which could be equivalently treated by a superposition of double two - level systems with two gaps @xmath128 and @xmath142 .
we have also found that a direct fitting to a three - level system is not as good as to double two - level systems .
the zero - field specific heat of a quantum ferrimagnet , which is believed as an intrinsic double - peak structure of topological origin and is different from the external field induced double - peak structure , was investigated by using the modified spin - wave ( msw ) theory@xcite , where it has been found that in the case of the af - af - f - f bac , there are four distinct branches of spin - wave excitations , say , the lower two ferromagnetic bands construct the low - temperature bump , while an upper ferromagnetic band and an antiferromagnetic band contribute to a schottky - type peak at mid temperatures .
the field - induced double - peak structure of the specific heat in a quantum ferrimagnet has also been studied by using the linear spin - wave theorymaisinger . in the case of spin-@xmath143 ferrimagnet ,
a simple picture was given by noting that the zeeman term introduces a gap to the f excitations that increases with increasing the field , and a gap to the af excitations that decreases with increasing the field , and the specific heat really reflects the dual structure of the antiferromagnetic and ferromagnetic excitations for all fields .
the msw theory can also be applied to the present @xmath22 af - af - f and f - f - af chains in absence of the external field so that our tmrg results of the thermodynamic quantities may be used to compare with the corresponding msw results . in the conventional spin - wave scheme ,
the spin deviations in each sublattice diverge in the one - dimensional ( 1d ) antiferromagnets , but the quantum as well as thermal divergence of the number of bosons can be overcome in the takahashi scheme@xcite that will be applied to the present antiferromagnetic f - f - af chain . the af - af - f chain is a ferrimagnet , whose magnetization should be nonzero in the ground state but zero at finite temperature , leading to that we can apply the yamamoto schemeyama4,yama5 , where the lagrange multiplier was introduced directly in the free energy , to our present ferrimagnetic af - af - f chain .
the detail derivations of the msw formalism are collected in appendix a , where the linear modified spin - wave ( lmsw ) theory , which is up to the order of @xmath144 , and the perturbational interacting modified spin - wave ( pimsw ) theory , which is up to the order of @xmath145 , are included .
our results show that the msw results may describe well the low - temperature behavior of the system , but they are not good in agreement with the tmrg results at high temperature . figs .
[ msw - aaf](a ) and ( b ) present the zero - field excitation spectra which were calculated by the lmsw and pimsw methods within the yamamoto scheme for the af - af - f chain .
it is seen that this trimerized system has three branches of excitation spectra , one gapless excitation spectrum and two gapful excitation spectra . when @xmath76 is small , the lower f gapless spectrum and the upper af gapful spectrum look coincidence , while the mid f branches are slightly shifted ; when @xmath76 becomes large , the lower f branch is far separated from the mid and upper excitation spectra for the pimsw results , while for the lmsw results , the excitation spectra are more dispersive , and the mid branch has two crossing points with the upper branch , showing that the lmsw and pimsw methods give different results when the af interaction is more dominant .
[ msw - aaf](c ) and ( d ) show the comparison between the tmrg and lmsw and pimsw results of the temperature dependence of the zero - field susceptibility for the af - af - f chain with different @xmath76 . when @xmath76 is large , the lmsw and pimsw results of @xmath77 are in agreement with the tmrg result , as shown in figs .
[ msw - aaf](d ) ; when @xmath76 is small , say , the f interaction takes predominant , the low - temperature behavior of @xmath77 calculated by the lmsw and pimsw theories coincides with the tmrg result , while the high - temperature behavior shows a somewhat difference , as shown figs .
[ msw - aaf](c ) . figs .
[ msw - aaf](e ) and ( f ) give the comparison of the lmsw and pimsw results with the tmrg result of the zero - field specific heat @xmath146 for the af - af - f chain with different @xmath76 .
it can be seen that the lmsw and pimsw theories can be applied to describe the specific heat of the af - af - f chain at lower temperature , where the pimsw result can fairly recover the first peak position of @xmath146 at lower temperature , but they can not describe well the high - temperature behavior where both msw results look only qualitatively similar to the tmrg result .
however , it can be stated that the double - peak structure of the specific heat is intimately related to the three branch excitations of the trimerized system , as manifested by the msw results . figs .
[ msw - ffa](a ) and ( b ) present the excitation spectra calculated by the lmsw and pimsw methods within the takahashi scheme for the f - f - af chain with @xmath147 and @xmath87 in absence of the magnetic field , respectively .
there are also three branch excitations , i.e. , one f gapless excitation spectrum , and two gapful excitation spectra , which are responsible for the thermodynamic behaviors of the system at low temperature .
when the af interaction is weak , the lmsw and pimsw theories give a similar result , as shown in fig .
[ msw - ffa](a ) ; when the af interaction becomes stronger , both methods produce quite different results particularly for the upper excitation branches , say , the two upper excitation spectra calculated from the pimsw method have two crossings , while those calculated by the lmsw method are separated , as indicated in fig .
[ msw - ffa](b ) , showing that the lmsw and pimsw methods should be cautiously applied when the f interaction of the system is weaker . figs .
msw - ffa(c ) and ( d ) show a comparison between the tmrg and msw results of the temperature dependence of the zero - field susceptibility for the f - f - af chain with @xmath147 and @xmath87 , respectively .
one may see that when @xmath104 is small , the low - temperature peak of @xmath77 can be fairly recovered , although there are somewhat differences ; when @xmath76 is large , say , the f interaction is weak , the position of low - temperature peak of @xmath77 can be clearly reproduced , although there are quantitative changes , indicating that the lmsw and pimsw methods could capture the main characteristics of the low - lying excitations of the present trimerized system at low temperature . figs .
[ msw - ffa](e ) and ( f ) give the comparison between the tmrg and msw results of the zero - field specific heat @xmath148 for the f - f - af chain with @xmath147 and @xmath87 , respectively .
it is seen that the double - peak structure of @xmath146 can also be obtained by the lsmsw and pimsw methods , which can produce the main features of the specific heat of the system at low temperature , but have large deviations at high temperature .
in this paper , we have numerically studied the thermodynamics of the @xmath22 af - af - f and f - f - af trimerized quantum heisenberg chains by means of the tmrg method and the msw theory .
the magnetization process , the low - lying excitations , the susceptibility and the specific heat of both systems have been explored .
it is found that the tmrg results of the temperature dependence of the susceptibility as well as the specific heat can be nicely fitted by a superposition of double two - level systems , and two fitting equations for the peak structures are proposed . for the af - af - f chain with strong f interactions ,
@xmath77 can be well fitted by a polynomial . for the af - af - f chain which is a ferrimagnet
, it is observed that the magnetic curve @xmath51 reveals some peculiar behaviors where a magnetization plateau at @xmath1 is observed at low temperature @xmath72 , while the plateau is smeared out when @xmath72 is increased , and the temperature dependence of the magnetization @xmath54 shows various behaviors in different ranges of the magnetic field .
the susceptibility @xmath77 exhibits different behaviors in different fields : at @xmath80 , @xmath77 goes to divergent at lower @xmath72 and decreases with increasing @xmath72 for @xmath79 ; at @xmath149 , @xmath150 shows a low - temperature peak for @xmath61 except @xmath151 and @xmath59 , and decreases at high @xmath72 , where the peak height and position of @xmath132 vary under different @xmath28 .
the behavior of @xmath78 against @xmath72 for a spin-@xmath97 af - af - f chain compound @xmath95 $ ] looks qualitatively similar to the tmrg result of a spin-@xmath0 af - af - f chain .
the specific heat @xmath146 is found to show a double - peak structure , with the peak height and position varying for different @xmath76 as well as underdifferent @xmath28 .
the results of @xmath54 , @xmath77 and @xmath146 show that the system under interest could be in several different states under different magnetic fields . for the f - f - af chain which is an antiferromagnet
, it is found that the magnetization process shows an interesting behavior , and a magnetization plateau at @xmath1 is also seen at low temperature which is smeared out at high temperature .
@xmath54 displays different behaviors in different regimes of the magnetic field , which is further confirmed by the tmrg results of the susceptibility @xmath77 where a peak is observed at low temperature , but the peak height and position are found dependent on different ranges of @xmath28 .
the experimental data of @xmath77 for the spin-@xmath0 f - f - af chain compound @xmath96 is compared with the corresponding tmrg result , which is found quite agreement .
the specific heat as a function of temperature shows a double - peak structure , whose shapes change with different @xmath104 as well as in different magnetic fields . the msw theory within the yamamoto scheme as well as the takahashi scheme is also applied to study the thermodynamics of the present trimerized quantum heisenberg af - af - f chain and f - f - af chain , respectively .
three branch excitations are found for both systems , say , one gapless excitation spectrum and two gapful excitation spectra , that come from the competition between the f and af interactions . a perfect fitting of @xmath77 and @xmath146 to a superposition of double two - level systems
could be reasonably understood within the framework of the msw theory .
it is three branch excitations with two gaps in the long wave - length limit that are responsible for the peak structures of the susceptibility and the specific heat observed by the tmrg calculations .
it is found that the msw theory could to some extent describe the low - temperature behavior of the systems , and gives only a qualitative description for @xmath77 and @xmath146 of af - af - f chain and @xmath77 of f - f - af chain at high temperature , but has a large deviation for @xmath146 of the f - f - af chain at high temperature .
finally , we would like to state that the present tmrg and msw results could be helpful for understanding the magnetization process as well as the thermodynamic properties of the af - af - f and f - f - af trimerized quantum spin chains . as there are few experimental data available for such systems at present ,
we expect that more experiments can be performed in near future to verify our theoretical findings .
the formalism of the msw theory for the af - af - f chain can be found in ref . @xcite . in this appendix ,
the msw theory for the f - f - af chain is formulated .
the hamiltonian of the 1d spin-@xmath0 f - f - af chain can be written as @xmath152where @xmath153 is the length of the f - f - af chain and @xmath154 , @xmath155 .
we start from the holstein - primakoff ( hp ) transformation @xmath156where @xmath22 , @xmath157 , @xmath86 , @xmath158 and @xmath159 , @xmath160 , @xmath161 .
the lmsw method treats the hp transformation up to o(@xmath162 ) , while the pimsw method treats is the hp transformation up to o(@xmath163 ) where the interactions between spin waves are handled perturbatively . in the conventional spin - wave scheme ,
the spin deviations in each sublattice , @xmath164 , diverge in the antiferromagnetic ground state but stay finite in the ferrimagnetic ground state . in the af system , the quantum as well as the thermal divergences of the number of bosons
are all suppressed : @xmath165 as the f - f - af is an antiferromagnet , the takahashi scheme@xcite will be invoked . in order to enforce the constraint ( [ staggt ] ) ,
a lagrange multiplier @xmath166 is introduced to the hamiltonian @xmath167the fourier transformations are defined by @xmath168where @xmath169 and @xmath170 .
the hamiltonian , up to the order of @xmath171 , takes the form of @xmath172where @xmath173 , @xmath174 , @xmath175@xmath176 , @xmath177@xmath178 , @xmath179 , @xmath180 , @xmath181 , and @xmath182 . via the bogoliubov transformation @xmath183where @xmath184 and @xmath185
, the hamiltonian can be diagonalized as @xmath186where @xmath187 for @xmath184 and @xmath188 for @xmath189 , and @xmath190 $ ] .
the coefficients of the bogoliubov transformation can be found through equations of motion @xmath191 $ ] : @xmath192for a given @xmath193 and @xmath166 , the eigenvalues @xmath194 and eigenvectors @xmath195 can be numerically calculated by the driver @xmath196 of the @xmath197 , which is available on the website@xcite . at finite temperature
, the lagrange multiplier @xmath198 is determined through the constraint ( [ staggt ] ) : @xmath199where @xmath169 , @xmath170 , and @xmath200 .
defining @xmath201 , the spin - wave distribution function , we have @xmath202^{-1}$ ] with @xmath187 for @xmath184 and @xmath188 for @xmath189 .
the internal energy and the magnetic susceptibility can be expressed as taka2 @xmath203where @xmath187 for @xmath184 and @xmath188 for @xmath189 .
@xmath194 , @xmath75 and @xmath40 within the lmsw framework can be thus calculated numerically , as shown in figs.[msw - ffa ] . for the pimsw method ,
the hp transformation is treated , up to the order of o(@xmath163 ) , such that @xmath204the interactions may be handled in the perturbational way : @xmath205define @xmath206 , @xmath207 and @xmath208 .
we obtain @xmath209and @xmath210where @xmath211 in ( [ up2s0 ] ) is the up - to-@xmath145 correction to @xmath194 from the term @xmath212 , and the other terms have similar contributions to @xmath194 . the only difference between lmsw and pimsw
is @xmath194 , say , the former is treated up to o(@xmath162 ) , while the latter up to o(@xmath163 ) .
accordingly , @xmath194 , @xmath75 and @xmath40 within the pimsw scheme can thus be calculated numerically , as shown in figs.msw-ffa .
the authors are grateful to h. f. mu , x. y. wang , q. b. yan and q. r. zheng for helpful discussion and assistance .
this work is supported in part by the national science foundation of china ( grant nos .
90403036 , 20490210 , 10247002 ) .
j. c. livermore , r. d. willett , r. m. gaura and c. p. landee , inorg .
* 21 * , 1403 ( 1982 ) ; y. ajiro , t. asano , t. inami , h. aruga - katori and t. goto , j. phys .
japan * 63 * , 859 ( 1994 ) ; t. toto , y. ajiro , t. asano , t. inami and h. aruga - katori , physica b * 216 * , 294 ( 1996 ) .
m. hagiwara , y. narumi , k. minami and k. kindo , physica b * 294 * , 30 ( 2001 ) ; m. hagiwara , y. narumi , k. minami , k. kindo , h. kitazawa , h. suzuki , n. tsujii and h. abe , j. phys .
japan * 72 * , 943 ( 2003 ) .
t. xiang and x. wang , _ density - matrix renormalization _ , lecture notes in physics , vol .
528 , edited by i. peschel , x. wang , m. kaulke and k. hallberg ( springer - verlag , new york , 1999 ) ; u. schollwock , rev .
mod . phys . *
77 * , 259 ( 2005 ) . | the magnetization process , the susceptibility and the specific heat of the spin-@xmath0 af - af - f and f - f - af trimerized quantum heisenberg chains have been investigated by means of the transfer matrix renormalization group ( tmrg ) technique as well as the modified spin - wave ( msw ) theory .
a magnetization plateau at @xmath1 for both trimerized chains is observed at low temperature .
the susceptibility and the specific heat show various behaviors for different ferromagnetic and antiferromagnetic interactions and in different magnetic fields .
the tmrg results of susceptibility and the specific heat can be nicely fitted by a linear superposition of double two - level systems , where two fitting equations are proposed .
three branch excitations , one gapless excitation and two gapful excitations , for both systems are found within the msw theory .
it is observed that the msw theory captures the main characteristics of the thermodynamic behaviors at low temperatures .
the tmrg results are also compared with the possible experimental data . |
multiplication in finite fields is a fundamental operation in arithmetic and finding efficient multiplication methods remains a topical issue .
let @xmath1 be a prime power , @xmath2 the finite field with @xmath1 elements and @xmath3 the degree @xmath4 extension of @xmath2 .
if @xmath5 is a basis of @xmath3 over @xmath6 then for @xmath7 and @xmath8 , we have the product @xmath9 where @xmath10 @xmath11 being some constants .
the complexity of a multiplication algorithm in @xmath3 depends on the number of multiplications and additions in @xmath6 .
there exist two types of multiplications in @xmath6 : the scalar multiplication and the bilinear multiplication .
the scalar multiplication is the multiplication by a constant ( in @xmath12 ) which does not depend on the elements of @xmath13 that are multiplied .
the bilinear multiplication is a multiplication of elements that depend on the elements of @xmath13 that are multiplied .
the bilinear complexity is independent of the chosen representation of the finite field .
for example , the direct calculation of @xmath14 using ( [ calculdirect ] ) requires @xmath15 non - scalar multiplication @xmath16 , @xmath17 scalar multiplications and @xmath18 additions .
more precisely , the multiplication of two elements of @xmath13 is an @xmath6-bilinear application from @xmath19 onto @xmath13
. then it can be considered as an @xmath6-linear application from the tensor product @xmath20 onto @xmath13 .
consequently , it can also be considered as an element @xmath21 of @xmath22 where @xmath23 denotes the dual .
when @xmath21 is written @xmath24 where the @xmath25 elements @xmath26 as well as the @xmath25 elements @xmath27 are in the dual @xmath28 of @xmath13 while the @xmath25 elements @xmath29 are in @xmath13 , the following holds for any @xmath30 : @xmath31 the decomposition ( [ tensor ] ) is not unique .
every expression @xmath32 defines a bilinear multiplication algorithm @xmath33 of bilinear complexity @xmath34 .
the minimal number of summands in a decomposition of the tensor @xmath21 of the multiplication is called the bilinear complexity of the multiplication and is denoted by @xmath35 : @xmath36 where @xmath33 is running over all bilinear multiplication algorithms in @xmath13 over @xmath6 . the bilinear complexity of the multiplication in @xmath3 over @xmath2 has been widely studied . in particular , it was proved in @xcite that it is uniformly linear with respect to the degree @xmath4 of the extension .
it follows from a clever algorithm performing multiplication : the so - called multiplication algorithm of chudnovsky and chudnovsky .
the original chudnovsky - chudnovsky algorithm was introduced in 1987 by d.v . and
chudnovsky @xcite and is based on the interpolation on some algebraic curves . from now on
, we will denote this algorithm by ccma .
there is benefit having a low bilinear complexity when considering hardware implementations mainly because it reduces the number of gates in the circuit .
in fact , in the so - called non - scalar model ( denoted ns ) , only the bilinear complexity is taken into account and it is assumed that all scalar operations are free . indeed , this model does not reflect the reality and since the bilinear complexity is not the whole complexity of the algorithm , the complexity of the linear part of the algorithm should also be taken into account . in this paper , we consider two other models .
the model s1 , which takes into account the number of multiplications without distinguishing between the bilinear ones and the scalar ones .
the model s2 which takes into account all operations ( multiplications and additions ) in @xmath6 .
notice that so far , practical implementations of multiplication algorithms over finite fields have failed to simultaneously optimize the number of scalar multiplications , additions and bilinear multiplications .
regarding exponentiation algorithms , the use of a normal basis is of interest because the @xmath37 power of an element is just a cyclic shift of its coordinates .
a remaining question is , how to implement multiplication efficiently in order to have simultaneously fast multiplication and fast exponentiation . in 2000 , gao et al .
@xcite show that fast multiplication methods can be adapted to normal bases constructed with gauss periods .
they show that if @xmath13 is represented by a normal basis over @xmath6 generated by a gauss period of type @xmath38 , the multiplication in @xmath13 can be computed with @xmath39 and the exponentiation with @xmath40 operations in @xmath6 ( @xmath1 being small ) .
this result is valuable when @xmath41 is bounded .
however , in the general case @xmath41 is upper - bounded by @xmath42 . in 2009 ,
couveignes and lercier construct in ( * ? ? ?
* theorem 4 ) two families of basis ( called elliptic and normal elliptic ) for finite field extensions from which they obtain a model @xmath43 defined as follows .
to every couple @xmath44 , they associate a model , @xmath45 , of the degree @xmath4 extension of @xmath6 such that the following holds : there is a positive constant @xmath46 such that the following are true : @xmath47 elements in @xmath13 are represented by vectors for which the number of components in @xmath6 is upper bounded by @xmath48 @xmath47 there exists an algorithm that multiplies two elements at the expense of @xmath49 multiplications in @xmath6 .
@xmath47 exponentiation by @xmath1 consists in a circular shift of the coordinates .
therefore , for each extension of finite field , they show that there exists a model which allows both fast multiplication and fast application of the frobenius automorphism .
their model has the advantage of existing for all extensions .
however , the bilinear complexity of their algorithm is not competitive compared with the best known methods , as pointed out in ( * ? ? ?
* section 4.3.4 ) .
indeed , it is clear that such a model requires at least @xmath50 bilinear multiplications .
note that throughout the paper , efficiency of algorithms is described in terms of parallel time ( depth of the circuit , in number of multiplications ) , number of processors ( width ) and total number of multiplications ( size ) .
we have width @xmath51 size @xmath51 depth@xmath52width .
we propose another model with the following characteristics : - our model is based on ccma method , thus the multiplication algorithm has a bilinear complexity in @xmath53 , which is optimal .
- our model is tailored to parallel computation .
hence , the computation time used to perform a multiplication or any exponentiation can easily be reduced with an adequate number of processors . since our method has a bilinear complexity of multiplication in @xmath53 , it can be parallelized to obtain a constant time complexity using @xmath54 processors . the previous aforementioned works ( @xcite and @xcite ) do not give any parallel algorithm ( such an algorithm is more difficult to conceive than a serial one ) . - exponentiation by @xmath1 is a circular shift of the coordinates and can be considered free .
thus , efficient parallelization can be done when doing exponentiation .
- the scalar complexity of our exponentiation algorithm is reduced compare to a basic exponentiation using ccma algorithm thanks to a suitable basis representation of the riemann - roch space @xmath55 in the second evaluation map .
more precisely , the normal basis representation of the residue class field is carried in the associated riemann - roch space @xmath56 , and the exponentiation by @xmath1 consists in a circular shift of the @xmath4 first coordinates of the vectors lying in the riemann - roch space @xmath55 .
- our model uses coppersmith - winograd @xcite method ( denoted cw ) or any variants thereof to improve matrix products and to diminish the number of scalar operations .
this improvement is particularly efficient for exponentiation . in term of complexity
, we can state the following results , depending on the chosen model ( ns , s1 and s2 ) .
[ lns ] in the non - scalar model ns , there exist multiplication and exponentiation algorithms in @xmath13 such that : * multiplication is done in parallel time in @xmath57 multiplications in @xmath6 with @xmath54 processors , for a total in @xmath54 multiplications . *
exponentiation is done in parallel time in @xmath58 multiplications in @xmath6 with @xmath59 processors , for a total in @xmath60 multiplications .
when considering models s1 and s2 , two cases can be distinguished for the multiplication complexity . we might be interested either by the complexity of one multiplication or by the average ( amortized ) complexity of one multiplication when many multiplications are done simultaneously .
regarding exponentiation , a wise use of cw method allows the complexity to be improved .
we can state the followings : [ ls1 ] in the model s1 , there exist multiplication and exponentiation algorithms in @xmath13 such that : * multiplication : * * one multiplication is done in parallel time in @xmath57 multiplications in @xmath6 with @xmath61 processors , for a total in @xmath61 multiplications ; * * in the amortized sense , the parallel time is in @xmath57 multiplications in @xmath6 with @xmath62 processors , for a total in @xmath62 multiplications where the value of @xmath63 is approximately @xmath64 for the best known matrix product methods ; * exponentiation is done in a parallel time of @xmath58 multiplications in @xmath6 with @xmath65 processors , for a total in @xmath66 multiplications .
[ ls2 ] in the model s2 , there exist multiplication and exponentiation algorithms in @xmath13 such that : * multiplication : * * one multiplication is done in parallel time in @xmath58 operations in @xmath6 with @xmath60 processors , for a total in @xmath61 operations ; * * in the amortized sense , the parallel time is in @xmath58 operations in @xmath6 with @xmath67 processors , for a total in @xmath62 operations ; recall that the value of @xmath63 is approximately @xmath64 for the best matrix product methods ; * exponentiation is done in a parallel time of @xmath68 operations in @xmath6 with @xmath69 processors , for a total in @xmath66 operations . after some background on ccma algorithm
, we describe in subsection [ construction ] our method which leads to an effective algorithm that can directly be implemented .
our algorithm reveals the use of matrix - vector products that can easily be parallelized . in section [ secexp ]
, we use this algorithm to tackle the problem of computing @xmath70 where @xmath71 and @xmath72 and we derive an exponentiation algorithm from the work of von zur gathen @xcite . in section
[ effective ] , we focus on the multiplication in @xmath73 and we explain how to construct our algorithm . a magma @xcite implementation of the multiplication algorithm in @xmath74 is given in appendix .
first , we present the ccma algorithm on which is based our method .
let @xmath75 be an algebraic function field over the finite field @xmath2 of genus @xmath76 .
we denote by @xmath77 the number of places of degree one of @xmath78 over @xmath2 .
if @xmath79 is a divisor , @xmath80 denotes the riemann - roch space associated to @xmath79 .
we denote by @xmath81 the valuation ring of the place @xmath82 and by @xmath83 its residue class field @xmath84 which is isomorphic to @xmath85 where @xmath86 is the degree of the place @xmath82 . the following theorem that makes effective the original algorithm groups some results of @xcite .
[ mainth ] let @xmath87 be an algebraic function field of genus @xmath76 defined over @xmath6 and @xmath4 an integer .
let us suppose that there exists a place @xmath82 of degree @xmath4 . then , if @xmath88 there is an effective divisor @xmath79 of degree @xmath89 such that : 1 .
@xmath82 is not in the support of @xmath79 , 2 .
the evaluation map @xmath90 defined by @xmath91 is an isomorphism of vector spaces over @xmath2 , 3 .
there exist @xmath92 places of degree one @xmath93 which are not in the support of @xmath79 such that the multi - evaluation map @xmath21 defined by @xmath94 is an isomorphism .
the chosen framework is the original ccma algorithm , namely using only places of degree one and without derivated evaluation ( cf .
we transform this algorithm in order that it be adapted to both multiplication and exponentiation computations . in this context , the construction of this algorithm is based on the choice of the place @xmath82 of degree @xmath4 , the effective divisor @xmath79 ( cf .
@xcite ) and the bases @xmath95 and @xmath96 .
recall some notions on normal bases .
the finite field @xmath3 will be considered as a vector space of dimension @xmath4 over the finite field @xmath2 .
let @xmath97 be an element of @xmath3 such that @xmath98 is a basis of @xmath3 over @xmath2 .
such a basis is called a normal basis of @xmath3 over @xmath2 and @xmath97 is called a cyclic element .
thus , a normal basis is composed of all conjugates of a cyclic element @xmath97 .
there is always a normal basis and furthermore , there is always a primitive normal basis .
we call a normal polynomial of degree @xmath4 over @xmath2 , a polynomial in @xmath99 $ ] , irreducible over @xmath2 , and having for roots in @xmath3 the @xmath4 conjugates of a cyclic element @xmath97 .
we refer to @xcite and @xcite for a detailed presentation .
when @xmath13 is represented by a normal basis , the @xmath1th power of an element is just a cyclic shift of its coordinates .
the repeated use of this operation allows exponentiation to be efficiently parallelized . without normal basis @xcite ,
precomputation should be stored for a same base @xmath100 .
this makes sense only when many exponentiation have to be done with this same base and in this case , precomputations are not considered in the running time .
the use of a normal basis has the following benefits : * substitute lookup table accesses by circular shifts . *
reduce prior storage . *
avoid the constraint of fixing a base .
the construction of the algorithm is based on the choice of the place @xmath82 of degree @xmath4 , the effective divisor @xmath79 of degree @xmath89 ( cf .
@xcite ) , the bases of spaces @xmath101 and @xmath55 and the basis of the residue class field @xmath102 of the place @xmath82 . in practice , as in @xcite , we take as a divisor @xmath79 one place of degree @xmath89 .
it has the advantage to solve the problem of the support of divisor @xmath79 ( condition ( 1 ) of theorem [ mainth ] ) as well as the problem of the effectivity of the divisor @xmath79 .
furthermore , we require additional properties . to build the good places , we draw them at random and we check that they satisfy the required conditions namely : 1 .
we draw at random an irreducible polynomial @xmath103 of degree @xmath4 in @xmath104 $ ] and check that this polynomial is : 1 .
2 . normal .
3 . totally decomposed in the algebraic function field @xmath87 ( which implies that there exists a place @xmath82 of degree n above the polynomial @xmath103 ) .
we choose a place @xmath82 of degree n among the @xmath4 places lying above the polynomial @xmath103 .
we draw at random a place @xmath79 of degree @xmath89 and check that @xmath105 is a non - special divisor of degree @xmath106 , _ i.e. _ @xmath107 . in practice
, it is easy to find @xmath82 and @xmath79 satisfying these properties in our context since there exist many such places .
however , it is not true in the general case ( it is sufficient to consider an elliptic curve with only one rational point ) . a sufficient condition for the existence of at least one place of degree @xmath4 is given by the following inequality : @xmath108 when @xmath109 , we are sure of the existence of a non - special divisor of degree @xmath106 @xcite .
the larger @xmath1 , the larger the probability to draw a non - special divisor of degree @xmath106 becomes ( proposition 5.1 @xcite ) but not necessarily as a difference of two places ( this is an open problem ) .
however , in practice , such divisors are easily found . when we take a place @xmath82 of degree @xmath4 lying above a normal polynomial in @xmath104 $ ] , we mean that the residue class field is the finite field @xmath13 for which we choose as a representation basis the normal basis @xmath110 generated by a root @xmath97 of the polynomial @xmath103 .
[ changebasis ] suppose that the context requires the use of a representation basis of @xmath13 which is not the basis @xmath111 of the residue class field of the place @xmath82 .
then , we can easily avoid the problem by a change of basis .
this requirement may happen when additional properties on the basis @xmath111 ( cf . section [ chudnb ] ) are required .
in particular , it would be the case in our context if we could not find a place @xmath82 of degree @xmath4 above a normal polynomial @xmath112 $ ] .
as the residue class field @xmath102 of the place @xmath82 is isomorphic to the finite field @xmath13 , from now on we identify @xmath13 to @xmath102 .
notice that the choice of @xmath79 and @xmath82 of section [ si ] are such that the map @xmath90 of theorem [ mainth ] is an isomorphism .
indeed , @xmath114 , @xmath115 yet @xmath116 in particular , we choose for basis of @xmath80 , the reciprocal image @xmath117 of the basis @xmath118 of @xmath102 by the evaluation map @xmath90 , namely @xmath119 .
note that as the divisor @xmath79 is an effective divisor , we have @xmath121 .
let @xmath122 be the map from @xmath55 to @xmath55 defined in the following way : if @xmath123 then @xmath124 is in the residue field @xmath102 of the place @xmath82 ; define @xmath125 where @xmath126 is the injection map from @xmath80 into @xmath55 .
then @xmath122 is a linear map from @xmath55 into @xmath55 whose image is @xmath80 .
more precisely , @xmath122 is a projection from @xmath55 onto @xmath80 .
let @xmath127 be the kernel of @xmath122
. then @xmath128 [ rem_proj ] from the definition of @xmath122 we remark that 1 .
@xmath129 , 2 . for any @xmath123
, we have @xmath130 . as @xmath131 ,
the divisor @xmath132 is non - special and @xmath133 hence , we define as basis of @xmath55 , the basis @xmath134 defined by : @xmath135 where @xmath136 is a basis of @xmath127 and @xmath137 is the basis of @xmath80 defined in section [ baseld ] .
[ mainrem ] as a consequence of the choice of the basis @xmath138 for @xmath55 , if @xmath139 then @xmath140 in this section , we use as representation basis of spaces @xmath102 , @xmath80 , @xmath55 , the basis defined in section [ choixbases ] .
the product of two elements in @xmath3 is computed by the algorithm of chudnovsky and chudnovsky .
let @xmath141 and @xmath142 be two elements of @xmath3 given by their components over @xmath143 relative to the chosen basis @xmath111 . according to the previous notation
, we can consider that @xmath100 and @xmath144 are identified to the following elements of @xmath80 : @xmath145 the product @xmath146 of the two elements @xmath147 and @xmath148 of @xmath80 is their product in the valuation ring @xmath81 .
this product lies in @xmath55 .
we will consider that @xmath100 and @xmath144 are respectively the elements @xmath147 and @xmath148 of @xmath55 where the @xmath89 last components are @xmath149 .
now it is clear that knowing @xmath100 or @xmath147 by their coordinates is the same thing .
let us consider the following hadamard product in @xmath150 : @xmath151 the product of @xmath100 by @xmath144 is such that @xmath152 indeed , from the definition of @xmath21 the following holds @xmath153 then @xmath154 by remark [ rem_proj ] we conclude @xmath155 we can now present the setup algorithm and the multiplication algorithm .
note that the setup algorithm is only done once .
@xmath157 1 .
the elements @xmath100 of the field @xmath3 are known by their components relatively to a fixed basis : @xmath141 ( where @xmath158 ) .
2 . the function field @xmath159 , the place @xmath82 , the divisor @xmath79 and the points @xmath160 are as in theorem [ mainth ] .
3 . construct a basis @xmath161 of @xmath55 where @xmath162 is the basis of @xmath80 defined in section [ baseld ] and @xmath163 a basis of @xmath127 .
any element @xmath164 in @xmath3 is identified to the element @xmath165 of @xmath80 .
5 . compute the matrices @xmath21 and @xmath166 .
@xmath168 1 . compute @xmath169 2 .
compute @xmath170 where @xmath171 .
3 . compute @xmath172 .
4 . return@xmath173 ( remark that in the previous step we just have to compute the @xmath4 first components of @xmath174 ) . in terms of number of multiplications in @xmath6 , the complexity of this multiplication algorithm is as follows : calculation of @xmath175 and @xmath176 needs @xmath177 multiplications , calculation of @xmath178 needs @xmath92 multiplications and calculation of @xmath174 needs @xmath179 multiplications .
the total complexity is bounded by @xmath180 .
the asymptotic analysis of our method needs to consider infinite families of algebraic function fields defined over @xmath6 with increasing genus ( or equivalently of algebraic curves ) having the required properties .
the existence of such families follows from that of recursive towers of algebraic function fields of type garcia - stichtenoth @xcite reaching the drinfeld - vladut bound .
note that , because of the drinfeld - vladut bound , in the case of small basis fields @xmath181 respectively @xmath182 with @xmath183 , we select the method of the quartic respectively quadratic embedding .
in these cases , instead of the embeddings , it would be possible to use places of degree four and respectively two ( cf .
@xcite and @xcite ) but it requires to generalize our algorithm and it generates significant complications ( cf .
@xcite , @xcite ) .
note also that , in the case of the quartic respectively quadratic embedding of the small fields , the operations are precomputed and do not increase the complexity
. then it is proved @xcite from a specialization of the original chudnovsky algorithm on these families that the bilinear complexity of the multiplication in any degree @xmath4 extension of @xmath6 is uniformly linear in @xmath1 with respect to @xmath4 .
hence , the number of bilinear multiplications is in @xmath53 and the genus @xmath184 of the required curves also necessarily increases in @xmath53 .
consequently , the total number of multiplications / additions / subtractions of our method is in @xmath61 and the total number of bilinear multiplications is in @xmath54 . on some occasions in the paper
, the cw algorithm @xcite will be used to decrease the number of scalar operations . given two square matrices of size @xmath4 , the product can be computed in @xmath185 multiplications where @xmath186 .
in fact , if we consider a parallel version of the algorithm in the model s1 , a product can be performed in @xmath57 multiplications in @xmath6 using @xmath185 processors .
this is a consequence of strassen s normal form theorem @xcite . in the model s2 where scalar multiplications and additions / subtractions have same costs
, the depth becomes @xmath58 and a rescheduling technique @xcite allows the reduction of the width in @xmath187 .
now , we focus on the parallel complexity of our multiplication algorithm .
this actually consists of determining the parallel complexity of a constant number of matrix - vector products and the product coordinate - wise of two vectors .
first , let us consider the ns model in which a round represents the time interval for a processor to perform one bilinear multiplication in @xmath182 ( scalar operations are considered as free ) , a multiplication can be carry out in a constant number of rounds with only @xmath54 processors , as stated in theorem [ lns ] . regarding the two other models , two cases have to be considered , the non amortized case and the amortized one . in the model s1 ( additions and subtractions in @xmath182
are considered as free ) , if a round represents the time interval for a processor to perform one multiplication , the product can be performed in a constant number of rounds with @xmath61 processors , as stated in theorem [ ls1]a .
this width corresponds to the asymptotic number of scalar multiplications . in the model s2 where we take into account all scalar operations in @xmath182 ,
a multiplication in @xmath188 can be performed in parallel time @xmath58 operations in @xmath182 using @xmath60 processors , as stated in theorem [ ls2]a . if we have @xmath189 multiplications in @xmath188 to perform , the width can be decreased ( in an amortized sense ) by using an optimized matrix multiplication method .
it consists in grouping the operands in order that they become the columns of square matrices allowing the use of an efficient method like the cw one .
more precisely , the method is as follows : 1 .
store the @xmath190 operands ( vectors ) as columns in square matrices of size @xmath92 , denoted @xmath191 ; 2 .
compute the products @xmath192 where @xmath193 represent the columns of @xmath194 , by using the coppersmith - winograd algorithm ( or another efficient method ) ; 3 .
perform the @xmath195 bilinear products @xmath196 ; 4 .
store the @xmath195 results as columns in square matrices , denoted @xmath197 ; 5 .
compute the products @xmath198 with the cw method .
the results are then stored in the columns of the resulting matrices . in the model s1 , the necessary width to perform step 3 is in @xmath199 processors .
the overall width is in fact dominated by steps 2 and 5 .
all these products are performed in constant time using @xmath200 processors .
this represents , in an amortized sense , @xmath201 processors per multiplication , as stated in theorem [ ls1]b . in the model s2 ,
a multiplication in @xmath188 can be performed in parallel time @xmath58 using ( thanks to the cw matrix product ) @xmath202 processors , as stated in theorem [ ls2]b .
this last result is obtained using a rescheduling technique @xcite which allows the parallel computation of the matrix product to be done in @xmath58 operations in @xmath182 using @xmath203 processors , instead of @xmath204 processors without rescheduling @xcite .
the previous algorithm can be iterated in order to obtain the product of three elements @xmath205 in @xmath80 ( or equivalently in @xmath3 ) .
@xmath207 1 . compute @xmath208 2 .
compute @xmath209 , 3 . then compute @xmath210 ( this is the hadamard product in @xmath150 ) , 4 . and finally the result is @xmath211 . in terms of number of multiplications in @xmath6 , the complexity of this algorithm
is as follows : the matrix @xmath212 can be precomputed and the matrix - vector product needs @xmath213 multiplications .
the total complexity is then @xmath214 including @xmath215 bilinear multiplications ( this number is almost doubled compare to the preceding algorithm ) .
note that the precomputation of @xmath216 is of interest for the parallel computations .
asymptotically , the complexity is the same as in the previous case .
this algorithm will be used to construct our exponentiation algorithms in section [ secexp ] .
the form of the involved matrices is analysed in the next subsection . obviously , the crucial part of the algorithm consuming more time is the iterative call to @xmath216 , namely the iterative call to the composition @xmath217 .
the matrix @xmath21 is @xmath218 the matrix @xmath122 is @xmath219 where @xmath220 is the unit matrix of size @xmath221 .
let us define the following blocks : @xmath222 where @xmath223 is a @xmath221-matrix , @xmath224 is a @xmath225-matrix , @xmath226 is a @xmath227-matrix and @xmath228 is a @xmath229-matrix . in the same way ,
we can write @xmath166 in the following form : @xmath230 then @xmath231 moreover ,
the following relations hold : @xmath232 @xmath233 @xmath234 @xmath235 an important problem , which is out of the scope of this paper , is to choose a basis @xmath236 of @xmath55 and places of degree one @xmath93 in order to obtain a `` simple '' matrix @xmath216 .
indeed , a sparse matrix may reduce the number of multiplications . having a matrix @xmath21 ( or @xmath216 )
, the product @xmath237 for all possible @xmath238 can be efficiently precomputed .
we choose an integer @xmath239 dividing @xmath4 and we set @xmath240 . for all @xmath241 such that @xmath242 ,
we denote by @xmath243 a variable of the form @xmath244 , where : @xmath245 thus , we have @xmath246 .
we can store in a table @xmath247 $ ] the products @xmath248 for the @xmath249 possible values of @xmath243 . after having stored all the precomputations in tables @xmath250)_{i=0 \ldots k-1}$ ]
, we can compute @xmath251 by evaluating @xmath252 + tab^t_1[y_{l,1 } ] + \cdots + tab^t_{k-1}[y_{l , k-1}]$ ] . as an example , for @xmath253 we can choose @xmath254 , leading to the storage of @xmath255 tables of @xmath256 values plus one table of @xmath257 values .
in this section , our aim is to point out the interest , in terms of parallel running time and number of processors involved , of computing an exponentiation in finite fields ( _ i.e. _ @xmath70 in @xmath3 with @xmath258 ) based on our model .
first , it is natural to consider the well known square and multiply algorithm ( with say , the method `` right - to - left '' ) .
we describe this basic algorithm , showing the use of matrices @xmath21 , @xmath122 and @xmath216 . in a second time
, we consider a more advanced algorithm , based on an idea from von zur gathen @xcite that we embed in our model .
we show that this algorithm reaches optimal depth in terms of operations in @xmath6 .
let @xmath46 be the unidimensional array of length @xmath259 containing the bits of @xmath41 .
this array will be indexed by @xmath260 .
more precisely @xmath2612^i.\ ] ] in order to bind operations we must iterate the use of the operator @xmath216 .
we obtain algorithm [ sam_algo ] : @xmath262 @xmath70 @xmath263 @xmath264 @xmath265 @xmath266 @xmath267 * return * @xmath268 within a same loop turn , operations under the condition `` if '' are not used in the subsequent operations .
we consider two sets of processing units @xmath269 and @xmath270 , the set @xmath270 running the operations under the condition `` if '' while the set @xmath269 deals with the other calculations .
we assume that the amounts of resources of @xmath269 and @xmath270 are the same . as an example , let us describe the steps in the calculation of @xmath271 .
figure [ exponentiation_wo_precomput ] depicts the operations made in parallel .
we remark that at each step , the set @xmath269 or @xmath270 ( or both ) perform(s ) a vector product in @xmath272 and subsequently a matrix - vector product , except for the first and the last step for which only a matrix - vector product has to be performed . at step 0
, the set @xmath269 ( or @xmath270 ) computes @xmath273 .
the set @xmath269 computes @xmath274 at step @xmath275 . for its part , the set @xmath270 computes @xmath276 at step 2 , @xmath277 at step 3 and @xmath278 at step 4 .
a last step is needed to retrieve the result @xmath279 .
then we can remark that three products in @xmath272 are made in parallel , to which must be added the product performed by @xmath269 at the step 1 , for an overall computation time of four products in @xmath272 .
we now consider operations over @xmath6 and the amount of processing units used . in the model ns ,
the computation time is in @xmath54 with only @xmath54 processors . in the model s1 ,
the computation time is in @xmath54 with @xmath61 processors and in the model s2 , the computation time is in @xmath280 with @xmath60 processors . with the use of a normal basis
, the base @xmath100 does not have to be fixed anymore .
moreover , the number of multiplications in @xmath13 is reduced and it becomes possible to obtain an `` overall '' parallel time in @xmath58 without prior storage . to make use of normal bases ,
let us now consider the @xmath1-ary representation @xmath46 of the exponent ( composed of @xmath4 terms according to fermat s little theorem ) by writing @xmath281q^i=\sum_{i=0}^{n-1 } \sum_{j=0}^{t } k[i , j]2^jq^i,\ ] ] where @xmath282 $ ] for @xmath283 are the bits of @xmath284 $ ] .
let @xmath285 be the function such that @xmath286 right shifts @xmath241 times the vector @xmath100 .
then , we can express @xmath70 as : @xmath287q^i } & = & \prod_{i=0}^{n-1}\sigma(x , i)^{k[i ] } & = & \prod_{i=0}^{n-1}\prod_{j=0}^{t}\sigma(x , i)^{k[i , j]2^j}\\ & = & \prod_{i=0}^{n-1}\sigma(x^{k[i]},i ) & = & \prod_{i=0}^{n-1}\sigma\left(\prod_{j=0}^{t}x^{k[i , j]2^j},i\right).\end{aligned}\ ] ] this gives two ways of rewriting the square and multiply algorithm .
in fact , since we are no longer limited to the use of two ( sets of ) processors , there exist more efficient algorithms , as shown by von zur gathen @xcite .
the idea is that short patterns might occur repeatedly in the @xmath288ary representation of the exponent @xmath41 .
hence , precomputation of all patterns of a given short length @xmath289 allows the overall cost to be lower . by setting @xmath290 and writing @xmath291 with @xmath292 for all @xmath241
, von zur gathen obtains optimal depth for an appropriate choice of @xmath25 .
in particular it is shown that this result is reachable with width in @xmath293 processors .
lee et al .
@xcite introduced a rescheduling technique to reduce the number of processors at the counterpart of near - optimal depth . in
what follows , we describe our own variant of von zur gathen algorithm , achieving the same asymptotical efficiencies than the one from lee et al . , while being simpler .
we set @xmath294 and @xmath295 , and rewrite the exponent in the following form : @xmath296 thus , @xmath290 is in @xmath297 and @xmath298 is in @xmath58 .
the algorithm is divided into five steps : 1 . for @xmath299 ,
compute @xmath300 , 2 . for @xmath301 and @xmath302 , compute @xmath303 , 3 . for @xmath301 ,
compute @xmath304 , 4 . for @xmath301 ,
compute @xmath305 , 5 .
return @xmath306 .
we first examine the parallel time complexities in terms of multiplication in @xmath188 .
we can use a binary tree of multiplications ( executed from root to leaves ) to compute step 1 in @xmath307 multiplications using @xmath308 processors .
step 2 is free . in step 3 , each @xmath309 for @xmath301 can be computed by distinct processors in @xmath310 multiplications .
step 4 is free .
step 5 can be computed with a binary multiplication tree ( executed from leaves to root ) in @xmath311 multiplications using @xmath312 processors . by summing the times of each step
, the overall depth can be upper bounded by @xmath313 for sufficiently large @xmath4 .
moreover , by using an optimisation from von zur gathen , the parallel execution time of step 1 can be reduced to approximately @xmath314 .
thus , the overall depth is in @xmath58 multiplications .
it can be noticed that at each parallel step , the number of processors involved stays in @xmath297 .
consequently , this overall depth is achieved with a width in @xmath297 processors . from now on , we scale up the number of processors to optimize the running time in terms of operations in @xmath6 .
when considering this algorithm in our chudnovsky model , we consider @xmath297 sets of @xmath54 processors if the scalar multiplications in @xmath182 are considered free .
otherwise , we consider sets of @xmath61 processors .
@xmath315 @xmath316 for @xmath317 @xmath318 @xmath319 @xmath320 @xmath320 * return * @xmath321 in terms of bilinear multiplications in @xmath12 , step 1 , described in algorithm [ precomputevzg ] , is performed in depth @xmath58 and width @xmath322 .
step 2 and 4 consist in the shift of the first @xmath4 coordinates and are thus considered free .
step 3 , the details of which are left to the reader , is performed in depth @xmath58 and width @xmath322 .
the last step , which consists in computing @xmath323 , is a binary multiplication tree .
it is executed for a cost of @xmath58 bilinear multiplications using @xmath322 processors .
overall , in the model ns , this algorithm is done in depth @xmath58 , width @xmath322 , and size @xmath60 , as stated in theorem [ lns ] .
we let the reader deduce that , @xmath241 ) in the model s1 , in terms of multiplications in @xmath182 , this algorithm is done in depth @xmath58 , width @xmath324 , and size @xmath325 ; @xmath326 ) in the model s2 , in terms of any operations in @xmath182 , this algorithm is done in depth @xmath68 , width @xmath327 , and size @xmath325 .
+ [ [ reducing - the - number - of - scalar - operations . ] ] * reducing the number of scalar operations . *
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we have seen that in the scalar model , the high number of processors is due to the high number of scalar operations ( in particular the matrix - vector products @xmath237 , @xmath328 and @xmath329 ) . at each step of the computation ,
the number of multiplications in @xmath188 done in parallel is in @xmath297 in the worst case .
thus , at a step of the computation involving multiple parallel matrix - vector products , instead of performing @xmath297 separate matrix - vector products , we write the @xmath297 vectors as columns of a matrix @xmath330 , then complete this matrix with zero columns in order to obtain a square matrix .
now , we have a product of two square matrices that can be performed using the coppersmith - winograd method , thus reducing the number of scalar operations . in the s1 model
, the coppersmith - winograd product can be performed in @xmath57 multiplications in @xmath6 using @xmath204 processors .
this optimization allows the exponentiation to be done in depth @xmath58 , width @xmath204 and size @xmath331 . in the model s2
, the coppersmith - winograd product can be performed in @xmath58 operations in @xmath6 using @xmath203 processors .
the optimization allows the exponentiation to be done in depth @xmath68 , width @xmath203 and size @xmath331 .
remark that the number of zero columns in the matrix @xmath330 may not be negligible .
thus , instead of filling @xmath330 with zero columns in order to obtain a square matrix , we can slightly improve the complexity by using the cw method in the following way : consider a matrix @xmath330 containing only @xmath332 vectors .
let us denote by @xmath333 the number of columns of @xmath330 ( @xmath333 is then in @xmath332 ) .
we partition @xmath330 in square submatrices of size @xmath333 so that @xmath330 is seen as a column @xmath334 of blocks ( with @xmath335 in @xmath68 ) . in the same way , we partition @xmath216 in square submatrices of size @xmath333 so that a row @xmath241 of blocks of @xmath216 is represented as @xmath336 .
the sum of products @xmath337 corresponds to the @xmath241-th block of the resulting column of blocks .
a product @xmath338 is done using cw in @xmath339 scalar multiplications .
since we have @xmath340 such products to compute @xmath341 , the overall number of scalar multiplications is in @xmath342 .
consequently , in the model s1 the product @xmath341 can be computed in @xmath57 multiplications in @xmath182 using @xmath342 processors , whereas in the model s2 , it can be computed in @xmath58 operations in @xmath182 using @xmath343 processors .
we can substitute these last results in the case of the parallel exponentiation to obtain the stated complexities of theorem [ ls1 ] and theorem [ ls2 ] , with the current exponent for the best optimized matrix product : in the model s1 , @xmath70 can be computed in @xmath58 multiplications in @xmath182 using @xmath342 processors for a size of @xmath344 multiplications in @xmath182 , whereas in the model s2 , @xmath70 can be computed in @xmath68 operations in @xmath182 using @xmath343 processors for a size of @xmath344 operations in @xmath182 .
set @xmath345 and @xmath346 . from now on
, @xmath87 denotes the algebraic function field associated to the hyper elliptic curve @xmath347 with plane model @xmath348 , of genus two .
this curve has 33 rational points , which is maximal over @xmath6 according to the hasse - weil bound .
we represent @xmath349 as the field @xmath350/(p(x))$ ] where @xmath351 is the irreducible polynomial @xmath352 and @xmath353 denotes a primitive root of @xmath352 .
let us give the projective coordinates @xmath354 of rational points of the curve @xmath347 : @xmath355 it is sufficient to take a place @xmath356 of degree @xmath4 in the rational function field @xmath357 , which totally splits in @xmath87 .
it is equivalent to choose a monic irreducible polynomial @xmath358 $ ] of degree
n such that its roots @xmath359 in @xmath13 satisfy @xmath360 for @xmath361 where the map @xmath362 denotes the classical function trace over @xmath363 by ( * ? ? ?
* theorem 2.25 ) .
in fact , it is sufficient to verify that this property is satisfied for only one root since a finite field is galois .
moreover , in the context of our method , we require that this irreducible polynomial @xmath103 corresponds to a normal polynomial ( cf .
section [ basefq ] ) .
for example , for the extension @xmath364 , we choose the primitive normal polynomial @xmath365 let @xmath366 be one primitive root of @xmath103 .
it is easy to check that @xmath367 , hence the place @xmath368 of @xmath369 is totally splitted in the algebraic function field @xmath87 , which means that there exist two places of degree @xmath4 in @xmath87 lying over the place @xmath368 of @xmath369 . for the place @xmath82 of degree @xmath4 in the algebraic function field @xmath87
, we consider one of the two places in @xmath87 lying over the place @xmath368 of @xmath369 , namely the orbit of the @xmath0-rational point @xmath370 where @xmath359 is a root of @xmath103 and @xmath371 for @xmath372 .
notice that the second place is given by the conjugated points @xmath373 for @xmath372 . for the divisor @xmath79 of degree @xmath89 , we choose a place @xmath79 of degree 14 according to the method used for the place @xmath82 .
we consider the orbit of the @xmath374-rational point @xmath375 where @xmath376 is a root of @xmath377 and @xmath378 for @xmath379 .
notice that the second place is given by the conjugated points @xmath380 for @xmath379 . the place @xmath82 and the divisor @xmath79 satisfy the good properties since the dimension of the divisor @xmath105 is zero which means that the divisor @xmath105 is non - special of degree @xmath106 .
we choose as basis of the residue class field @xmath102 the normal basis @xmath381 associated to the place @xmath82 obtained in section [ q13 ] .
we choose as basis of the riemann - roch space @xmath113 the basis @xmath382 such that @xmath383 is a normal basis of @xmath102 as in section [ baseld ] .
any element @xmath384 of @xmath385 is such that @xmath386 where @xmath387 $ ] . to simplify , we set @xmath388 .
let us give the elements of @xmath385 : @xmath389 , + @xmath390 , + @xmath391 , + @xmath392 , + @xmath393 , + @xmath394 , + @xmath395 , + @xmath396 , + @xmath397 , + @xmath398 , + @xmath399 , + @xmath400 , + @xmath401 , + let @xmath402 be a basis of @xmath120 as defined in section [ basel2d ] . since the divisor @xmath79 is effective , we can complete the basis @xmath403 of @xmath113 in @xmath120
. then any element @xmath404 of @xmath184 is such that @xmath405 for @xmath361 and @xmath406 ( which we will denote @xmath407 ) , with @xmath408 $ ] for @xmath409 .
moreover , we require the completion of @xmath113 by the kernel of the map @xmath122 .
hence , the set @xmath410 is a basis of the kernel of the restriction map over @xmath120 of the canonical projection from the valuation ring of the place @xmath82 in its residue class field @xmath102 .
@xmath411 , + @xmath412 , + @xmath413 , + @xmath414 , + @xmath415 , + @xmath416 , + @xmath417 , + @xmath418 , + @xmath419 , + @xmath420 , + @xmath421 , + @xmath422 , + @xmath423 , + @xmath424 .
.... // construction of the function field n:=13 ; g:=2 ; q:=16 ; f16<a>:=gf(16 ) ; kx < x > : = functionfield(f16 ) ; r < x>:=polynomialring(f16 ) ; kxy < y > : = polynomialring(kx ) ; f:=y^2 + y + x^5 ; f < c > : = functionfield(f ) ; // construction of the place q and divisor d q : = x^13+a^6*x^12+a^5*x^11+a^11*x^10+x^9+a^12*x^8+a^7*x^7 + a^7*x^5+a^2*x^4+a^11*x^3+a^8*x^2+a^6*x+a^14 ; dd:=x^14+a^9*x^13+a^6*x^12+a^7*x^11+a^11*x^10+a^12*x^9 + a^10*x^8+a^6*x^7+a^7*x^6+a^10*x^5+a^14*x^4+x^3+x^2+a^3*x+a ; p:=decomposition(f , zeros(kx!q)[1 ] ) ; q:=p[1 ] ; d:=decomposition(f , zeros(kx!dd)[1])[1 ] ; d:=1d ; isspecial(1d ) ; // false dimension(q - d ) ; // 0 isspecial(d - q ) ; // false //
construction of the residue class field // and the degree one places k < b>:=residueclassfield(q ) ; lp:=places(f,1 ) ; // construction of the riemann space ld , h : = riemannrochspace(d ) ; l2d , h2 : = riemannrochspace(2d ) ; baseld:=[(h(v))@@h2 : v in basis(ld ) ] ; base:=extendbasis(baseld , l2d ) ; l2d:= [ ] ; for i in[1 .. 2*n+g-1 ] do l2d:=append(l2d , h2(base[i ] ) ) ; end for ; ml2d:=matrix(2*n+g-1,1,l2d ) ; // construction of e : e = evalf(q ) l:= [ ] ; for i in [ 1 .. n ] do l:=append(l , elementtosequence(evaluate(l2d[i],q ) ) ) ; end for ; e:=transpose(matrix(l ) ) ; // we use a normal basis pp:= [ ] ; for i:=0 to n-1 do pp:=append(pp , elementtosequence(b^(q^i ) ) ) ; end for ; nc:=transpose(matrix(f16,n , n , pp ) ) ; nci:=nc^-1 ; // x.m = fx m:=matrix(f , e^-1*nc ) ; ev:=matrix(1,n,[l2d[i ] : i in [ 1 .. n ] ] ) ; //
construction of l(2d ) with the required properties el2d:=matrix(2*n+g-1,1 , [ evaluate(l2d[i],q ) : i in [ 1 .. 2*n+g-1 ] ] ) ; bel2d:=matrix(f16,2*n+g-1,n , [ elementtosequence(el2d[i][1 ] ) : i in [ 1 .. 2*n+g-1 ] ] ) ; mm:=parent(zeromatrix(f , n+g-1,2n+g-1 ) ) !
matrix(basis(nullspace(bel2d)))*ml2d ; for i in [ 1 .. n ] do l2d[i]:=transpose(evm)[i,1 ] ; end for ; //
rows of ml2d form a basis of l(2d ) ml2d:=matrix(2*n+g-1,1 , [ l2d[i ] : i in [ 1 .. n ] ] cat [ mm[i,1 ] : i in [ 1 .. n+g-1 ] ] ) ; // we can check that evaluation in //
q of the last n+g-1 gives 0 : //
[ evaluate(ml2d[i,1],q ) : i in [ 1 .. 2*n+g-1 ] ] ; // construction of t and t^-1 st:= [ ] ; for j:=1 to 2*n+g-1 do for i:=1 to 2*n+g-1 do st:=append(st , evaluate(ml2d[i,1 ] , lp[j ] ) ) ; end for ; end for ; t:=matrix(2*n+g-1 , st ) ; ti:=t^-1 ; // construction of p p:=verticaljoin(horizontaljoin(scalarmatrix(f16,n,1 ) , zeromatrix(f16,n , n+g-1 ) ) , zeromatrix(f16,n+g-1,2*n+g-1 ) ) ; // matrix t1
t1:=t*p*ti ; // x and y are the elements to multiply // represented in a normal basis x:=matrix(f16,13,1,[a,1,0,0,0,0,0,0,0,0,0,0,0 ] ) ; //
example y:=matrix(f16,13,1,[1,a , a,0,0,0,0,0,0,0,0,0,0 ] ) ; //
example fx:=verticaljoin(x , zeromatrix(f16,n+g-1,1 ) ) ; fy:=verticaljoin(y , zeromatrix(f16,n+g-1,1 ) ) ; // u = t(fx)t(fy ) u:= matrix(2*n+g-1,1 , [ ( tfx)[i][1]*(tfy)[i][1 ] : i in [ 1 .. 2*n+g-1 ] ] ) ; //
fz = mm(p*ti*u ) fz:=matrix(n,1,[(p*ti*u)[i][1 ] : i in [ 1 .. n ] ] ) ; //
fz gives x*y in the normal basis ....
d. coppersmith and s. winograd .
matrix multiplication via arithmetic progressions . in _ proceedings of the nineteenth annual acm symposium on theory of computing _
, stoc 87 , pages 16 , new york , ny , usa , 1987 . acm . | thanks to a new construction of the so - called chudnovsky - chudnovsky multiplication algorithm , we design efficient algorithms for both the exponentiation and the multiplication in finite fields . they are tailored to hardware implementation and they allow computations to be parallelized while maintaining a low number of bilinear multiplications .
we give an example with the finite field @xmath0 . |
recently , a new sub - discipline of condensed matter physics , spintronics , is emerging rapidly and generating great interests.@xcite the spin current , the most important physical quantity in spintronics , has been extensively studied .
many interesting and fundamental phenomena , such as the spin hall effect@xcite and the spin precession@xcite in systems with spin - orbit coupling , have been discovered and are under further study . as for the charge current ,
the definition of the local charge current density @xmath4 $ ] and its continuity equation @xmath5 is well - known in physics . here
@xmath6 is the electronic wave function , @xmath7 is the velocity operator , and @xmath8 is the charge density .
this continuity equation is the consequence of charge invariance , i.e. when an electron moves from one place to another , its charge remains the same .
however , in the spin transport , there are still a lot of debates over what is the correct definition for spin current.@xcite the problem stems from that the spin @xmath9 is no invariant quantity in the spin transport , so that the conventional defining of the spin current @xmath10 is no conservative .
recently , some studies have begun investigation in this direction@xcite , e.g. a semi - classical description of the spin continuity equation has been proposed@xcite , as well as studies introducing a conserved spin current under special circumstances.@xcite in this paper , we study the definition of local spin current density .
we find that due to the spin is vector and it has the translational and rotational motion , one has to use two quantities , the linear spin current and the angular spin current , to completely describe the spin transport . here
the linear spin current describe the translational motion of a spin , and the angular spin current is for the rotational motion .
the conventional linear spin current has been extensively studied .
however , the physical meaning of the angular spin current is given for the first time .
the definition of two spin current densities are given and they appear naturally in the quantum spin continuity equation .
moreover , we predict that the angular spin current can generate an electric field similar as with the linear spin current , and thus contains physical consequences . the paper is organized as follows . in section
ii , we first discuss the flow of a classical vector . the flow of a quantum spin is investigated in section iii . in section iv , we study the problem of electric fields induced by spin currents .
finally , a brief summary is given in section v.
before studying the spin current in a quantum system , we first consider the classical case . consider a classical particle having a vector @xmath11 ( e.g. the classical magnetic moment , etc . ) with its magnitude @xmath12 fixed under the particle motion . to completely describe this vector flow ( see fig.1c ) , besides the local vector density @xmath13 one needs two quantities : the linear velocity @xmath14 and the angular velocity @xmath15 . here
@xmath16 is the particle density , and @xmath17 and @xmath18 describe the translational and rotational motions , respectively ( see fig.1a and b ) .
in contrast with the flow of a scalar quantity in which one only needs one quantity , namely the local velocity @xmath19 , to describe the translational motion , it is essential to use two quantities , @xmath14 and @xmath15 , for the vector flow .
we emphasize that it is impossible to use one vector to describe both translational and rotational motions altogether .
since @xmath12 is a constant , the change of the local vector @xmath20 in the volume element @xmath21 with the time from @xmath22 to @xmath23 can be obtained ( see fig.1c ) : @xmath24 \nonumber \\ & & + \vec{\omega}({\bf r},t)\times \vec{m}({\bf r},t ) \delta v dt.\end{aligned}\ ] ] the first and the second terms on the right describe the classical vector flowing in or out the volume element @xmath25 and its rotational motion respectively , and both can cause a change in the local vector density .
when @xmath25 goes to zero , we have the vector continuity equation:@xcite @xmath26 where @xmath27 is a tensor , and its element @xmath28 .
notice that this continuity equation is from the kinematics and the invariance of @xmath12 , in particular it is independent of the dynamic laws.@xcite this is completely same with the continuity equation with a scalar quantity ( e.g. charge ) . introducing @xmath29 and @xmath30 , then eq.(2 ) reduces to : @xmath31 here @xmath32 is from the translational motion of the classical vector @xmath11 , and @xmath33 describes its rotational motion.@xcite since @xmath17 and @xmath18 are called as the linear velocity and the angular velocity respectively , it is natural to name @xmath34 and @xmath35 as the linear and the angular current densities . in order to describe a scalar ( e.g. charge ) flow ,
one local current @xmath36 is sufficient .
why is it required to introduce two quantities instead of one to describe a vector flow ?
the reason is that the scalar quantity only has the translational motion , but the vector quantity has two kinds of motion , the translational and the rotational .
so one has to use two quantities , the velocity @xmath17 and the angular velocity @xmath18 , to describe the motion of a single vector .
correspondingly , two quantities @xmath37 and @xmath33 are necessary to describe the vector flow . in the steady state case
, the scalar ( e.g. charge ) continuity equation reduces into @xmath38 , so the scalar current @xmath39 is a conserved quantity .
but for a vector flow , the linear vector current @xmath34 is not conserved since @xmath40 . whether it is possible to have a conserved vector current through redefinition ?
of course , this redefined vector current should have a clear physical meaning and is measurable . in our opinion , this is almost impossible in the 3-dimensional space .
the reasons are : ( i ) one can not use a 3-dimensional vector to combine both @xmath17 and @xmath18 .
therefore one can also not use a 3-dimensional tensor to combine both @xmath37 and @xmath33 .
( ii ) consider an example , as shown in fig.3a , a one - dimensional classical vector flowing along the x - axis .
when @xmath41 , the vector s direction is in @xmath42 axis . at @xmath43 ,
the vector rotates in accompany with its translational motion .
when @xmath44 , its direction is along @xmath45 axis .
since for @xmath41 and @xmath44 the vector has no rotational motion , the definition of the vector current is unambiguous , and the non - zero element is @xmath46 for @xmath41 , and @xmath47 for @xmath44 .
therefore , the vector current is obviously different for @xmath41 and @xmath44 , and the vector current is non - conservative . in our opinion ,
@xmath37 and @xmath33 already have clear physical meanings .
they also completely and sufficiently describe a vector flow .
one may not need need to enforce a conserved current .
in particular , as shown in the example of fig.3a , sometimes it is impossible to introduce a conserved current .
now we study the electronic spin @xmath9 in the quantum case .
consider an arbitrary wave function @xmath6 .
the local spin density @xmath9 at the position @xmath48 and time @xmath22 is : @xmath49 , where @xmath50 with @xmath51 being the pauli matrices .
the time - derivative of @xmath52 is : @xmath53 \hat{\vec{\sigma } } \psi + \psi^{\dagger } \hat{\vec{\sigma } } \frac{d}{dt } \psi \right\}.\ ] ] from the schrodinger equation , we have @xmath54 and @xmath55^{\dagger}$ ] .
notice here the transposition in the symbol @xmath56 only acts on the spin indexes . by using the above two equations , the eq.(4 ) changes into : @xmath57/2i.\ ] ] in the derivation below , we use the following hamiltonian : @xmath58 note that our results are independent of this specific choice of the hamiltonian.@xcite in eq.(6 ) the 1st and 2nd terms are the kinetic energy and potential energy .
the 3rd term is the zeeman energy due to a magnetic field , and the last term is the rashba spin - orbit coupling@xcite , which has been extensively studied recently@xcite .
next we substitute the hamiltonian eq.(6 ) into eq.(5 ) , and the eq.(5 ) reduces to : @xmath59 \hat{\vec{\sigma}}\psi \ } \nonumber\\ & + & re\ { \psi^{\dagger } [ \vec{b } + \frac{\alpha}{\hbar}\vec{p}\times\hat{z}]\times\hat{\vec{\sigma } } \psi \}.\end{aligned}\ ] ] introducing a tensor @xmath60 and a vector @xmath61 : @xmath62 \hat{\vec{s}}\psi \}\\ \vec{j}_{\omega}({\bf
r},t ) & = & re \{\psi^{\dagger } \frac{2}{\hbar } [ \vec{b } + \frac{\alpha}{\hbar}(\vec{p}\times\hat{z})]\times\hat{\vec{s}}\psi \},\end{aligned}\ ] ] then eq.(7 ) reduces to : @xmath63 or it can also be rewritten in the integral form : @xmath64 due to the fact that @xmath65 and @xmath66 = \frac{2}{\hbar}[\vec{b } + \frac{\alpha}{\hbar } \vec{p}\times\hat{z}]\times \hat{\vec{\sigma}}$ ] , eqs.(8 ) and ( 9 ) become : @xmath67 where @xmath68 $ ] is the angular velocity operator.@xcite eq.(10 ) is the spin continuity equation , which is same with the classic vector continuity equation ( 3 ) although the derivation process is very different . in some previous works , this equation has also been obtained in the semiclassical case.@xcite here we emphasize that this spin continuity equation ( 10 ) is the consequence of invariance of the spin magnitude @xmath69 , i.e. when an electron makes a motion , either translation or rotation , its spin magnitude @xmath70 remains a constant . and the eq.(10 ) should be independent with the force ( i.e. the potential ) and the torque , as well the the dynamic law .
the two quantities @xmath60 and @xmath71 in eq.(10 ) , which are defined in eqs.(12,13 ) respectively , describe the translational and rotational motion ( precession ) of a spin at the location @xmath48 and the time @xmath22 .
they will be named the linear and the angular spin current densities accordingly@xcite , similar as @xmath17 and @xmath18 are called the linear and the angular velocities .
in fact , the linear spin current @xmath60 is identical with the conventional spin current investigated in recent studies@xcite .
however , we give the physical meaning of @xmath61 for the first time .
next , we discuss certain properties of @xmath72 and @xmath61 .
notice that @xmath71 which describe the rotational motion ( precession ) of the spin plays a parallel role in comparison with the conventional linear spin current @xmath72 for the spin transport .
( 1 ) in similar with the classical case , it is necessary to introduce the two quantities @xmath60 and @xmath61 to completely describe the motion of a quantum spin .
( 2 ) the linear spin current is a tensor .
its element , e.g. @xmath73 , represents an electron moving along the @xmath74 direction with its spin in the @xmath75 direction ( see fig.2a ) .
the angular spin current @xmath35 is a vector . in fig.2b ,
its element @xmath76 describes the rotational motion of the spin in the @xmath75 direction and the angular velocity @xmath77 in the @xmath78 direction .
( 3 ) from the linear spin current density @xmath72 , one can calculate ( or say how much ) the linear spin current @xmath79 flowing through a surface @xmath80 ( see fig.2d ) : @xmath81 .
however , the behavior for the angular spin current is different . from the density @xmath82 ,
it is meaningless to determine how much the angular spin current flowing through a surface @xmath80 , because the angular spin current describes the rotational motion not the movement . on the other hand
, one can calculate the total angular spin current @xmath83 in a volume @xmath84 from @xmath35 : @xmath85 .
( 4 ) it is easy to prove that the spin currents in the present definitions of eqs.(12,13 ) are invariant under a space coordinate transformation as well the gauge transformation .
( 5 ) if the system is in a steady state , @xmath34 and @xmath35 are independent of the time @xmath22 , and @xmath86 .
then the spin continuity equation ( 10 ) reduces to : @xmath87 or @xmath88 .
this means that the total linear spin current flowing out of a closed surface is equal to the total angular spin current enclosed .
if to further consider a quasi one dimensional ( 1d ) system ( see fig.2d ) , then one has @xmath89 .
( 6 ) the linear spin current density @xmath90 gives both the spin direction and the direction of spin movement , so it completely describes the translational motion . however , the angular spin current density , @xmath91 involves the vector product of @xmath92 , not the tensor @xmath93 .
is it correct or sufficient to describe the rotational motion ?
for example , the rotational motion of fig.2b with the spin @xmath9 in the @xmath75 direction and the angular velocity @xmath18 in the @xmath78 direction is different from the one in fig.2c in which @xmath9 is in the @xmath94 direction and @xmath18 is in the @xmath75 direction , but their angular spin currents are completely the same .
shall we distinguish them ?
it turns out that the physical results produced by the above two rotational motions ( fig.2b and 2c ) are indeed the same .
for instance , the induced electric field by them is identical since a spin @xmath9 has only the direction but no size ( see detail discussion below ) .
thus , the vector @xmath35 is sufficient to describe the rotational motion , and no tensor is necessary .
now we give an example of applying the above formulas , eqs.(12,13 ) , to calculate the spin currents .
let us consider a quasi 1d quantum wire having the rashba spin orbit coupling , and its hamiltonian is : @xmath95 + \frac{\hbar^2 k_r^2}{2 m } , \ ] ] where @xmath96 .
@xmath97 for @xmath41 and @xmath44 , and @xmath98 while @xmath43 .
the other rashba term @xmath99 is neglected because z - direction is quantized@xcite .
let @xmath100 be a stationary wave function @xmath101 where @xmath102 is the bound state wave function in the confined @xmath75 and @xmath94 directions .
@xmath103 represents the spin motion as shown in fig.3a , in which the spin moves along the @xmath74 axis , as well the spin precession in the @xmath74-@xmath75 plane in the region @xmath43.@xcite using eqs.(12,13 ) , the spin current densities of the wave function @xmath103 are easily obtained .
there are only two non - zero elements of @xmath104 : @xmath105 the non - zero elements of @xmath106 are : @xmath107 where @xmath108 .
those spin current densities confirm with the intuitive picture of an electron motion , precession in the @xmath74-@xmath75 plane in @xmath43 and movement in the @xmath74 direction ( see fig.3a ) .
in particular , in the region of @xmath41 and @xmath44 , @xmath109 , and @xmath110 is a good quantum number , hence , @xmath111 . in this case
, the definition of the spin current @xmath34 is unambiguous .
however , the spin currents are different in @xmath41 and @xmath44 except for @xmath112 ( @xmath113 ... ) .
this is clearly seen from fig.3a .
therefore , through this example , one can conclude that it is sometimes impossible to define a conserved spin current .
the example of fig.3a indeed exists and has been studied before@xcite . above discussion
shows that the linear spin current @xmath114 and the angular spin current @xmath115 have clear physical meanings , representing the translational motion and the rotational motion ( precession ) respectively .
they completely describe the flow of a quantum spin .
any physical effects of the spin currents , such as the induced electric field , can be expressed by @xmath34 and @xmath35.@xcite
recently , theoretic studies have suggested that the ( linear ) spin current can induce an electric field @xmath0@xcite . can the angular spin current also induce an electric field ? if so this gives a way of detecting the angular spin current .
following , we study this question by using the method of equivalent magnetic charge@xcite .
let us consider a steady - state angular spin current element @xmath116 at the origin .
associated with the spin @xmath9 , there is a magnetic moment ( mm ) @xmath117 where @xmath118 is the bohr magneton .
thus , corresponding to @xmath35 , there is also a angular mm current @xmath119 . from above discussions , we already know that @xmath120 ( or @xmath35 ) comes from the rotational motion of a mm @xmath11 ( or @xmath9 )
( see fig.2b and 2c ) , and @xmath121 ( or @xmath122 ) . under the method of equivalent magnetic charge ,
the mm @xmath11 is equivalent to two magnetic charges : one with magnetic charge @xmath123 located at @xmath124 and the other with @xmath125 at @xmath126 ( see fig.2e ) .
@xmath127 is the unit vector of @xmath11 and @xmath128 is a tiny length .
the angular mm current @xmath120 is equivalent to two magnetic charge currents : one is @xmath129 at the location @xmath124 , the other is @xmath130 at @xmath126 ( see fig.2e ) , with @xmath131 being the unit vector of @xmath120 and @xmath132 the angle between @xmath18 and @xmath11 . in our previous work,@xcite
we have given the formulae of the electric field induced by a magnetic charge current .
the electric field induced by @xmath133 can be calculated by adding the contributions from the two magnetic charge currents .
let @xmath134 , and note that @xmath135 and @xmath136 , we obtain the electric field @xmath137 generated by an element of the angular spin current @xmath138 : @xmath139 we also rewrite the electric field @xmath140 generated by an element of the linear spin current using the tensor @xmath141:@xcite @xmath142 below we emphasize three points : ( i ) in the large @xmath2 case , the electric field @xmath137 decays as @xmath1 .
note that the field from a linear spin current @xmath140 goes as @xmath3 .
in fact , in terms of generating an electric field , the angular spin current is as effective as a magnetic charge current .
( ii ) in the steady - state case , the total electric field @xmath143 contains the property : @xmath144 , where @xmath145 is an arbitrary close contour not passing through the region of spin current . however , for each @xmath137 or @xmath140 , @xmath146 or @xmath147 can be non - zero . (
iii ) as mentioned above , a angular spin current @xmath35 may consist of different @xmath18 and @xmath9 ( see fig.2b and 2c ) .
however , the resulting electric field only depends on @xmath148 .
this is because a spin vector contains only a direction and a magnitude , but not a spatial size ( i.e. the distance @xmath128 approaches to zero ) . in the limit @xmath149 ,
both magnetic charge currents @xmath150 reduce to @xmath151 at the origin .
therefore , the overall effect of the rotational motion is only related to @xmath152 , not separately on @xmath18 and @xmath11 .
hence it is enough to describe the spin rotational motion by using a vector @xmath153 , instead of a tensor @xmath154 . due to the fact that the direction of @xmath9 can change during the particle motion , the linear spin
current density @xmath34 is not a conserved quantity .
it is always interesting to uncover a conserved physical quantity from both theoretical and experimental points of view .
let us apply @xmath155 acting on two sides of eq.(10 ) , we have : @xmath156 where @xmath157 means that @xmath158 acts on the second index of @xmath34 , i.e. @xmath159 with @xmath160 .
define @xmath161 , the above equation reduces into:@xcite @xmath162 this means that the current @xmath163 of the spin divergence is a conserved quantity in the steady state case .
in fact , @xmath164 represents an equivalent magnetic charge , so @xmath163 can also be named the magnetic charge current density .
moreover , the total electric field produced by @xmath34 and @xmath35 can be rewritten as : @xmath165 so the total electric field @xmath166 only depends on the current @xmath163 of the spin divergence .
note that @xmath166 can be measured experimentally in principle . through the measurement of @xmath167 , @xmath163 can be uniquely obtained . in the following ,
let us calculate the induced electric fields at the location @xmath168 by the spin currents in the example of fig.3a . substituting the spin currents of eqs.(16 - 19 ) into eqs.(20,21 ) and assuming the transverse sizes of the 1d wire are much smaller than @xmath169 , the induced fields @xmath137 and @xmath140 can be obtained straightforwardly .
then the total electric field @xmath170 is : @xmath171^{3/2 } } dx ' \nonumber\\ & = & a \nabla \int ( \vec{\cal{v}}\times\vec{\cal{s}})\bullet \frac{{\bf r}-{\bf r'}}{|{\bf r}-{\bf r'}|^3 } dx'\end{aligned}\ ] ] where @xmath172 , @xmath173 , @xmath174 , the constant @xmath175 , and @xmath176 is the linear density of moving electrons under the bias of an external voltage .
the total electric field @xmath166 represents the one generated by a 1d wire of electric dipole moment @xmath177 at the @xmath74 axis ( see fig.3b ) , where @xmath178 is a constant .
it is obvious that @xmath179 , i.e. @xmath144 .
however , in general @xmath146 and @xmath147 are separately non - zero .
finally , we estimate the magnitude of @xmath166 .
we use parameters consistent with realistic experimental samples .
take the rashba parameter @xmath180evm ( corresponding to @xmath181=1/100 nm for @xmath182 ) , @xmath183/m ( i.e. one moving electron per 1000 nm in length ) , and @xmath184/m .
the electric potential difference between the two points a and b ( see fig.3b ) is about @xmath185 , where the positions of a and b are @xmath186 and @xmath187 .
this value of the potential is measurable with today s technology.@xcite furthermore , with the above parameters the electric field @xmath166 at a or b is about @xmath188v / m which is rather large .
in summary , we find that in order to completely describe the spin flow ( including both classic and quantum flows ) , apart from the conventional spin current ( or linear spin current ) , one has to introduce another quantity , the angular spin current .
the angular spin current describes the rotational motion of the spin , and it plays a parallel role in comparison with the conventional linear spin current for the spin translational motion .
moreover we point out that the angular spin current can also induce an electric field and its @xmath137 field scales as @xmath1 at large @xmath2 .
in addition , a conserved quantity , the current @xmath163 of the spin divergence , is discovered , and the total electric field only depends on @xmath163 .
we gratefully acknowledge financial support from the chinese academy of sciences and nsfc under grant no . 90303016 and no .
xcx is supported by us - doe under grant no .
de - fg02 - 04er46124 , nsf ccf-0524673 and nsf - mrsec under dmr-0080054 . s .- q .
shen , m. ma , x.c .
xie , and f.c .
zhang , phys .
lett . * 92 * , 256603 ( 2004 ) ; cond - mat/0410169 ( 2004 ) ; s. murakami , n. nagaosa , and s .- c .
zhang , phys .
lett . * 93 * , 156804 ( 2004 ) ; s .- p .
qi , and z .- y . weng , cond - mat/0412146 ( 2004 ) . if to consider that there are many particles in the volume element @xmath25 ( i.e. the volume element @xmath25 is very small macroscopically but very large microcosmically ) , the vector direction @xmath189 , the velocity @xmath190 , and the angular velocity @xmath191 for each particle may be different , however , the vector continuity equation ( 3 ) is still valid : @xmath192 with @xmath193 and @xmath194 . here
@xmath72 and @xmath61 still describe the translational and the rotational motions of the classical vector .
the scalar ( e.g. charge @xmath195 ) continuity equation @xmath196 is from the kinematics and the invariance of the charge @xmath195 .
it is independent of the external force @xmath197 as well dynamic laws . in other words , even if the acceleration @xmath198 , the continuity equation still survives .
it is complete same with the vector continuity equation ( 2 ) or ( 3 ) .
this equation is also from the kinematics and the invariance of @xmath12 . in particular , it is independent of the external force and the torque acting on the vector , as well as the dynamic laws . in other words ,
the vector continuity equation does not depend on the changes in the velocity and the angular velocity under the actions of the forces and the torques .
note that those results , eqs.(10 , 12 , and 13 ) , are valid in general .
they are independent of the special choice of hamiltonian ( 6 ) .
for example , in the case with a vector potential @xmath199 , the general spin - orbit coupling @xmath200 $ ] , and so on , the results still hold .
@xmath35 is called the spin torque in the work@xcite in which the semi - classical approach to the spin continuity equation is discussed . here
we consider that @xmath35 describes the rotational motion of a spin , thus named the angular spin current . from @xmath72 ,
the total linear spin current along @xmath201 direction ( @xmath202 ) is : @xmath203 . to assume @xmath204 independent on @xmath22 and @xmath201 ( e.g. in the case of the steady state and without spin flip ) , one has : @xmath205 , where @xmath206 is sample length in @xmath201-direction .
this definition is the same as in recent publications@xcite .
here we consider another physical effect , the heat produced by the spin currents .
assume a uniform isotropic conductor having a linear spin current @xmath34 and a charge current @xmath39 , and considering the simple case that there exists no spin flip process ( i.e. @xmath9 is conserved ) so that @xmath111
. then the produced heat @xmath207 in unit volume and in unit time is @xmath208 = \rho \left ( \sum\limits_{ij } j^2_{s , ij } + \sum\limits_i ( j_{ei})^2 \right ) \equiv \rho ( { \bf j}_s^2 + \vec{j}_e^2)$ ] , where @xmath209 is the resistivity .
so the produced heat by the spin current can be expressed by @xmath34 ( for the case of @xmath111 ) .
note the produced heat @xmath207 depends on @xmath210 , whereas the induced electric field depends on @xmath211 . | we find that in order to completely describe the spin transport , apart from spin current ( or linear spin current ) , one has to introduce the angular spin current .
the two spin currents respectively describe the translational and rotational motion ( precession)of a spin .
the definitions of these spin current densities are given and their physical properties are discussed .
both spin current densities appear naturally in the spin continuity equation .
moreover we predict that the angular spin current can also induce an electric field @xmath0 , and in particular @xmath0 scales as @xmath1 at large distance @xmath2 , whereas the @xmath0 field generated from the linear spin current goes as @xmath3 . |
parallel computing is important for speeding up computation .
the design of an interconnection network is the first thing to be considered .
many topologies have been proposed in the literature @xcite , and the desirable properties of an interconnection network include symmetry , relatively small degree , small diameter , embedding capabilities , scalability , robustness , and efficient routing . among those proposed interconnection networks ,
the hypercube is a popular interconnection network with many attractive properties such as regularity , symmetry , small diameter , strong connectivity , recursive construction , partition ability , and relatively low link complexity @xcite .
the architecture of an interconnection network is usually represented by a graph .
we will use graphs and networks interchangeably .
the @xmath0-dimensional twisted cube @xmath1 , an important variation of the hypercube , was first proposed by hilbers et al .
@xcite and possesses some properties superior to the hypercube .
in fact , the twisted cube is derived from the hypercube by twisting some edges . due to these twisted edges ,
the diameter , wide diameter , and fault diameter of @xmath1 are about half of those of the comparable hypercube @xcite .
an @xmath0-dimensional twisted cube is @xmath4-hamiltonian connected @xcite and @xmath5-pancyclic @xcite , whereas the hypercube is not .
moreover , its performance is superior to that of the hypercube even if it is asymmetric @xcite .
recently , some interesting properties of the twisted cube @xmath1 were investigated .
let @xmath6 be a graph .
we denote by @xmath7 and @xmath8 the node set and the edge set of @xmath6 , respectively .
a graph @xmath6 is _ pancyclic _ if , for every latexmath:[$4 \leqslant l \leqslant @xmath6 is _ edge - pancyclic _ ( resp .
_ node - pancyclic _ ) if , for any edge @xmath10 ( resp .
node @xmath11 ) of @xmath6 and every @xmath12 , @xmath6 has a cycle of length @xmath13 containing @xmath10 ( resp .
@xmath11 ) .
yang et al . showed that , with @xmath14 , a faulty @xmath1 still contains a cycle of length @xmath13 for every @xmath15 , where @xmath16 and @xmath17 are the numbers of faulty edges and faulty nodes in @xmath1 , respectively @xcite . in @xcite
, fu showed that @xmath1 can tolerate up to @xmath18 edge faults , while retaining a fault - free hamiltonian cycle .
fan et al . showed that the twisted cube @xmath1 , with @xmath3 , is edge - pancyclic and provided an @xmath19-time algorithm to find a cycle of length @xmath13 containing a given edge of the twisted cube @xcite . in @xcite
, the author also asked if @xmath1 is edge - pancyclic with @xmath4 faults for @xmath3 .
yang answered the question and showed that @xmath1 is not edge - pancyclic with only one faulty edge for any @xmath20 , and that @xmath1 is node - pancyclic with @xmath21 faulty edges for every @xmath3 @xcite .
in addition , lai et al . embedded a family of 2-dimensional meshes into a twisted cube @xcite .
two hamiltonian cycles in a graph are said to be _ edge - disjoint _ if they do not share any common edge .
the edge - disjoint hamiltonian cycles can provide advantage for algorithms that make use of a ring structure @xcite .
the following application about edge - disjoint hamiltonian cycles can be found in @xcite .
consider the problem of all - to - all broadcasting in which each node sends an identical message to all other nodes in the network .
there is a simple solution for the problem using an @xmath0-node ring that requires @xmath22 steps , i.e. , at each step , every node receives a new message from its ring predecessor and passes the previous message to its ring successor .
if the network admits edge - disjoint rings , then messages can be divided and the parts broadcast along different rings without any edge contention .
if the network can be decomposed into edge - disjoint hamiltonian cycles , then the message traffic will be evenly distributed across all communication links .
edge - disjoint hamiltonian cycles also form the basis of an efficient all - to - all broadcasting algorithm for networks that employ warmhole or cut - through routing @xcite . recently ,
conditional link faults of interconnected networks were discussed in @xcite .
the edge - disjoint hamiltonian cycles in @xmath23-ary @xmath0-cubes and hypercubes has been constructed in @xcite .
barden et al . constructed the maximum number of edge - disjoint spanning trees in a hypercube @xcite .
petrovic et al .
characterized the number of edge - disjoint hamiltonian cycles in hyper - tournaments @xcite .
hsieh et al . constructed edge - disjoint spanning trees in locally twisted cubes @xcite .
although the existence of a hamiltonian cycle in twisted cubes has been shown @xcite , it is not clear to generate edge - disjoint hamiltonian cycles in twisted cubes . in this paper , we show that , for any odd integer @xmath2 , there are two edge - disjoint hamiltonian cycles in the @xmath0-dimensional twisted cube @xmath1 .
further , we also construct two equal node - disjoint cycles in the twisted cube .
two cycles in a graph are said to be _ equal _ and _ node - disjoint _ if they contain the same number of nodes , there is no common node in them , and every node is in one cycle exactly once .
finding two equal node - disjoint cycles in a interconnected network is equivalent to decompose the network into two disjoint components with the same number of nodes such that each component contains a hamiltonian cycle .
then , two distinct algorithms that require a ring structure can be preformed in the two components simultaneously . in this paper , we show that , for any odd integer @xmath3 , there are two equal node - disjoint cycles in the @xmath0-dimensional twisted cube @xmath1 . the rest of the paper is organized as follows . in section [ preliminaries ] , the structure of the twisted cube is introduced , and some definitions and notations used throughout this paper are given . section [ edgedisjointhc ] shows the construction of two edge - disjoint hamiltonian cycles in the twisted cube . in section [ nodedisjointcyles ] ,
we construct two equal node - disjoint cycles in the twisted cube .
finally , we conclude this paper in section [ conclusion ] .
we usually use a graph to represent the topology of an interconnection network . a graph
@xmath24 is a pair of the node set @xmath25 and the edge set @xmath26 , where @xmath25 is a finite set and @xmath26 is a subset of @xmath27 is an unordered pair of @xmath28 .
we also use @xmath7 and @xmath8 to denote the node set and the edge set of @xmath6 , respectively .
if @xmath29 is an edge in a graph @xmath6 , we say that @xmath11 _ is adjacent to _ @xmath30 .
neighbor _ of a node @xmath30 in a graph @xmath6 is any node that is adjacent to @xmath30 .
moreover , we use @xmath31 to denote the neighbors of @xmath30 in @xmath6 .
the subscript ` @xmath6 ' of @xmath31 can be removed from the notation if it has no ambiguity .
a path @xmath32 , represented by @xmath33 , is a sequence of distinct nodes such that two consecutive nodes are adjacent .
the first node @xmath34 and the last node @xmath35 visited by @xmath32 are called the _ path - start _ and _ path - end _ of @xmath32 , denoted by @xmath36 and @xmath37 , respectively , and they are called the _
end nodes _ of @xmath32 .
path @xmath38 is called the _ reversed path _ , denoted by @xmath39 , of @xmath32 .
that is , @xmath39 visits the vertices of @xmath32 from @xmath37 to @xmath36 sequently .
in addition , @xmath32 is a cycle if @xmath40 and @xmath37 is adjacent to @xmath36 .
a path @xmath41 may contain other subpath @xmath42 , denoted as @xmath43 , where @xmath44 .
a path ( or cycle ) in @xmath6 is called a _ hamiltonian path _ ( or _ hamiltonian cycle _ ) if it contains every node of @xmath6 exactly once . two paths ( or cycles ) @xmath45 and @xmath46 connecting a node @xmath11 to a node @xmath30
are said to be _ edge - disjoint _
iff @xmath47 . two paths ( or cycles ) @xmath48 and @xmath49 of graph @xmath6
are called _ node - disjoint _
iff @xmath50 . two node - disjoint paths ( or cycles ) @xmath48 and @xmath49 of graph @xmath6
are said to be _ equal _ iff @xmath51 and @xmath52 .
two node - disjoint paths @xmath48 and @xmath49 can be _ concatenated _ into a path , denoted by @xmath53 , if @xmath54 is adjacent to @xmath55 .
now , we introduce twisted cubes .
the node set of the @xmath0-dimensional twisted cube @xmath1 is the set of all binary strings of length @xmath0 , where @xmath0 is odd .
a binary string @xmath56 of length @xmath0 is denoted by @xmath57 , where @xmath58 is the most significant bit .
we denote the complement of @xmath59 by @xmath60 . to define @xmath1 , a @xmath61-th bit _ parity function _
@xmath62 is introduced .
let @xmath63 be a binary string .
for @xmath64 , @xmath65 , where @xmath66 is the exclusive - or operation . note that @xmath67 and @xmath68 .
we then give the recursive definition of the @xmath0-dimensional twisted cube @xmath1 for any odd integer @xmath69 as follows .
@xcite[def_twisted - cube ] @xmath70 is the complete graph with two nodes labeled by 0 and 1 , respectively . for an odd integer @xmath20 ,
@xmath1 consists of four copies of @xmath71 .
we use @xmath72 to denote an @xmath5-dimensional twisted cube which is a subgraph of @xmath1 induced by the nodes labeled by @xmath73 , where @xmath74 .
edges that connect these four subtwisted cubes can be described as follows : each node @xmath75 is adjacent to @xmath76 and @xmath77 if @xmath78 ; and to @xmath76 and @xmath79 if @xmath80 . according to definition [ def_twisted - cube ]
, @xmath1 is an @xmath0-regular graph with @xmath81 nodes and @xmath82 edges .
the parameter @xmath0 is always an odd integer if it is a dimension of the twisted cube .
in addition , @xmath1 is decomposed into four subtwisted cubes @xmath83 , @xmath84 , @xmath85 , @xmath86 , where @xmath72 consists of those nodes @xmath56 with @xmath87 and @xmath88 . for each @xmath89
, @xmath72 is isomorphic to @xmath71 .
for example , fig .
[ fig_tq_3 ] shows @xmath90 and fig .
[ fig_tq_5 ] depicts @xmath91 containing four subtwisted cubes @xmath92 , @xmath93 , @xmath94 , @xmath95 . .
] containing @xmath92 , @xmath93 , @xmath94 , @xmath95 . ]
let @xmath56 is a binary string @xmath96 of length @xmath97 .
we denote @xmath98 the new binary string obtained by repeating @xmath56 string @xmath61 times .
then , the length of @xmath98 is @xmath99 .
for instance , @xmath100 and @xmath101 .
obviously , @xmath90 has no two edge - disjoint hamiltonian cycles since each node is incident to three edges .
our method for constructing two edge - disjoint hamiltonian cycles of @xmath1 , with odd integer @xmath2 , is based on an inductive construction .
we will construct two edge - disjoint hamiltonian paths @xmath32 and @xmath42 in @xmath1 , with @xmath2 , such that @xmath102 , @xmath103 , @xmath104 , and @xmath105 .
the basic idea is described as follows .
initially , we construct two edge - disjoint hamiltonian paths @xmath32 and @xmath42 in @xmath91 such that @xmath106 , @xmath107 , @xmath108 , and @xmath109 . by the definition of parity function @xmath110 , @xmath111 and @xmath112 . by definition [ def_twisted - cube ] ,
@xmath113 and @xmath114 .
thus , @xmath32 and @xmath42 are two edge - disjoint hamiltonian cycles .
consider that @xmath0 is an odd integer with @xmath115 .
we first partition @xmath1 into four subtwisted cubes @xmath83 , @xmath84 , @xmath85 , @xmath86 .
assume that @xmath116 and @xmath117 are two edge - disjoint hamiltonian paths in @xmath72 , for @xmath74 , such that @xmath118 , @xmath119 , @xmath120 , and @xmath121 .
we then discover six distinct edges to concatenate these eight edge - disjoint paths into two edge - disjoint hamiltonian paths @xmath32 and @xmath42 of @xmath1 such that @xmath102 , @xmath103 , @xmath104 , and @xmath105 . by definition [ def_twisted - cube ] ,
@xmath32 and @xmath42 are two edge - disjoint hamiltonian cycles in @xmath1 since @xmath122 and @xmath123 .
the concatenating process will be shown in lemma [ 2hp - tq ] .
now , we first show that @xmath91 contains two edge - disjoint hamiltonian paths in the following lemma .
[ 2hp - tq_5 ] there are two edge - disjoint hamiltonian paths @xmath32 and @xmath42 in @xmath91 such that @xmath106 , @xmath107 , @xmath108 , and @xmath109 .
we prove this lemma by constructing such two paths .
let + @xmath12400000 @xmath125 00001 @xmath125 00101 @xmath125 00100 @xmath125 10100 @xmath125 10101 @xmath125 10001 @xmath125 10000 @xmath125 10110 @xmath125 10010 @xmath125 00010 @xmath125 00011 @xmath125 10011 @xmath125 10111 @xmath125 00111 @xmath125 00110 @xmath125 11110 @xmath125 11010 @xmath125 01010 @xmath125 01011 @xmath125 11011 @xmath125 11111 @xmath12601111 @xmath125 01110 @xmath125 01000 @xmath125 01001 @xmath125 01101 @xmath125 01100 @xmath125 11100 @xmath125 11101 @xmath125 11001 @xmath125 11000@xmath127 , and let + @xmath12800100 @xmath125 00000 @xmath125 10000 @xmath125
10100 @xmath125 10010 @xmath125 10011 @xmath125 10001 @xmath125 00001 @xmath125 00011 @xmath125 00111 @xmath125 00101 @xmath125 10101 @xmath125 10111 @xmath125 10110 @xmath125 00110 @xmath125 00010 @xmath125 01010 @xmath125 01110 @xmath125 11110 @xmath125 11111 @xmath125 11101 @xmath125 01101 @xmath125 01111 @xmath125 01011 @xmath125 01001 @xmath125 11001 @xmath125 11011 @xmath125 11010 @xmath125 11100 @xmath125 11000 @xmath125 01000 @xmath125 01100@xmath127 .
[ fig_tq_5 - 2hp ] depicts the constructions of @xmath32 and @xmath42 .
clearly , @xmath32 and @xmath42 are edge - disjoint hamiltonian paths in @xmath91
. , where solid arrow lines indicate a hamiltonian path and dotted arrow lines indicate the other edge - disjoint hamiltonian path . ] by definition [ def_twisted - cube ] , nodes 00000 and 11000 are adjacent , and nodes 00100 and 01100 are adjacent .
the following corollary immediately holds true from lemma [ 2hp - tq_5 ] .
[ 2hc - tq_5 ] there are two edge - disjoint hamiltonian cycles in @xmath91 . using lemma [ 2hp - tq_5
] , we prove the following lemma .
[ 2hp - tq ] for any odd integer @xmath2 , there are two edge - disjoint hamiltonian paths @xmath32 and @xmath42 in @xmath1 such that @xmath102 , @xmath103 , @xmath104 , and @xmath105 .
we prove this lemma by induction on @xmath0 . by lemma [ 2hp - tq_5 ]
, the lemma holds true when @xmath129 .
assume that the lemma holds when @xmath130 .
we will prove that the lemma holds true for @xmath131 .
we first partition @xmath132 into four subtwisted cubes @xmath133 , @xmath134 , @xmath135 , @xmath136 . by the induction hypothesis
, there are two edge - disjoint hamiltonian paths @xmath116 and @xmath117 , for @xmath74 , in @xmath137 such that @xmath138 , @xmath139 , @xmath140 , and @xmath141 . by the definition of parity function @xmath110 , @xmath142 , @xmath143 , and @xmath144 . according to definition [ def_twisted - cube ] , we have that + @xmath145 , @xmath146 , @xmath147 , + @xmath148 , @xmath149 , and @xmath150
. + let @xmath151 and let @xmath152 , where @xmath153 , @xmath154 , @xmath155 , and @xmath156 are the reversed paths of @xmath157 , @xmath158 , @xmath159 , and @xmath160 , respectively .
then , @xmath32 and @xmath42 are two edge - disjoint hamiltonian paths in @xmath132 such that @xmath161 , @xmath162 , @xmath163 , and @xmath164 .
[ fig_tq_k-2hp ] shows the constructions of such two edge - disjoint hamiltonian paths in @xmath132 .
thus , the lemma hods true when @xmath131 . by induction
, the lemma holds true .
, with @xmath165 , where dotted arrow lines indicate the paths and solid arrow lines indicate concatenated edges . ] by definition [ def_twisted - cube ]
, nodes @xmath102 and @xmath103 are adjacent , and nodes @xmath104 and @xmath105 are adjacent .
it immediately follows from lemma [ 2hp - tq ] that the following corollary holds true .
[ 2hc - tq ] for any odd integer @xmath2 , there are two edge - disjoint hamiltonian cycles in @xmath1 .
in this section , we will construct two equal node - disjoint cycles @xmath32 and @xmath42 in a @xmath0-dimensional twisted cube @xmath1 with an odd integer @xmath3 . our method for constructing two equal node - disjoint cycles of @xmath1
is also based on an inductive construction . for any odd integer @xmath3
, we will construct two equal node - disjoint paths @xmath32 and @xmath42 in @xmath1 such that @xmath166 , @xmath167 , @xmath168 , and @xmath169 .
the basic idea is similar to that of constructing two edge - disjoint hamiltonian paths and is described as follows .
initially , we construct two equal node - disjoint paths @xmath32 and @xmath42 in @xmath90 such that @xmath170 , @xmath171 , @xmath172 , and @xmath173 . by definition [ def_twisted - cube ] ,
@xmath32 and @xmath42 are also node - disjoint cycles with the same length since their end nodes are adjacent .
consider that @xmath0 is an odd integer with @xmath174 .
we first partition @xmath1 into four subtwisted cubes @xmath83 , @xmath84 , @xmath85 , @xmath86 .
assume that @xmath116 and @xmath117 are two equal node - disjoint paths in @xmath72 , for @xmath74 , such that @xmath175 , @xmath176 , @xmath177 , and @xmath178 .
we then concatenate them into two equal node - disjoint paths @xmath32 and @xmath42 of @xmath1 such that @xmath166 , @xmath167 , @xmath168 , and @xmath169 . by definition [ def_twisted - cube ] ,
@xmath32 and @xmath42 are also two equal node - disjoint cycles of @xmath1 since @xmath113 and @xmath114 .
the concatenating process will be presented in lemma [ 2pc - tq ] . for @xmath90 ,
let @xmath124001 @xmath125 101 @xmath125 111 @xmath125 011@xmath127 and let @xmath128000 @xmath125 100 @xmath125 010 @xmath125 110@xmath127 . then
, @xmath32 and @xmath42 are two equal node - disjoint paths in @xmath90 . by definition [ def_twisted - cube ] , @xmath113 and @xmath179 .
thus , the following lemma holds true .
[ 2pc - tq_3 ] there are two equal node - disjoint paths @xmath32 and @xmath42 in @xmath90 such that @xmath170 , @xmath171 , @xmath172 , and @xmath173 .
moreover , @xmath32 and @xmath42 are two equal node - disjoint cycles of @xmath90 .
[ 2pc - tq_3 ] there are two equal node - disjoint paths @xmath32 and @xmath42 in @xmath90 such that @xmath170 , @xmath171 , @xmath172 , and @xmath173 .
moreover , @xmath32 and @xmath42 are two equal node - disjoint cycles of @xmath90 .
using lemma [ 2pc - tq_3 ] , we prove the following lemma .
[ 2pc - tq ] for any odd integer @xmath3 , there are two equal node - disjoint paths @xmath32 and @xmath42 in @xmath1 such that @xmath166 , @xmath167 , @xmath168 , and @xmath169 .
we prove this lemma by induction on @xmath0 . by lemma [ 2pc - tq_3 ] , the lemma holds true when @xmath180 .
assume that the lemma holds when @xmath181 .
we will prove that the lemma holds true for @xmath131 .
we first partition @xmath132 into four subtwisted cubes @xmath133 , @xmath134 , @xmath135 , @xmath136 . by the induction hypothesis
, there are two equal node - disjoint paths @xmath116 and @xmath117 , for @xmath74 , in @xmath137 such that @xmath182 , @xmath183 , @xmath184 , and @xmath185 . by the definition of parity function @xmath110 , @xmath186 , @xmath187 , and @xmath188 . according to definition [ def_twisted - cube ] , we have that + @xmath145 , @xmath189 , @xmath190 , + @xmath148 , @xmath191 , and @xmath192
. + let @xmath193 and let @xmath194 , where @xmath153 , @xmath195 , @xmath155 , and @xmath196 are the reversed paths of @xmath157 , @xmath197 , @xmath159 , and @xmath198 , respectively .
then , @xmath32 and @xmath42 are two equal node - disjoint paths in @xmath132 such that @xmath199 , @xmath200 , @xmath201 , and @xmath202 .
[ fig_tq_k-2pc ] depicts the constructions of such two equal node - disjoint paths @xmath32 and @xmath42 in @xmath132 .
thus , the lemma hods true when @xmath131 . by induction
, the lemma holds true . , with @xmath203 , where dotted arrow lines indicate the paths and solid arrow lines indicate concatenated edges . ] by definition [ def_twisted - cube ] , nodes @xmath166 and @xmath167 are adjacent , and nodes @xmath168 and @xmath169 are adjacent .
it immediately follows from lemma [ 2pc - tq ] that the following corollary holds true .
[ 2cc - tq ] for any odd integer @xmath3 , there are two equal node - disjoint cycles in @xmath1 .
in this paper , we construct two edge - disjoint hamiltonian cycles ( paths ) of a @xmath0-dimensional twisted cubes @xmath1 , for any odd integer @xmath2 . on the other hand ,
we also construct two equal node - disjoint cycles ( paths ) of @xmath1 , for any odd integer @xmath3 . in the construction of two edge - disjoint hamiltonian cycles ( paths ) of @xmath1 , some edges are not used .
it is interesting to see if there are more edge - disjoint hamiltonian cycles of @xmath1 for @xmath115 .
we would like to post it as an open problem to interested readers . | the hypercube is one of the most popular interconnection networks since it has simple structure and is easy to implement .
the @xmath0-dimensional twisted cube , denoted by @xmath1 , an important variation of the hypercube , possesses some properties superior to the hypercube .
recently , some interesting properties of @xmath1 were investigated . in this paper
, we construct two edge - disjoint hamiltonian cycles in @xmath1 for any odd integer @xmath2 .
the presence of two edge - disjoint hamiltonian cycles provides an advantage when implementing two algorithms that require a ring structure by allowing message traffic to be spread evenly across the twisted cube .
furthermore , we construct two equal node - disjoint cycles in @xmath1 for any odd integer @xmath3 , in which these two cycles contain the same number of nodes and every node appears in one cycle exactly once . in other words ,
we decompose a twisted cube into two components with the same size such that each component contains a hamiltonian cycle . +
* keywords : * edge - disjoint hamiltonian cycles , equal node - disjoint cycles , twisted cubes |
recently there has been a lot of effort going into the investigation of various quantum information inequalities . we shall present a unified view that will cover most of the already known results and add some more .
the notion of metric adjusted skew information was introduced by the third author @xcite as a measure of quantum information generalizing the wigner - yanase - dyson skew informations @xcite .
the metric adjusted skew information is a convex function on the manifold of states and also satisfies other requirements , suggested by wigner and yanase , for a measure of the information content of a state with respect to a conserved observable .
these requirements include additivity with respect to the aggregation of isolated subsystems and time independence for an isolated system .
one more requirement , super additivity , fails spectacularly @xcite for all the measures , including the wigner - yanase skew information .
we introduce an order relation @xmath0 on the set of functions representing quantum fisher information that renders the set into a lattice with a maximal and a minimal element .
the order relation induces an order relation on the set of metric adjusted skew informations that renders also this set into a lattice with the sld - information as maximal element .
there is no minimal element .
the wigner - yanase - dyson skew informations given by @xmath1[\rho^{1-p},a]\qquad 0<p<1\ ] ] are shown to be increasing in the parameter @xmath2 with respect to this new order relation in the interval @xmath3 $ ] and decreasing in the interval @xmath4 there is thus maximum in the wigner - yanase skew information for @xmath5 more elementary , the function @xmath6 is increasing in @xmath7 $ ] and decreasing in @xmath8 for fixed @xmath9 and @xmath10 the metric adjusted skew information is defined by setting @xmath11 c(l_\rho , r_\rho)i[\rho , a]\ ] ] for every positive definite @xmath12 density matrix @xmath9 and every @xmath13 matrix @xmath14 the function @xmath15 of two variables is a so called regular morozova - chentsov function which may be written on the form @xmath16 where @xmath17 is an operator monotone function defined on the positive half - axis with @xmath18 satisfying the functional equation @xmath19 for @xmath20 the regularity condition means that @xmath21 the operators @xmath22 and @xmath23 are the commuting positive definite left- and right- multiplication operators by @xmath24 if we want to emphasize the dependence of the representing operator monotone function @xmath17 we may also denote the metric adjusted skew information by @xmath25 the metric adjusted skew information is thus proportional to the metric length , as measured by the quantum fisher information , of the commutator @xmath26 .
$ ] in the case of the wigner - yanase - dyson information this proportionality was already noticed in @xcite .
the constant is however important as it can be chosen such that the metric adjusted skew information coincides with the variance on pure states .
we have tacitly extended the definition of the metric adjusted skew information to include the case where @xmath27 may not be self - adjoint .
this does not directly make sense in physical applications , but it turns out to be a useful mathematical tool .
the symmetry of @xmath15 implies that @xmath28 for self - adjoint @xmath27 and @xmath29 in particular , the metric adjusted skew information is convex in @xmath9 also for non self - adjoint `` observables '' .
the basic quantum information inequality is given by @xmath30 where the symmetrized variance @xmath31 the inequality was stated and proved in @xcite for self - adjoint @xmath27 by noting that the metric adjusted skew information is a convex function in @xmath9 while the covariance is concave , and that the two measures coincide on pure states .
gibilisco et al .
* proposition 9.2 ) subsequently gave another proof .
the inequality generalizes an earlier result by luo for the wigner - yanase skew information @xcite .
since the symmetrized variance satisfies @xmath32 for self - adjoint @xmath27 and @xmath33 we immediately obtain the inequality ( [ basic quantum information inequality ] ) also for non self - adjoint @xmath14 the above observations effectively reduce the multi - dimensional versions of the quantum information inequalities given by a number of authors @xcite to the main inequality ( [ basic quantum information inequality ] ) .
let @xmath34 be a tuple of ( possibly non self - adjoint ) matrices .
we then obtain the inequality @xmath35 where @xmath36 c(l_\rho , r_\rho)i[\rho , b]\ ] ] and @xmath37 indeed , we only have to notice the identity @xmath38 for arbitrary complex numbers @xmath39 and the similar identity for the covariance matrix .
this reduces the matrix version ( [ matrix inequality ] ) to the basic quantum information inequality ( [ basic quantum information inequality ] ) .
in particular , we obtain the determinant inequality @xmath40 by the well known formula @xmath41 since the logarithm is operator monotone is increasing for any increasing function @xmath42 .
this version has been coined the `` dynamic uncertainty principle '' .
we introduce various methods to compare measures of quantum information for one observable .
the generalization to several observables is then obtained as explained in the introduction
. the metric adjusted skew information may be written ( * ? ? ?
* proposition 3.4 ) on the form @xmath43 where @xmath44 is operator concave . since @xmath45 is homogeneous of degree one we may write @xmath46 where the function @xmath47 since @xmath48 is operator concave and defined on the positive half - axis it is also operator monotone @xcite . with this notation the metric adjusted skew
information takes the form @xmath49 as it is studied in @xcite .
then @xmath17 admits a canonical representation @xmath54 where the weight function @xmath55\to[0,1 ] $ ] is measurable .
the equivalence class containing @xmath56 is uniquely determined by @xmath57 any function on the given form maps the positive half - axis into itself and satisfy the conditions @xmath58 @xmath59 and @xmath60 the third author ( * ? ? ?
* theorem 1 ) gave an exponential representation of the functions @xmath61 satisfying @xmath62 and @xmath59 of the form @xmath63 where @xmath55\to[0,1 ] $ ] is measurable and @xmath64 setting @xmath65 and solving for @xmath66 we obtain @xmath67 which reduces to the expression in the theorem . the uniqueness of @xmath68 ( up to equivalence ) and the sufficiency of the condition follows from the reference .
following several authors we denote by @xmath69 the set of functions characterized in theorem [ theorem : canonical representation of f ] .
a function @xmath70 is said to be regular if its extension to the closed positive half line satisfies @xmath71 if @xmath72 is regular we set @xmath73 and may write @xmath74 \check c(l_\rho , r_\rho)i[\rho , a].\ ] ] notice that @xmath75 is a symmetric function in two positive variables , and that the metric adjusted skew information @xmath76 is increasing in the transform @xmath77 cf . also @xcite .
the representing function @xmath78 of the wigner - yanase - dyson skew information @xmath79 with parameter @xmath2 is given by @xmath80 we observe that the transform @xmath81 is increasing in @xmath82 $ ] and decreasing in @xmath83 it follows that the wigner - yanase - dyson skew information @xmath84 for fixed @xmath9 and @xmath85 is an increasing function of @xmath2 in the interval @xmath7 $ ] and a decreasing function of @xmath2 in the interval @xmath8 with maximum in the wigner - yanase skew information . [ representation of tilde f ] the transform @xmath86 of a regular function @xmath70 has a canonical representation @xmath87 where the weight function @xmath55\to[0,1 ] $ ] is measurable and @xmath88 the equivalence class containing @xmath56 is uniquely determined by @xmath89 any function on the given form is the transform @xmath86 of a regular function @xmath17 in @xmath90 the integrability condition for a weight function @xmath56 of a regular function @xmath91 follows from theorem [ theorem : canonical representation of f ] .
furthermore , @xmath92 and thus @xmath93 which reduces to the expression in the proposition .
the function @xmath99 obviously satisfy @xmath100 and @xmath101 for @xmath20 the condition in the definition is thus equivalent to operator monotonicity of @xmath102 setting @xmath103 it is known that @xmath104 for every function @xmath105 but we realize that also @xmath106 this is so since @xmath107 is reduced to operator monotonicity of the function @xmath108 while @xmath109 is reduced to operator monotonicity of @xmath17 itself .
since as noted @xmath110 for @xmath111 we realize that @xmath112 in particular , @xmath97 and @xmath113 implies @xmath114 [ theorem : partial order relation on f_op ] the relation @xmath0 is a partial order relation on @xmath115 rendering @xmath116 into a lattice with @xmath117 as the minimal element and @xmath118 as the maximal element .
furthermore , if @xmath119 then @xmath120 where @xmath121 and @xmath122 respectively are representing functions of @xmath17 and @xmath123 as in theorem [ theorem : canonical representation of f ] .
the function @xmath99 in definition [ definition : preceq ] has the representation @xmath124 and is therefore operator monotone , according to theorem [ theorem : canonical representation of f ] , if and only if @xmath125 for almost all @xmath126 .
$ ] therefore , @xmath97 if and only if @xmath127 almost everywhere .
it follows that @xmath0 is an order relation . indeed , if @xmath128 and @xmath129 then , for representing functions , @xmath130 and @xmath131 almost everywhere . consequently ,
@xmath132 almost everywhere and thus @xmath133 for arbitrary @xmath95 with representing functions @xmath121 and @xmath122 we define @xmath134 as the function in @xmath69 with representing function @xmath135 similarly , we define @xmath136 as the function in @xmath69 with representing function @xmath137 it follows that @xmath138 if furthermore @xmath139 and @xmath140 for a @xmath141 it follows that @xmath142 if similarly @xmath143 and @xmath144 it follows that @xmath145 therefore @xmath146 is a lattice .
it also follows that @xmath154 and that @xmath155 is the unique fixpoint of the involution @xmath156 in @xmath157 furthermore , @xmath97 implies @xmath158 for functions @xmath159 we notice that @xmath150 is not regular for a regular @xmath160 all of these assertions may also be verified by calculating the representing weight functions .
[ corollary : order relation on metric adjusted skew informations ] the restriction of the order relation @xmath0 to the regular part of @xmath69 induces an order relation on the set of metric adjusted skew informations . the order relation on the set of metric adjusted skew informations introduced in corollary [ corollary : order relation on metric adjusted skew informations ] is tractable in the sense that we only have to study and compare the representing functions of the associated quantum fisher informations as they are given in theorem [ theorem : canonical representation of f ] .
the function @xmath167 representing the wigner - yanase - dyson skew information with parameter @xmath168 has weight function @xmath169 according to the representation given in theorem [ theorem : canonical representation of f ] , cf .
* theorem 2.7 ) .
we consider a fixed @xmath173 set @xmath174 for a small @xmath175 and obtain @xmath176 where @xmath177 & = \arg(1-z_0^p)+\arg(1-z_0^{1-p } ) .
\end{array}\ ] ] obviously , we have @xmath178 .
thus , to show that @xmath179 is increasing over the left half @xmath180 $ ] of the domain , it suffices to show that the first term @xmath181 is concave in @xmath182 over the interval @xmath183 $ ] .
we can show that the function @xmath184 is concave over @xmath183 $ ] for any @xmath185 in the `` lune '' @xmath186 , including @xmath187 for sufficiently small @xmath188 since @xmath189 , the second derivative @xmath190 is given by @xmath191 and we have to show that this is non - positive for @xmath192 and @xmath193 .
in fact , it is enough to do this for @xmath194 only , since for @xmath195 and @xmath193 , then @xmath196 too . the imaginary part of a complex number is non - positive if and only its complex argument is between @xmath197 and @xmath198 .
thus we need to show @xmath199\qquad \forall z\in\mathcal i.\ ] ] putting @xmath200 , with @xmath201 and @xmath202 , @xmath203 & \displaystyle = \pi+ \theta + 2\arctan\frac{\theta}{\log r } + 2\arctan\frac{r\sin\theta}{1-r\cos\theta}\ , .
\end{array}\ ] ] for @xmath204 ( @xmath185 real ) @xmath205 is obviously @xmath206 . for @xmath207 @xmath208 for @xmath209 .
we can now show that for fixed @xmath210 , @xmath211 increases with @xmath212 , which implies that @xmath213 is indeed between @xmath197 and @xmath198 for @xmath214 .
the first derivative is @xmath215 because of the inequality @xmath216 , we obtain a lower bound @xmath217 & = \displaystyle\frac{\phi(r)\log r + \theta^2 \psi(r)}{((1-r)^2+r\theta^2)((\log r)^2+\theta^2 ) } \end{array}\ ] ] where @xmath218 \psi(r)&=1-r^2 + 2r\log r. \end{array}\ ] ] the first derivative @xmath219 and the second derivative @xmath220 is positive .
therefore @xmath221 is increasing and since @xmath222 we derive that @xmath221 is negative and thus @xmath223 is decreasing .
but since @xmath224 we derive that @xmath225 for @xmath226 we furthermore calculate @xmath227 \phi''(r)&=\displaystyle\frac{1}{r^2}(r^2 - 2r^2\log r-1 ) .
\end{array}\ ] ] the parenthesis in the expression for @xmath228 has derivative @xmath229 and value zero in @xmath230 it is thus non - positive , so @xmath231 is concave and @xmath232 therefore decreasing . since @xmath233 we derive that @xmath234 so @xmath231 is increasing and since @xmath235 we conclude that @xmath231 is non - positive .
we have consequently proved @xmath236 for @xmath237 and @xmath238 and the proof is complete .
the graphs for different @xmath240 look very much alike except for a change in units . an equivalent formulation of the result is that the function @xmath241 is operator monotone ( and thus belongs to @xmath242 for @xmath243 we have thus proved that @xmath244 for @xmath245 and @xmath246 for @xmath247 the wigner - yanase skew information is therefore the maximal element among the wigner - yanase - dyson informations with respect to the order relation inherited from @xmath248 another example is the variant bridge considered in @xcite with representing functions @xmath249 with weight functions according to the representation in theorem [ theorem : canonical representation of f ] given by @xmath250 p , & \lambda&\ge 1-p \end{array}\right.\qquad 0\le p\le 1.\ ] ] this family of weight functions is decreasing in the parameter @xmath251 .
$ ] the corresponding quantum fisher informations are therefore increasing in the parameter @xmath2 and connects the sld - metric for @xmath252 with the bures metric for @xmath253 and they are regular for @xmath254 the corresponding metric adjusted skew informations , defined for @xmath255 are increasing with respect to the order relation @xmath256 koenraad audenaert : mathematics department of royal holloway , university of london egham hill , egham tw20 0ex , uk . + liang cai : academy of mathematics and systems science , chinese academy of sciences , beijing 100080 , china . + frank hansen : department of economics , university of copenhagen , studiestraede 6 , dk-1455 copenhagen k , denmark . | we study quantum information inequalities and show that the basic inequality between the quantum variance and the metric adjusted skew information generates all the multi - operator matrix inequalities or robertson type determinant inequalities studied by a number of authors .
we introduce an order relation on the set of functions representing quantum fisher information that renders the set into a lattice with an involution .
this order structure generates new inequalities for the metric adjusted skew informations .
in particular , the wigner - yanase skew information is the maximal skew information with respect to this order structure in the set of wigner - yanase - dyson skew informations .
+ * key words and phrases : * quantum covariance , metric adjusted skew information , robertson - type uncertainty principle , operator monotone function , wigner - yanase - dyson skew information . |
khovanov @xcite introduced a cohomology theory for oriented classical links which values in graded @xmath1-modules and whose graded euler characteristic is the jones polynomial .
we denote khovanov s cohomology groups of an oriented link @xmath2 by @xmath3 @xmath4 .
his theory is powerful for classical links ; for example , bar - natan @xcite and wehrli @xcite showed that khovanov s cohomology is stronger than the jones polynomial , and rasmussen @xcite gave a combinatorial proof of the milnor conjecture by using a variant of khovanov s theory defined by lee @xcite .
jacobsson @xcite and khovanov @xcite proved that khovanov s theory is functorial for link cobordisms in the following sense : a link cobordism @xmath5 $ ] between classical links @xmath6 and @xmath7 induces a homomorphism @xmath8 , well - defined up to overall minus sign , under ambient isotopy of @xmath9 rel @xmath10 .
a _ surface - knot _
@xmath11 is a closed connected oriented surface embedded locally flatly in @xmath12 , and can be considered as a link cobordism between empty links .
then the induced map @xmath13 , up to overall minus sign , gives an invariant of the surface - knot @xmath11 .
since the cohomology group @xmath14 of empty link @xmath15 is @xmath1 , the map @xmath16 is an endomorphism of @xmath1 .
hence we obtain an invariant of the surface - knot @xmath11 defined as @xmath17 and denote it by @xmath18 .
this invariant is called the _ khovanov - jacobsson number _ in @xcite .
as far as the author knows , there are a few result on the computation of the khovanov - jacobsson numbers of surface - knots ; see @xcite , for example .
it follows from a simple observation that @xmath19 for any surface - knot @xmath11 with @xmath20 and that @xmath21 for a trivial @xmath0-knot @xmath11 ( a surface - knot with @xmath22 ) .
it seems to be hard to compute the khovanov - jacobsson number in general , but carter , saito and satoh @xcite proved that @xmath21 for any @xmath0-knot @xmath11 obtained from a spun / twist - spun @xmath23-knot by attaching a @xmath24-handle .
they also proved that @xmath21 for any pseudo - ribbon @xmath0-knot @xmath11 . in this paper
, we define an invariant @xmath25 $ ] of a surface - knot @xmath11 by using a variant of khovanov s theory defined by bar - natan @xcite .
this invariant is a generalization of the khovanov - jacobsson number such that @xmath26 the main result of this paper is that the invariant @xmath27 is trivial for any surface - knot @xmath11 , and hence it turns out that the khovanov - jacobsson number is trivial for any @xmath0-knot using lee s theory . ] : [ th - main ] for any surface - knot @xmath11 of genus @xmath28 @xmath29 , we have the following . * if @xmath28 is an even integer , then we have @xmath30 . *
if @xmath28 is an odd integer , then we have @xmath31 .
[ cor - kj ] for any @xmath0-knot , we have @xmath21 .
this paper is organized as follows . in section [ sec - kh ] and [ sec - kj ] , we review khovanov s cohomology theory for oriented classical links and the khovanov - jaconsson numbers of surface - knots respectively . in section [ sec - bn ]
, we define the surface - knot invariant @xmath27 . section [ sec - pf ] is devoted to the proof of theorem [ th - main ] .
in this section , we briefly recall khovanov s cohomology theory for oriented classical links @xcite .
see also @xcite , @xcite , ( * ? ? ?
* section @xmath32 ) , ( * ? ? ?
* section @xmath32 ) , for example .
let @xmath33 be a free graded @xmath1-module of rank two generated by @xmath35 and @xmath36 with @xmath37 we give the graded @xmath1-module @xmath33 a frobenius algebra structure with a multiplication @xmath38 , a comultiplication @xmath39 , a unit @xmath40 , and a counit @xmath41 defined by @xmath42 the structure maps @xmath38 , @xmath39 , @xmath40 and @xmath41 are graded maps of degree @xmath43 , @xmath43 , @xmath24 and @xmath24 respectively .
khovanov s cohomology theory is based on a @xmath44-dimensional tqft @xmath34 , a monoidal functor from oriented @xmath44-cobordisms to graded @xmath1-modules , associated to @xmath33 .
the frobenius algebra @xmath33 defines @xmath34 by assigning @xmath1 to an empty @xmath24-manifold , @xmath33 to a single circle , @xmath45 to a disjoint union of two circles , and so on .
the structure maps are assigned to elementary cobordisms such that @xmath46{cob_2 - 1.eps } \end{minipage}\right)= m : v \otimes v \rightarrow v , & \mathcal{f}\left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{cob_0 - 1.eps } \end{minipage}\right)= \iota : { \ensuremath{\mathbb{z}}}\rightarrow v , \\ \mathcal{f}\left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{cob_1 - 2.eps } \end{minipage}\right)= \delta : v \rightarrow v \otimes v , & \mathcal{f}\left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{cob_1 - 0.eps } \end{minipage}\right)= \epsilon : v \rightarrow { \ensuremath{\mathbb{z}}}\ .\end{array}\ ] ] let @xmath2 be an oriented link , and @xmath47 an oriented link diagram of @xmath2 with @xmath48 crossings labeled by @xmath49 . a double point of @xmath47 can be resolved in two ways : one is the @xmath50-smoothing ( 10,10 ) ( 1,1)(1,1)(9,9 ) ( 1,9)(1,9)(3,7 ) ( 7,3)(7,3)(9,1 ) @xmath51 ( 10,10 ) ( 1,1)(7,5)(1,9 ) ( 9,1)(3,5)(9,9 ) , and the other is the @xmath24-smoothing ( 10,10 ) ( 1,1)(1,1)(9,9 ) ( 1,9)(1,9)(3,7 ) ( 7,3)(7,3)(9,1 ) @xmath51 ( 10,10 ) ( 1,1)(5,7)(9,1 ) ( 1,9)(5,3)(9,9 ) .
we construct an @xmath48-dimensional cube @xmath52^n$ ] , called the _ cube of resolutions _ , from @xmath47 such that each vertex @xmath53 is decorated by a complete resolution of @xmath47 and each edge @xmath54 is decorated by a cobordism . to each vertex @xmath53 of the cube
, we associate the complete resolution @xmath55 as follows : the @xmath56th crossing is resolved by the @xmath50-smoothing if the @xmath56th coordinate of @xmath53 is @xmath50 , and by the @xmath24-smoothing if it is @xmath24 .
then @xmath55 is a collection of simple closed curves . each edge @xmath54 of the cube
can be represented by sequences in @xmath57 with just one @xmath58 , and has the two end vertices @xmath59 and @xmath60 : the vertex @xmath59 is obtained by substituting @xmath50 to @xmath58 , and the vertex @xmath60 is obtained by substituting @xmath24 to @xmath58 .
to an edge @xmath54 for which the @xmath61th coordinate is @xmath58 , we associate the cobordism @xmath62 as follows : we remove a neighborhood of the @xmath61th crossing , assign a product cobordism , and fill the saddle cobordism between the @xmath50- and @xmath24-smoothings around the @xmath61th crossing .
the cobordism @xmath63 is either of the following two types : ( i ) two circles of @xmath64 merge into one circle of @xmath65 , or ( ii ) one circle of @xmath64 splits into two circles of @xmath65 .
applying @xmath34 to the cube of resolutions , we construct another @xmath48-dimensional cube @xmath52^n$ ] , called the _ cube of modules_. each vertex @xmath53 is replaced by a graded @xmath1-module @xmath66 and each edge @xmath54 is replaced by a homomorphism @xmath67 the homomorphism @xmath68 is induced by a map @xmath69 if the cobordism @xmath63 is of type ( i ) , and is induced by a map @xmath70 if it is of type ( ii ) . for a graded @xmath1-module @xmath71 and an integer @xmath72 , let @xmath73 denote the graded @xmath1-module of which the @xmath61th graded component is the @xmath74th graded component of @xmath71 .
we note that @xmath75 is identical to @xmath71 as @xmath1-modules .
khovanov s cochain complex @xmath76 of graded @xmath1-modules is obtained from the cube of modules as follows .
the underlying group of @xmath76 @xmath77 is defined by @xmath78 where @xmath79 is the sum of all coordinates of @xmath53 and @xmath80 ( resp .
@xmath81 ) is the number of positive ( resp .
negative ) crossings of @xmath47 .
the coboundary map @xmath82 for an element @xmath83 of @xmath66 is defined by @xmath84 where @xmath85 is the sum of all coordinates of an edge @xmath54 after @xmath58 .
we give some remarks about the coboundary map @xmath82 in section [ subsec - remark ] .
the cohomology groups @xmath86 of the cochain complex @xmath76 are graded @xmath1-modules .
khovanov @xcite proved that @xmath76 and @xmath87 are cochain homotopic to each other for any other diagram @xmath88 of @xmath2 . to prove it
, he constructed cochain maps @xmath89{r1.eps } \end{minipage } , \mathcal{f } \right ) \rightarrow c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r1-line.eps } \end{minipage } , \mathcal{f } \right ) , & g_1 : c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r1-line.eps } \end{minipage } , \mathcal{f } \right ) \rightarrow c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r1.eps } \end{minipage } , \mathcal{f } \right ) , \\ f_2 : c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r2.eps } \end{minipage } , \mathcal{f } \right ) \rightarrow c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r2 - 01.eps } \end{minipage } , \mathcal{f } \right ) , & g_2 : c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r2 - 01.eps } \end{minipage } , \mathcal{f } \right ) \rightarrow c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r2.eps } \end{minipage } , \mathcal{f } \right ) , \\ f_3 : c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r3-u.eps } \end{minipage } , \mathcal{f } \right ) \rightarrow c \left(\begin{minipage}{15pt } \includegraphics[width=15pt]{r3-d.eps } \end{minipage } , \mathcal{f } \right ) , & g_3 : c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r3-d.eps } \end{minipage } , \mathcal{f } \right ) \rightarrow c \left ( \begin{minipage}{15pt } \includegraphics[width=15pt]{r3-u.eps } \end{minipage } , \mathcal{f } \right ) , \end{array}\ ] ] for reidemeister moves such that the maps @xmath90 and @xmath91 are cochain homotopic to the identities for each @xmath56 @xmath92 , but we omit the precise definition of these maps .
( refer to @xcite for more details . )
this implies that the isomorphism classes of the cohomology groups @xmath86 are invariants of @xmath2 .
then we denote cohomology groups of an oriented link @xmath2 by @xmath3 @xmath4 .
we give some remarks about khovanov s cochain complex . *
the frobenius algebra @xmath33 has the following properties : the map @xmath38 is associative , the map @xmath39 is coassociative , and @xmath93 .
these properties can be interpreted as the commutativity of saddle point moves in oriented @xmath44-cobordisms . *
it follows from ( i ) that each two dimensional face of the cube of modules is commutative .
if we replace the map @xmath94 by the map @xmath95 for each edge @xmath54 , then each two dimensional face becomes anti - commutative . *
the above ( anti-)commutativity of the cube of modules ensures that the map @xmath82 satisfies @xmath96 .
a _ link cobordism _ @xmath9 between classical links @xmath6 and @xmath7 is a compact oriented surface embedded properly and locally flatly in @xmath97 $ ] such that @xmath98 for each @xmath56 @xmath99 . in this section
, we briefly recall the map @xmath100 induced by the link cobordism @xmath9 and the khovanov - jaconsson numbers of surface - knots ( cf .
@xcite ) . for a fixed projection @xmath101 \rightarrow { \ensuremath{\mathbb{r}}}^2 \times [ 0,1]$ ] ,
each intersection @xmath102 of a link cobordism @xmath9 is called a _ still _ and denoted by @xmath103 for each @xmath104 @xmath105)$ ] . by perturbing @xmath9 if necessary , we may assume that each still is a classical link diagram for all but finitely many critical values and contains at most one singular point .
when @xmath104 passes through a critical value , we see one of the following changes , called _ elementary string interaction_s ( or _ esi_s , for short ) : reidemeister moves ( r1 , r2 and r3 ) and morse moves ( birth , death and saddle ) .
the collection @xmath106}$ ] of stills is called a _ movie _ of @xmath9 .
refer to @xcite for more details .
we construct a map @xmath107 for a link cobordism @xmath9 represented by a single elementary string interaction . for the @xmath56th reidemeister move , the map @xmath108 is defined to be @xmath109 or @xmath110 for each @xmath56 @xmath92 . for a birth ( resp .
death ) move , the map @xmath108 is induced by @xmath40 ( resp .
@xmath41 ) . for a saddle move ,
the map @xmath108 is induced by either @xmath38 or @xmath39 on each component of the cochain complex , depending on whether the saddle move merges two circle or splits one circle into two circles at the cube of resolutions level .
we note that the map @xmath108 is a graded map of degree @xmath50 for a reidemeister move , degree @xmath24 for a birth / death move , and degree @xmath43 for a saddle move . for a link cobordism @xmath9 , take a movie presentation of @xmath9 .
as mentioned above , the movie presentation of @xmath9 is represented as a collection of elementary string interactions : @xmath111 .
the map @xmath108 is defined to be the composite @xmath112 of the maps induced by elementary cobordisms @xmath113 .
then the map @xmath108 is a graded map of degree @xmath114 , where @xmath115 is the euler characteristic of @xmath9 .
jacobsson @xcite and khovanov @xcite proved that a link cobordism @xmath9 between classical links @xmath6 and @xmath7 induces a homomorphism @xmath116 well - defined up to overall minus sign , under ambient isotopy of @xmath9 rel @xmath10 .
( jacobsson @xcite pointed out that if we do not assume rel @xmath10 , then the statement does not hold in general . ) a _ surface - knot _
@xmath11 is a closed connected oriented surface embedded locally flatly in @xmath12 .
when we consider the surface - knot @xmath11 as an oriented cobordism between empty links , the induced map @xmath13 , up to overall minus sign , gives an invariant of @xmath11 .
since the cohomology group @xmath14 of empty link @xmath15 is @xmath1 , the map @xmath16 is an endomorphism of @xmath1 .
hence we obtain an invariant of the surface - knot @xmath11 defined as @xmath17 and denote it by @xmath18 .
this invariant is called the _ khovanov - jacobsson number _ in @xcite .
it is easy to see that @xmath19 for any surface - knot @xmath11 with @xmath20 , since the map @xmath16 is a graded map of degree @xmath117 .
bar - natan @xcite defined several variants of khovanov s theory .
let @xmath118 be a free graded @xmath119$]-module of rank two generated by @xmath35 and @xmath36 with @xmath120 we give @xmath118 a frobenius algebra structure with a multiplication @xmath121 , a comultiplication @xmath122 , a unit @xmath123 , and a counit @xmath124 defined by @xmath125 one of his cohomology theories , implicitly defined in ( * ? ? ?
* section 9.2 ) , is based on a @xmath44-dimensional tqft @xmath126 associated to a frobenius algebra @xmath118 , and we denote cohomology groups of an oriented link @xmath2 by @xmath127 @xmath128 . here
the cohomology groups @xmath86 are graded @xmath119$]-modules .
we note that the cohomology theory associated to @xmath126 is essentially the same as that associated to the frobenius system @xmath129 in @xcite , but the notational conventions are slightly different .
he proved that the cohomology theory associated to @xmath126 is also functorial for link cobordisms up to sign indeterminacy ( cf .
@xcite , ( * ? ? ?
* proposition @xmath130 ) ) .
given a surface - knot @xmath11 , the induced map @xmath131 becomes an endomorphism of @xmath119 $ ] .
hence we obtain an invariant of the surface - knot @xmath11 defined as @xmath132,\ ] ] and denote it by @xmath27 .
it follows from the definition of @xmath33 and @xmath118 that bar - natan s theory recovers khovanov s theory by adding the relation @xmath133 , and hence we have @xmath26 we remark here that bar - natan s theory also recovers lee s theory @xcite by adding the relation @xmath134 .
for a surface - knot @xmath11 , taking an arbitrary point @xmath135 of @xmath11 and cutting off a small neighborhood of @xmath135 which is homeomorphic to the standard disk pair @xmath136 , we obtain a link cobordism between an empty link and a trivial knot
. then we can define the following two maps @xmath137 where @xmath138 stands for a trivial knot and the cohomology group @xmath139 of a trivial knot is @xmath118 , and these two maps satisfy @xmath140 for the connected sum @xmath141 of two surface - knots @xmath142 and @xmath143 , the map @xmath144 can be decomposed into the composite of two maps such that @xmath145 the following two lemmas are direct consequences of the fact that @xmath146 we note that the map @xmath147 corresponds to a link cobordism between trivial knots induced by a trivial @xmath0-knot with two holes .
[ lem - odd ] if the surface - knot @xmath11 of genus @xmath148 @xmath149 is trivial , then we have @xmath150 .
[ lem - even ] if the surface - knot @xmath11 of genus @xmath151 @xmath149 is trivial , then we have @xmath152 since the map @xmath153 induced by a surface - knot @xmath11 is a graded map of degree @xmath117 and the degree of @xmath104 is @xmath154 , it is easy to see the following : * if the genus of a surface - knot @xmath11 is @xmath151 @xmath155 , then we have @xmath30 . *
if the genus of a surface - knot @xmath11 is @xmath148 @xmath155 , then there exists some nonnegative integer @xmath156 such that @xmath157 .
it is sufficient to prove that the above integer @xmath156 is equal to @xmath158 for any surface - knot @xmath11 of genus @xmath148 .
it follows from @xmath159 that @xmath160 let @xmath161 denote a trivial surface - knot of genus @xmath28 .
we consider the connected sum @xmath162 of @xmath11 and @xmath163 for a nonnegative integer @xmath121 . by lemma [ lem - even ]
, we have @xmath164 and hence we have @xmath165 . if we take the integer @xmath121 such that @xmath166 is greater than the unknotting number @xcite of @xmath11 , then the surface - knot @xmath162 is ribbon - move equivalent to @xmath167 . when two surface - knots are related by ribbon - moves , it is known that the induced maps on the cohomology groups are the same ( cf .
hence we have @xmath168 by lemma [ lem - odd ] .
this implies @xmath169 .
the author would like to express his sincere gratitude to yukio matsumoto for encouraging him .
he would also like to thank seiichi kamada , masahico saito , shin satoh for helpful comments , magnus jacobsson for telling me that jacob rasmussen has a different proof of corollary [ cor - kj ] , isao hasegawa and yoshikazu yamaguchi for stimulating discussions on an early version of this study . | khovanov introduced a cohomology theory for oriented classical links whose graded euler characteristic is the jones polynomial . since khovanov s theory is functorial for link cobordisms between classical links , we obtain an invariant of a surface - knot , called the _ khovanov - jacobsson number _ , by considering the surface - knot as a link cobordism between empty links . in this paper
, we define an invariant of a surface - knot which is a generalization of the khovanov - jacobsson number by using bar - natan s theory , and prove that any @xmath0-knot has the trivial khovanov - jacobsson number . |
in this supplementary material we show how the multi - configurational time - dependent hartree for bosons ( mctdhb ) theory attacks time - dependent and static problems , and how the many - body theory boils down to the multi - orbital mean - field .
these relations allow us to understand that the self - induced effective potentials @xmath79 introduced in the main manuscript at the mean - field level have the many - body origin behind .
then we show how to analyze the many - body solutions at hand and demonstrate how a normal ( standard ) condensate is transforming to multi - hump fragmented state with an increase of the inter - particle repulsion .
finally , we address the question whether it is possible to verify the predicted physics within presently available experimental setups .
the multi - configurational time - dependent hartree method for bosons ( mctdhb ) is used to solve the time - dependent schrdinger equation : @xmath80 here @xmath8 and @xmath81 is the generic many - body hamiltonian defined in the main text .
the mctdhb _ ansatz _ for the many - body wave - function @xmath82 is taken as a linear combination of time - dependent permanents : @xmath83 where the summation runs over all possible configurations whose occupations @xmath84 preserve the total number of bosons @xmath3 .
so , the fock state @xmath55 studied in the main text is one of the terms in this sum .
the expansion coefficients @xmath85 and shapes of the orbitals @xmath86 are variational time - dependent parameters of the mctdhb method , determined by the time - dependent variational principle . to solve a time - dependent problem means to specify an initial many - body state and to find how the many - body function evolves .
the initial mctdhb many - body state is given by initial expansion coefficients @xmath87 and by initial shapes of the orbitals @xmath88 . to determine their evolution one has to solve the respective governing equations which we list below .
the equations - of - motion for expansion coefficients @xmath85 read : @xmath89 where @xmath90 is the hamiltonian matrix with elements @xmath91 .
the equations - of - motion for the orbitals @xmath92 , @xmath93 are : @xmath94 , \eqno{\mathrm { ( s2)}}\ ] ] where @xmath95 is the projector , @xmath96 and @xmath97 are elements of the reduced one- and two - body density matrices available at every point in time from the many - body wave - function @xmath82 : @xmath98 @xmath99 @xmath100 @xmath101 and @xmath102 are _ local _ time - dependent potentials which we analyze later . here
we are interested in static solutions , i.e. , in many - body eigenstates of the time - independent schrdinger equation . within the mctdhb theory
they are obtained by propagating the mctdhb equations - of - motion in imaginary time .
this procedure relaxes an initial state to an eigenstate , typically to the ground - state , and , thus , is called relaxation . to find many - body eigenstates of the time - independent schrdinger equation , i.e.
, static solutions , one can boil down the full time - dependent mctdhb theory to its time - independent version to multi - configurational hartree for bosons ( mchb ) method [ 27 ] . from mathematical point of view
this is done by switching off all time - dependency in the respective equations of motion ( s1 , s2 ) .
the equations - of - motion for the expansion coefficients eq .
( s1 ) are reduced to standard diagonalization procedure : @xmath103 where @xmath104 is the total energy corresponding to the eigenstate @xmath105 .
the static version of the equations - of - motion ( s2 ) can be rewritten without projections : @xmath106 where @xmath107 are scaled lagrange multipliers introduced to ensure orthonormality of the orbitals ( fragments ) @xmath108 .
the optimal shapes of the now time - independent orbitals @xmath108 , @xmath93 and optimal @xmath105 have to be determined by solving the coupled system of eqs .
( s5 , s6 ) self - consistently .
now our goal is to demonstrate how the full many - body mctdhb theory boils down to the multi - orbital ( best ) mean - field ( bmf ) . from formal mathematical point of view
this reduction is reached by taking a single term ( permanent ) in the mctdhb ansatz eq .
( s0 ) and putting all other to zero .
physically it means that the many - body solution is assumed to be well - described by @xmath74 separated sub - clouds ( fragments ) carrying @xmath109 bosons .
moreover , it also implies that there is no exchange of bosons between the fragments .
the multi - orbital mean field ansatz reads : @xmath110 the mc(td)hb equations - of - motions for the expansion coefficients eqs .
( s1 , s5 ) become redundant , because we have only a single expansion coefficient .
all off - diagonal elements of @xmath96 and @xmath97 corresponding to the single - permanent state are zero , while the diagonal ones are known in advance @xmath111 , @xmath112 and @xmath113 .
the above facts considerably simplify the static equations - of - motion ( s6 ) for orbitals @xmath108 , @xmath93 : @xmath114 \left|\phi_j\right > + \sum^m_{i \ne j } n_j \lambda_0 \hat{w}_{ij } \left|\phi_i\right > = \sum^m_{i=1 } \mu_{ji } \left|\phi_i\right>. \eqno{\mathrm { ( s7)}}\ ] ] hence , for a given set of the occupations numbers @xmath115 by solving this system one determines the optimal shapes of the fragments .
the _ local _ potentials @xmath116 defined in eq .
( s4 ) are generally complicated time - dependent objects which depend on inter - particle interaction and on shapes of the involved time - dependent orbitals @xmath92 , @xmath93 . however , in the static cases , see eqs .
( s6 , s7 ) , they become time - independent functions : @xmath117 in the present study the mean - field physics dictates localization of the fragments @xmath108 , @xmath93 at different places of the trap potential ( see e.g. fig . 2 ) .
it also means that the overlap between the different fragments is negligible .
hence , contribution from @xmath118 terms in @xmath119 cases can be neglected , because the integral over two functions with zero overlap is zero .
this fact helps us to derive from eqs .
( s7 ) the mean - field equations for a two - fold fragmented system with well - isolated left and right fragments discussed in the main text . to facilitate a formal connection between external trapping potential @xmath4 and effective potentials
we redefine them as @xmath120 , with @xmath121 .
the multi - configurational time - dependent hartree method has been successfully formulated [ s1 ] in terms of reduced one- and two - body densities and _ local _ potentials @xmath116 also for systems of mixed bosons , fermions as well as for mixture thereof .
so , the proposed analysis of the multi - hump solutions in terms of _ local _ self - induced effective potentials @xmath79 is expected to be universal and can be applied to all many - body systems .
having the mc(td)hb(@xmath74 ) many - body wave function @xmath122 at hand we can analyze it . the reduced one - body density matrix available from eq .
( s3 ) can be diagonalized : ( , ;t)=^m_k , q=1 _
kq(t)^_k(,t)_q(,t ) = ^m_k=1 n_k(t ) _
k^no(,t ) _ k^no(,t ) . the obtained eigenvalues @xmath123 and eigenvectors @xmath124 are called natural occupation numbers and natural orbitals , respectively .
this diagonalization procedure is often referred to as natural analysis .
the natural occupation numbers @xmath123 can be considered as average numbers of bosons residing in @xmath125 .
the natural orbital analysis is used to characterize the system : the system is condensed [ s2 ] when only a single natural orbital has a macroscopic occupation and _ fragmented _ if several natural orbitals are macroscopically occupied [ 10 ] . usually , one plots only the diagonal part @xmath126 often referred to as the density . into natural fragments . * the four - hump density of @xmath23 bosons interacting via @xmath127 of range @xmath24 with @xmath25 in asymmetric linear trap is formed by four contributing natural sub - fragments @xmath128 , @xmath129 . the first fragment , carrying @xmath42 bosons is localized at the trap minimum and forms the most intense hump .
the left - most hump is formed by @xmath43 bosons residing in the third well - localized natural orbital .
the two other humps at the right are formed by @xmath44 and @xmath45 bosons residing in the second and fourth natural orbitals which are slightly delocalized . shown
are normalized reduced one - particle and partial densities corresponding to the natural fragments which are shifted vertically for a better visualization .
the trap potential is also rescaled to fit the figure .
all quantities shown are dimensionless.,width=642 ] in fig .
s1 we plot the results of the natural analysis done for the ground state ( static solution ) of the system with @xmath23 bosons interacting via @xmath127 of range @xmath24 and @xmath25 confined in asymmetric linear trap @xmath130 , studied in the main text .
the ground - state density obtained from mctdhb(4 ) solution has four well - pronounced picks / humps .
the natural analysis applied shows that this density is formed by four contributing natural sub - fragments , i.e. , @xmath131 . in fig .
s1 the normalized densities of the natural fragments , @xmath132 ( @xmath129 ) are shifted vertically for better visualization .
the trap potential is also rescaled to fit the figure .
the natural orbitals are sorted according to the numerical values of the computed natural occupation numbers .
the first fragment with @xmath42 bosons is localized at the trap minimum and forms the most intense hump .
the left - most hump is formed by @xmath43 bosons residing in the third well - localized natural orbital .
two other humps at the right are formed by @xmath44 and @xmath45 bosons residing in the second and fourth natural orbitals .
the two left fragments are well separated from the right ones .
the non - integer occupations of these right fragments indicate that the physics here is slightly beyond mean - field . indeed ,
the sub - clouds forming these humps are not completely localized , i.e. , the respective natural orbitals have two - hump profiles and non - zero density between the humps .
this means that there is an exchange / flow of the bosons between the fragments .
summarizing , in this example the many - body ground state is four - fold fragmented .
the phase information between two left fragments is completely lost because they are very - well separated .
the two right - most fragments are not entirely isolated from each other and , therefore , can communicate .
the phase contrast experiments where the quantum state is repeatedly prepared and released to form the interference patterns would reveal different phase distribution patterns for the left and right sub - clouds .
a perfect random phase distribution between the left fragments would contrast with a deformed phase distribution between the right sub - clouds .
let us now show how an increase of the strength @xmath6 of the inter - particle interaction modifies the ground state properties of the trapped systems . in fig .
s2 we plot the results of the natural analysis done for the ground state ( static solution ) of the same system as in fig .
s1 for several other values of @xmath133 . for weak interactions
the normal condensation is of course recovered i.e. , only single natural orbital is macroscopically occupied @xmath134 .
however , for stronger repulsive finite - range interactions to minimize the total energy the density profile develops multi - hump structures and the systems become multi - fold fragmented . for @xmath135
two - fold ( two - hump ) fragmented state is formed by two localized fragments with @xmath136 ( left ) and @xmath137 ( right ) bosons correspondingly . for @xmath138 three localized fragments with @xmath139 ( center ) , @xmath140 ( right ) and @xmath141 ( left )
form three - fold ( three - hump ) fragmented state . of @xmath23 bosons confined in the asymmetric linear trap ( same as in fig .
s1 ) are computed for several different interaction strengths @xmath142 .
the results obtained for @xmath127 interaction potential of range @xmath24 and a pure contact interaction @xmath143 are contrasted . in the regime of a weak repulsion
@xmath144 irrespective to the particular form of interactions both systems have a single - hump density profile and are fully condensed @xmath134 . for a stronger repulsion @xmath135
the contact interaction gives a condensed single - hump state while the finite - range interaction a two - fold ( two - hump ) fragmented state with localized fragments occupied by @xmath136 and @xmath137 bosons . for stronger repulsion @xmath138 to minimize the total energy the system with finite - range interaction forms a three - hump ( three - fold ) fragmented state , the contact interaction still result in a single - hump condensate .
the structure of the four - fold fragmented ground state for the finite - range interaction with @xmath25 is discussed in more detail in fig .
the corresponding system with contact interaction is a slightly depleted single - hump condensate @xmath145 .
the normalized reduced one - particle densities @xmath146 are shifted vertically for a better visualization .
the trap potential @xmath147 is also rescaled to fit the figure .
all quantities shown are dimensionless.,width=642 ] for comparison we compute the ground state for the same systems but with a pure contact inter - particle interaction @xmath148 . in all the presented cases ,
i.e. , @xmath142 the systems with contact interaction are essentially condensed and have single - hump densities .
the condensed fraction of the system with n=108 bosons and @xmath25 is @xmath145 . for smaller @xmath6
the condensed fractions are even larger .
to observe the predicted effects the first necessary condition is that the length / range of the inter - particle interaction should be comparable with the length of trapping potentials .
the second necessary condition is that the interaction should be strong enough .
there are at least two generically different ultra - cold systems where one expects to observe the formation of few - hump fragmented ground states : the dipolar condensates made of atomic or molecular species with non - zero electric / magnetic moments and ultra - cold systems made of rydberg - dressed atoms [ 10 , 11 , s3 ] let us first consider the dipolar condensates .
the dipolar systems apart from the long - range dipole - dipole interaction ( ddi ) also posses a residual contact contribution , so , in this respect , the main obstacle in the preparation of a strongly interacting dipolar system is the contact term .
however , luckily , the short- and long - range terms do originate to different physics and , hence , can be individually manipulated .
it was proposed to use the famous feshbach - resonance technique to demolish the contribution of the short - range terms , providing thereby opportunity to study pure dipolar systems .
the interaction between two polarized ( oriented ) dipoles reads : @xmath149 where @xmath150 is the separation between the dipoles , @xmath151 defines their orientation and @xmath152 is the strength of the dipole - dipole interaction .
@xmath153 for electric dipoles and @xmath154 for magnetic ones .
here @xmath155 is electric dipole moment in debye , @xmath156 , @xmath157 is the land factor , @xmath158 is the vacuum permeability , @xmath159 is the dielectric permittivity of classical vacuum , @xmath160 is the bohr magneton . to connect these real " inter - particle interaction potentials eq .
( s9 ) to dimensionless " ones used in the main text we introduce a change of the variables @xmath161 . here
@xmath162 is the length of the external harmonic trap of frequency @xmath163 associated with the size of the harmonic oscillator ground - state wave function , @xmath164 the mass of a trapped particle .
this change of variables rescales the hamiltonian @xmath165 and respective kinetic energy @xmath166 , single - particle trapping potential @xmath167 , and , the inter - particle interaction @xmath168 here @xmath169 is the rescaled interaction strength [ s4 ] .
the length scale ( range ) associated to the pure ddi is @xmath170 [ s4 ] . taking the values of magnetic moments
@xmath171 for @xmath0 and @xmath172 for @xmath1 and @xmath2 one obtains the respective ddi length scales ( ranges ) : @xmath173 , @xmath174 , @xmath175 . for these atoms
a typical trap with @xmath176 would result in @xmath177 0.5- , i.e. , the length of this trap is few orders of magnitude larger then the @xmath178 .
fortunately , with optical lattices one can create traps with much stronger confinement .
for example , in recent experiments with cr reported in ref .
[ s5 ] , the vibrational spacing in a lattice was about .
this lattice confinement results in @xmath179 of the corresponding trap length and much better ratio @xmath180 .
if a similar confinement could be created for dy and er atomic clouds , one would get : @xmath181 and @xmath182 correspondingly .
hence , the necessary ratio between the lengths of the ddi and of the confinement potentials can be reached in modern experiments with dipolar dy and er becs in optical lattices .
let us now verify how strong are the ddis in these systems .
the lattice - induced trap rescales the inter - particle ddi interaction strengths @xmath169 and results in : @xmath183 for cr , @xmath184 for dy and @xmath185 for er .
it is worthwhile to recall that the multi - fold fragmented states depicted in figs .
s1,s2 were reached with @xmath186 .
the electric ddi can , in principle , be realized with ultra - cold clouds ( bec ) of oriented molecules .
according to ref .
[ s4 ] the ddi length of polar molecules of mass 100 a.m.u with an electric dipole of 0.2 debye is @xmath187 .
a lattice - induced trap would result in @xmath188 , so , formally , @xmath189 .
the strength of the electric ddi in this trap is @xmath190 .
a tighter trap would results in a shorter @xmath191 and stronger ddi .
concluding , the predicted multi - fold fragmented states can , in principle , be observed within presently available technologies in dipolar ultra - cold atomic / molecular becs trapped in tight optical traps .
let us now consider the opportunity to realize the predicted multi - hump muti - fold fragmented physics in rydberg - dressed systems .
the off - resonant optical coupling of ground state atoms to highly excited rydberg states [ 10 , 11 ] allows to modify and control the shape and strength of effective two - body interactions .
the advantage of this technique is that a small component of the rydberg state is admixed to the ground state of atoms , providing thereby an additional degree of manipulation of the interaction strength .
the shape of the potential energy of the two many - electron atoms excited to a rydberg state depends very much on details of the electronic structure .
the most common long - range behavior , however , is of a standard van der waals ( @xmath192 ) type with possible admixture of terms of other degrees , e.g. , a pure dipole interaction ( @xmath193 ) .
these additional contributions are responsible for the tails of the interaction potentials .
the dressing " of two rydberg atoms with van der waals type of interaction results refs .
[ 10 , 11 ] in a two - body effective inter - particle potential : @xmath194 a change of variables @xmath195 allows us to rewrite it in the units of external confining potential @xmath162 as : @xmath196 where @xmath197 is the rescaled interaction strength and @xmath198 . in the present study
we have used a similar one - dimensional inter - particle interaction @xmath199 , see figs .
1a,1d and figs .
the strength @xmath6 of the inter - particle interaction depends on experimentally tunable parameters : @xmath200 a two - photon rabi frequency between the involved atomic levels , @xmath201 detunings of lasers with respect to the atomic transitions , and the external confinement @xmath163 .
the screening " constant @xmath202 defines the critical distance , below which the interaction is constant , i.e. , originates to the blockade phenomenon [ s3 ] .
this range depends on the detuning @xmath201 and on a pure spectroscopic @xmath203 coefficient governed by the electronic structure of the excited state , i.e. , it can be manipulated by a proper choice of an atomic rydberg level .
what are the presently available / reachable experimental conditions for rydberg excitations ? in recent experiments ref .
[ s6 ] with rb becs a few tens of rydberg atoms in a quasi-1d trap have been successfully detected .
the reported blockade radii were between 5- while the radial dipole trap frequencies used were around resulting in radial dimension of order of 1- , i.e. , ratio @xmath204 .
it means that the range of the interaction was larger then the the size of the trap in the transverse direction .
the strength of the interaction of the dressed rydberg atoms is defined by @xmath205 ratio and can be tuned between 0.1- , implying that in a trap the corresponding dimensionless interaction strength @xmath197 can be @xmath206 1 - 100 .
so , formally , the lengths / ranges and interaction strength needed to observe the predicted multi - hump muti - fold fragmented states are already reachable with ultra - cold dressed " rydberg atoms . finally , the results discussed in this work have been nicely explained at the multi - orbital mean - field level in terms of self - induced effective barriers .
this also implies that the underlying physics is defined not by a strong interaction strength @xmath6 between individual particles but , rather , by the @xmath207 interaction mean - field parameter ( also known as non - linearity ) .
so , the interesting physics of the multi - hump fragmented states reported and presented in this study for @xmath23 bosons can , in principle , be also realized in the systems with larger particle numbers and weaker inter - particle interactions .
alon , o. e. , streltsov , a. i. , sakmann , k. , lode , a. u. , grond , j. & cederbaum , l. s. recursive formulation of the multiconfigurational time - dependent hartree method for fermions , bosons and mixtures thereof in terms of one - body density operators .
_ * 401 * , 2 ( 2012 ) . | recent progress in cooling and trapping of polarized clouds of chromium @xmath0 , dysprosium @xmath1 and erbium @xmath2 opens a road - map to quantum systems where shapes of inter - particle interactions can be customized .
the main purpose of this work is to get a deeper insight on a role which the overall shape of the inter - particle interaction plays in a context of trapped ultra - cold bosons .
we show that strong inter - particle repulsion inevitably leads to multi - hump fragmentation of the ground state .
the fragmentation phenomenon is universal
it takes place in traps of different dimensionality and topologies and for very broad classes of repulsive inter - particle potentials .
the physics behind is identified and explained . nowadays ,
dilute ultra - cold atomic / molecular clouds are considered as toolboxes to probe static and dynamical properties of many - particle hamiltonians @xcite .
consequently , some phenomena which are very difficult or even impossible to study in their natural appearance / environment can explicitly be reconstructed and modeled in the ultra - cold atomic systems . however , until recent times one substantial ingredient was missed - a control on the overall shape of inter - particle interactions .
recent experiments with ultra - cold polarized clouds of chromium @xmath0 @xcite , dysprosium @xmath1 @xcite and erbium @xmath2 @xcite have ultimately shown that the short - range inter - particle interaction potential alone can not describe the observed physics and it should be augmented by an additional long - range term which usually takes on the form of a dipole - dipole interaction @xcite .
another venue to manipulate the effective inter - particle interaction is to admix a small component of an excited rydberg state to the ground state of ultra - cold atoms @xcite via the off - resonant optical coupling , creating thereby so - called dressed " rydberg systems .
these steps towards control and manipulation of the overall shape of the inter - particle interaction open a road - map to mimic / simulate static and dynamical phenomena and effects appearing in the context of other fields of physics where shapes of inter - particle interactions play crucial roles , e.g. in nuclear physics .
the perspective of working with quantum systems where the inter - particle interaction is customized encourages us to get a deeper insight on a role it plays .
the role of the sign is evident : if it is negative the system is attractive , if it is positive repulsive , but what roles are playing its range and tails ? what physical phenomena or properties they are envisioned to impact ?
the main phenomenon predicted and intensively studied in the field of ultra - cold atoms is condensation , manifesting itself in a peaked density with a profile similar to the lowest - in - energy eigenstate of the confining potential .
condensation also means coherence between weakly - interacting particles @xcite . on the other hand
, the existence in the ground - state density of multi - hump features might indicate on possible ( strong ) correlations in the system which usually destroy condensation and break partially or completely coherence between the particles .
the multi - hump structure of the ground - state density can be caused by applied external potential barriers . in a double - well for instance , a sufficiently high barrier can split the density into two well - separated sub - clouds .
such a system is then called two - fold fragmented @xcite implying that coherence between the sub - fragments is lost .
if more barriers are available , as in optical lattices , the system can be multi - fold fragmented , see in this respect the famous mott - insulating phases @xcite .
complimentary , modulations of density profiles and loss of an inter - particle coherence can be caused by strong repulsive inter - particle interaction .
in one - dimensional systems with contact inter - particle interactions it originates to the famous fermionization phenomenon @xcite for inter - boson interactions of other shapes to solid - like states see , e.g. coulomb bosons @xcite and dipolar ( screened coulomb - like ) bosons @xcite . in two - dimensions strong repulsion is predicted to be responsible for so - called crystallization @xcite appearances of stable rims in the density profiles and/or its partial factorization into multi - hump structures , see e.g. refs .
@xcite .
the main goal of the present work is to investigate the microscopic details of how repulsive inter - particle potentials create the non - trivial features ( humps ) in the densities of ultra - cold systems confined in simple barrier - less traps .
we also want to understand which characteristic of a general inter - particle interaction function favors these density modulations and controls , thereby , the accompanying developments of correlations and fragmentation .
let us consider a generic many - body hamiltonian of @xmath3 identical bosons trapped in an external trap potential @xmath4 and interacting via general inter - particle interaction potential : @xmath5 \!+\!\sum_{j < k}^n \lambda_0 w(\r_j\!-\!\r_k ) . \nonumber $ ] here @xmath6 defines the strength of the interaction and @xmath7 its shape . in this work @xmath8 , @xmath9 . to start with we consider several one - dimensional barrier - less traps : a standard harmonic @xmath10 , a non - harmonic @xmath11 and asymmetric linear @xmath12 .
we examine the following inter - particle interaction functions : exponential @xmath13 $ ] , screened coulomb @xmath14 and sech - shaped @xmath15 $ ] of half - width @xmath16 with @xmath17 and their _ sharper _ analogous with @xmath18 . in the following we use the shorthand notations @xmath19 $ ] , @xmath20 and @xmath21 $ ] . to solve the respective many - boson schrdinger equation
numerically we use the recently developed multi - configurational time - dependent hartree method for bosons ( mctdhb ) @xcite , see @xcite for more details .
this method is capable of providing numerically - exact solutions @xcite . and trapping @xmath22 potentials used .
shown are the densities of one - dimensional systems made of @xmath23 bosons and corresponding trapping potentials which are scaled and shifted for better presentation .
upper panels : asymmetric linear trap and screened coulomb inter - boson interactions of half - width @xmath24 and strength @xmath25 with @xmath17 ( * a * ) and @xmath18 ( * d * ) .
middle panels : harmonic trap and sech - shaped interactions with @xmath26 and @xmath27 , @xmath17 ( * b * ) , @xmath18 ( * e * ) .
lower panels : non - harmonic trap and exponential inter - particle interactions with @xmath28 and @xmath29 , @xmath17 ( * c * ) , @xmath18 ( * f * ) . inter - particle interaction potentials with
_ sharper _ edges ( @xmath18 ) enhance fragmentation , see text for details .
all quantities shown are dimensionless.,width=642 ] the left panels of fig .
[ fig1 ] plot the ground - state densities obtained for one - dimensional systems with @xmath30 , @xmath31 $ ] , @xmath32 $ ] inter - particle interaction functions of different widths ( ranges ) confined in different traps .
the main fascinating observation is that irrespective to the shapes of inter - particle and trapping potentials used a strong enough inter - particle repulsion leads to formation of the multi - hump localized structures and , therefore , indicates on possible correlations / fragmentation in the systems . and
interacting via inter - particle interaction potential of rectangular shape with half - width @xmath28 .
the left panels show that for strong repulsion @xmath33 ( * b , c * ) it costs energy to keep @xmath23 bosons in the middle of the trap ( @xmath34 ) and to minimize the energy the system fragments into two well - separated sub - clouds .
the right panels show how to control the fragmentation in strongly repulsive systems ( @xmath27 ) by varying the length @xmath35 of the box .
for @xmath36 the ground state is three - fold fragmented ( * e * ) and for @xmath37 four - fold ( * f * ) . shown
are the one - particle densities and trap potentials .
all quantities shown are dimensionless.,width=642 ] to shed more light on what characteristic parameters of the inter - particle interaction function drive and control the number of humps and correlations / fragmentation we have computed the ground - states properties of the same systems as above , but with _ sharper _ inter - particle interactions @xmath38 , @xmath39 $ ] , @xmath40 $ ] .
the results presented in the right panels of fig . [ fig1 ] show that for sharper versions of the inter - particle interactions the multi - hump structures are retained and become even more pronounced .
the many - body analysis of these multi - hump solutions reveals that they are not condensed but multi - fold fragmented , see @xcite for more details . in the asymmetric linear trap the four - hump density of @xmath23 bosons depicted in fig .
[ fig1]d is formed by four contributing natural sub - fragments @xmath41 .
the first fragment with @xmath42 bosons is localized at the trap minimum and forms the most intense hump .
the left - most hump is formed by @xmath43 bosons residing in a well - localized third natural orbital .
two other humps at the right are formed by @xmath44 and @xmath45 bosons residing in the second and fourth natural orbitals which are slightly delocalized . in the harmonic trap and sech - shaped interaction ,
see fig .
[ fig1]e , the three symmetric fragments carry 30 , 48 , and 30 bosons , respectively . in the non - harmonic trap
a gaussian interaction results in a two - fold fragmented ground state with 97.2 and 10.8 bosons per fragment , see fig .
[ fig1]f . here
it is worthwhile to stress that for weak repulsions the regime of normal condensation is , of course , recovered irrespective to the particular form of interactions , see @xcite for more details .
the above observed diversity of the fragmentation is caused by the interplay between the inter - particle and trapping potentials . to distinguish and isolate effects originating from the range / width of the inter - particle interaction , particular shape of its tails and strength of the inter - particle repulsion let us consider a box - shaped trap and inter - particle interaction potential of a rectangular shape @xmath46 . in this system
an increase of the strength of the inter - particle repulsion does not change neither the effective length of the trap nor the range of the inter - particle interaction potential .
the left panels of fig .
[ fig2 ] show how the ground - state densities of the system made of @xmath23 bosons with rectangular inter - particle interaction of half - width @xmath28 , trapped in a box - shaped trap of length @xmath34 , response when the repulsion strength is increased @xmath47 .
we see that in the non - interacting case the density is broad and has a maximum at the trap center .
the strong inter - particle repulsion , in contrast , leads to localization of the density at the edges of the box .
basic `` electrostatic '' arguments can explain this behavior . for strong repulsion
it costs energy to keep the bosons in the middle of the trap , so , they repeal and push each other away the cloud starts to form a minimum in the center of the trap .
if we increase the repulsion further the total energy is increased , but the system can not expand further apart due to the box - shaped topology of the trap . to minimize the energy the density
is split into two well - separated fragments .
the right panels of fig .
[ fig2 ] present a complimentary study where we keep the inter - particle interaction @xmath27 , @xmath28 fixed and increase the box size @xmath48 providing , thereby , more room for the bosons .
we see that starting from some critical length of the trap ( box size ) , to minimize the repulsion the system of @xmath23 bosons is split into three sub - clouds with 36 bosons per fragment . to split the system into four fragments with 27 bosons per sub - cloud one has to increase the trap s length further on .
so , for strong enough inter - particle repulsions ground state fragmentation is an inevitable property of trapped bosonic systems .
the interplay between the width of the finite - range part of the inter - particle interaction function and the length of the trap defines the particular fragmentation scenario .
produced by a rectangular inter - particle interaction potential @xmath49 of half - width @xmath16 acting on a gaussian - shaped function @xmath50 of half - width @xmath51 .
lower panel : a rectangular inter - particle potential acting on the sub - cloud @xmath52 produces an effective barrier @xmath53 .
the superposition of this interaction - induced effective barrier and the external trap ( here it is a square box of length @xmath35 ) results in a well where the @xmath54 sub - cloud is confined .
middle panel : the action of the same finite - range @xmath49 on @xmath54 produces an effective well capable of confining the left sub - cloud @xmath52 .
all quantities shown are dimensionless.,width=642 ] to get a deeper insight into the physics behind the fragmentation let us rely on an idealized picture of a two - fold fragmented state .
the total density of such a state is formed by a sum of two isolated and independent sub - clouds ( fragments ) .
its many - body wave - function is then a fock state @xmath55 represented by a single symmetrized permanent @xmath56 with @xmath57 bosons residing in the left @xmath52 and @xmath58 in the right @xmath54 fragment , respectively .
the optimal shapes of the fragments are determined self - consistently by solving the multi - orbital best mean - field ( bmf ) equations @xcite which we rewrite as ( see @xcite for details ) : @xmath59\!\phi_l\!\!=\!\ ! \mu_{11}\phi_l\!+\!\mu_{12}\phi_r \nonumber \\
\!\!\left[\hat h\!+\!\lambda_0(n_r\!-\!1 ) v^{eff}_{\phi_r}(\br)\!+\!\lambda_0n_lv^{eff}_{\phi_l}(\br ) \right]\!\phi_r\!\!=\!\ ! \mu_{21}\phi_l\!+\!\mu_{22}\phi_r .
% \label{bmf } % \eeqn \end{array}\ ] ] here @xmath60 is a single - particle hamiltonian .
the @xmath61 terms play the roles of effective _ self - consistent _ potentials their profiles depend on a given shape of inter - particle function @xmath62 and on the left / right densities @xmath63 , @xmath64 of the involved sub - clouds .
the upper panel of fig .
[ fig3 ] exemplifies schematically how effective barrier @xmath65 is produced by a rectangular inter - particle interaction potential @xmath49 of range @xmath16 acting on a gaussian - shaped cloud @xmath50 of half - width @xmath51 .
this effective barrier depicted by a solid line with circles is centered at the cloud origin .
the profile of this barrier is flat over the extent of the cloud @xmath50 and drops like @xmath66-function at the @xmath67 edges .
two other panels of fig .
[ fig3 ] show how a rectangular inter - particle interaction of half - width @xmath16 induces the effective potentials in the above studied one - dimensional bosonic system trapped in a box - shaped trap .
the lower panel of fig .
[ fig3 ] depicts by a solid line with circles the interaction - induced effective potential @xmath53 .
it is constant over the extent of the @xmath68 fragment and , therefore , in the above bmf equations it just shifts the left - localized sub - cloud upwards energetically without changing its shape : @xmath69 .
the action of @xmath70 on the right fragment @xmath71 , in contrast , is dramatic it induces the effective barrier .
a superposition of this effective barrier and the external trap ( box ) is capable of confining the right sub - cloud @xmath71 .
similarly , the action of @xmath49 on the right sub - cloud creates an effective barrier @xmath72 confining the left fragment , see middle panel of fig .
[ fig3 ] .
we have arrived here at a microscopic , self - consistent picture of _ self - induced _ fragmentation which requires as a prerequisite localized sub - clouds , finite - range inter - particle repulsion and a trap of a finite length .
the action of a finite - range repulsive inter - particle potential on a cloud produces an effective barrier .
the profile of this barrier depends on the density of the cloud , number of the particles in it and on the shape of @xmath73 .
when several localized clouds / fragments are present each of them creates its own effective potential seen by the other sub - clouds as an effective barrier .
the superposition of these self - induced barriers and external trap results in a multi - well potential confining the fragmented system as a whole object .
the multi - hump structure of the density is driven by the fragmentation phenomenon which has been observed experimentally @xcite and is well - understood theoretically @xcite . from this perspective , variations of shapes of @xmath62 can modify the induced effective potentials and , thereby , the fragmentation in the following manner : a ) by decreasing range of the inter - particle repulsion we increase the overlap between the fragments which reduces the fragmentation and leads to the development of coherence between them @xcite ; b ) by increasing strength of the inter - particle repulsion and keeping its range fixed we increase the heights of the induced barriers which isolates the fragments and enhances fragmentation ; c ) by flattening the tails of the inter - particle interaction function one stimulates the overlap between the sub - clouds which melts the humps and blurs the fragmentation .
comparison of the left and right panels of fig .
[ fig1 ] confirms this analysis .
a simple geometrical relation between the effective length of the trap @xmath35 and width / range @xmath16 of the finite - range part of the inter - particle interaction function defines the number @xmath74 of available humps(fragments ) .
for @xmath75 we have a system of non - interacting particles because irrespective to the shape and size of the trapped cloud , the created effective barrier is geometrically broader than the trap ( box ) itself .
when @xmath76 there is enough room for two fragments to be trapped by the self - induced effective barriers , to accommodate @xmath74 fragments the length of the trap should be @xmath77 , see e.g. fig .
[ fig2 ] .
the obtained geometrical picture of localized fragments and self - induced effective potentials is universal and can be applied to systems of bosons , fermions and distinguishable particles also at higher dimensions . in the left and right panels of fig .
[ fig4 ] we plot illustrative numerical examples of almost ideal self - induced two - fold fragmented ground states of bosonic systems with @xmath78 particles interacting via sharp repulsive sech - shaped inter - particle potentials and confined in slightly elongated two- and three - dimensional harmonic traps . summarizing , we predict that the density of trapped repulsive ultra - cold bosons inevitably fragments into multi - hump structures to minimize strong inter - particle repulsion .
the physics behind is an interplay between classical
electrostatic " repulsion which pushes the bosons from the trap center towards its edges , provoking thereby the formation of multi - hump structures in the density , and quantum mechanics which governs loss of inter - hump coherence , i.e. , fragmentation .
finally , for experimental verification of the predicted phenomena with ultra - cold systems one needs that ( i ) the range of inter - particle interaction should be comparable with the length of trapping potentials ; ( ii ) repulsive interaction should be strong enough .
both these requirements are already reachable within presently available technologies in dipolar ultra - cold atomic / molecular becs trapped in tight optical traps and in trapped ultra - cold systems of dressed " rydberg atoms , see @xcite for discussion .
i am grateful to oksana i. streltsova for 2d and 3d mctdhb simulations , to ofir e. alon , lorenz s. cederbaum , luis santos and michael fleischhauer for fruitful discussions .
computation time on the bwgrid and k100 clusters and financial support within the framework of the `` enable fund '' of the excellence initiative at heidelberg university are greatly acknowledged
. 64 f. dalfovo _
et al . _ , rev .
mod .
phys . * 71 * , 463 - 512 ( 1999 ) .
a.j .
leggett , rev .
mod .
phys . * 73 * , 307 - 356 ( 2001 ) .
t. khler _
et al . _ , rev .
mod .
phys . * 78 * , 1311 - 1361 ( 2006 ) .
a. griesmaier , j. werner , s. hensler , j. stuhler , and t. pfau , phys .
rev .
lett .
* 94 * , 160401 ( 2005 ) .
j. stuhler _
et al . _ ,
phys .
rev .
lett . * 95 * , 150406 ( 2005 ) .
m. lu , n. q. burdick , s. h. youn , and b. l. lev , phys .
rev .
lett . * 107 * , 190401 ( 2011 ) .
k. aikawa _
et al . _ ,
phys .
rev .
lett . * 108 * , 210401 ( 2012 ) .
m. a. baranov , phys .
rep . * 464 * , 71 - 111 ( 2008 ) .
t. lahaye _
et al . _ ,
rep .
. phys . * 72 * , 126401 ( 2009 ) .
phys .
rev .
a * 82 * , 033412 ( 2010 ) .
phys .
rev .
lett . * 104 * , 195302 ( 2010 ) .
nozires _ et al .
_ , j. phys .
france * 43 , * 1133 - 1148 ( 1982 ) .
m. greiner _
et al . _ , nature ( london ) * 415 * , 39 - 44 ( 2002 ) .
m. girardeau , j. math . phys . * 1 * , 516 - 523 ( 1960 ) .
t. kinoshita , t. wenger , and d. s. weiss , phys .
rev .
lett .
* 95 * , 190406 ( 2005 ) .
g. e. astrakharchik and m. d. girardeau , phys .
rev .
b * 83 * , 153303 ( 2011 ) .
f. deuretzbacher , j. c. cremon , and s. m. reimann , phys .
rev .
a * 81 * , 063616 ( 2010 ) .
yu .
e. lozovik _
et al . _ , jetp lett . * 79 * , 473 ( 2004 ) .
h. p. bchler
_ et al . _ ,
phys .
rev .
lett . * 98 * , 060404 ( 2007 ) .
a. i. streltsov , o. e. alon , and l. s. cederbaum , phys .
rev .
lett .
* 99 , * 030402 ( 2007 ) .
o. e. alon , a. i. streltsov , and l. s. cederbaum , phys .
rev .
a * 77 , * 033613 ( 2008 ) .
see epaps document no .
xxxx for supplemental physical analysis of the quantum systems with customized inter - particle interactions . for more information on epaps ,
see http://www.aip.org/pubservs/epaps.html .
a. u. j. lode , k. sakmann , o. e. alon , l. s. cederbaum , and a. i. streltsov , phys .
rev .
a * 86 * , 063606 ( 2012 ) .
l. s. cederbaum and a. i. streltsov , phys .
lett .
a * 318 * , 564 - 569 ( 2003 ) .
o. e. alon , a. i. streltsov , and l. s. cederbaum , phys .
lett .
a * 347 * , 88 ( 2005 ) . c. orzel _ et al . _ ,
science * 291 * , 2386 - 2389 ( 2001 ) .
s. hofferberth _
et al . _ , nature physics * 2 * , 710 - 716 ( 2006 ) .
r. w. spekkens and j. e. sipe , phys .
rev .
a * 59 * , 3868 ( 1999 ) .
a. i. streltsov , o. e. alon , and l. s. cederbaum , , phys .
rev .
a * 73 , * 063626 ( 2006 ) .
e. j. mueller , t. -l .
ho , m. ueda , and g. baym , phys . rev .
a * 74 , * 033612 ( 2006 ) .
o. e. alon , a. i. streltsov , and l. s. cederbaum , phys .
lett .
a * 373 * , 301 - 304 ( 2009 ) . |
correlation functions are fundamental objects for statistical analysis , and are thus ubiquitous in most kinds of scientific inquires and their applications @xcite . in physics ,
correlation functions have an important role for research in areas such as quantum optics and open systems @xcite , phase transitions and condensed matter physics @xcite , and quantum field theory and nuclear and particle physics @xcite .
another area in which correlation functions are omnipresent is quantum information science ( qis ) , an interdisciplinary field that extends the applicabilities of the classical theories of information , computation , and computational complexity @xcite .
investigations about the quantum correlations in physical systems have been one of the main catalyzers for developments in qis @xcite .
there are several guises of quantum correlations , and quantum discord stands among the most promising quantum resources for fueling the quantum advantage @xcite .
when computing or witnessing quantum discord , or other kinds of correlation or quantumness quantifiers , we are frequently faced with the need for calculating coherence vectors and correlation matrices @xcite . and
it is the main aim of this article to provide formulas for these functions that are amenable for more efficient numerical calculations when compared with the direct implementation of their definitions . in order to define coherence vectors and correlation matrices ,
let us consider a composite bipartite system with hilbert space @xmath0 .
hereafter the corresponding dimensions are denoted by @xmath1 for @xmath2 .
in addition , let @xmath3 , with @xmath4 be a basis for the special unitary group @xmath5 .
any density operator describing the state of the system @xmath6 can be written in the local basis @xmath7 as follows : @xmath8 where @xmath9 and @xmath10 is the identity operator in @xmath11 .
one can readily verify that the components of the coherence ( or bloch s ) vectors @xmath12 and @xmath13 and of the correlation matrix @xmath14 are given by : @xmath15 it is worthwhile mentioning that the mean value of _ any observable _ in @xmath11 , for @xmath16 , can be obtained using these quantities . in https://github.com/jonasmaziero/libforq.git , we provide fortran code to compute the coherence vectors , correlation matrices , and quantum discord quantifiers we deal with here . besides these functions , there are other tools therein that may be of interest to the reader .
the instructions on how to use the software are provided in the readme file .
related to the content of this section , the subroutine ` bloch_vector_gellmann_unopt(d_{s } , \rho_{s } , \mathbf{s } ) ` returns the coherence vectors @xmath17 or @xmath18 and the subroutine ` corrmat_gellmann_unopt(d_{a } , d_{b } , \rho , c ) ` computes the correlation matrix @xmath19 .
now , let us notice that if calculated directly from the equations above , for @xmath20 , the computational complexity ( cc ) to obtain the coherence vectors @xmath17 and @xmath18 or the correlation matrix @xmath19 is : @xmath21 the remainder of this article is structured as follows . in sec .
[ coh_vec ] , we obtain formulas for @xmath17 , @xmath18 , and @xmath19 that are amenable for more efficient numerical computations . in sec .
[ discord ] we test these formulas by applying them in the calculation of hilbert - schmidt quantum discords . in sec .
[ conclusion ] we make some final remarks about the usefulness and possible applications of the results reported here .
the partial trace function @xcite can be used in order to obtain the reduced states @xmath22 and @xmath23 and to write the components of the bloch vectors in the form : @xmath24 thus , when computing the coherence vectors of the parties @xmath25 and @xmath26 , we shall have to solve a similar problem ; so let s consider it separately .
that is to say , we shall regard a generic density operator written as @xmath27 where @xmath28 now , and for the remainder of this article , we assume that the matrix elements of regarded density operator @xmath29 in the standard computational basis are given .
we want to compute the _ bloch s vector _
@xcite : @xmath30 for the sake of doing that , a particular basis @xmath3 must be chosen . here
we pick the generalized gell mann matrices , which are presented below in three groups @xcite : @xmath31 which are named the diagonal , symmetric , and antisymmetric group , respectively .
the last two groups possess @xmath32 generators each .
any one of these matrices can be obtained by calling the subroutine ` gellmann(d_{s } , g , k , l , \gamma_{(k , l)}^{s(g ) } ) ` .
for the first group , @xmath33 , we make @xmath34 and , in this case , one can set @xmath35 to any integer .
it is straightforward seeing that , for the generators above , the corresponding components of the bloch s vector can expressed directly in terms of the matrix elements of the density operator @xmath36 as follows : @xmath37 these expressions were implemented in the fortran subroutine ` bloch_vector_gellmann(d_{s } , \rho_{s } , \mathbf{s } ) ` . with this subroutine , and the partial trace function @xcite ,
we can compute the coherence vectors @xmath17 and @xmath18 . we observe that after these simple algebraic manipulations the computational complexity of the bloch s vector turns out to be basically the cc for the partial trace function .
hence , from ref .
@xcite we have that for @xmath20 , @xmath38 one detail we should keep in mind , when making use of the codes linked to this article , is the convention we apply for the indexes of the components of @xmath39 . for the first group of generators , @xmath40 , naturally , @xmath41 .
we continue with the second group of generators , @xmath42 , by setting @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 , @xmath45 . the same convention is used for the third group of generators , @xmath48 , but here we begin with @xmath49 next we address the computation of the _ correlation matrix _ @xmath14 , which is a @xmath50 matrix that we write in the form : @xmath51 with the sub - matrices given as shown below . for convenience ,
we define the auxiliary variables : @xmath52 the matrix elements of @xmath53 , whose dimension is @xmath54 , correspond to the diagonal generators for @xmath25 and diagonal generators for @xmath26 : @xmath55 the matrix elements of @xmath56 , whose dimension is @xmath57 , correspond to the diagonal generators for @xmath25 and symmetric generators for @xmath26 : @xmath58 the matrix elements of @xmath59 , whose dimension is @xmath57 , correspond to the diagonal generators for @xmath25 and antisymmetric generators for @xmath26 : @xmath60 the matrix elements of @xmath61 , whose dimension is @xmath62 , correspond to the symmetric generators for @xmath25 and diagonal generators for @xmath26 : @xmath63 the matrix elements of @xmath64 , whose dimension is @xmath65 , correspond to the symmetric generators for @xmath25 and symmetric generators for @xmath26 : @xmath66 the matrix elements of @xmath67 , whose dimension is @xmath65 , correspond to the symmetric generators for @xmath25 and antisymmetric generators for @xmath26 : @xmath68 the matrix elements of @xmath69 , whose dimension is @xmath62 , correspond to the antisymmetric generators for @xmath25 and diagonal generators for @xmath26 : @xmath70 the matrix elements of @xmath71 , whose dimension is @xmath65 , correspond to the antisymmetric generators for @xmath25 and symmetric generators for @xmath26 : @xmath72 the matrix elements of @xmath73 , whose dimension is @xmath65 , correspond to antisymmetric generators for @xmath25 and antisymmetric generators for @xmath26 : @xmath74 we remark that when implementing these expressions numerically , for the sake of mapping the local to the global computational basis , we utilize , e.g. , @xmath75 the subroutine ` corrmat_gellmann(d_{a } , d_{b } , \rho , c ) ` returns the correlation matrix @xmath14 , as written in eq .
( [ eq : corrmat ] ) , associated with the bipartite density operator @xmath29 and computed using the gell mann basis , as described in this section .
the convention for the indexes of the matrix elements @xmath76 is defined in the same way as for the coherence vectors . the computational complexity for @xmath19 , computed via
the optimized expressions obtained in this section , is , for @xmath20 , @xmath77 by generating some random density matrices @xcite , we checked that the expressions and the corresponding code for the unoptimized and optimized versions of @xmath17 , @xmath18 , and @xmath19 agree .
additional tests shall be presented in the next section , where we calculate some quantum discord quantifiers .
the calculation of quantum discord functions ( qd ) usually involves hard optimization problems @xcite . in the last few years
, a large amount of effort have been dedicated towards computing qd analytically , with success being obtained mostly for low dimensional quantum systems @xcite .
although not meeting all the required properties for a _ bona fide _ qd quantifier @xcite , the hilbert - schmidt discord ( hsd ) @xcite , @xmath78 is drawing much attention due to its amenability for analytical computations , when compared with most other qd measures . in the last equation , the minimization is performed over the classical - quantum states @xmath79 with @xmath80 being a probability distribution , @xmath81 an orthonormal basis for @xmath82 , @xmath83 generic density operators defined in @xmath84 , and @xmath85 is the hilbert - schmidt norm of the linear operator @xmath86 , with @xmath87 being the transpose conjugate of @xmath86 . in this article , as a basic test for the fortran code provided to obtain coherence vectors and correlation matrices , we shall compute the following lower bound for the hsd @xcite : @xmath88 where @xmath89 are the eigenvalues , sorted in non - increasing order , of the @xmath90 matrix : @xmath91 in the equation above @xmath92 stands for the transpose of a vector or matrix .
we observe that the other version of the hsd , @xmath93 , can be obtained from the equations above simply by exchanging @xmath25 and @xmath26 and using @xmath94 instead of @xmath95 .
it is interesting regarding that , as was proved in ref .
@xcite , a bipartite state @xmath29 , with polarization vectors @xmath17 and @xmath18 and correlation matrix @xmath19 , is classical - quantum if and only if there exists a @xmath96-dimensional projector @xmath97 in the space @xmath98 such that : @xmath99 based on this fact , an ameliorated version for the hilbert - schmidt quantum discord ( ahsd ) was proposed @xcite : @xmath100 with the matrix @xmath101 defined as @xmath102 where @xmath103 and @xmath104 are arbitrary functions of @xmath105 .
then , by setting @xmath106 and using the purity , @xmath107 to address the problem of non - contractivity of the hilbert - schmidt distance , the following analytical formula was presented @xcite : @xmath108 thus both discord quantifiers @xmath109 and @xmath110 are , in the end of the day , obtained from the eigenvalues @xmath89 .
and the computation of these eigenvalues requires the knowledge of the coherence vector @xmath17 ( or @xmath18 ) and of the correlation matrix @xmath19 .
these qd measures were implemented in the fortran functions ` discord_hs(ssys , d_{a } , d_{b } , \rho ) ` and ` discord_hsa(ssys , d_{a } , d_{b } , \rho ) ` , where ` ssys = s ` , with @xmath111 , specifies which version of the quantum discord is to be computed . as an example , let us use the formulas provided in this article and the associated code to compute the hsd and ahsd of werner states in @xmath112 ( with @xmath113 : @xmath114 where @xmath115 $ ] and @xmath116 the reduced states of @xmath117 are @xmath118 , whose purity is @xmath119 .
the results for the hsd and ahsd of @xmath117 are presented in fig .
[ werner ] . ) .
the lines are the corresponding values of the ahsd plotted via the analytical formula : @xmath120 . due to the symmetry of @xmath117 , here @xmath121 in the inset
is shown the difference between the times taken by the two methods to compute the ahsd for a fixed value of @xmath122 .
we see clearly here that our optimized algorithm gives a exponential speedup against the brute force calculation of the bloch s vectors and correlation matrix . ]
in this article , we addressed the problem of computing coherence vectors and correlations matrices .
we obtained formulas for these functions that make possible a considerably more efficient numerical implementation when compared with the direct use of their definitions .
we provided fortran code to calculate all the quantities regarded in this paper . as a test for our formulas and code , we computed hilbert - schmidt quantum discords of werner states .
it is important observing that , although our focus here was in quantum information science , the tools provided can find application in other areas , such as e.g. in the calculation of order parameters and correlations functions for the study of phase transitions in discrete quantum or classical systems .
the author declares that the funding mentioned in the acknowledgments section do not lead to any conflict of interest .
additionally , the author declares that there is no conflict of interest regarding the publication of this manuscript .
this work was supported by the brazilian funding agencies : conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) , processes 441875/2014 - 9 and 303496/2014 - 2 , instituto nacional de cincia e tecnologia de informao quntica ( inct - iq ) , process 2008/57856 - 6 , and coordenao de desenvolvimento de pessoal de nvel superior ( capes ) , process
6531/2014 - 08 .
i gratefully acknowledges the hospitality of the physics institute and laser spectroscopy group at the universidad de la repblica , uruguay .
j. hutchinson , j. p. keating , and f. mezzadri , on relations between one - dimensional quantum and two - dimensional classical spin systems , http://dx.doi.org/10.1155/2015/652026[adv .
2015 , 652026 ( 2015 ) ] .
m. d. reid , p. d. drummond , w. p. bowen , e. g. cavalcanti , p. k. lam , h. a. bachor , u. l. andersen , and g. leuchs , the einstein - podolsky - rosen paradox : from concepts to applications , http://dx.doi.org/10.1103/revmodphys.81.1727[rev .
81 , 1727 ( 2009 ) ] .
l. c. cleri , j. maziero , and r. m. serra , theoretical and experimental aspects of quantum discord and related measures , http://dx.doi.org/10.1142/s0219749911008374[int .
j. quantum inf .
9 , 1837 ( 2011 ) ] .
k. modi , a. brodutch , h. cable , t. paterek , and v. vedral , the classical - quantum boundary for correlations : discord and related measures , http://dx.doi.org/10.1103/revmodphys.84.1655[rev .
84 , 1655 ( 2012 ) ] .
d. o. soares - pinto , r. auccaise , j. maziero , a. gavini - viana , r. m. serra , and l. c. cleri , on the quantumness of correlations in nuclear magnetic resonance , http://dx.doi.org/10.1098/rsta.2011.0364 [ phil .
r. soc . a 370 , 4821 ( 2012 ) ] .
m. vila , g. h. sun , and a. l. salas - brito , scales of time where the quantum discord allows an efficient execution of the dqc1 algorithm , http://dx.doi.org/10.1155/2014/367905[adv .
2014 , 367905 ( 2014 ) ] .
m piani , v narasimhachar , and j calsamiglia , quantumness of correlations , quantumness of ensembles and quantum data hiding , http://dx.doi.org/10.1088/1367-2630/16/11/113001[new j. phys .
16 , 113001 ( 2014 ) ] .
b. dakic , y. o. lipp , x. ma , m. ringbauer , s. kropatschek , s. barz , t. paterek , v. vedral , a. zeilinger , .
brukner , and p. walther , quantum discord as resource for remote state preparation , http://dx.doi.org/10.1038/nphys2377[nat .
8 , 666 ( 2012 ) ] .
d. girolami , a. m. souza , v. giovannetti , t. tufarelli , j. g. filgueiras , r. s. sarthour , d. o. soares - pinto , i. s. oliveira , and g. adesso , quantum discord determines the interferometric power of quantum states , http://dx.doi.org/10.1103/physrevlett.112.210401[phys .
112 , 210401 ( 2014 ) ] .
t. k. chuan , j. maillard , k. modi , t. paterek , m. paternostro , and m. piani , quantum discord bounds the amount of distributed entanglement , http://dx.doi.org/10.1103/physrevlett.109.070501[phys .
109 , 070501 ( 2012 ) ] .
r. auccaise , j. maziero , l. c. celeri , d. o. soares - pinto , e. r. deazevedo , t. j. bonagamba , r. s. sarthour , i. s. oliveira , and r. m. serra , experimentally witnessing the quantumness of correlations , http://dx.doi.org/10.1103/physrevlett.107.070501[phys .
107 , 070501 ( 2011 ) ] .
f. m. paula , i. a. silva , j. d. montealegre , a. m. souza , e. r. deazevedo , r. s. sarthour , a. saguia , i. s. oliveira , d. o. soares - pinto , g. adesso , and m. s. sarandy , observation of environment - induced double sudden transitions in geometric quantum correlations , http://dx.doi.org/10.1103/physrevlett.111.250401[phys .
111 , 250401 ( 2013 ) ] .
i. a. silva , d. girolami , r. auccaise , r. s. sarthour , i. s. oliveira , t. j. bonagamba , e. r. deazevedo , d. o. soares - pinto , and g. adesso , measuring bipartite quantum correlations of an unknown state , http://dx.doi.org/10.1103/physrevlett.110.140501[phys .
110 , 140501 ( 2013 ) ] .
g. h. aguilar , o. jimnez faras , j. maziero , r. m. serra , p. h. souto ribeiro , and s. p. walborn , experimental estimate of a classicality witness via a single measurement , http://dx.doi.org/10.1103/physrevlett.108.063601[phys .
108 , 063601 ( 2012 ) ] .
m. cianciaruso , t. r. bromley , w. roga , r. lo franco , and g. adesso , universal freezing of quantum correlations within the geometric approach , http://dx.doi.org/10.1038/srep10177[sci .
rep . 5 , 10177 ( 2015 ) ] .
t. r. bromley , m. cianciaruso , r. lo franco , and g. adesso , unifying approach to the quantification of bipartite correlations by bures distance , http://dx.doi.org/10.1088/1751-8113/47/40/405302[j .
phys . a : math .
47 , 405302 ( 2014 ) ] .
b. aaronson , r. lo franco , and g. adesso , comparative investigation of the freezing phenomena for quantum correlations under nondissipative decoherence , http://dx.doi.org/10.1103/physreva.88.012120[phys .
a 88 , 012120 ( 2013 ) ] .
b. bellomo , g. l. giorgi , f. galve , r. lo franco , g. compagno , and r. zambrini , unified view of correlations using the square norm distance , http://dx.doi.org/10.1103/physreva.85.032104[phys .
a 85 , 032104 ( 2012 ) ] .
b. bellomo , r. lo franco , and g. compagno , dynamics of geometric and entropic quantifiers of correlations in open quantum systems , http://dx.doi.org/10.1103/physreva.86.012312[phys . rev . a 86 , 012312 ( 2012 ) ] .
r. lo franco , b. bellomo , e. andersson , and g. compagno , revival of quantum correlations without system - environment back - action , http://dx.doi.org/10.1103/physreva.85.032318[phys .
a 85 , 032318 ( 2012 ) ] .
xu , k. sun , c .- f .
li , x .- y .
xu , g .- c .
guo , e. andersson , r. lo franco , and g. compagno , experimental recovery of quantum correlations in absence of system - environment back - action , http://dx.doi.org/10.1038/ncomms3851[nat .
commun . 4 , 2851 ( 2013 ) ] .
r. lo franco , b. bellomo , s. maniscalco , and g. compagno , dynamics of quantum correlations in two - qubit systems within non - markovian environments , http://dx.doi.org/10.1142/s0217979213450537[int . j. mod
b 27 , 1345053 ( 2013 ) ] .
t. chanda , t. das , d. sadhukhan , a. k. pal , a. sen de , and u. sen , reducing computational complexity of quantum correlations , http://dx.doi.org/10.1103/physreva.92.062301[phys .
a 92 , 062301 ( 2015 ) ] .
a. beggi , f. buscemi , and p. bordone , analytical expression of genuine tripartite quantum discord for symmetrical x - states , http://dx.doi.org/10.1007/s11128-014-0882-z[quantum inf . process .
14 , 573 ( 2015 ) ] .
g. li , y. liu , h. tang , x. yin , and z. zhang , analytic expression of quantum correlations in qutrit werner states undergoing local and nonlocal unitary operations , http://dx.doi.org/10.1007/s11128-014-0888-6[quantum inf . process .
14 , 559 ( 2015 ) ] .
p. c. obando , f. m. paula , and m. s. sarandy , trace - distance correlations for x states and the emergence of the pointer basis in markovian and non - markovian regimes , http://dx.doi.org/10.1103/physreva.92.032307[phys .
a 92 , 032307 ( 2015 ) ] . | abstract + coherence vectors and correlation matrices are important functions frequently used in physics
. the numerical calculation of these functions directly from their definitions , which involves kronecker products and matrix multiplications , may seem to be a reasonable option .
notwithstanding , as we demonstrate in this article , some algebraic manipulations before programming can reduce considerably their computational complexity .
besides , we provide fortran code to generate generalized gell mann matrices and to compute the optimized and unoptimized versions of the associated bloch s vectors and correlation matrix , in the case of bipartite quantum systems . as a code test and application example , we consider the calculation of hilbert - schmidt quantum discords . |
the presence of renormalon singularities in the pole mass , @xmath16 , implies that the intuitive definition @xmath17 does not correspond to a physical quantity .
thus , if @xmath0 is to be used in phenomenological applications , a definition free of renormalon ambiguities has to be given .
one can define the b meson binding energy from the experimentally measured value of some physical quantity , for example the inclusive @xmath18 lepton spectrum .
it is important to notice that such a definition , although correct , depends on the order of the perturbation series that determines the theoretical prediction for the experimental quantity .
lattice simulations of the hqet provide the opportunity to compute @xmath0 non perturbatively by using the definition of @xmath0 proposed in ref.@xcite which is free of renormalon ambiguities . on the lattice ,
renormalons are absent but matrix elements of lattice operators may contain power ultraviolet divergences . consider the two point correlation function @xmath19 for large values of @xmath20 @xmath21 where @xmath22 creates a b meson from the vacuum .
it is tempting to define @xmath23 .
however , the `` bare '' binding energy @xmath24 can not be a physical quantity because it diverges linearly as @xmath25 , @xmath26 being the lattice spacing @xcite . the origin of the linear divergence is the mixing of the operator @xmath27 with the lower dimensional operator @xmath28 .
as pointed out in @xcite , the linear divergence must be subtrated non perturbatively in order not to reintroduce renormalon ambiguities .
this divergence can be eliminated from any correlation function by defining a renormalized action @xmath29 of the form @xcite |hd^r_4 h = |h d_4 h - ( 1-(-a|m ) ) |hh where @xmath30 is a mass counter
term which can be chosen in many different ways .
our preferred definition of @xmath30 is @xcite [ eq : defdm ] -|m_t ( ) where @xmath31 is the heavy
quark propagator in the landau gauge which can be computed in a numerical simulation of the hqet .
in words , ( [ eq : defdm ] ) is the condition that the subtracted heavy
quark propagator has no exponential fall at large times or equivalently that @xmath32 .
due to confinement effects , it is not clear that ( [ eq : defdm ] ) has a finite limit as @xmath33 .
our numerical results obtained from three independent high statistics simulations at @xmath34 and @xmath15 support our assumption on the infrared bahaviour of the heavy
quark propagator : no confining effects are visible up to distances of about @xmath35 fm , which is the largest distance which we can reach @xcite .
an alternative definition of @xmath36 which does not require any assumption about the large time behaviour of the heavy quark propagator is given in refs.@xcite .
however , with this definition @xmath0 is not a parameter of @xmath37 .
notice also that the divergent part of @xmath30 must be gauge invariant .
moreover , the gauge dependence of possible finite parts is eliminated from physical quantities up to the order of perturbation theory at which the calculation is performed .
the renormalized binding energy @xmath0 which is of order @xmath37 , free of renormalon singularities and independent of the renormalization scale is defined as : @xmath38 .
our best estimate of @xmath0 is @xmath39 mev ( see ref.@xcite for details ) .
we can also compute the @xmath2 running mass of the b
quark , @xmath3 . by comparing the heavy
quark propagator in the continuum with the one on the lattice , we obtain @xcite _ b(_b)= ( m_b - ) _
m(_s , |m ) where @xmath40 is the qcd hqet matching constant ( see ref.@xcite for details ) .
taking @xmath41 gev , we find @xmath42 gev @xcite , where the first error is statistical and the second is an estimation of the error in the truncation at @xmath43 of the perturbative factor @xmath40 , derived from the bubble
resummation approximation @xcite .
this error can be reduced to about @xmath44 mev by computing perturbative corrections at two loop order .
this calculation requires the evaluation of the heavy quark propagator in the continuum , already done , and on the lattice at @xmath45 .
the langevin approach of ref.@xcite can be used to perform the computation on the lattice .
like @xmath0 , @xmath1 has no physical meaning since it depends on the definition that one decides to adopt for the kinetic operator . on the lattice ,
matrix elements of this operator contain linear and quadratic divergences @xcite .
the finite kinetic operator , which is free of power divergences , has the form |h_r^2 h = |h^2 h - |h d^r_4 h - |h h where @xmath46 and @xmath47 are subtraction constants which we define by imposing the `` physical '' non perturbative renormalization prescription @xcite h(=0)|h^2_r hh(=0 ) = 0 in the landau gauge .
this is equivalent to the condition @xmath48 where [ rhorat ] _
^2(t ) with @xmath49 and @xmath50 . to determine @xmath46 and @xmath47
, we fit @xmath51 for large @xmath20 to a straight line .
our numerical results show that the infrared limit of the ratio in eq.([rhorat ] ) does indeed exist , in spite of the confinement effects in the heavy quark propagator .
@xmath47 can also be obtained directly by eliminating the sum over @xmath52 in eq.([rhorat ] ) and searching for a plateau in @xmath20 and @xmath52 .
we find that , after subtracting the contribution of contact terms , both methods give indistinguishable results ( see ref.@xcite for details ) . notice also that , for the same reason as for @xmath36 , @xmath47 must be gauge independent .
the renormalized kinetic energy @xmath1 is then given by : @xmath53 , where @xmath54 and @xmath55 is a gauge invariant matching constant known up to one loop order @xcite .
a convenient way to extract @xmath56 is to study the large time behaviour of the ratio of the two point and the three
point , with an insertion of the kinetic operator , b
meson correlation functions .
the significant reduction of the statistical errors with respect to our previous study allows for a much better identification of the plateau region , which could not be really observed in ref.@xcite .
we find @xmath57 gev@xmath58 and for the physical parameter @xmath59 gev@xmath58 ( see ref.@xcite for details ) . in order to make a meaningful comparison of our results with other theoretical determinations , a perturbative calculation of the relation between the different definitions of @xmath1 , and also of @xmath0 ,
is needed at a sufficient degree of accuracy @xcite .
this calculation is missing to date .
the chromo magnetic operator that enters in the calculation of @xmath6 is only logarithmically divergent in the ultraviolet cut off so that there is no need of a non perturbative subtraction .
moreover , @xmath6 has a physical meaning because it corresponds directly to the measurable @xmath4@xmath5 mass splitting up to higher power corrections in @xmath60 m_b^*^2 - m_b^2 = 4 _ 2 = 4 z _ _
2^bare where @xmath61 is the renormalization constant necessary to remove the logarithmic divergence present in the bare operator .
the procedure to extract @xmath62 is the same as for @xmath63 ( see ref.@xcite for details ) .
we find @xmath64 gev@xmath58 corresponding to a mass splitting which is about one half of the experimental value . in our opinion , there are three possible explanations of the discrepancy , which is common to all lattice results .
the first is the fact that , at one loop order , @xmath61 is very large indicating that higher order terms may modify this factor significantly .
a second source of systematic error is quenching .
the final possibility is that the discrepancy is due to the fact that the mass splitting receives significant contributions from higher terms in the hqet .
further investigation is needed on this subject .
we have also computed the parameter @xmath65 , which enters the theoretical prediction of the @xmath7@xmath8 mixing parameters @xmath66 and @xmath67 .
we measure @xmath10 and @xmath9 from the ratios of three and two point functions with the @xmath68 operator fixed at the origin . due to the high statistics we have accumulated
, we can study this ratios at large time distances at which the ground state has been unambiguosly isolated .
we find a very good signal for all the correlation functions and ratios .
however , the main uncertainty in our results comes from the evaluation of the renormalization factors relating the effective operators on the lattice to those in the continuum . for @xmath9
, we use the new calculation of the nlo wilson coefficients of the @xmath43 full theory hqet matching , which takes into account previously missed contributions @xcite . in our opinion ,
a non perturbative computation of these factors is crucial to reduce the rather big systematic errors .
this work is in progress now @xcite .
our best estimates are @xcite : @xmath69 ( rgi b parameter ) , @xmath70 and @xmath71 , where we use the boosted perturbation theory to evaluate the renormalization constants .
99 g. martinelli and c. t. sachrajda , phys .
b354 * ( 1995 ) 423 . l. maiani , g. martinelli and c. t. sachrajda , nucl . phys .
* b368 * ( 1992 ) 281 .
m. crisafulli , v. gimenez , g. martinelli and c. t. sachrajda , nucl .
phys . + * b457 ( 1995 ) 594 .
v. gimenez , g. martinelli and c. t. sachrajda , cern - th/96 - 163 , + ( hep - lat/9607018 ) .
g. martinelli and c. t. sachrajda , cern - th/96 - 117 , ( hep - ph/9605336 ) . f. di renzo , g. marchesini and e. onofri , nucl . phys . *
b457 * ( 1995 ) 202 .
v. gimenez , g. martinelli and c. t. sachrajda , cern - th/96 - 175 , + ( hep - lat/9607055 ) .
m. ciuchini , e. franco and v. gimenez , cern - th/96 - 206 , ( hep - ph/9608204 ) .
v. gimenez , g. martinelli , c. t. sachrajda and m. talevi , in preparation .
v. gimenez and g. martinelli , in preparation . * | we present a short review of our most recent high statistics lattice determinations in the hqet of the following important parameters in b physics : the b meson binding energy , @xmath0 and the kinetic energy of the b quark in the b meson , @xmath1 , which due to the presence of power divergences require a non perturbative renormalization to be defined ; the @xmath2 running mass of the b quark , @xmath3 ; the @xmath4@xmath5 mass splitting , whose value in the hqet is determined by the matrix element of the chromo magnetic operator between b meson states , @xmath6 ; the b parameter of the @xmath7@xmath8 mixing , @xmath9 , and the decay constant of the b meson , @xmath10 .
all these quantities have been computed using a sample of @xmath11 gauge field configurations on a @xmath12 lattice at @xmath13 . for @xmath0 and @xmath3
, we obtain our estimates by combining results from three independent lattice simulations at @xmath13 , @xmath14 and @xmath15 on the same volume . |
many properties of quantum spin chains depend crucially on whether the hamiltonian is gapped or gapless in the thermodynamic limit .
ground states of gapped hamiltonians are weakly entangled , as quantified by the entanglement area law @xcite , and exhibit an exponential decay of correlation functions @xcite . for such systems
the ground energy and the ground state itself can be efficiently computed using algorithms based on matrix product states @xcite . on
the other hand , ground states of gapless spin chains can exhibit drastic violations of the entanglement area law @xcite , and computing the ground energy can be quantum - np hard @xcite .
spin chain models studied in physics usually become gapless along quantum phase transition lines separating distinct gapped phases @xcite . deciding whether a given family of hamiltonians is gapped or gapless in the thermodynamic limit is therefore a fundamental problem . in this paper
we provide a complete solution of this problem for a class of translation - invariant chains of qubits with nearest - neighbor interactions .
let @xmath7 be a fixed two - qubit state with @xmath8 .
consider a chain of @xmath1 qubits with open boundary conditions and define a hamiltonian @xmath9 here each term is a rank-@xmath3 projector onto @xmath0 applied to a consecutive pair of qubits
. we shall refer to @xmath0 as the _ forbidden state _ since the hamiltonian penalizes adjacent qubits for being in the state @xmath0 . as we will see in section [ sec : ground space ] , the hamiltonian @xmath2 is frustration - free for any choice of @xmath0 , that is , ground states of @xmath2 are zero eigenvectors of each individual projector @xmath10 and the ground energy of @xmath2 is zero .
futhermore , the ground state degeneracy of @xmath2 is equal to @xmath11 for almost all choices of @xmath0 .
we are interested in the spectral gap separating the ground states and the excited states of @xmath2 or , equivalently , the smallest non - zero eigenvalue of @xmath2 .
to state our results , define a @xmath5 matrix @xmath12 here @xmath13 is the standard basis of @xmath14 . as we will see , the matrix @xmath15 is crucial for understanding the structure of the ground space of @xmath2 . in this paper
we prove that the eigenvalues of @xmath15 determine if @xmath2 is gapped or gapless is basis - dependent , eigenvalues of @xmath15 are invariant under global @xmath16 rotations .
in particular , one can check that a transformation @xmath17 with a single - qubit unitary operator @xmath18 maps @xmath15 to @xmath19 . ] .
our main result is the following .
[ thm : main ] let @xmath0 be an arbitrary two - qubit state .
suppose the eigenvalues of @xmath15 have equal non - zero absolute value .
then the spectral gap of @xmath2 is at most @xmath4 .
otherwise the spectral gap of @xmath2 is lower bounded by a positive constant independent of @xmath1 , which depends only on the forbidden state @xmath0 .
we now motivate our choice of the model eq .
( [ h ] ) , highlight previous work on related models , and provide some intuition for why the eigenvalues of @xmath15 appear in the statement of the theorem .
an informal sketch of the proof is provided in section [ sketch ] .
below we write @xmath20 for the spectral gap of @xmath2 . the family of hamiltonians defined in eq .
( [ h ] ) includes some well - known quantum models as special cases .
for example , choosing @xmath0 proportional to @xmath21 ( the singlet state ) one can easily check that @xmath2 coincides with the ferromagnetic heisenberg chain up to an overall energy shift . for this model
@xmath2 has spectral gap @xmath22 which decays as @xmath23 for large @xmath1 @xcite .
note that in this case @xmath15 is proportional to the identity matrix , so theorem [ thm : main ] gives an upper bound @xmath24 .
koma and nachtergaele studied a one - parameter deformation of the heisenberg chain known as the ferromagnetic xxz chain with kink boundary conditions @xcite . in this example
@xmath0 is proportional to @xmath25 for @xmath26 and the spectral gap of @xmath2 is given by @xmath27 for all @xmath28 , see @xcite for details .
one can check that @xmath15 is a diagonal matrix with eigenvalues @xmath29 and @xmath30 .
it follows that @xmath31 for any @xmath32 and theorem [ thm : main ] asserts that @xmath2 has a constant spectral gap .
we note that in the two special cases considered above the hamiltonian has a symmetry which enables an exact computation of the spectral gap .
such symmetries are not available for a general state @xmath0 .
the exact results summarized above may suggest that the hamiltonian @xmath2 is gapless if @xmath0 is a maximally entangled state and gapped otherwise .
theorem [ thm : main ] demonstrates that this naive intuition is wrong .
indeed , choose @xmath0 proportional to @xmath33 for some @xmath34
. then the matrix @xmath15 has eigenvalues @xmath35 and theorem [ thm : main ] implies that @xmath2 is gapless for all @xmath36 as above .
as a simple application of theorem [ thm : main ] , we now map out the phase diagram of @xmath2 restricted to the subset of _ real _ states @xmath37 . using the schmidt decomposition any real two - qubit state can be written as @xmath38 , \quad \quad r(\theta)\equiv \left ( \begin{array}{cc } \cos{(\theta ) } & \sin{(\theta ) } \\
-\sin{(\theta ) } & \cos{(\theta ) } \\ \end{array }
\right ) , \label{realstates}\ ] ] for some @xmath39 and @xmath40 $ ] . since the spectrum of @xmath41 is invariant under a simultaneous rotation of all qubits , the spectral gap depends only on two parameters @xmath42 and @xmath36 .
one can easily check that the eigenvalues of @xmath43 have equal non - zero magnitude iff @xmath44 on the other hand , the eigenvalues of @xmath45 have equal non - zero magnitude iff either @xmath46 , or @xmath47 and @xmath34 .
these conditions determine the gapless phase of the model for the special case of real states @xmath0 .
the gapped and gapless regions for @xmath48 as a function of @xmath49 are shown in fig .
[ fig : real ] .
a surprising feature is that the gapless phase occupies a finite volume in the parameter space .
in contrast , most of the models studied in physics only become gapless along phase transition lines which have zero measure in the parameter space . where @xmath0 has real amplitudes .
we use the parameterization @xmath50 $ ] and we show the gapped and gapless phases for @xmath48 as a function of @xmath51 $ ] and @xmath52 $ ] .
the phase diagram is symmetric under flipping the sign of @xmath42 and under the transformation @xmath53 . the sector corresponding to @xmath54 is not shown since it has a simple description : @xmath54 is in the gapless phase iff either @xmath46 , or @xmath47 and @xmath34 .
[ fig : real ] ] while it is possible to construct frustration - free translation - invariant hamiltonians on qubits which are composed of projectors of rank @xmath55 or @xmath56 , one can show that there are only a handful of such examples . in the appendix
we describe them and for each we determine if the system is gapped or gapless . taken together with our main result
, this gives a complete classificaction of gapped and gapless phases for frustration - free translation - invariant qubit chains with nearest - neighbor interactions .
note that the restriction to hamiltonians which are sums of projectors is without loss of generality .
where @xmath57 has smallest eigenvalue zero ( which can be arranged by adding a constant times the identity ) . since @xmath58 is frustration - free , it has the same null space as @xmath59 where @xmath60 projects onto the range of @xmath57 . using this fact and the inequality @xmath61 , where @xmath62 is the smallest non - zero eigenvalue of @xmath57 , we see that @xmath58 is gapped if and only if @xmath63 is . ]
there are several open questions related to our work .
we do not know if the gapless phase of the model eq .
( [ h ] ) can be connected to some known universality class of critical spin chains and what is the actual scaling of the spectral gap in the gapless phase . in particular
, we do not expect that the the upper bound @xmath4 on the spectral gap in theorem [ thm : main ] is tight .
it is a challenging open problem to generalize our results to qudits , i.e. , to map out the phase diagram of translation - invariant frustration - free spin chains for @xmath64-dimensional spins with @xmath65 .
a natural analogue of the hamiltonian defined in eq .
( [ h ] ) is @xmath66 where @xmath60 is a rank-@xmath67 projector acting on @xmath68 .
it was shown by movassagh et al @xcite that such chains are frustration - free for any @xmath60 and @xmath1 whenever @xmath69 .
their results also suggest that the hamiltonian @xmath70 may be generically frustrated for @xmath71 and @xmath1 sufficiently large . on the other hand ,
when @xmath71 , the hamiltonian is frustration - free for certain special choices of @xmath60 and @xmath1 ; examples include the famous aklt model @xcite ( with @xmath72 , @xmath73 ) , the model based on motzkin paths @xcite ( with @xmath74 ) , and the `` product vacua with boundary states '' models @xcite ( with @xmath75 ) .
in general there is no efficient algorithm for testing whether @xmath70 is frustration - free for a given @xmath1 and there are indications that this problem may be computationally hard @xcite .
it is therefore natural to focus on the case @xmath69 , where the chain is guaranteed to be frustration - free .
a next step could be to investigate the phase diagram of a chain of qutrits ( @xmath72 ) with projectors of rank @xmath76 .
finally , if one moves from frustration - free one - dimensional chains to general two - dimensional systems , the problem of distinguishing between gapped and gapless phases of translation - invariant hamiltonians becomes undecidable @xcite which leaves no hope for mapping out the full phase diagram of such systems .
_ gapless phase . _ in section [ sec : gapless ] we consider the case when eigenvalues of @xmath15 have the same non - zero magnitude and prove that the spectral gap of @xmath2 is at most @xmath4 .
the proof uses a result of knabe @xcite relating the spectral gap of @xmath2 to that of the following hamiltonian @xmath77 which describes the chain with periodic boundary conditions .
the other ingredient in the proof is a detailed understanding of the ground state degeneracy of @xmath78 .
we will see that @xmath78 is always frustration - free , but its ground state degeneracy can be smaller than that of @xmath2 .
in particular , if @xmath79 is not proportional to the identity operator then @xmath78 has a two - dimensional ground space whereas @xmath2 has an @xmath11-dimensional ground space . otherwise , if @xmath80 , then both hamiltonians @xmath2 and @xmath78 have ground space degeneracy @xmath11 .
we now sketch how these two ingredients can be used to prove the stated result .
for ease of presentation we focus on the example considered above , where @xmath48 is of the form given in eq . .
recall that the spectrum of @xmath81 depends only on the two parameters @xmath82 .
we can plot the ground state degeneracy of @xmath83 as a function of these two parameters .
as described in the previous paragraph , this function takes the value @xmath55 or @xmath11 depending on whether or not @xmath84 is proportional to the identity .
the black lines in figures [ fig : tncurves](a ) and [ fig : tncurves](b ) show the curves where the ground state degeneracy is equal to @xmath11 , for @xmath85 and @xmath86 respectively .
everywhere else ( @xmath87 and @xmath51 $ ] ) the ground state degeneracy is @xmath55 . for reference
we show the red and blue regions from figure [ fig : real ] , which correspond to the gapped and gapless phases of the open boundary chain in the thermodynamic limit .
as one might guess by looking at the figure , the black curves become dense in the blue region when @xmath88 . if we consider a point @xmath48 in this blue region which does not sit directly on one of the black curves then the eigenvalue gap of @xmath89 is equal to its third smallest eigenvalue .
however as @xmath88 this point @xmath48 becomes arbitrarily close to a black curve , where the hamiltonian has ground state degeneracy @xmath11 and third smallest eigenvalue equal to zero . using a bound on its derivative
one can show that as a result the third eigenvalue of @xmath89 takes arbitrarily small values as @xmath88 .
finally , knabe s result implies that this can occur only if the spectral gap of @xmath81 is at most @xmath4 .
this argument has to be modified slightly for states @xmath48 which , for some @xmath1 , lie directly on one of the black curves and for general ( complex ) states @xmath0 .
_ gapped phase .
_ in sections [ sec : mono],[sec : gapped ] we prove that the spectral gap of @xmath2 is lower bounded by a positive constant independent of @xmath1 if the eigenvalues of @xmath15 have distinct magnitudes or if both eigenvalues are equal to zero .
our starting point is a general method for bounding the spectral gap of frustration - free spin chains due to nachtergaele @xcite , see lemma [ lemma : n ] in section [ sec : gapped ] . to apply this method one has to manipulate expressions that involve the projector onto the ground space of @xmath2 which we denote @xmath90 .
the main technical difficulty that we had to overcome is a lack of an explicit expression for @xmath90 which prevents us from straightforwardly applying nachtergaele s bound .
our proof is therefore indirect and is based on establishing some features of the ground space which allow us to control @xmath90 sufficiently well .
the key technical ingredient is a new operator inequality which expresses a monotonicity of the ground space projectors under the partial trace .
more precisely , we show that @xmath91 where the partial trace is taken over the @xmath1-th qubit .
using the fact that the hamiltonians @xmath2 are frustration - free one can easily check that @xmath92 and @xmath93 have the same support , that is , eq . ( [ mintro ] ) is equivalent to saying that all non - zero eigenvalues of @xmath92 are at least one .
our proof of this monotonicity property , presented in section [ sec : mono ] , applies to general frustration - free chains of qubits composed of rank-@xmath3 projectors .
neither translation - invariance nor the conditions of theorem [ thm : main ] are needed for the proof of eq .
( [ mintro ] ) .
we note that eq .
( [ mintro ] ) differs from the well - known monotonicity property @xmath94 .
the latter follows trivially from the fact that @xmath2 is frustration - free , whereas eq .
( [ mintro ] ) holds for more subtle reasons .
we proceed by showing that a quantum state which is completely mixed over the ground space of the @xmath1-qubit chain ( i.e. , proportional to the projector @xmath90 ) exhibits an exponential decay of correlations for certain local observables , see lemma [ lemma : decay ] in section [ subs : corrfunc ] . in section
[ subs : regionexclusion ] we use the decay of correlations and eq .
( [ mintro ] ) to prove several region exclusion " lemmas . here
we consider a partition of the chain into three or more regions and define local ground space projectors associated with each region . loosely speaking
, the region exclusion lemmas state that the global ground space projector associated with the entire chain can be approximated by a certain operator built from the local ground space projectors .
the latter are defined on subsets of qubits where some of the chosen regions are excluded from the chain ( hence the name of the lemmas ) . by repeatedly applying the region exclusion lemmas in section [ subs : finalgappedproof ]
we arrive at the condition used in nachtergaele s bound , thus proving a constant lower bound on the gap .
in this section we describe the ground spaces of @xmath2 and @xmath95 , the hamiltonians for the chain with open and periodic boundary conditions respectively ( defined in eqs .
( [ h],[hcirc ] ) ) .
we first consider the hamiltonian @xmath2 for the chain with open boundary conditions .
we begin with the simple case where @xmath96 is a product state .
it is always possible to choose the basis states @xmath97 and @xmath98 so that @xmath99 where @xmath100 for each @xmath101 define an @xmath1-qubit state @xmath102 . also define @xmath103 . for example , choosing @xmath104 one gets @xmath105 loosely speaking , the states @xmath106 can be viewed as domain walls " where @xmath97 and @xmath107 represent two different values of a magnetization . by direct inspection
we see that @xmath108 are pairwise orthogonal ground states of @xmath2 .
[ prop : prod_basis ] suppose @xmath109 .
then the states @xmath108 form an orthonormal basis for the ground space of @xmath2 .
it suffices to show that the ground space of @xmath2 has dimension at most @xmath11 .
define @xmath110 and @xmath111 . given any binary string @xmath112 ,
define @xmath113 .
note that @xmath114 and @xmath115 are linearly independent since @xmath109 .
therefore the states @xmath116 , @xmath117 form a basis ( non - orthonormal ) for the hilbert space of @xmath1 qubits .
suppose @xmath118 is a ground state of @xmath2 .
then @xmath119 for some complex coefficients @xmath120 .
a simple calculation shows that @xmath121 for any @xmath122 . on the other hand , @xmath123 since @xmath118 is a ground state of @xmath2 .
this is possible only if @xmath124 for all strings @xmath125 that contain at least one consecutive pair @xmath126 .
thus @xmath118 belongs to a subspace spanned by vectors @xmath127 , where @xmath128 .
this shows that the ground subspace of @xmath2 has dimension at most @xmath11 .
now consider the case where @xmath0 is entangled .
in this case we can still construct the ground space of @xmath2 although , in contrast with the product state case , we are not able to obtain an orthonormal basis . the matrix @xmath15 defined in eq .
( [ tpsi ] ) plays a crucial role .
one can easily check that @xmath129 whenever @xmath0 is entangled and @xmath130 where @xmath131 is the antisymmetric state of two qubits .
this shows that the ground space of @xmath132 is the image of the @xmath55-qubit symmetric subspace under the map @xmath133 .
a similar characterization holds for @xmath2 with @xmath134 .
in particular , define @xmath135 the following proposition is a special case of a result presented in @xcite ( and has been used previously in , e.g. , @xcite ) .
[ gsdegen ] suppose @xmath129 .
then the ground space of @xmath2 is the image of the @xmath1-qubit symmetric subspace under the linear map @xmath136 . using eq . and the fact that @xmath137 we get @xmath138 for each @xmath139 , where @xmath140 is a positive operator acting on qubits in the set @xmath141\setminus\{j , j+1\}$ ] . from eq .
we see that the nullspace of @xmath142 is equal to the symmetric subspace .
the result follows since @xmath143 is invertible .
combining propositions [ prop : prod_basis],[gsdegen ] and noting that the symmetric subspace of @xmath1 qubits has dimension @xmath11 , we conclude that the ground space of @xmath2 has dimension @xmath11 for almost any choice of @xmath0 ( the only exception is when @xmath144 and @xmath0 is a symmetric product state ) .
we now consider the hamiltonian @xmath95 for the chain with periodic boundary conditions .
it is well - known that @xmath145 is frustration - free for any choice of @xmath0 , see for instance @xcite .
however in this paper we will only need to deal with periodic boundary conditions in the case where @xmath0 is an entangled state .
accordingly , in this section we assume that @xmath146 . for any such @xmath0
we compute the dimension of the zero energy ground space of @xmath145 .
we will see that it takes different values depending on the choice of @xmath0 .
this contrasts with the open boundary chain which has ground space dimension @xmath11 whenever @xmath0 is entangled .
here and throughout the paper we use the symbol @xmath147 to mean proportional to .
[ gs_counting ] suppose @xmath148 .
then the ground space of @xmath145 has dimension @xmath11 .
otherwise , @xmath145 has a two - fold degenerate ground space . note that @xmath145 has the same rank as @xmath149 where @xmath136 is given by eq . .
both terms on the right - hand side are positive semidefinite and , by proposition [ gsdegen ] , the nullspace of the first term is the symmetric subspace .
if @xmath148 then the second term in eq . can be written as @xmath150 where @xmath151 is positive and @xmath131 . since this term annihilates every state in the symmetric subspace we see that in this case the nullspace of eq .
is @xmath152-dimensional .
if @xmath153 is not proportional to the identity we show that there are exactly two states in the symmetric subspace which are annihilated by the second term in eq . .
we consider two cases depending on whether or not @xmath153 is defective ( has only one eigenvector ) .
first consider the case where @xmath153 has two linearly independent eigenvectors @xmath154 .
note that the last term in eq .
( [ eq : q_periodic ] ) projects qubits @xmath155 onto a state @xmath156 the last equality makes it clear that @xmath157 and @xmath158 are linearly independent whenever @xmath153 is not proportional to the identity .
thus the nullspace of eq .
( [ eq : q_periodic ] ) is spanned by @xmath1-qubit symmetric states that are orthogonal to @xmath157 on any pair of qubits .
one can easily check that the only two - qubit symmetric states orthogonal to @xmath157 are @xmath159 and @xmath160 .
likewise , one can check that the only @xmath1-qubit symmetric states orthogonal to @xmath157 on any pair of qubits are linear combinations of @xmath161 and @xmath162 . thus eq .
( [ eq : q_periodic ] ) has a two - dimensional nullspace and therefore the same is true for @xmath95 .
next suppose @xmath153 is defective , i.e. , has only one eigenvector .
let us work in a basis where @xmath97 is this eigenvector , so @xmath163 for some @xmath164 with @xmath165 .
then the last term in eq .
( [ eq : q_periodic ] ) projects onto a state @xmath166 since @xmath167 , the states @xmath157 and @xmath158 span the same subspace as @xmath168 and @xmath158 .
therefore the nullspace of eq .
( [ eq : q_periodic ] ) is spanned by @xmath1-qubit symmetric states that are orthogonal to @xmath168 on any pair of qubits .
one can easily check that the only such states are linear combinations of @xmath169 and the @xmath1-qubit w - state @xmath170 thus eq .
( [ eq : q_periodic ] ) has a two - dimensional nullspace and therefore the same is true for @xmath95 .
in this section we prove the first part of theorem [ thm : main ] , namely , [ gaplessthm ] suppose the eigenvalues of @xmath171 have the same non - zero absolute value
. then @xmath172 for all @xmath28 .
recall that @xmath20 denotes the smallest non - zero eigenvalue of @xmath2 .
in addition we write @xmath173 for the smallest non - zero eigenvalue of the hamiltonian @xmath145 with periodic boundary conditions , see eq .
( [ hcirc ] ) . to prove the gapless phase theorem we use the following lemma , proven by knabe @xcite , which relates the smallest non - zero eigenvalues of the chains with periodic and open boundary conditions .
knabe s result , presented in section 2 of reference @xcite , applies to more general frustration - free spin chains but here we specialize to the case at hand .
[ knabe @xcite]for all @xmath174 , @xmath175 [ lemma : knabe ] this lemma was originally proposed as a technique for proving that the periodic chain is gapped in the thermodynamic limit .
this follows from the lemma if one can show that there exists a finite @xmath1 for which the open chain has a gap strictly larger than @xmath4 .
here we apply the lemma in the opposite direction .
we use the following strategy which works for some ( but not all ) @xmath0 satisfying the conditions of the gapless phase theorem .
first we apply the argument sketched in section [ sketch ] to show that @xmath176 can take arbitrarily small values for large enough @xmath177 .
then we apply knabe s lemma to infer that @xmath24 for any @xmath134 since otherwise eq .
( [ knabe ] ) would provide a constant lower bound on @xmath176 for all @xmath178 , leading to a contradiction .
note also that @xmath179 since @xmath180 . for some states @xmath0 we are not able to use the above strategy directly
; however in these cases we choose a state @xmath181 which can be taken arbitrarily close to @xmath0 for which the strategy can be applied .
the result for @xmath0 then follows by continuity . in order to handle these cases
( and for other portions of the proof ) we will need the following straightforward bound on how much the eigenvalues of @xmath2 ( or @xmath182 can change as @xmath0 varies .
write @xmath183 for the eigenvalues of @xmath2 and @xmath184 respectively .
[ weyl]let @xmath0 and @xmath181 satisfy @xmath185 .
then @xmath186 for each @xmath187 the proof of the two inequalities is almost identical so here we prove only the first one .
we use the weyl inequality for perturbed eigenvalues ( see for example corollary iii.2.6 of reference @xcite ) which in this case says @xmath188 to complete the proof we bound @xmath189 where in the last line we used the fact that @xmath190 ( since @xmath191 and @xmath157 are normalized ) .
we now proceed to the proof of the gapless phase theorem .
first we claim that the eigenvalues of @xmath2 and absolute values of the eigenvalues of @xmath15 are invariant under a transformation @xmath192 where @xmath18 is an arbitrary single - qubit unitary operator .
indeed , let @xmath193 . then eq .
( [ h ] ) implies @xmath194 and eq .
( [ teps ] ) implies @xmath195 .
thus the eigenvalues of @xmath196 and magnitudes of the eigenvalues of @xmath197 do not depend on @xmath18 .
we shall use the freedom in choosing @xmath18 to bring @xmath0 into a certain canonical form as defined below .
[ prop : canonical ] for any @xmath198 there exists a single - qubit unitary @xmath18 such that @xmath199 for some real coefficients @xmath200 .
since the proof is rather straightforward , we shall postpone it until the end of this section . from now on we can assume that @xmath0 has the canonical form as in the right - hand side of eq .
( [ psi_canonical ] ) . substituting this canonical form into eq .
( [ tpsi ] ) one gets @xmath201 the eigenvalues of this matrix are @xmath202 and @xmath203 , with magnitudes @xmath204 and @xmath205 . by assumption of the theorem
, the eigenvalues have the same magnitude and thus @xmath206 .
we consider the two cases @xmath206 separately and we show that @xmath207 in each case . * case 1 : @xmath208 * fix @xmath1 and let @xmath209 be even . setting @xmath210 in eq . and
taking the square we see that @xmath211 .
since @xmath177 is even we get @xmath212 and therefore @xmath213 by proposition [ gs_counting ] .
let @xmath214 be a normalized state which satisfies @xmath215 and such that the eigenvalues of @xmath216 have different magnitudes and are both non - zero .
this guarantees that @xmath217 is not proportional to the identity and @xmath218 .
then , by propositions [ gs_counting ] and [ weyl ] , @xmath219 applying lemma [ lemma : knabe ] gives @xmath220 for all even @xmath209 , and using propositions [ gs_counting ] and [ weyl ] again we have @xmath221 the result follows since @xmath209 can be taken arbitrarily large .
* case 2 : @xmath222 * let @xmath223 where @xmath224 and @xmath67 are positive real numbers . here
@xmath225 since @xmath0 is normalized and @xmath226 since the eigenvalues of @xmath15 are assumed to be non - zero .
we first consider the case where @xmath224 is irrational . in this case
the convergents of the continued fraction expansion of @xmath224 give sequences of positive integers @xmath227 and @xmath228 with @xmath229 @xmath230 , and where @xmath228 diverges .
here we shall omit the first two convergents obtained by the standard continued fraction expansion , in order to guarantee that the sequence @xmath231 is strictly increasing and @xmath232 for all @xmath233 .
define @xmath234 , and let @xmath235 then @xmath236 where we used the inequality @xmath237 , eq . , and the fact that @xmath225 . note that @xmath238 has eigenvalues @xmath239 and @xmath240 .
we have @xmath241 , which follows from the fact that @xmath242 is not an integer , since @xmath232 .
thus @xmath243 is diagonalizable .
furthermore , @xmath244 , and therefore ( using the fact that it is diagonalizable ) @xmath245 .
hence @xmath246 by proposition [ gs_counting ] . on the other hand @xmath247 is not proportional to the identity since @xmath224 is irrational , hence @xmath248 . using these facts and proposition [ weyl ] we have @xmath249 now for all @xmath233 such that @xmath250 we get @xmath251 and hence @xmath252 since the sequence @xmath228 diverges and the second term can be made arbitrarily small .
it remains to consider the case where @xmath224 is rational . in this case , for any @xmath253 we may choose @xmath254 to be an irrational number satisfying @xmath255 . letting @xmath256 we may now apply the above proof to get @xmath257 . now using proposition [ weyl ] we get @xmath258 where in the second line we used the same reasoning as eq . .
since @xmath253 can be chosen arbitrarily small we get @xmath252 .
finally , let us prove proposition [ prop : canonical ] .
recall that a transformation @xmath259 maps @xmath15 to @xmath19 . here
@xmath18 is an arbitrary unitary operator .
we shall choose a sequence of such transformations that bring @xmath15 into the canonical form defined in eq .
( [ transfer ] ) which is equivalent to eq .
( [ psi_canonical ] ) .
first , choose @xmath18 such that @xmath97 is an eigenvector of @xmath15 .
this is always possible since any complex matrix has at least one eigenvector .
now we can assume that @xmath260\ ] ] for some complex coefficients @xmath261 .
next choose @xmath262 where the phase @xmath224 satisfies @xmath263 .
this maps @xmath15 to @xmath264 and now we can assume that @xmath265 .
finally , choosing @xmath266 one can map @xmath267 to @xmath268 without changing @xmath269 .
thus we can make @xmath267 real .
this brings @xmath15 into the canonical form defined in eq .
( [ transfer ] ) .
in this section we establish a relationship between the ground space projectors describing a chain of @xmath1 and @xmath270 qubits .
our result holds in a more general setting than considered elsewhere in this paper since we do not assume translation invariance .
let @xmath271 be an arbitrary sequence of normalized two - qubit states . for each @xmath272 define a hamiltonian @xmath273 which describes a chain of @xmath1 qubits .
this hamiltonian is frustration - free for any choice of @xmath274 ( this follows directly from proposition [ dimincreasing ] given below ) .
let @xmath275 be the ground space of @xmath276 and @xmath90 be the projector onto @xmath275 .
we adopt the convention @xmath277 and @xmath278 .
first , we note that @xmath279 due to the fact that the considered hamiltonians are frustration - free .
this results in a trivial monotonicity property @xmath94 .
below we prove that one also has a different type of monotonicity , namely @xmath280 , where @xmath281 represents the partial trace over the @xmath1-th qubit .
for each @xmath272 one has [ lemma : mono ] @xmath282 given the simplicity and generality of eq .
( [ mono ] ) , one may be tempted to ask whether it holds for some trivial reason unrelated to the structure of the considered hamiltonians .
we have observed numerically that eq .
( [ mono ] ) can be false if @xmath93 and @xmath90 are chosen as projectors onto random linear subspaces @xmath283 and @xmath284 , even if the dimensions of @xmath285 and @xmath275 match those of the ground subspaces of @xmath286 and @xmath287 .
thus any proof of the monotonicity property must exploit the special structure of the projectors @xmath90 . in the absence of an explicit formula for @xmath90
, one has to rely on some indirect arguments in order to derive eq .
( [ mono ] ) .
this partially explains why the proof of the lemma given below is rather cumbersome .
we use induction in @xmath1 .
the base of the induction is @xmath288 . in this case @xmath289 and thus @xmath290 .
here we used the fact that the partial trace of any two - qubit state is a density matrix which has eigenvalues at most one .
we now prove the induction step . for brevity denote @xmath291 such that the last term in @xmath287 is @xmath292 .
first consider the case where @xmath0 is unentangled , that is , @xmath293 for some single - qubit states @xmath294 . in this case
the result follows trivially without using the inductive hypothesis , since @xmath295 which implies @xmath296 and thus @xmath297 . in the remainder of the proof we consider the case where @xmath0 is entangled ( i.e. , not a product state ) .
write the schmidt decomposition of @xmath0 as @xmath298 and @xmath299 with @xmath300 .
let @xmath301 .
obviously , @xmath302 .
furthermore , the trivial monotonicity @xmath279 implies that @xmath303 has all of its support on @xmath285 , that is , @xmath304 .
define an operator @xmath305 the above implies that @xmath306 thus it suffices to prove that @xmath307 .
choose an arbitrary orthonormal basis @xmath308 also choose an arbitrary orthonormal basis @xmath309 in general the dimensions @xmath67 and @xmath310 of the spaces @xmath311 and @xmath285 will depend on the states @xmath312 but we will not need an explicit expression for them .
we will however need to use the fact that @xmath313 , which we now establish .
the following proposition is a special case of the result presented in @xcite ; we include a proof here for completeness .
[ dimincreasing ] let @xmath314 be the dimension of @xmath315 .
then @xmath316 for all @xmath317 .
recall our convention that @xmath278 , so @xmath318 . on the other hand @xmath319 and @xmath320 , which confirms @xmath321 .
we now establish that @xmath322 for all @xmath323 .
this is sufficient to complete the proof since it implies @xmath324 .
let @xmath181 be a general state in @xmath275 , with @xmath323 .
let @xmath325 be an orthonormal basis for @xmath285 and let @xmath326 be an orthonormal basis for @xmath311 .
we can write @xmath327 for some complex coefficients @xmath328 .
the fact that the hamiltonian @xmath287 is frustration - free implies @xmath329 and thus the dimension of @xmath275 is the number of linearly independent solutions to the equations @xmath330 this is a set of @xmath331 linear equations for the @xmath332 variables @xmath328 .
the number of linearly independent solutions satisfies @xmath333 , or equivalently @xmath322 .
define @xmath334 matrices @xmath335 where @xmath336 are defined in eq .
( [ psischmidt ] ) .
the trivial monotonicity @xmath337 implies @xmath338 here and below @xmath339 denotes the identity matrix of dimension @xmath340 .
furthermore , expressing @xmath341 one gets @xmath342 since @xmath343 by the induction hypothesis , we infer that @xmath344 the usefulness of the matrices @xmath345 comes from the following facts . recall that we defined @xmath346 .
[ prop : r ] suppose @xmath0 is entangled .
then the matrix of the operator @xmath347 in the chosen basis @xmath348 of @xmath285 can be written as @xmath349 where @xmath350 are defined by eq .
( [ psischmidt ] ) .
[ prop : tricky ] let @xmath351 be arbitrary positive integers with @xmath352 .
let @xmath345 be arbitrary matrices of size @xmath334 satisfying @xmath353 and @xmath354 .
let @xmath355 be any real positive numbers .
then the operator @xmath347 defined by eq .
( [ rsimple1 ] ) satisfies @xmath307 . combining propositions [ prop : r],[prop : tricky ] , the inequality @xmath313 proved in proposition [ dimincreasing ] , and eqs .
( [ norm1],[norm2 ] ) one gets @xmath356 .
the lemma follows from eq .
( [ r1 ] ) . in the rest of this section
we prove the above propositions . denoting
@xmath357 one can rewrite @xmath347 as @xmath358 by symmetry we can assume that @xmath359 .
then eq .
( [ norm2 ] ) implies @xmath360 since the function @xmath361 is operator monotone , we arrive at @xmath362 , where @xmath363 hence it suffices to prove that @xmath364 . from eq .
( [ norm1 ] ) one infers that @xmath365 and @xmath366 . since @xmath367 has @xmath368 fewer rows than columns , it must have at least this many linearly independent vectors in its nullspace . from eq .
( [ norm1 ] ) one infers that for any @xmath369 with @xmath370 and @xmath371 we have @xmath372 . thus @xmath373 has at least @xmath368 singular values equal to @xmath3 .
likewise , @xmath367 has at least @xmath368 singular values equal to @xmath3 .
note that this implies that @xmath374 since @xmath375 is an @xmath376 matrix with at least @xmath368 eigenvalues equal to @xmath3 .
furthermore , conditions eqs .
( [ norm1],[norm2 ] ) and the norm of @xmath377 are invariant under a transformation @xmath378 , where @xmath379 and @xmath380 are arbitrary unitary matrices .
we can always choose @xmath379 and @xmath380 to bring @xmath373 into a diagonal form such that the diagonal matrix elements of @xmath373 are non - negative and non - increasing .
thus we can assume without loss of generality that @xmath381\ ] ] where @xmath382 ( if @xmath383 the above equation should read @xmath384 . ) .
here and below @xmath385 denotes an all - zeros matrix of size @xmath386 .
it follows that @xmath387 and thus @xmath388 , \quad \mbox{where } \quad \tilde{s}_1 = x\hat{d}^2 ( i_{r}+(x-1)\hat{d}^2)^{-1},\ ] ] and @xmath389 the above arguments also show that @xmath390 is in the nuillspace of @xmath367 iff @xmath181 has support on basis vectors @xmath391 with @xmath392 .
thus we can assume wlog that @xmath393,\ ] ] where @xmath394 is some matrix of size @xmath376 .
substituting eqs .
( [ m1],[m0 ] ) into eq .
( [ norm1],[norm2 ] ) yields @xmath395 and @xmath396 using the polar decomposition of @xmath394 and eq .
( [ mdagm ] ) we obtain the parameterization @xmath397 , where @xmath18 is unitary .
then eq .
( [ mmdag ] ) is equivalent to @xmath398 , or @xmath399 using the definition of @xmath400 one gets @xmath401 , \quad \mbox{where } \quad \tilde{s}_0 = m^\dag ( i_{r } + ( x-1 ) d^2)^{-1 } m.\ ] ] combining eqs .
( [ eq2],[eq3 ] ) , it suffices to show that @xmath402 . using the chosen parameterization of @xmath394 one
gets @xmath403 now eq . ( [ mmdag1 ] ) implies @xmath404 .
since @xmath361 is an operator monotone function , it follows that @xmath405 that is , @xmath406 this proves that @xmath402 which implies @xmath407 and thus @xmath307 .
we first show that @xmath408 note that the two tensor products in @xmath409 and in @xmath410 refer to two different partitions of the chain .
we use the following two general properties of the orthogonal complement ( here @xmath411 are subspaces of a hilbert space ) @xmath412 we have @xmath413 applying eqs .
( [ comp1],[comp2 ] ) gives @xmath414 where to get the last line we absorbed the second and fourth terms into the first , using the fact that @xmath415 .
choose an arbitrary orthonormal basis @xmath416 from eq .
( [ ground1 ] ) one infers that @xmath417 is spanned by @xmath418 where @xmath419 are the schmidt vectors of @xmath0 as defined in eq .
( [ psischmidt ] ) .
we now show that the gram matrix @xmath420 defined by @xmath421 is invertible , which implies that @xmath422 are linearly independent .
we note that @xmath420 has the following simple form @xmath423\ ] ] where @xmath424 define @xmath425 $ ] , @xmath426 , and @xmath427 .
note that @xmath428 and @xmath429 have the same non - zero eigenvalues .
also note that @xmath420 is invertible if none of these eigenvalues are equal to @xmath3 , since in this case @xmath430.\ ] ] to show that @xmath420 is invertible it therefore suffices to show that @xmath431 is invertible .
using eqs .
( [ gram],[bdef ] ) and the identity @xmath432 we get @xmath433 so @xmath434 to prove that @xmath431 is invertible we show that this operator is positive definite : @xmath435 where we used eq .
( [ norm2 ] ) and the fact that @xmath436 are both positive .
this completes the proof that @xmath420 is invertible and establishes that @xmath437 are linearly independent . since we have shown that @xmath437 are a basis for @xmath417 , we have @xmath438 substituting eqs .
( [ ground3],[invgamma ] ) into @xmath439 and noting that @xmath440 with @xmath441 are the only basis vectors of @xmath417 which are not orthogonal to @xmath442 , we arrive at @xmath443 substituting @xmath341 and @xmath444 into the above equation yields @xmath445 replacing the last two factors by matrix elements of @xmath345 defined in eq .
( [ gram ] ) and @xmath431 by eq .
( [ iminusx ] ) one arrives at eq .
( [ rsimple1 ] ) .
in this section we prove the second part of theorem [ thm : main ] , namely suppose the eigenvalues of @xmath15 have different magnitudes or both eigenvalues are equal to zero .
then the spectral gap of @xmath2 is lower bounded by a positive constant independent of @xmath1 .
let us first consider the simple case when both eigenvalues of @xmath15 are equal to zero . using the canonical form of @xmath0 established in proposition [ prop : canonical ] and eq .
( [ transfer ] ) one can check that this is possible only if @xmath446 for some single - qubit unitary operator @xmath18 .
thus the hamiltonian @xmath2 is a sum of pairwise commuting projectors and @xmath447 for all @xmath1 which proves the desired lower bound . in the rest of this section
we assume that the eigenvalues of @xmath15 have distinct magnitudes . in this case
the eigenvectors of @xmath15 must be linearly independent .
let us first introduce some notation .
suppose @xmath448 $ ] is a consecutive block of qubits .
we shall write @xmath449 for the projector onto the ground space of the truncated hamiltonian @xmath450 obtained from @xmath2 by retaining only the terms fully contained in @xmath451 .
the projector @xmath449 acts trivially on all qubits in the complement of @xmath451 .
note that @xmath452 in the case where @xmath451 is the entire chain . our starting point is a general lower bound on the gap of 1d frustration - free hamiltonians due to nachtergaele @xcite .
specializing theorem 3 of ref .
@xcite to our case one gets the following lemma .
[ lemma : n ] suppose there exists an integer @xmath453 and a real number @xmath454 such that for all large enough @xmath1 and for the partition @xmath141=abc$ ] with @xmath455 and @xmath456 one has @xmath457 .
then @xmath458 for all large enough @xmath1 .
note that the right - hand side of eq .
( [ delta_n ] ) is a positive constant independent of @xmath1 .
thus lemma [ lemma : n ] reduces the problem of lower bounding the spectral gap of @xmath2 to that of upper bounding the quantity @xmath459 .
our main technical result is an upper bound on this quantity that decays exponentially with the size of @xmath460 .
[ thm : nbound ] let @xmath269 be the eigenvalues of @xmath15 such that @xmath461 . define @xmath462 .
let @xmath62 be the inner product between the normalized eigenvectors of @xmath15 .
consider any partition @xmath141=abc$ ] such that @xmath463 .
then @xmath464 where the constant coefficients in @xmath465 depend only on the forbidden state @xmath0 .
if @xmath466 then holds with a formal replacement @xmath467 which sets the first term to zero .
note that @xmath468 since the eigenvectors of @xmath15 are linearly independent .
furthermore , since @xmath469 , the right - hand side of eq .
( [ nbound ] ) is an exponentially decaying function of @xmath67 .
therefore we can choose a constant @xmath67 depending only on the forbidden state @xmath0 such that @xmath470 for all @xmath471 . substituting this into lemma [ lemma : n ] one gets @xmath472 for all large enough @xmath1 which proves the gapped phase theorem .
in the rest of this section we prove theorem [ thm : nbound ] .
we shall first consider the case where @xmath0 is an entangled state ( @xmath473 ) .
the main technical difficulty we had to overcome is a lack of explicit formulas for the projectors @xmath474 , and @xmath475 . at a high level ,
our approach is to develop a set of identities relating the global ground space projector such as @xmath476 and the local ones such as @xmath477 or @xmath478 .
these identities hold with a small error controlled by the size of the regions .
our proof of the theorem uses three identities of this type which are stated as region exclusion " lemmas in section [ subs : regionexclusion ] .
we use the region exclusion lemmas to decompose the operator @xmath479 in eq .
( [ nbound ] ) into a sum of several terms and to show that the norm of each term is exponentially small in @xmath67 , see section [ subs : finalgappedproof ] .
the proof of the region exclusion lemmas combines two ingredients : monotonicity of the ground space projectors under the partial trace ( established in section [ sec : mono ] ) and the fact that certain correlation functions in the ground space decay exponentially ( established in section [ subs : corrfunc ] ) .
in section [ subs : simple ] we consider the case where @xmath0 is a product state ( @xmath466 ) . in this case
the orthonormal basis for the ground space constructed in section [ subs : open ] gives an explicit formula for the ground space projector .
we use this formula to establish the region exclusion lemmas ( for the @xmath466 case ) in a more direct way . the corresponding special case of the theorem then follows from the region exclusion identities . before proceeding ,
let us establish some notation and conventions .
recall that a local unitary transformation @xmath259 preserves eigenvalues of @xmath2 and maps @xmath15 to @xmath19 , see section [ sec : gapless ] .
we shall choose the unitary @xmath18 to fix one of the eigenvectors of @xmath15 .
specifically , in the rest of this section we shall assume that @xmath480 for some state @xmath481 note that @xmath62 is the inner product between the eigenvectors of @xmath15 , as defined in the statement of theorem [ thm : nbound ] .
it will also be convenient to define a state @xmath482 given a set of qubits @xmath451 and a single - qubit state @xmath483 we shall write @xmath484 for the product state @xmath485
. we shall write @xmath486 for the projector onto this state and @xmath487 . in this section
we show that certain ground space correlation functions decay exponentially
. specifically , define @xmath488 and @xmath489 where @xmath490 . for notational convenience
we have suppressed the dependence of these functions on the forbidden state @xmath0 .
our main result in this section is as follows .
[ lemma : decay ] the sequence @xmath491 is monotonically increasing and has a finite limit @xmath492 such that @xmath493 furthermore , @xmath494 here the constant coefficients in @xmath465 depend only on the forbidden state @xmath0 .
let us define yet another correlation function @xmath495 where the maximum is over normalized ground states , that is , @xmath370 .
[ prop : easyfact ] @xmath496 for @xmath490 .
define @xmath497 using eq .
we see that @xmath498 and @xmath499 .
therefore @xmath500 where @xmath501 ( recall that @xmath147 means proportional to ) . comparing eq .
( [ teps ] ) and eq .
one infers that @xmath502 is the forbidden state , that is , @xmath503 .
let @xmath504 be a normalized state ( i.e. , @xmath505 ) for which the maximum in eq .
is achieved , so @xmath506 since @xmath504 , by proposition [ gsdegen ] it can be written @xmath507 where @xmath508 belongs to the symmetric subspace . using this fact and eq .
we see that @xmath509 for all integers @xmath510 . writing @xmath511\setminus{i}}+|v\rangle_{i}|\phi_{1}^{i}\rangle_{[n]\setminus{i}}\ ] ] and substituting this into eq .
( [ ijpsi ] ) one gets @xmath512 which implies @xmath513 using the fact that @xmath157 is normalized we have @xmath514 we upper bound the magnitude of the third term using the cauchy - schwarz and the arithmetic / geometric mean inequality : @xmath515 substituting this into eq .
( [ eq : normalized ] ) yields @xmath516 and hence @xmath517 . using this fact and eq .
( [ eq : implication ] ) we obtain @xmath518 and thus @xmath519 now we are ready to prove eq .
( [ limit ] ) .
first , applying the monotonicity lemma ( lemma [ lemma : mono ] ) to the left - right flipped chain yields @xmath520 .
therefore @xmath521 that is , @xmath522 is monotonically increasing . inserting the identity decomposition @xmath523 on the first qubit in eq .
( [ tau1point ] ) one gets @xmath524 the first term in eq .
( [ tau01 ] ) is upper bounded by @xmath525 since we can decompose @xmath526 using some orthonormal basis @xmath527 of @xmath275 and use the fact that @xmath528 for each individual state @xmath529 . the second term in eq .
( [ tau01 ] ) is upper bounded by @xmath530 , which follows from @xmath531 .
thus @xmath532 proposition [ prop : easyfact ] implies @xmath533 , that is , @xmath534 this shows that @xmath522 has a finite limit @xmath492 at @xmath535 . summing up the series produces the desired bound eq .
( [ limit ] ) .
the proof of eq .
( [ decay ] ) follows a similar strategy .
first consider the case @xmath536 .
the same argument used above shows that @xmath537 next suppose @xmath538 and @xmath539 .
inserting the identity decomposition @xmath540 on the first qubit , using the fact that @xmath541 , and noting that @xmath542 , one gets @xmath543 this shows that @xmath544 substituting eq .
( [ 11n ] ) into this bound and summing up the series , we get @xmath545 ( here we included the case @xmath546 which was handled in eq .
( [ 11n ] ) ) . finally , consider the case @xmath547 .
inserting the identity decomposition @xmath548 on the @xmath1-th qubit , using the fact that @xmath94 , and noting that @xmath549 , one gets @xmath550 this shows that @xmath551 combining this with eq .
( [ inn ] ) leads to the desired bound eq .
( [ decay ] ) . to perform manipulations with ground space projectors that involve several regions
we now prove three region exclusion lemmas .
these lemmas allow one to exclude one of the regions from certain operators built from ground space projectors .
the first region exclusion lemma states that @xmath552 .
[ lemma : c1 ] let @xmath141=abc$ ] with @xmath553 .
then @xmath554 here the constant coefficients in @xmath465 depend only on the forbidden state @xmath0 .
define @xmath555 and @xmath556 .
using the fact that @xmath557 one can write the quantity we wish to bound as @xmath558 define @xmath559 one can rewrite eq .
( [ x1eq2 ] ) as @xmath560 where @xmath561 , @xmath562 , and @xmath563 .
representing @xmath564 on the last qubit in eq .
( [ x1eq3 ] ) and using the monotonicity property @xmath280 from lemma [ lemma : mono ] one gets @xmath565 where @xmath566 from eq .
we get @xmath567 and plugging into eq .
gives @xmath568 to complete the proof we now show that @xmath569 has an upper bound which is exponentially small in @xmath67 and does not depend on @xmath1 .
partition the chain as @xmath141=a'b'b''c'$ ] , where @xmath570 , @xmath571 , @xmath572 , and @xmath573 .
using the fact that @xmath574 we get @xmath575 using the second part of lemma [ lemma : decay ] we have @xmath576 note that @xmath577 for any @xmath578 .
let @xmath579 .
it follows that @xmath580 thus @xmath581 has almost all its weight on the basis vector @xmath582 and an exponentially small weight @xmath583 on the space orthogonal to @xmath582 .
furthermore , the first part of lemma [ lemma : decay ] implies that @xmath584 .
combining this fact and eq .
( [ x1eq12 ] ) results in @xmath585 therefore @xmath586 finally , substituting this into eq . gives @xmath587 the second region exclusion lemma concerns the operator @xmath588 ( recall that @xmath589 ) .
[ lemma : c2 ] consider any partition @xmath141=abcd$ ] with @xmath553 .
then @xmath590 here the constant coefficient in @xmath465 depends only on the forbidden state @xmath0
. for brevity denote @xmath591 and @xmath592 . then @xmath593 taking into account that @xmath594 gives @xmath595 substituting @xmath596 in the last term results in @xmath597 applying the monotonicity lemma ( lemma [ lemma : mono ] ) one gets @xmath598 .
this shows that @xmath599 and therefore @xmath600 defined in section [ subs : corrfunc ] .
using the second part of lemma [ lemma : decay ] one gets @xmath601 here we denoted @xmath602 so that @xmath603 .
we also used the fact that the number of pairs @xmath604 with @xmath605 and @xmath606 such that @xmath607 is at most @xmath608 .
the third and final region exclusion lemma involves operators built from the ground space projectors as follows .
given any bipartition @xmath141=ab$ ] , where @xmath609 and @xmath460 are consecutive blocks of qubits , define @xmath610 [ lemma : x ] consider any partition @xmath141=abc$ ] with @xmath553 . then @xmath611 and @xmath612 here the constant coefficients in @xmath465 depend only on the forbidden state @xmath0 .
we first show that the bound eq .
( [ x2 ] ) follows from eq .
( [ x1 ] ) and thus it suffices to prove the latter . to see this ,
consider horizontally flipping the chain so that the vertices previously labeled @xmath613 are now @xmath614 .
the new forbidden state is @xmath615 where @xmath616 is the unitary transformation which interchanges the two qubits .
the new matrix @xmath197 is proportional to @xmath617 . from this
we see that @xmath618 and @xmath619 , and that @xmath620 . using these facts
we can see that eq .
( [ x2 ] ) is just eq .
( [ x1 ] ) applied to the left - right flipped chain .
it remains to prove eq .
( [ x1 ] ) .
let @xmath181 be any normalized state of @xmath621 such that @xmath622 partition the region @xmath460 as @xmath623 , where @xmath624 and @xmath625 .
define states @xmath626 one can easily check that the above states are pairwise orthogonal , @xmath627 we shall bound the contributions to the right - hand side of eq .
( [ xstep1 ] ) coming from @xmath628 , and @xmath629 separately .
let us start with @xmath630 . using the definitions of @xmath631 and @xmath632 one
gets @xmath633 since @xmath634 and @xmath635 , this results in @xmath636 here the last inequality follows from lemma [ lemma : c2 ] , where we set @xmath637 . next let us consider @xmath638 . using the definitions of @xmath631 and @xmath632 one
gets @xmath639 we can bound both terms in the right - hand side of the above equations using lemma [ lemma : c2 ] .
one should choose the four regions in the statement of lemma [ lemma : c2 ] as @xmath640 for the first term and @xmath641 for the second term .
this results in @xmath642 finally , let us consider @xmath629 .
we have @xmath643 it follows that @xmath644 here we applied lemma [ lemma : c1 ] choosing the regions in the statement of the lemma as @xmath645 .
likewise , @xmath646 here we applied lemma [ lemma : c1 ] choosing the regions in the statement of the lemma as @xmath647 .
substituting eqs .
( [ xstep2],[xstep3],[xstep5 ] ) into eq .
( [ xstep1 ] ) and using the triangle inequality one arrives at the desired bound eq .
( [ x1 ] ) .
let us now prove theorem [ thm : nbound ] for the case where @xmath0 is entangled , that is , @xmath648 . in section [ subs : simple ] we will see how to modify this proof to handle the product state case @xmath466 .
partition region @xmath460 as @xmath623 , where @xmath649 .
first we note that @xmath650 and @xmath651 here we used the notation from lemma [ lemma : x ] . applying lemma [ lemma : x ] to exclude region @xmath609 from @xmath631 one gets @xmath652 where @xmath653 denotes some operator such that @xmath654 substituting the identity @xmath655 one gets @xmath656 applying lemma [ lemma : x ] to exclude region @xmath657 from @xmath658 one gets @xmath659 using @xmath660 , @xmath661 , and @xmath662 , one arrives at @xmath663 finally , partition @xmath664 with @xmath665 and @xmath666 ( so that each part has size @xmath667 ) .
denote @xmath668 applying lemma [ lemma : c2 ] where the four regions are chosen as @xmath669 one concludes that @xmath670 replacing @xmath671 in eq .
( [ tstep5 ] ) by @xmath672 and using eq .
( [ tstep6 ] ) results in @xmath673 this completes the proof of theorem [ thm : nbound ] for the case when @xmath0 is an entangled state .
finally consider the case @xmath466 .
this implies that @xmath674 , that is , @xmath0 is a product state . using the notation from section [ subs : open ] write @xmath675 where @xmath676 . here
@xmath677 which follows from the fact that @xmath678 .
it is also easily checked that @xmath97 and @xmath107 are eigenvectors of @xmath15 corresponding to eigenvalues @xmath679 respectively .
we now show that the region exclusion identities presented in lemmas [ lemma : c1],[lemma : c2],[lemma : x ] ( for the case of entangled @xmath0 ) become exact equalities . indeed , as was shown in section [ subs : open ]
, the ground space of @xmath2 has an orthonormal basis @xmath108 , where @xmath103 and @xmath102 for @xmath680 .
thus @xmath681
we start with lemma [ lemma : c2 ] .
let @xmath141=abcd$ ] be an arbitrary partition such that @xmath460 and @xmath682 are non - empty .
we have to prove that @xmath683 note that @xmath684 for all @xmath685 .
this implies @xmath686 substituting eq .
( [ gexplicit ] ) for @xmath687 and using the above identity yields @xmath688 for all @xmath689 .
this is equivalent to eq .
( [ c2exact ] ) .
next consider lemma [ lemma : c1 ] .
let @xmath141=abc$ ] be any partition such that @xmath460 is non - empty .
we have to prove that @xmath690 note that for any @xmath691 one has @xmath692 for any @xmath693 .
therefore @xmath694 likewise , @xmath695 for any @xmath696 .
therefore @xmath697 comparing the last two identities one arrives at eq . ( [ c1exact ] ) .
finally , note that the proof of lemma [ lemma : x ] only uses lemmas [ lemma : c1],[lemma : c2 ] , and the fact that eq .
is equivalent to eq .
applied to the left - right flipped chain . in the proof of lemma [ lemma : x ] we use the fact that @xmath15 is invertible to establish this latter fact . in the case at hand @xmath15
is not invertible but since @xmath698 we immediately see that eq .
is just the left - right flipped version of eq . .
so both inequalities in lemma [ lemma : x ] become exact equalities .
the proof of theorem [ thm : nbound ] from section [ subs : finalgappedproof ] uses the region exclusion lemmas to establish the result .
since we have shown that each of these lemmas holds ( with exact equality ) for product states @xmath0 , we see that the proof of theorem [ thm : nbound ] also applies in this case if one formally sets @xmath467 in all error terms .
dg thanks edward farhi , jeffrey goldstone , and sam gutmann for many discussions about this problem when he was a graduate student .
dg was supported in part by nserc and by aro .
iqc is supported in part by the government of canada and the province of ontario .
sb would like to thank ramis movassagh for helpful discussions .
sb acknowledges nsf grant ccf-1110941 .
10 m. b. hastings .
an area law for one - dimensional quantum systems .
, 08:p08024 , 2007 .
i. arad , a. kitaev , z. landau , and u. vazirani . an area law and sub - exponential algorithm for 1d systems . , 2013 .
f. brando and m. horodecki .
an area law for entanglement from exponential decay of correlations .
, 9(11):721726 , 2013 .
m. b. hastings and t. koma .
spectral gap and exponential decay of correlations .
, 265(3):781804 , 2006 .
g. vidal .
efficient classical simulation of slightly entangled quantum computations .
, 91(14):147902 , 2003 .
d. perez - garcia , f. verstraete , m. wolf , and j. i. cirac .
matrix product state representations . , 7:401 , 2007 .
z. landau , u. vazirani , and t. vidick .
a polynomial - time algorithm for the ground state of 1d gapped local hamiltonians . ,
2013 . c. t. chubb and s. t. flammia . computing the degenerate ground space of gapped spin chains in polynomial time . , 2015 .
d. gottesman and m.b .
entanglement vs. gap for one - dimensional spin systems .
, 12:025002 , 2009 .
s. irani .
ground state entanglement in one dimensional translationally invariant quantum systems .
, 51:022101 , 2010 .
s. bravyi , l. caha , r. movassagh , d. nagaj , and p. shor .
criticality without frustration for quantum spin-1 chains . , 109(20):207202 , 2012 .
s. bachmann and b. nachtergaele .
product vacua with boundary states .
, 86:035149 , 2012 .
r. movassagh and p. shor
. power law violation of the area law in quantum spin chains . , 2014 .
d. aharonov , d. gottesman , s. irani , and j. kempe .
the power of quantum systems on a line .
, 287(1):4165 , 2009 .
s. hallgren , d. nagaj , and s. narayanaswami .
the local hamiltonian problem on a line with eight states is qma - complete .
, 13(9 - 10):721750 , 2013 .
subir sachdev . .
cambridge university press , 2011 .
t. koma and b. nachtergaele .
the spectral gap of the ferromagnetic xxz - chain .
, 40(1):116 , 1997 .
r. movassagh , e. farhi , j. goldstone , d. nagaj , t. j. osborne , and p. w. shor .
unfrustrated qudit chains and their ground states . , 82:012318 , 2010 .
i. affleck , t. kennedy , e. h. lieb , and h. tasaki .
rigorous results on valence - bond ground states in antiferromagnets . , 59:799802 , 1987 .
d. gottesman and s. irani .
the quantum and classical complexity of translationally invariant tiling and hamiltonian problems . in _
50th annual ieee symposium on _ , pages 95104 .
ieee , 2009 .
t. cubitt , d. garcia - perez , and m. wolf .
undecidability of the spectral gap . .
s. knabe .
energy gaps and elementary excitations for certain vbs - quantum antiferromagnets . , 52(3 - 4):627638 , 1988 .
b. nachtergaele . the spectral gap for some spin chains with discrete symmetry breaking .
, 175(3):565606 , 1996 .
s. bravyi . .
published in _ contemporary mathematics _ , vol .
536 , p. 33
, 2011 .
n. de beaudrap , t.j .
osborne , and j. eisert .
ground states of unfrustrated spin hamiltonians satisfy an area law .
, 12:095007 , 2010 .
c. r. laumann , r. moessner , a. scardicchio , and s. sondhi .
phase transitions and random quantum satisfiability . , 10(1):00010015 , 2010 .
s. bravyi , c. moore , and a. russell .
bounds on the quantum satisfiability threshold . in _ proc .
1st symposium on innovation in computer science ( ics 2010 ) _
, page 482 , 2010 .
rajendra bhatia . .
graduate texts in mathematics .
springer new york , 1997 .
in the main body of the paper we considered qubit chains where @xmath60 is rank-@xmath3 and @xmath70 , defined in eq . , is guaranteed to be frustration - free .
here we consider qubit chains where @xmath60 is rank-@xmath55 or rank-@xmath56 and we determine which projectors @xmath60 correspond to frustration - free chains .
for each frustration - free chain we determine if the system is gapped or gapless .
we say that @xmath70 is gapped if its spectral gap , denoted @xmath699 , is lower bounded by a positive constant independent of @xmath1 ( otherwise it is gapless ) .
we shall write @xmath700 for the null space of @xmath70 and @xmath90 for the projector onto this space .
the case where @xmath60 is rank-@xmath56 is trivial , so we consider it first . in this case
there is a unique two - qubit state @xmath508 in the null space of @xmath60 .
if @xmath70 is frustration - free then there exists an @xmath1-qubit state @xmath0 with reduced state on each pair of consecutive qubits @xmath701 supported entirely on @xmath702 .
it follows that the rank-@xmath56 chain is frustration - free if and only if @xmath703 for some single - qubit state @xmath224 .
thus @xmath704 and @xmath70 is a sum of pairwise commuting projectors .
this shows that @xmath70 has unique ground state @xmath705 , and its eigenvalue gap is equal to @xmath3 , for all @xmath706 .
the rank-@xmath55 case is slightly more interesting .
there is a trivial case where @xmath60 is a @xmath3-local projector , i.e. , @xmath707 or @xmath708 ; in that case @xmath709 for all @xmath706 .
the following theorem handles all other cases .
[ thm : rank2 ] suppose @xmath60 is a two - qubit , rank-2 projector which can not be written as @xmath710 or @xmath711 for some projector @xmath712 .
then the dimension of the null space of @xmath70 satisfies @xmath713 for all @xmath714 . moreover , exactly one of the following holds : 1 .
@xmath715 for some single - qubit state @xmath716 .
2 . @xmath717 for some linearly independent single - qubit states @xmath294 .
3 . @xmath718 for some linearly independent single - qubit states @xmath294 .
4 . @xmath719 for some orthonormal single - qubit states @xmath720 and non - zero @xmath721 .
@xmath722 is empty . in cases 1 , 2 , 3 , and 4
, the hamiltonian @xmath70 is frustration - free for all @xmath706 , whereas in case 5 it is frustrated for @xmath714 .
@xmath70 is gapped in cases 1 , 2 , and 3 , and it is gapped in case 4 if @xmath723 . if @xmath724 in case 4 then the hamiltonian is gapless , with spectral gap upper bounded as @xmath725 . to establish lower bounds on spectral gaps we shall use nachtergaele s criterion @xcite .
recall that lemma [ lemma : n ] states this criterion for the case where @xmath60 is rank-@xmath3 .
more generally if @xmath70 is frustration - free for @xmath706 then the same bound holds for its spectral gap ( with @xmath20 and @xmath726 replaced by @xmath699 and @xmath727 in the statement of the lemma ) . for our purposes it will be sufficient to use the following weaker version of the bound .
[ lem : gapcor ] suppose @xmath70 is frustration - free for @xmath706 .
let @xmath141=abc$ ] with @xmath456 , @xmath463 , and @xmath728 .
suppose there exist constants @xmath729 and @xmath730 such that for all sufficiently large @xmath1 we have @xmath731 .
then @xmath70 is gapped .
we first establish that the range of @xmath60 is spanned by two linearly independent states @xmath735 which are both entangled .
it is easy to check that the only two dimensional subspaces of @xmath736 which contain only product states are of the form @xmath737 or @xmath738 for some single qubit state @xmath702 . by the hypothesis of the theorem @xmath60
can not be written as @xmath710 or @xmath711 , which implies @xmath739 does not have this form ; thus it contains at least one entangled state @xmath181 .
let @xmath740 be linearly independent from @xmath181 .
it is easy to see that we can always choose @xmath741 for some nonzero @xmath742 so that @xmath0 is entangled .
so @xmath743 where @xmath744 are both entangled ( equivalently , @xmath745 and @xmath15 are both invertible ) .
furthermore , an @xmath1-qubit state @xmath702 is in the null space of @xmath70 if and only if it is in the null space of @xmath746 and @xmath747 for all @xmath122 . to complete the proof
we now suppose that @xmath722 is nonempty and we consider two cases depending on whether or not @xmath748 has two linearly independent eigenvectors .
the theorem follows directly from the following propositions which handle the two cases .
suppose @xmath60 is a two - qubit projector such that @xmath743 where @xmath744 are both entangled .
suppose that @xmath748 has linearly independent eigenvectors @xmath749 and that @xmath722 is nonempty .
then one of the cases 1,2 , or 3 from theorem [ thm : rank2 ] occurs .
moreover , @xmath70 is frustration - free for all @xmath706 and it is gapped .
its ground space dimension satisfies @xmath750 for all @xmath714 .
[ prop : teig1 ] suppose @xmath60 is a two - qubit projector such that @xmath743 where @xmath744 are both entangled .
suppose that @xmath748 has only one linearly independent eigenvector @xmath751 and that @xmath722 is nonempty .
then case 4 from theorem [ thm : rank2 ] occurs .
moreover , @xmath70 is frustration - free and has ground space dimension equal to @xmath55 for all @xmath706 .
it is gapped if and only if @xmath723 ; if @xmath724 then @xmath725 .
[ prop : teig2 ] * proof of proposition [ prop : teig1 ] * we first establish that @xmath752 is spanned by @xmath753 and @xmath754 .
these states are linearly independent ( since @xmath294 are ) . since @xmath752 is @xmath55-dimensional it suffices to establish that it contains both of these states .
clearly each is orthogonal to @xmath755 since @xmath756 ( recall @xmath131 ) . to check that they are orthogonal to @xmath757 use the fact that @xmath758 and that @xmath294 are eigenvectors of @xmath748
now consider @xmath759 .
since @xmath722 is nonempty , there exists a state @xmath760 , which by proposition [ gsdegen ] can be written as @xmath761 for some state @xmath762 in the @xmath56-qubit symmetric subspace .
since the chain is frustration free , the first two qubits of @xmath702 have all of their support in @xmath752 .
this implies @xmath763 where @xmath764 are not both zero . by symmetry
we assume without loss of generality that @xmath167 .
next , imposing orthogonality to @xmath765 gives @xmath766 since @xmath294 are linearly independent , both terms must be zero . using the fact that any two - qubit state orthogonal to @xmath253 is symmetric , and that @xmath167 , we get : now recall that @xmath294 are linearly independent eigenvectors of @xmath748 .
note that @xmath748 is not proportional to the identity ( since this would imply that @xmath181 is proportional to @xmath0 ) and therefore @xmath294 are the only eigenvectors of @xmath748 .
hence @xmath767 is proportional to one of the states @xmath770 , and if @xmath768 then the same holds for @xmath771 .
however , since @xmath15 is invertible , it can not be the case that @xmath772 .
putting this together we see there are 4 subcases to consider ( below we show that the first three correspond to cases 1 . , 2 . , 3 .
, from the statement of the theorem and that the fourth does not occur ) : * * case ( i ) : @xmath773 and @xmath774 where @xmath775 are both non - zero.**in this case @xmath769 is not an eigenvector of @xmath748 , which implies ( by the facts established above ) that @xmath776 and thus @xmath777 is the only state in @xmath778 .
likewise , @xmath779 is the only state in @xmath700 for @xmath323 ( since it is the unique @xmath1-qubit state such that any three consecutive qubits @xmath780 have all of their support on @xmath778 . )
this establishes that we are in case 1 from theorem [ thm : rank2 ] .
note that for any partition @xmath141=abc$ ] with @xmath781 we have @xmath782 and so the conditions of lemma [ lem : gapcor ] are satisfied ( with @xmath783 ) and @xmath70 is gapped .
* case ( ii ) : @xmath773 and @xmath784 * in this case ( by the facts established above ) @xmath752 is spanned by @xmath785 and @xmath786 .
this implies that @xmath700 is spanned by @xmath779 and @xmath787 for all @xmath706 , so we are in case 2 of theorem [ thm : rank2 ] .
one can easily construct an orthonormal basis and confirm that @xmath788 .
using this expression three times and the triangle inequality we get @xmath789 where @xmath790 is a constant , for any partition @xmath141=abc$ ] .
hence the conditions of lemma [ lem : gapcor ] are satisfied ( with @xmath791 ) and @xmath70 is gapped . * * case ( iii ) : @xmath792 and @xmath793**in this case @xmath794 , which implies that for all @xmath706 , @xmath700 is spanned by product states @xmath795 and @xmath796 ( the last tensor product factors are instead @xmath797 respectively if @xmath1 is odd ) .
this shows that we are in case 3 of theorem [ thm : rank2 ] . now constructing an orthonormal basis we see that @xmath798 letting @xmath141=abc$ ] and using this expression we get @xmath789 for some constant @xmath790 .
hence the conditions of lemma [ lem : gapcor ] are satisfied ( with @xmath791 ) and @xmath70 is gapped . *
* case ( iv ) : @xmath792 and @xmath774 where @xmath775 are both non - zero.**in this final case @xmath776 and the only state in @xmath778 is @xmath799 .
any non - zero state @xmath800 must have its first three and last three qubits in the state @xmath702 .
this implies in particular that @xmath801 is a product state .
furthermore , the second qubit of @xmath801 must be in the state @xmath802 ( look at the first three qubits ) and also in the state @xmath716 ( look at the last three qubits ) , which is impossible . therefore @xmath722 is empty , which is a contradiction .
so case ( iv ) does not occur .
* proof of proposition [ prop : teig2 ] * define an orthonormal basis @xmath803 .
since @xmath804 is the only eigenvector of @xmath748 , in this basis we have @xmath805 for some @xmath806 with @xmath807 ( since @xmath808 ) .
first consider @xmath288 .
we claim that @xmath809 and @xmath810 span @xmath752 . to see this note that these states are linearly independent and orthogonal to @xmath811 . to show that they are also orthogonal to @xmath812 , use the fact that @xmath813 and eq . .
we now show that our assumption that @xmath722 ( and therefore also @xmath778 ) is nonempty implies @xmath814 is in @xmath778 . to reach a contradiction ,
assume @xmath815 .
by proposition [ gsdegen ] any state @xmath816 satisfies @xmath817 for some @xmath762 in the three - qubit symmetric subspace .
the first two qubits of @xmath762 must be supported entirely in @xmath752 . using the form of @xmath752 derived above
, we see that this implies @xmath762 is a superposition of the symmetric hamming weight zero and one states ( with respect to the @xmath818 basis ) .
since @xmath819 ( and since @xmath778 is nonempty ) , we have shown that @xmath778 is one - dimensional and contains a single state @xmath820 of the form @xmath821 where @xmath768 . by assumption
there exists a state in @xmath722 , which must have its last three and first three qubits each supported on @xmath778 , i.e. , it must be of the form @xmath822 for some single - qubit states @xmath823 .
however this is impossible since @xmath820 is not a product state whenever @xmath768 .
having reached a contradiction we conclude @xmath814 is in @xmath778 .
now @xmath824 implies that @xmath825 is an eigenvector of @xmath748 . by assumption
this operator has only one eigenvector @xmath826 , so @xmath827 .
thus @xmath804 is an eigenvector of both @xmath15 and @xmath745 .
therefore , in the @xmath828 basis we have @xmath829 for some @xmath830 . comparing with eq .
we see that @xmath831 .
now @xmath832 and @xmath833 comparing eqs . , we see that @xmath834 where @xmath835 and @xmath836 ( note that @xmath837 are both nonzero since @xmath831 ) .
moreover these states are orthonormal so @xmath838 . by proposition [ gsdegen ] , a basis for the zero energy ground space of @xmath839
is given by @xmath840 where @xmath841 and @xmath842 .
it is then easy to see that the only states in this space orthogonal to @xmath843 are the basis vectors corresponding to @xmath844 and linear combinations thereof .
these two states span @xmath700 for all @xmath706 ; this shows that we are in case 4 of theorem [ thm : rank2 ] .
we now show that the system is gapped if @xmath845 . without loss of generality
we may assume @xmath846 ; note that if @xmath847 we may relabel the qubits @xmath848 as @xmath614 ( flipping the chain left to right ) which sends @xmath849 . define @xmath850 where @xmath851 noting that the ground space projector is @xmath852 and that @xmath853 we obtain @xmath854 now let a partition @xmath141=abc$ ] be given with @xmath855 and @xmath456 . using the above bound three times , the triangle inequality , and the facts that @xmath856 and @xmath857 , we obtain @xmath858 using explicit expressions for @xmath859 , and
@xmath860 we see that @xmath861 and the first term above is zero .
this shows that the conditions of lemma [ lem : gapcor ] are satisfied with @xmath862 , and @xmath70 is gapped .
finally , we establish that @xmath70 is gapless if @xmath724 . using eq .
we see that @xmath70 commutes with the total hamming weight operator @xmath863 and is therefore block diagonal with a block for each hamming weight @xmath864 .
the spectral gap @xmath699 is upper bounded by the smallest nonzero eigenvalue within any given block .
the matrix of the block with hamming weight @xmath3 , in the orthonormal basis @xmath865 , is given by @xmath866 this is @xmath6 times the laplacian matrix of the path graph of length @xmath1 , and its spectrum is known .
in particular , its smallest nonzero eigenvalue is @xmath867 , which upper bounds @xmath699 if @xmath724 . | we consider a family of translation - invariant quantum spin chains with nearest - neighbor interactions and derive necessary and sufficient conditions for these systems to be gapped in the thermodynamic limit .
more precisely , let @xmath0 be an arbitrary two - qubit state .
we consider a chain of @xmath1 qubits with open boundary conditions and hamiltonian @xmath2 which is defined as the sum of rank-@xmath3 projectors onto @xmath0 applied to consecutive pairs of qubits .
we show that the spectral gap of @xmath2 is upper bounded by @xmath4 if the eigenvalues of a certain @xmath5 matrix simply related to @xmath0 have equal non - zero absolute value .
otherwise , the spectral gap is lower bounded by a positive constant independent of @xmath1 ( depending only on @xmath0 ) . a key ingredient in the proof is a new operator inequality for the ground space projector which expresses a monotonicity under the partial trace .
this monotonicity property appears to be very general and might be interesting in its own right . as an extension of our main result , we obtain a complete classification of gapped and gapless phases of frustration - free translation - invariant spin-@xmath6 chains with nearest - neighbor interactions . |
quantum confined model systems incorporating different repulsive or attractive potential terms have become the subject of increasing research interest @xcite .
a number experiments involving , e.g. atoms and molecules under pressure , quantum dots , and atoms in metallofullerenes , are analysed using such models , which are essentially defined by the given shape and strength of the confining potential .
further , the confined model potentials have been used to study the autoionization resonance states for the two - electron spherical quantum dots @xcite and two electron atomic states @xcite .
theoretical studies of confined systems generate valuable guidelines to the design and synthesis of new materials and future devices .
it is therefore of interest to derive new exact solutions of the quantum model systems involving such confining potentials under certain special conditions .
0.1true in we revisit a model by m. taut @xcite for a two - electron system in which the electrons are softly confined by means of oscillator potentials .
we shall consider the problem in @xmath1 spatial dimensions and find exact solutions for certain special choices of the parameters , and accurate numerical solutions in other cases .
we shall also study this system when it is confined to an impenetrable spherical box of radius @xmath2 in @xmath3 we shall refer to this as ` hard confinement ' , which persists even when the oscillator interaction is switched off .
thus we shall begin with only soft confinement in form of an attractive oscillator , and we go on later to implement hard confinement by applying to the same system a vanishing boundary condition at the surface of a finite sphere of radius @xmath0 0.1true in in order to place our work in perspective , we shall present here a brief review on the confined @xmath4-dimensional h atom @xmath5 and the isotropic harmonic oscillator @xmath6 .
the eigenspectrum of the spherically confined h atom ( scha ) within impenetrable walls is characterized by novel degeneracy @xcite effects of three different kinds .
two of them are generated from the specific choice of the radius of confinement @xmath2 , chosen exactly at the radial nodes corresponding to the free hydrogen atom ( fha ) wave functions . in the _ incidental degeneracy _
case , the confined @xmath7 state with the principal quantum number @xmath8 is iso - energic with @xmath9 state of the fha with energy @xmath10 atomic units ( a.u . ) , at an @xmath2 defined by the radial node in the fha .
for example , the @xmath7 state corresponding to the lowest energy value , when confined at the radius @xmath2 given by the radial node the first excited _ free _ state @xmath9 , increases in such a way that the confined - state energy becomes the same as excited free - state energy .
the specific node in question is given by @xmath11 .
such a degeneracy can be realized at similar choices for @xmath2 where multiple nodes exist in the second and higher excited states of a given @xmath12
. however , such closed analytical expressions for the radial nodes are not available in the case of higher excited states . in the _ simultaneous - degeneracy _ case , on the other hand , for all @xmath13 ,
each pair of confined states denoted by @xmath7 and @xmath14 state , confined at the common @xmath11 , become degenerate .
note that the pair of levels in the free state are nondegenerate .
both these degeneracies have been shown @xcite to result from the gauss relationship applied at a unique @xmath15 by the confluent hypergeometric functions that describe the general solutions of the scha problem .
0.1true in finally , the interdimensional degeneracy @xcite arises , as in the case of the free hydrogen atom , due to the invariance of the schrdinger equation under the transformation @xmath16 . in order to preserve the number of nodes in the radial function , it is necessary simultaneously to make the transformation @xmath17 .
the _ incidental degeneracy _ observed in the case of a spherically confined isotropic harmonic oscillator ( sciho ) is qualitatively similar to that of the scha .
for example , the only radial node in the first excited free state of any given @xmath12 for @xmath4-dimensional sciho is located at @xmath18 . for the multiple node states ,
the corresponding numerical values must be used .
however , the behavior of the two confined states at a common radius of confinement is found to be interestingly different @xcite . in particular , for the sciho the pairs of the confined states defined by @xmath19 and @xmath20 at the common @xmath18 a.u .
, display for all @xmath8 , a constant energy separation of _ exactly _ @xmath21 harmonic - oscillator units , @xmath22 , with the state of higher @xmath23 corresponding to the lower energy .
it is interesting to note that the two confined states at the common @xmath2 with @xmath24 , considered above contain different numbers of radial nodes .
the condition for interdimensional degeneracy@xcite due to the invariance of the schrdinger equation remains the same as before .
recently , the confined systems of the @xmath4-dimensional hydrogen atom @xcite and harmonic oscillator @xcite have been studied .
problems involving short - range potentials in @xmath4 dimensions have recently been considered @xcite . in the light of the above discussion , it is interesting to study the various aforementioned degeneracies in the free and spherically confined @xmath4-dimensional potential generally given by @xmath25 .
we consider now the model potential of our interest in the present work . for a single particle the @xmath4-dimensional schrdinger equation , in atomic units @xmath26 , with a spherically symmetric potential @xmath27
can be written as the following @xmath28\psi(r)=e\psi(r),\ ] ] where @xmath29 is the @xmath4-dimensional laplacian operator and @xmath30 .
if the positions of two particles are @xmath31 and @xmath32 , the hamiltonian for the system in which the particles interact by means of a repulsive coulomb term but are bound by oscillators is given by @xmath33 where @xmath34 and @xmath35 . by using a pair - distance
coordinate @xmath36 and a center - of - mass coordinate @xmath37 , the hamiltonian @xmath38 and potential @xmath39 for relative motion can be expressed in terms of @xmath40 and @xmath41 and may be written @xmath42 in order to transform ( [ eq1 ] ) to the @xmath4-dimensional spherical coordinates @xmath43 , we first separate variables @xcite using the representation @xmath44 where @xmath45 is a normalized spherical harmonic with characteristic value @xmath46 , and @xmath47 .
one then obtains the radial schrdinger equation in the form @xmath48u(r)=0,\quad\quad
\int_0^\infty u^2(r)dr=1 , u(0)=0,\ ] ] where @xmath49 .
since the spherical - harmonic factor @xmath50 has definite parity @xcite given by @xmath51 and the interaction in the present model does not depend on spin , the configuration - space wave function @xmath52 can be multiplied by a two - particle spin factor with complementary permutation symmetry ( singlet or triplet ) so that the overall wave function for relative motion is antisymmetric under the exchange of the particle indices . a center - of - mass factor
@xmath53 will not change this symmetry since it is in any case symmetric .
thus in the present model , all possible energy eigenstates must be considered for the two - body system in configuration space .
we assume that the potential @xmath27 is less singular than the centrifugal term so that @xmath54 we note that the hamiltonian and boundary conditions of ( [ eq3 ] ) are invariant under the transformation @xmath55 thus , given any solution for fixed @xmath4 and @xmath56 , we can immediately generate others for different values of @xmath4 and @xmath57 more particularly , the energy is unchanged if both @xmath58 and the number of nodes @xmath59 are held constant .
repeated application of this transformation produces a large collection of states , the only apparent limitation being a lack of interest in some values of @xmath4 ( see , for example @xcite ) . in the present work , we consider the repulsive coulomb plus a harmonic - oscillator potential @xmath60 where @xmath61 , and the coefficients @xmath62 and @xmath63 are both constant .
when we look at hard confinement inside a sphere of finite radius @xmath2 , we can also consider removing the oscillator binding by choosing @xmath64 : in both cases of confinement , the spectrum of @xmath38 is entirely discrete .
the present paper is organized as follows . in section 2 ,
we discuss some general spectral features and bounds . in section 3
we express the repulsive coulomb potential term @xmath65 as the envelope of a family @xmath66 of potential terms each of which admits an exact analytical solution .
this application of envelope theory @xcite leads to a spectral upper bound formula valid for all the soft - confined eigenvalues . in section 4
we briefly review the asymptotic iteration method ( aim ) for solving a second - order linear differential equation .
we also discuss necessary and sufficient conditions for certain classes of differential equations with polynomial coefficients to have polynomial solutions . in sections 5 and 6
, we use aim to study how the eigenvalues depend on the potential parameters @xmath67 and @xmath68 . in each of these sections ,
the results obtained are of two types : exact analytic results that are valid when certain parametric constraints are satisfied , and accurate numerical values for arbitrary sets of potential parameters . in section 5
there is only soft confinement effected by a non - zero oscillator term ; in section 6 we consider ` hard confinement ' , that is to say when the same system is confined to the interior of an impenetrable spherical box of radius @xmath0
we shall show shortly that the hamiltonian @xmath38 is bounded below .
indeed provided that @xmath63 or @xmath69 , the entire spectrum is discrete .
the eigenvalues of @xmath38 may therefore be characterized variationally .
we first discuss the open system where @xmath70 the eigenvalues @xmath71 are monotonic in each parameter @xmath67 and @xmath68 as a direct consequence of the monotonicity of the potential @xmath39 in these parameters .
indeed , since @xmath72 and @xmath73 , it follows from the hellmann - feynman theorem that @xmath74 elementary scaling arguments can be used to reduce the dimension of the parametric space .
for example we obtain @xmath75 the generalized heisenberg uncertainty relation may be expressed @xcite for dimension @xmath76 as the operator inequality @xmath77 this allows us to construct the following lower energy bound @xmath78.\ ] ] provided @xmath79 this lower bound is finite for all @xmath67 .
it also obeys the same scaling and monotonicity laws as @xmath80 itself .
but the bound is weak . for potentials such as @xmath27 that satisfy @xmath81 common
has shown @xcite for the ground state in @xmath82 dimensions that @xmath83 but the resulting energy lower bound is still weak .
lower and upper bounds on the ground - state energy @xmath80 are provided in @xmath4 dimensions , for example , by a gaussian trial wave function @xmath84 used here in two distinct ways . by the local energy theorem
@xcite we have ( for @xmath85 initially sufficiently small ) we find the lower energy bound : @xmath86 = \max_{0<\alpha<\alpha_1}\,\min_{r>0}\,\left[-\alpha^2 r^2 + \alpha d + \frac{a}{r } + b^2\,r^2\right].\ ] ] meanwhile , if the constant @xmath87 is chosen so that normalization ( apart from the angular terms ) , is provided by @xmath88 then by the rayleigh - ritz theorem , we obtain the upper bound @xmath89.\ ] ] these bounds also bound the bottoms of the angular - momentum subspaces labelled by @xmath12 : this is a consequence of the degeneracies between the states with the same value of @xmath90 , as we have mentioned above : practically speaking , if @xmath91 one can increase @xmath4 to @xmath92 and set @xmath93 in order to use the bound formulas ( [ elet],[err ] ) for the energies . for the @xmath4-dimensional oscillator @xmath94
, we observe that the @xmath4-dimensional bounds coalesce to the exact energy @xmath95= db.$ ] a specific example is provided by the case @xmath96 and @xmath97 for which we find @xmath98 where the accurate estimate @xmath80 was found numerically , by shooting methods .
more generally , we shall sometimes use the convention of atomic physics in which , even for non - coulombic central potentials , a ` principal quantum number ' @xmath8 is used and defined by @xmath99 now we turn to the case of hard confinement @xmath100 and we write @xmath101 the monotonicity with respect to the box size @xmath2 may be proved by a variational argument .
let us consider two box sizes , @xmath102 and an angular momentum subspace labelled by a fixed @xmath57 we extend the domains of the wave functions in the finite - dimensional subspace spanned by the first @xmath103 radial eigenfunctions for @xmath104 so that the new space @xmath105 may be used to study the case @xmath106 .
we do this by defining the extended eigenfunctions so that @xmath107 for @xmath108 we now look at @xmath38 in @xmath105 with box size @xmath109 .
the minima of the energy matrix @xmath110 $ ] are the exact eigenvalues for @xmath111 and , by the rayleigh - ritz principle , these values are one - by - one upper bounds to the eigenvalues for @xmath112 thus , by formal argument we deduce what is perhaps intuitively clear , that the eigenvalues increase as @xmath2 is decreased , that is to say @xmath113 if a very special box is now considered , whose size @xmath2 coincides with any radial node of the corresponding @xmath114 problem , then these two problems share an eigenvalue exactly .
this is an example of a very general relation which exists between constrained and unconstrained eigensystems , and , indeed , also between two constrained systems with different box sizes .
we first consider the radial eigenvalue problem for a generalized harmonic oscillator we have @xmath115u(r)=eu(r),\quad\quad e = ( 4n + 2\lambda+3)b,\ ] ] where @xmath116 and @xmath117 thus , for example , in @xmath1 dimensions , we have @xmath118 experience with the envelope method @xcite suggests that we express the repulsive coulomb term @xmath65 as the envelope of a family of curves of the form @xmath119 where @xmath120 is the point of contact between the target @xmath121 and a tangential curve @xmath122 the idea behind this is to think of @xmath65 as a smooth transformation @xmath123 of @xmath124 whose tangents @xmath125 all lie above @xmath65 because @xmath126 is concave ; meanwhile the tangential potentials are schrdinger soluble . in order to find the coefficients @xmath85 and @xmath127 we require at contact point @xmath128 that @xmath129 and @xmath125 agree in value and slope .
thus we have @xmath130 meanwhile , since @xmath131 is concave , we know that each tangential curve lies above @xmath129 , that is to say @xmath132 the geometrical situation is illustrated for the problem at hand in fig .
[ figenv ] . in the potential
is represented as the envelope of an upper family of tangential potentials @xmath133 . when a tangential potential @xmath125 is used in place of @xmath129 ,
schrdinger s equation is exactly soluble . in the envelope method
, the best choice is made by optimizing over the point of contact @xmath134 . ] in order to obtain the upper energy bound it remains to replace the coulomb term in by the right side of and solve the resulting equation for the energy , shifted by @xmath135 we first determine @xmath136 from the equation @xmath137 the final expression for this bound is as follows : @xmath138^{\half } + \frac{ab}{2t}.\ ] ] the function @xmath139 is u - shaped with one minimum that is easy to find .
what we have in effect is a continuous family of upper bounds , and the minimization in selects the best of these for the given parameter set and eigenvalue . for the ground state with @xmath140
we obtain the higher bound @xmath141 than the @xmath142 provided by a scale - optimized gaussian trial function in , but of course the simple general formula is valid for every eigenvalue . by using aim
we shall find very accurate numerical eigenvalues for soft confinement in section 5 , where some are tabulated in table [ table : difdn3 ] .
we exhibit now two more upper bounds otained by , along with accurate values in parentheses from table [ table : difdn3 ] : @xmath143 ; @xmath144
the asymptotic iteration method ( aim ) was originally introduced @xcite to investigatae the solutions of differential equations of the form @xmath145 where @xmath146 and @xmath147 are @xmath148differentiable functions .
a key feature of this method is to note the invariant structure of the right - hand side of ( [ eq5 ] ) under further differentiation . indeed ,
if we differentiate ( [ eq5 ] ) with respect to @xmath149 , we obtain @xmath150 where @xmath151 and @xmath152 if we find the second derivative of equation ( [ eq5 ] ) , we obtain @xmath153 where @xmath154 and @xmath155 thus , for @xmath156 and @xmath157 derivative of ( [ eq5 ] ) , @xmath158 , we have @xmath159 and @xmath160 respectively , where @xmath161 from ( [ eq8 ] ) and ( [ eq9 ] ) we have @xmath162 clearly , from ( [ eq11 ] ) if @xmath163 , the solution of ( [ eq5 ] ) , is a polynomial of degree @xmath59 , then @xmath164 .
further , if @xmath165 , then @xmath166 for all @xmath167 . in an earlier paper @xcite we proved the principal theorem of aim , namely 0.1 in * theorem 1 @xcite . *
_ given @xmath168 and @xmath169 in @xmath170 the differential equation ( [ eq5 ] ) has the general solution @xmath171\ ] ] if for some @xmath172 @xmath173 where @xmath174 and @xmath175 are given by ( [ eq10 ] ) .
_ 0.1true in recently , it has been shown @xcite that the termination condition ( [ eq13 ] ) is necessary and sufficient for the differential equation ( [ eq5 ] ) to have polynomial - type solutions of degree at most @xmath59 , as we may conclude from eq.([eq11 ] ) . thus , using theorem 1 , we can now find the necessary and sufficient conditions @xcite for the polynomial solutions of the differential equation @xmath176 where @xmath177 are constants .
these conditions have been reported in the following theorem : 0.1true in * theorem 2 [ @xcite theorem 5 ] . * in the next sections , we shall apply the result of theorem 2 to study the possible quasi - exact analytic solutions for the @xmath4-dimension schrdinger equation ( [ eq3 ] ) for unconstrained and constrained coulomb plus harmonic oscillator potential ( [ eq4 ] ) .
we shall also apply aim , theorem 1 , to obtain _ accurate _
approximations for arbitrary potential parameters , again , for the unconstrained and constrained @xmath4-dimensional schrdinger equation ( [ eq3 ] ) .
in this section , we consider the @xmath4-dimensional schrdinger equation @xmath178u_{nl}^d(r)=e_{nl}^d u_{nl}^d(r),\quad 0<r<\infty.\ ] ] in order to solve this equation by means of aim , the first step is to transform ( [ eq25 ] ) into the standard form ( [ eq5 ] ) . to this end
, we note that the differential equation ( [ eq25 ] ) has one regular singular point at @xmath179 and an irregular singular point at @xmath180 .
since for large @xmath149 , the harmonic oscillator term dominates , the asymptotic solution of ( [ eq25 ] ) as @xmath181 is @xmath182 ; meanwhile the indicial equation of ( [ eq25 ] ) at the regular singular point @xmath183 yields @xmath184 which is solved by @xmath185 the value of @xmath186 , in eq.([eq26 ] ) , determines the behavior of @xmath187 for @xmath188 , and only @xmath189 is acceptable , since only in this case is the mean value of the kinetic energy finite @xcite .
thus , the exact solution of ( [ eq25 ] ) may assume the form @xmath190 where we note that @xmath191 as @xmath188 . on substituting this ansatz wave function into ( [ eq25 ] )
, we obtain the differential equation for @xmath192 as @xmath193 this equation has a regular singular point at @xmath183 with exponents @xmath194 and an irregular point at infinity .
we may note first that is a special case of the differential equation ( [ eq21 ] ) with @xmath195 , @xmath196 , @xmath197 , @xmath198 and @xmath199 .
thus , the necessary condition for the polynomial solutions of eq.([eq28 ] ) is @xmath200 and the sufficient condition follows from the vanishing of the tridiagonal determinant @xmath201 , @xmath202 , namely @xmath203 = [ cols= " < , < , < , < , < , < , < " , ] [ table : difdn5 ]
we study a model atom - like system @xmath204 which is confined softly by the inclusion of a harmonic - oscillator potential term @xmath205 and possibly also by the presence of an impenetrable spherical box of radius @xmath0 for @xmath206 or @xmath207 the entire spectrum @xmath208 is discrete .
we have studied these eigenvalues and we present an approximate spectral formula for the ` free ' case , @xmath209 . for
the general case of @xmath210 aim has been used to provide both a large number of exact analytical solutions , valid for certain special choices of the parameters @xmath211 and also very accurate numerical eigenvalues for arbitrary parametric data . in the cases where we have found analytic solutions for @xmath114 ,
the exact wave functions are no longer expressed in terms of known special functions , as is possible for the hydrogen atom . however , the exact solutions we have found for confining potentials correspond to confinement at the zeros of the unconfined case .
an interesting qualitative feature seems to be that @xmath208 , for large @xmath2 , is concave with respect to @xmath59 , @xmath12 , or @xmath4 , but becomes convex as @xmath2 is reduced ; this may arise because the reduction in @xmath2 perturbs the higher states more severely since , when free , they are naturally more spread out .
it is hoped that the work reported in the present paper will provide a useful addition to the growing body of results concerning the spectra of confined atomic systems in @xmath4 dimensions .
partial financial support of this work under grant nos .
gp3438 and gp249507 from the natural sciences and engineering research council of canada is gratefully acknowledged by two of us ( rlh and ns ) .
kds thanks the department of science and technology , new delhi , for the j.c .
bose fellowship award .
ns and kds are grateful for the hospitality provided by the department of mathematics and statistics of concordia university , where part of this work was carried out .
rlh is grateful for the hospitality provided by the school of chemistry , university of hyderabad 500046 , india , where part of this work was carried out .
j. r. sabin , e. j. brndas and s. a. cruz ( eds . ) , _ advances in quantum chemistry : theory of confined quantum systems - part one _ ( book * 57 * ) , academic press .
new york , 2009 .
k. d. sen ( ed . ) , _ electronic structure of quantum confined atoms and molecules _ , springer ( uk ) 2014 .
y. sajeev and n. moiseyev , phys .
b * 78 * , 075316 ( 2008 ) . m. genkin and e. lindroth , phys . rev .
b * 81 * , 125315 ( 2010 )
. s. chakraborty and y. k. ho , phys .
a * 84 * , 032515 ( 2011 ) .
m. taut , phys .
a * * 4**8 , 3561 ( 1993 ) .
a. i. pupyshev and a. v. scherbinin , chem .
lett . * 295 * , 217 ( 1998 ) ; phys .
a * 299 * , 371 ( 2002 ) .
d. r. herrick , j. math .
* 16 * , 281 ( 1975 ) .
d. r. herrick and f. h. stillinger , phys .
a * 11 * , 42 ( 1975 ) .
sen , h.e .
montgomery , jr . and n.a.aquino , int . j. quantum chem . * 107 * , 798 ( 2007 ) .
sen , v. i. pupyshev and h.e .
montgomery jr . , ad .
quantum chem . * 57 * , 25 ( 2009 ) .
muzaian a. shaqqor and sami m. al - jaber , int .
* 48 * , 2462 ( 2009 ) .
montgomery jr , g. campoy and n. aquino , phys .
scr . * 81 * , 045010 ( 2010 ) .
xiao - yan gu and jian - qiang sun , j. math .
* 51 * , 022106 ( 2010 ) .
d. agboola , pramana * 76 * , 875 ( 2011 ) .
j. d. louck , j. mol .
spectrosc . 4 ( 1960 ) 298 - 333 ; a. chatterjee , phys .
186 ( 1990 ) 249 - 370 .
k. atkinson and w. han,_spherical harmonics and approximations on the unit sphere : an introduction _
( springer , new york , 2012 ) .
d. j. doren and d. r. herschbach , j. chem .
85 ( 1986 ) 4557 .
v. a. fock , bull . acad .
sci ussr , phys .
* 2 * , 169 ( 1935 ) .
alliluev , sov .
jetp * 6 * , 156 ( 1958 ) . j. avery , _ hyperspherical harmonics : applications in quantum theory _ ,
kluwer academic , boston ( 1989 ) .
d. r. herschbach , j. avery , and o. goscinski ( eds . ) , dimensional scaling in chemical physics , kluwer academic , dordrecht ( 1993 ) . s. f. singer , _ linearity , symmetry , and prediction in the hydrogen atom _ , ( springer , new york , 2005 ) . [ @xmath212-symmetry of the hydrogen atom
is discussed in chapters 8 and 9 . ]
d. m. fradkin , amer .
* 33 * , 207 ( 1965 ) .
a. chatterjee , phys .
rep . * 186 * , 249 ( 1990 ) .
m.dunn and d.k .
watson , ann .
* 251 * , 266 ( 1996 ) .
xiao - yan gu and zhong - qi ma , j. math .
phys . 44,3763 ( 2003 ) .
nightingale and mervlyn moodley , j. chem .
123 , 014304 ( 2005 ) . c. e. shannon , a mathematical theory of communication , the bell system technical journal * 27 * , 379423 ( 1948 ) @xmath213 ibid ,
* 27 * , 623 ( 1948 ) .
i. bialynicki - birula and j. mycielski , commun .
* 44 * , 129 ( 1975 ) .
r. a. fisher , theory of statistical estimation , in : proceedings of the cambridge philosophical society , no .
22 , pp . 700725
elvira romera , p. snchez - moreno , and j. s. dehesa , j. math .
47 , 103504 ( 2006 ) , mol .
108 , 2527 ( 2010 ) .
r. l. hall , phys .
d * 22*,2062 ( 1980 ) .
r. l. hall , j. math .
* 24 * , 324 ( 1983 ) .
r. l. hall , j. math .
phys . * 25 * , 2708 ( 1984 ) .
r. l. hall , phys .
a * 39 * , 550 ( 1989 ) .
r. l. hall , j. math .
phys . * 33 * , 1710 ( 1992 ) . r. l. hall , j. math .
phys . * 34 * , 2779 ( 1993 ) . s. j. gustafson and i. m. sigal , _ mathematical concepts of quantum mechanics _ ,
( springer , new york , 2006 ) . [
the operator inequality is proved for dimensions
@xmath76 on page 32 . ]
m. reed and b. simon , _ methods of modern mathematical physics ii : fourier analysis and self - adjointness _ , ( academic press , new york , 1975 ) .
[ the operator inequality is proved for @xmath82 on p 169 ] .
a. k. common , j. phys .
a * 18 * , 2219 ( 1985 ) .
w. thirring , _ a course in mathematical physics 3 : quantum mechanics of atoms and molecules _ ,
( springer , new york / wien , 1990 ) .
the local energy theorem is discussed on p154 .
h. ciftci , r. l hall and n. saad , j. phys .
36 ( 2003 ) 11807 .
n. saad , r. l. hall , and h. ciftci , j. phys .
39 ( 2006 ) 13445 - 13454 .
h. ciftci , r. l. hall , n. saad , and e. dogu , j. phys .
a : math . theor . 43
( 2010 ) 415206 .
r. l .. hall , n. saad , and k. sen , _ discrete spectra for confined and unconfined @xmath214 potentials in @xmath4-dimensions _ , j. math .
52 ( 2011 ) 092103 l. d. landau and e. m. lifshitz , _ quantum mechanics : non - relativistic theory , _ pergamon , london , 1981
. f. m. arscott , _ periodic differential equations : an introduction to mathieu , lam , and allied functions , _ pergamon press ( 1964 ) .
a. hautot , _ sur les solutions polynomiales de lequation differentielle de heun _ , bull .
lige 38 ( 1969 ) 654 - 659 and 660 - 663 .
j. rovder , _ zeros of the polynomial solutions of the differential equation @xmath215 , _ mat . cs . * 24 * ( 1974 ) 15 - 20 .. | a model physical problem is studied in which a system of two electrons is subject either to soft confinement by means of attractive oscillator potentials or by entrapment within an impenetrable spherical box of finite radius @xmath0 when hard confinement is present the oscillators can be switched off .
exact analytical solutions are found for special parameter sets , and highly accurate numerical solutions ( 18 decimal places ) are obtained for general cases .
some interesting degeneracy questions are discussed at length .
# 1 # 1eq.([#1 ] ) cuqm-152 + |
the barred spiral galaxy ngc1097 has an active nucleus classified as liner ( low - ionization nuclear emission - line region , @xcite ) , whose spectrum was found in 1991 to show transient broad ( fwhm@xmath310,000kms@xmath1 ) double - peaked balmer lines @xcite .
the nucleus of ngc1097 was the first low - luminosity active galactic nucleus ( llagn ) in a liner discovered displaying such lines , and , because of these broad lines , has also been classified as a seyfert1 nucleus .
later , other llagns in liner nuclei were found to display double - peaked balmer lines , such as m81 @xcite , ngc 4203 @xcite , and ngc 4450 @xcite .
the double - peaked h@xmath2 profile of the ngc1097 nucleus was monitored for the following 11 years after the discovery @xcite ( sb03 ) , showing variability in its flux , width and relative intensity of the blue and red peaks .
these variations were attributed to gas emission from a thin keplerian accretion disk , with a non - axisymmetric perturbation around a supermassive black hole ( hereafter smbh ) , where the gas is ionized by a central source , @xcite .
the variability of the double - peaked profile of ngc1097 is not rare .
the first active nucleus found to display variable double - peaked broad balmer emission lines was the nucleus of the radio galaxy 3c390.3 @xcite .
this object can be considered prototypical of the class of double - peaked emitters , because many of the now well - known features of these objects , such as variability in the relative intensity of the blue and red peaks @xcite , were first observed in 3c390.3 .
another remarkably well studied double - peaked emitter is the active nucleus of arp102b . for this object
, a study by @xcite has shown that the double - peaked balmer lines can not be driven by local viscous dissipation in the line - emitting part of the disk , as the h@xmath2 luminosity is of the order or even exceeds the energy locally available to power the line .
the same conclusion was obtained for other double - peaked emitters by @xcite and @xcite .
these authors suggested that the lines are powered by illumination from an external source .
the ionizing source could be an geometrically thick accretion flow , located within the inner radius of the line - emitting portion of the disk . in the case of ngc1097
, @xcite found that the spectral energy distribution ( sed ) of the nucleus is indeed well described by such a structure , a radiatively inefficient accretion flow ( riaf , @xcite ) .
this has led to the conclusion that an inner riaf is the ionizing source of the accretion disk in ngc1097 .
@xcite and @xcite reported a long - term monitoring of 14 other double - peaked emitters and found similar variations to those observed for ngc1097 , 3c390.3 and arp102b .
these objects also display in at least one epoch the red peak more intense than the blue peak .
this indicates that most line - emitting disks are non - axisymmetric , as in an axisymmetric disk the blue peak should be always more intense than the red because of doppler boosting .
the disks could then be elliptical @xcite , or have a non - axisymmetric emitting structure , such as a hot spot or spiral arm @xcite .
another model which can reproduce double - peaked emission - line profiles is the one proposed by @xcite and @xcite , in which the ionizing source is the continuum from a thin accretion disk , and the origin of the double - peaked lines is the broad - line region ( blr ) which is proposed to be an inner extension of the torus , in the shape of a `` bird s nest '' .
this geometry is not that distinct from that of the outer parts of an accretion disk , and in this regard is in approximate agreement with our proposed geometry .
but , in the case of ngc1097 , the study by @xcite favors an riaf origin for the nuclear ionizing continuum instead of a hot and thin accretion disk .
the monitoring of double - peaked profiles is a valuable experiment in the investigation of the structure and dynamics of the accretion disks and how they evolve as the gas moves inwards to feed the smbh . in @xcite , observing the ngc1097 h@xmath2 double - peaked profile approximately once or twice a year , we concluded that the asymmetry of the double - peaked profile was due to a spiral arm perturbation in the disk , rotating with a period of 5.5 years .
but as relevant timescales for evolution of the innermost regions of agns can be as short as weeks and months , we decided to repeat our experiment on ngc1097 in order to probe shorter timescales . here
we report the results of this experiment : we indeed found short variability timescales .
we also verified that , although our last observations of 2001 showed that the profile was becoming fainter and fainter , it is now again strong and was even stronger in 2010 december than in all previous observations .
this paper is organized as follows : in section 2 we describe the observations and the data reduction in section 3 we present the observational results , in section 4 we describe the models and the results from the modeling . in section 5
we discuss the timescales of the accretion disk variability and the interpretation of the modeling .
the conclusions of this work are presented in section 6 .
long - slit optical spectra of the nucleus of ngc 1097 were obtained in queue mode in 11 epochs from 2010 march 4 to 2011 march 21 using the gemini south multi - object spectrograph ( gmos ) in the long - slit mode .
the allocation was part of the `` poor weather '' opportunity , which is ideal for spectral monitoring which does not need absolute flux calibration or regular cadence .
this is the case for the nucleus of ngc1097 , for which we use the integrated flux in the narrow emission lines to normalize the spectra to match previous observations , under the assumption that the narrow lines do not vary within the extraction aperture .
the first three observations were obtained at time intervals of 4 - 5 months , since our previous monitoring suggested significant variations on a timescale of years . however , our new observations showed strong variations in the nuclear spectrum on timescales of months , which motivated us to ask for new observations at shorter time intervals .
we were then able to get at least two observations a month from 2011 january to 2011 march , as shown in the observing log presented in table [ obslog ] .
this table lists the identification of the program , date of observation , position angle ( p.a . ) of the slit and the average seeing during the observations , measured as the fwhm of a spatial profile extracted along a range of columns containing only the broad line ( after subtraction of the underlying continuum ) .
the data were obtained with the gmos south with a slit width of 1@xmath40 ( @xmath5pc ) , using the b600 grating and the gg455 filter to block possible higher order contamination .
the spectral range covered was 60007000 for most spectra , at a spectral resolution of @xmath0 4.5(fwhm of the lines in the arc spectrum , @xmath6 km s@xmath1 ) .
six 600s exposures were obtained at each epoch , which were then combined into a final spectrum .
l l l l l gs-2009b - q-99 & 2010 mar 4 & 70 & 1.00 + gs-2010a - q-81 & 2010 aug 15 & 55 & 0.98 + gs-2010b - q-90 & 2010 dec 22 & 280 & 0.63 + gs-2010b - q-90 & 2011 jan 11 & 280 & 0.73 + gs-2010b - q-90 & 2011 jan 15 & 280 & 0.73 + gs-2010b - q-90 & 2011 jan 18 & 280 & 0.70 + gs-2010b - q-90 & 2011 feb 8 & 280 & 1.21 + gs-2010b - q-90 & 2011 feb 10 & 280 & 0.79 + gs-2010b - q-90 & 2011 feb 14 & 280 & 0.72 + gs-2010b - q-90 & 2011 mar 17 & 280 & 1.43 + gs-2010b - q-90 & 2011 mar 21 & 280 & 0.82 the slit width of 1@xmath40 was selected because it is usually larger than the average seeing during the observations and was also the slit width of previous observations ( sb03 ) .
using the same slit width allows us to normalize the new spectra so that they match the flux of the narrow emission lines observed in previous spectra and study the evolution of the flux and profile of the broad h@xmath2 line over @xmath320yr .
the data reduction was performed using the standard tasks from the _ gmos _ package in the image reduction and analysis facility software ( iraf ) .
we extracted the nuclear spectra using a window of 1@xmath40 @xmath71@xmath40 centered at the peak of the continuum emission , assumed to correspond to the galaxy nucleus .
this position coincides also with the location of the unresolved source of the double - peaked h@xmath2 lines .
figure [ acquisition ] shows the acquisition image of the observation of 2010 august 15 , where the nuclear extraction region is represented by the square labeled a. since the nuclear spectra show strong absorption lines from the underlying stellar population , we subtracted the stellar population contribution , in order to isolate and better study the emission line profile and in particular the broad h@xmath2 line .
a template stellar population spectrum was obtained as follows . for each epoch , in addition to the nuclear spectrum , we extracted two more spectra using two similar windows of 1@xmath40@xmath71@xmath40 , one centered at 1@xmath45 to the southwest ( figure [ acquisition ] , region b ) and another at 1@xmath45 to the northeast of the nucleus ( figure [ acquisition ] , region c ) . these extranuclear spectra do not show any broad h@xmath2 emission , but show some narrow - line emission which was edited out , using as guides , stellar population templates from previous studies @xcite .
averaging these two spectra , we obtain a stellar population spectrum from a location approximately 100pc away from the nucleus . under the assumption that this spectrum is representative of the stellar population at the nucleus , we subtracted it from the nuclear spectrum ( after normalizing the continuum to the corresponding value at the nucleus ) thus isolating the gas emission .
figure [ exemp1 ] illustrates the process for the spectrum of 2010 august 15 , which covers a broader spectral range than most spectra , from 4750 to 7200 .
we note that this process reveals the presence of a broad double - peaked h@xmath8 line which was not obvious prior to the subtraction , as well as some weak emission lines such as [ ni]@xmath95199and [ oi]@xmath96363 .
the naid absorption line in the residual spectrum is most probably interstellar , as all other stellar absorption lines did not show any residual .
0@xmath71@xmath40 extraction windows for the nucleus and stellar population spectra .
the square labeled a represents the nuclear extraction window , while squares b and c represent the extranuclear ones , from which we obtained the spectrum of the underlying stellar population . ] ) .
middle : stellar population obtained from averaging the spectra from regions b and c in figure [ acquisition ] .
bottom : nuclear spectrum after the subtraction of the stellar population . ]
we also note that after subtracting the stellar population spectrum from the nuclear one , we do not see any evidence for a significant contribution of an agn continuum , as can be seen in figure [ exemp1 ] .
the effect of a nuclear continuum , if present , would be to dilute the stellar absorption lines , making them shallower at the nucleus than outside .
the subtraction of the stellar population template , as described above , after normalization to the flux in the continuum of the nuclear spectrum would result in the presence of `` fake '' emission features in the residual spectrum and a possible slope ( in the case of a blue continuum , for example ) , which are not seen .
we have performed a sequence of tests in order to estimate the contribution from a featureless continuum ( fc ) that could be `` hidden '' in the nuclear spectrum as follows .
we added the contribution of the fc , assumed in the form @xmath10 , to the stellar population template described above , at increasing percent contributions of its flux at 5800 .
the resulting composite ( fc+stellar population ) spectrum was then renormalized to the same flux at 5800 , and we subtracted the stellar population template from it .
we concluded that for the short wavelength range and the signal - to - noise ratio of our observations , in order to have clear residuals showing fake emission features and a blue continuum we need the fc continuum to contribute with at least 15%20% of the flux at 5800 .
we thus can say that a possible fc contribution to the nuclear spectrum is lower than 20% of the continuum flux at 5800 .
we consider this value an upper limit , as a comparison between the stellar continuum from our spectra and the agn continuum obtained in the work of nemmen et al .
( 2006 ) shows that the stellar continuum in our aperture is about 30 times stronger than the agn continuum in the optical .
after subtracting the contribution of the stellar population , the final step before the analysis of the double - peaked profile was to normalize the spectra using as reference a previous flux - calibrated spectrum .
we used as reference the spectrum of 1991 november 2 @xcite , and assumed that the flux of the narrow lines h@xmath2 , [ nii]@xmath116548 , 6584 , and [ sii]@xmath126717 , 6731 did not vary during the last 20 years .
this is a common assumption in this type of study , as the narrow lines originate in extended regions with spatial scales much larger ( @xmath0 100pc ) than that of the accretion disk ( @xmath13pc ) .
thus , their response time to variations in the ionizing flux is correspondingly much larger than that for the line - emitting accretion disk , and can not respond to variations of the ionizing flux that are as fast as those we have detected .
the normalized spectra from our 201011 observations are compared with the reference spectrum of 1991 november in figure[results ] ( pairs of spectra are shifted from each other in the vertical direction for clarity ) . in this figure
we show the three spectra from 2011 january , the three from 2011 february , and the two from 2011 march , respectively , on top of each other because we did not find any significant variation within each group .
note that the double - peaked h@xmath2 profile from 2010 august is very different from that of 2010 march . in 2010
march , the red peak was stronger than the blue peak , while in 2010 august , the blue peak was stronger than the red .
this inversion of the relative intensity of the peaks occurred on a timescale of 5 months .
in addition , in 2010 december , the profile showed a marked increase in its integrated flux compared to that of 2010 august , while the red and blue peaks showed similar strengths . from 2010 december 22 to 2011 january 11 we observed a @xmath025% decrease in the integrated flux of the double - peaked profile .
the three observations of 2011 january were obtained within a time interval of seven days , and showed no significant variations . in 2011 february , the three observations were obtained within a time interval of six days , and again we did not see significant variations among the three profiles .
however , the total flux continued to decrease compared with that of january , with the blue and red peaks showing similar strengths . in 2011 march , we obtained two observations separated by a time interval of four days , between which we again did not find significant variation in the double - peaked profile .
but the total flux decreased further , and the profile showed the red peak somewhat stronger than the blue peak . in summary
, this series of observations obtained between 2010 march and 2011 march reveal : ( 1 ) a marked increase of the broad - line flux between 2010 august and 2010 december ; ( 2 ) a significant decrease of the flux on timescales longer than a week ; ( 3 ) no significant changes in the profile on time intervals shorter than a week ; ( 4 ) an inversion in the relative strength of the blue and red peaks on a timescale of 5 months .
we have quantified the observed variations by measuring the wavelengths of the blue and red peaks of the profiles , their peak flux densities , and the integrated flux of the broad line , as described below .
we list the results in table [ medidas ] . in order to isolate the broad double - peaked h@xmath2 emission and perform the measurements
, we first evaluated the contribution of the narrow lines h@xmath14 , [ nii]@xmath15 , [ sii]@xmath16 and [ oi]@xmath17 by fitting gaussians to these lines and then subtracting them .
the flux of the double - peaked line ( @xmath18 ) was measured by integrating the total flux under the profile after this subtraction .
the average uncertainty in @xmath18 is @xmath19 erg @xmath20 s@xmath1 .
the wavelengths of the blue and red peaks ( @xmath21 and @xmath22 ) and the corresponding peak flux densities ( @xmath23 and @xmath24 ) were obtained by fitting three gaussians to the double - peaked profile , one for the blue peak , another for the red peak and a third gaussian for the center of the profile .
@xmath23 and @xmath24 are the peak flux densities of the gaussians fitted to the blue and red peaks , respectively , while @xmath21 and @xmath22 are the corresponding wavelengths .
we have tried other methods as well , such as using splines to fit the peaks , but concluded that the fit of gaussians gave equivalent but more robust results .
the average uncertainties in the measurements are @xmath25for @xmath21 and @xmath22 and @xmath26 erg @xmath20 s@xmath1@xmath1 for @xmath23 and @xmath24 . from @xmath21 and @xmath22
we obtained the blue and red peak velocities @xmath27 and @xmath28 relative to the systemic velocity of the galaxy , adopted as the one corresponding to the narrow component of the h@xmath2 emission , as we have done in our previous studies ( sb03 ) .
figure [ vels ] shows the temporal evolution of @xmath27 and @xmath28 .
the velocities of the blue peak range between @xmath29 and @xmath30 km s@xmath1 while for the red peak they are between 4000 and 6000kms@xmath1 .
a comparison between these peak velocities with those from the period 19912001 ( sb03 ) shows that , in the new observations , the blue and red peaks always have larger peak velocities than the minimum values observed between 1991 and 2001 , which are @xmath31kms@xmath1 for the blue and @xmath32kms@xmath1 for the red , as indicated by the lower gray horizontal bars in both panels of figure[vels ] .
most velocities are instead similar or higher than the maximum values observed between 1991 and 2001 , @xmath33kms@xmath1 for the blue and @xmath34kms@xmath1 for the red peak .
these are indicated by the upper gray horizontal bars in figure[vels ] .
the values of v@xmath35 and v@xmath36 were obtained by using the expression for the relativistic doppler effect .
figure [ flarga ] shows the variation of the integrated broad h@xmath2 emission line flux from 2010 march to 2011 march .
the integrated flux is always above the minimum value observed in the period 19912003 ( lower gray bar in the figure ) , being close to the maximum value from 2010 march to 2011 january ( upper gray bar in the figure ) .
a maximum in the flux occurred between 2010 august 15 and 2010 december 22 , when the integrated flux rose above the maximum value observed between 1991 and 2001 .
after 2010 december 23 , there was an abrupt decrease in the broad - line flux , by @xmath325% of its maximum value in 20 days .
on the other hand , it can be observed that there is little variation on timescales shorter than a week ( the maximum time interval between the observations obtained in the same month ) .
thus significant variations in the broad - line flux occur on a timescale just above a week . from figure [ flarga ]
, one can derive a decrease of @xmath312% of the flux in 10 days .
figure [ fbfr ] shows the temporal evolution of the peak fluxes of the blue and red peaks as well as that of their ratio @xmath23/@xmath24 . on 2010 march ,
the red peak was stronger than the blue , while in 2010 august the blue peak was stronger .
thus , within a time interval of 5 months , there has been an inversion of the @xmath23/@xmath24 ratio , and , in the following 7 months this ratio decreased to a value which is just below 1 , while in 2010 march this value was 0.7 .
thus the change from a stronger red peak to a stronger blue peak was faster than the change back to a stronger red peak . from figure[fbfr ] we thus derive a timescale for the inversion of the relative intensity of the two peaks of @xmath35 months .
figure [ correlacao ] shows that there is an inverse correlation between the velocity separation of the red and blue peaks @xmath28 @xmath27 and @xmath18 , observed both in our recent observations , represented by the circles , as well as in the previous observations of sb03 , represented by the crosses . this inverse correlation between @xmath18 and @xmath37 @xmath27 can be interpreted as due to : ( 1 ) when the central ionizing source becomes brighter , the line emission from the outer radii become relatively more important , as the radiation of the ionizing source reaches larger distances in the disk , where the velocity is lower ; ( 2 ) when the central source is fainter , the inner radii are favored , as the ionizing radiation reaches smaller radii , where the velocity is larger .
a comparison of the two data sets in figure[correlacao ] shows that our recent data define a sequence parallel to that of the previous data , displaced from the previous data due to larger velocity separations between the two peaks and to higher fluxes .
the larger velocity separation of the two peaks suggests that the line emitting portion of the disk is closer to the ionizing source or has a larger contribution from gas closer to the source .
l c c c c c 2010 mar 4 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2010 aug 15 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2010 dec 22 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2011 jan 11 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2011 jan 15 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2011 jan 18 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2011 feb 8 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2011 feb 10 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2011 feb 14 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2011 mar 17 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2011 mar 21 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] line @xmath18 from 2010 march to 2011 march .
the gray horizontal bars show the minimum and maximum values ( plus uncertainties ) in the spectra from 1991 to 2001 ( sb03 ) . ] and @xmath28 @xmath39 @xmath27 .
circles represent the measures for this work and crosses represent the data from sb03 .
the inverse correlation for the new observations however is displaced to larger velocities and fluxes relative to previous observations ( sb03 ) .
this suggests that emission from the inner parts of the disk became more important relative to that of the outer parts . ]
in order to investigate the origin of the observed variations in the double - peaked profile , we have used accretion disk models to reproduce the profile .
figure [ fbfr ] shows that the double - peaked profile still shows variations in the relative intensities of the two peaks , alternating between a stronger blue peak and a stronger red peak , as previously observed by sb03 .
this pattern of variation supports the origin of the profile in an accretion disk whose emission is not axisymmetric . in sb03 ,
the relative strengths of the two peaks and their variation were better explained by emission from a circular accretion disk with a single spiral arm .
the enhanced emission in the arm , as it rotates in the disk , produces the observed asymmetry of the profile .
we will thus also adopt in our present modeling the single spiral arm model to fit the double - peaked profiles .
the source of radiation which ionizes the disk in order to drive the h@xmath2 emission is also relevant to the modeling . as pointed out in the introduction , in the case of ngc1097
, this source is an riaf internal to the line - emitting part of the disk @xcite . @xcite
have found that , in the case of an accretion disk illuminated by a non - thermal continuum source with a spectral energy distribution extending up to the hard x - ray band , there is a `` radius of maximum emission '' or `` saturation radius '' , at which the h@xmath2 emission reaches a maximum value .
thus we consider an additional model in which the disk emissivity may increase with radius up to the saturation radius and then decrease outward . with the above considerations in mind
, we then fitted the double - peaked h@xmath2 line profiles with two different models , described below . in both models
we assume a circular emitting disk , or wide ring ( as it has a hole in the middle ) , with a perturbation in the emissivity in the form a spiral arm . in the first model which we call _ spiral arm model _
the variations in the relative strengths of the peaks are controlled by the rotation of the spiral arm while the separation of the peaks is controlled by an emissivity law which is a simple power - law function of the radius . by changing the power - law index we can weight the emissivity of the disk toward inner regions ( leading to a broader profile ) or the outer regions ( leading to a narrower profile ) . in the second model which we call the _ saturated spiral model
_ we have changed the emissivity law to a broken power - law , which can produce a ring of maximum emissivity between the inner and outer radii . in this case , the separation of the peaks is controlled by the radius of this ring of maximum emissivity .
we adopt the formulation described in @xcite , in which : ( 1 ) the line emitting portion of the disk is circular and located between an inner radius @xmath40 and an outer radius @xmath41 ( where @xmath42 is the disk radius in units of the gravitational radius @xmath43 , @xmath44 is the light speed , @xmath45 is the gravitational constant and @xmath46 is the mass of the black hole ) ; ( 2 ) the disk has an inclination @xmath47 relative to the line of sight ( zero degrees is face on ) .
superimposed on the axisymmetric emissivity of the circular disk , there is a perturbation in the form of a spiral arm .
we adopt only one spiral arm , as we have already verified that two or more arms do not reproduce well the variations in the profile ( sb03 ) .
the total emissivity of the disk is given by : @xmath48 \right .
\\ \left .
+ \frac{a}{2 } \exp \left [ -\frac{4 \ln2}{\delta^{2 } } ( 2\pi - \phi + \psi_{0})^{2 } \right ] \right \},\end{gathered}\ ] ] where @xmath49 is the axisymmetric emissivity of the disk , @xmath50 is the brightness contrast between the spiral arm and the underlying disk , and the expression between square brackets represents the decay of the emissivity of the arm as a function of the azimuthal distance @xmath51 from the ridge line to both sides of the arm , assumed to be a gaussian function with fwhm @xmath52 ( azimuthal width ) .
the relation between the azimuthal angle @xmath53 and the angular position @xmath54 of the ridge of emissivity on the spiral arm is given by @xmath55 where @xmath56 is the azimuthal angle of the spiral pattern , @xmath57 is the _ pitch _ angle and @xmath58 is the innermost radius of the spiral arm . the specific intensity from each location in the disk , in the frame of the emitting particle is calculated as : @xmath59 where @xmath60 is the emission frequency , @xmath61 ( h@xmath2 6564.6 ) is the rest frequency and @xmath62 is the local `` broadening parameter '' @xcite . in the fits of the individual profiles we first used the same set of parameters of sb03 , namely : the inner and outer radii of the line emitting portion of the disk @xmath63 and @xmath64 , the inclination angle @xmath65 , broadening parameter @xmath66 km s@xmath1 , pitch angle @xmath67 and @xmath68 . in order to reproduce the evolution of the profile we varied the parameters @xmath50 , @xmath69 and @xmath53 .
the variation of @xmath53 regulates the evolution of the shape of the profile , while a variation in @xmath50 , together with @xmath53 , regulates the relative intensities of the two peaks .
the value of @xmath69 regulates the radius in the disk of the maximum line luminosity , which corresponds to the inner regions for the broadest profiles , or to the outer regions for the narrowest profiles .
( orientation of the spiral pattern ) , @xmath50 ( contrast between the spiral arm and the disk ) , and @xmath69 ( the slope of the emissivity law ) .
the values of the parameters are listed in table [ parspiral ] . ]
l c c c c 2010 mar 4 & 2.0 & @xmath392.0 & 130 & 0 + 2010 aug 15 & 1.0 & 1.0 & 250 & 0 + 2010 dec 22 & 0.3 & 1.0 & 330 & 0 + 2011 jan 11 & 0.5 & 1.5 & 340 & @xmath39460 + 2011 jan 15 & 0.5 & 1.5 & 340 & @xmath39460 + 2011 jan 18 & 0.5 & 1.5 & 340 & @xmath39460 + 2011 feb 8 & 1.0 & 2.0 & 360 & @xmath39460 + 2011 feb 10 & 1.0 & 2.0 & 360 & @xmath39460 + 2011 feb 14 & 1.0 & 2.0 & 360 & @xmath39460 + 2011 mar 17 & 2.0 & 2.0 & 400 & @xmath39460 + 2011 mar 21 & 2.0 & 2.0 & 400 & @xmath39460 the best - fitting models are compared to the observed line profiles in figure [ mod1 ] , while the corresponding parameter values ( @xmath50 , @xmath69 and @xmath53 ) are listed in table [ parspiral ] .
the contrast of the spiral arm @xmath50 ranged from 2.0 at the first epoch , decreased to a minimum value of 0.5 in 2011 january , and increased again to 2.0 in 2011 march , while @xmath69 increased from the first to the last epochs .
an abrupt variation in @xmath69 occurred between 2010 march and 2010 august when the value of @xmath69 changed from @xmath70 , favoring the outer parts of the disk , to 1.0 , favoring the inner parts of the disk .
the @xmath53 values were constrained to provide a monotonic rotation of the spiral pattern , from 130@xmath71 to 400@xmath71 .
the spiral arm has thus almost completed one revolution in one year . for the spectra from 2011 january to 2011 march ,
an improvement of the fit can be obtained by allowing a small , ad hoc blueshift of the central wavelength of the profile relative to the wavelength of the narrow h@xmath2 line ( 6564.6 ) . this blueshift is @xmath3 @xmath72 kms@xmath1 , as listed in the last column of table [ parspiral ] .
the improved fits with the blueshifts are shown as dashed lines in figure [ mod1 ] . in figure
[ phi1 ] we show the evolution of the @xmath53 values obtained from the fits , represented by the circles in the figure , while the black solid line is the best linear regression to the points and the gray band covers the uncertainty in the regression .
the linear regression has angular coefficient @xmath73 day@xmath1 , which represents the angular velocity of the arm and implies a rotation period of @xmath740.5 months .
values obtained from the fit of the spiral arm model .
the solid line is a linear fit to the data and the shaded bar covers the uncertainty in the fit . ] in an effort to improve the fit to the broad double - peaked h@xmath2 profiles , we modified the spiral arm model to allow a more flexible radial emissivity law .
we have chosen an emissivity law which allows the existence of a `` radius of maximum emissivity '' between the inner and outer radii of the disk at which the broad h@xmath2 line reaches a maximum intensity , as proposed by @xcite .
the emissivity increases with radius until the saturation radius and then decreases outward .
the radius of maximum emissivity was incorporated by changing the axisymmetric part of the emissivity of the disk ( equation [ epsilonzero ] ) as follows : @xmath75 where : the new parameter @xmath76 is the radius of maximum emissivity , or saturation radius , at which the emissivity law changes ; @xmath77 is the index of the emissivity law for the regions internal to @xmath76 ( @xmath78 ) ; @xmath79 is the index for the regions external to @xmath76 ( @xmath80 ) ; @xmath40 and @xmath41 are the inner and outer radii , respectively .
we kept the value @xmath81 , as discussed by @xcite , who showed that beyond the saturation radius the flux emitted by the line decreases as @xmath82 .
after many tests , @xmath83 gave the best fits , and was fixed at this value .
the negative value allows the increase of the emissivity until the saturation radius .
the parameter @xmath76 was allowed to vary .
we kept the parameters of the spiral arm at the values @xmath84 , @xmath67 , @xmath68 , as well as the inclination of the disk at @xmath85 .
we have varied the broadening parameter and found that the best value to reproduce the shape of most profiles was @xmath86 km s@xmath1 . in summary ,
in the saturated spiral model , the only parameters we needed to vary in order to reproduce the evolution of the profile were @xmath76 , @xmath50 , and @xmath53 .
the changes in the separation between the blue and red peaks were obtained by changing @xmath76 ; when @xmath76 is smaller , the separation between the blue and red peaks increases .
( orientation of the spiral pattern ) , @xmath50 ( contrast between the disk and the spiral arm ) and @xmath76 ( radius of maximum emissivity ) .
the values of the parameters are listed in table [ parshibrid ] . ]
l c c c c 2010 mar 4 & 3.0 & 1200 & 140 & 0 + 2010 aug 15 & 1.0 & 800 & 220 & 0 + 2010 dec 22 & 0.2 & 1000 & 330 & 0 + 2011 jan 11 & 1.0 & 650 & 340 & @xmath39500 + 2011 jan 15 & 1.0 & 650 & 340 & @xmath39500 + 2011 jan 18 & 1.0 & 650 & 340 & @xmath39500 + 2011 feb 8 & 1.0 & 550 & 360 & @xmath39550 + 2011 feb 10 & 1.0 & 550 & 360 & @xmath39550 + 2011 feb 14 & 1.0 & 550 & 360 & @xmath39550 + 2011 mar 17 & 1.5 & 600 & 400 & @xmath39230 + 2011 mar 21 & 1.5 & 600 & 400 & @xmath39230 the best - fitting models are compared with the observed line profiles in figure [ mod3 ] , and the corresponding parameters values ( @xmath76 , @xmath50 and @xmath53 ) are listed in table [ parshibrid ] .
the parameter @xmath76 ranged from a maximum value of 1200 in 2010 march when the separation between the peaks was the smallest , to a minimum value of 550600 in 2011 february march , when the separation between the peaks was the largest .
the parameter @xmath50 assumed the value 3.0 in 2010 march , when the red peak was significantly stronger than the blue , as seen in figure [ mod3 ] , while in 2010 december this parameter reached the minimum value of 0.2 , when the peaks had similar strengths .
similarly to the previous model , @xmath53 varied from 140@xmath71 in the first epoch to 400@xmath71 in the last .
as in the case of the spiral arm model , from 2011 january to 2011 march , the fit was improved by introducing a blueshift which is listed in the last column of table [ parshibrid ] .
this blueshift ranges from @xmath3 @xmath87 to @xmath88 kms@xmath1 , and the improved fits with the blueshifts are shown as dashed lines in figure [ mod3 ] . in figure [ modelo2 ]
we show emissivity maps constructed using the parameters of the best fits using the saturated spiral model . in 2010
march the radius of maximum emissivity @xmath89 was closer to @xmath41 , when the outer parts where brighter , while in 2011 march @xmath89 was closer to @xmath40 and the inner parts were brighter .
in order to discuss the physical processes driving the changes in the profiles , we first review four relevant timescales for the accretion disk in ngc1097 : the _ viscous timescale _ ( @xmath90 ) , the _ sound - crossing timescale _ ( @xmath91 ) , the _ dynamical timescale _ ( @xmath92 ) and the _ light travel timescale _ ( @xmath93 ) .
these timescale are presented in @xcite and can be modified in order to express the disk radius in terms of the gravitational radius @xcite : @xmath94 @xmath95 @xmath96 @xmath97 where @xmath98 , m@xmath99 , @xmath100 is the shakura
sunyaev viscosity parameter , @xmath101 is the temperature in units of 10@xmath102 k , and @xmath103 is the mass accretion rate in units of 0.1 @xmath104yr@xmath1 . for ngc 1097 , @xmath105 @xcite , and @xmath106 ( in eddington units , @xcite ) , corresponding to @xmath107 m@xmath108yr@xmath1 .
adopting typical values of @xmath109 and t@xmath110 , a characteristic radius for the disk is @xmath111 . adopting the above values of @xmath46 and @xmath112 , we obtain the following estimates for the timescales : 1 .
@xmath113 10@xmath114 years : this large value for the viscous timescale does not support that the observed changes in the profile on timescales from weeks to years are due to inward transfer of mass due to viscous torques ; 2 . @xmath115 265 years : this is also much larger than the observed variations timescales ; 3 .
@xmath116 months : the dynamical timescale depends on the radius , with the inner parts of the disk having shorter timescales than the outer parts .
adopting @xmath117 for the inner radius , as obtained from our modeling , the corresponding dynamical timescale is just above 2 months .
the new observations presented here show that the variation timescale of the relative strengths of the blue and red peaks ( @xmath23/@xmath24 ) is @xmath35 months ( see figure [ fbfr ] ) , thus compatible with the dynamical time . in our modeling
, this variation was reproduced by the rotation of the spiral pattern around the disk , whose period is close to the dynamical time at @xmath118 ; 4 .
@xmath119 days : considering the range of radii of the disk , this timescale ranges from @xmath34 to @xmath312 days .
this is the shortest timescale of the disk . according to our data ( see figure [ flarga ] ) ,
the smallest timescale of the profile variations namely variations in the integrated flux of the double - peaked line occur in @xmath1207 days .
this supports the interpretation that the variability in the integrated flux @xmath18 occurs in the light travel timescale , as a result of variations in the ionizing flux .
similarly , the variations in @xmath27 and @xmath28see figure [ vels]which correlate with @xmath18 , also occur in the light travel timescale . according to the models , these are due to changes in the region of maximum emissivity of the disk .
the timescales discussed above provide constraints on the mechanisms that cause the changes in the double - peaked profile .
our data clearly reveal two distinct timescales : ( 1 ) the first corresponding to changes in the integrated flux of the double - peaked profile , which is just over 7 days ; and ( 2 ) the second corresponding to the changes in relative strengths of the blue and red peaks , which is in the range 56 months .
we have identified the first timescale with the _ light travel timescale_. in order to model the corresponding variations in the flux of the double - peaked line , and at the same time reproduce the variation of the width of the profiles it was necessary to vary the disk emissivity . when the flux is higher , the profile is narrower , and the emissivity is higher in the outer parts of the disk , while when the flux is lower , the profile is broader and the emissivity is relatively higher in the inner parts of the disk . in the saturated spiral model ,
what varies is the radius of maximum emissivity .
the physical scenario we favor is the following : the double - peaked profile originates in gas rotating in a thin accretion disk , which is ionized by an riaf @xcite located inside the inner radius of the line - emitting portion of the disk . as the variations occur on the _ light travel timescale _ , our interpretation is that the variation in @xmath18 can be considered a `` reverberation '' effect produced by variations in the ionizing flux from the central riaf , and allows a test for the values of the radii obtained via the models . calculating the values for the inner and outer radii of the disk in light days we obtain for @xmath121@xmath33 light days and for @xmath118@xmath311 light days , which is consistent with the variability timescale of @xmath37 days that we observe . in order to further confirm the reverberation hypothesis
, it would be necessary to do a proper `` reverberation mapping campaign '' ( e.g. , @xcite ) of the accretion disk via simultaneous observations of the riaf emission ( e.g. , in x - rays ) and the double - peaked profile .
we plan to pursue such a monitoring campaign in the future , including a more extended spectral coverage to include the blue part of the optical spectrum which will also allow us to test other models for the origin of the continuum , such as the one proposed by @xcite .
we have identified the second timescale of the variations in the relative strengths of the blue and red peaks with the _
dynamical timescale_. this means that the asymmetry in the profile is due to a non - axisymmetric emitting structure modeled as a single spiral arm rotating in the disk , with a rotation period of @xmath318 months .
this timescale and its interpretation differs from our previous one ( sb03 ) . in that paper
we concluded that the variation of the relative strengths of the two peaks occurred in the _ sound - crossing timescale_. this distinct interpretation was due to the fact that the period we have derived then for the variation was 5.5yr , and at that time our estimate for the smbh mass was @xmath310@xmath122 @xmath123 , giving a smaller value for the sound - crossing timescale , consistent with the derived period of variation . with a revised smbh mass two orders of magnitude larger ,
the revised sound - crossing timescale is much larger than the timescale of the variation of the relative fluxes of the blue and red peaks we have now derived from our more frequent observations .
is important to point out that the perturbation does not need to be a spiral arm , even though we used the spiral arm prescription to model it . in the images of the disk ,
constructed using the best model parameters for the different epochs , shown in figure [ modelo2 ] , the spiral arm actually looks more like an azimuthal arc .
therefore , it could be that this arc is actually a clump that was produce by self - gravity and then sheared ( see discussion in @xcite ) .
another alternative is that this feature is an irradiation - induced warp @xcite .
but although the irradiation - induced warp model can reproduce the observed variability of the line profiles @xcite , the required rotation timescale is incompatible with the mass of the smbh in ngc 1097 ( see equation ( 4 ) in @xcite ) . on the other hand ,
the sheared clump would orbit the smbh on the dynamical timescale and therefore could reproduce the 5 month variation that we observe .
figure[correlacao ] shows that the profiles got broader and brighter in the recent ( 20102011 ) observations when compared to the previous ones ( 1991 - 2001 ) , although at both epochs there is an inverse correlation between the flux and width of the profiles .
this inverse correlation is expected in the reverberation scenario discussed above , but there is an offset between the previous and present correlations .
a comparison between the parameters of the present and previous spiral arm model shows that the emissivity law in the recent epochs favors a higher contribution of the inner regions relative to the outer regions of the disk , leading to a broader profile .
one possibility to explain this increase in emissivity of the innermost regions is an increase in the gas density in these regions of the disk in the most recent epochs when compared to the period 19912001 .
a comparison between figures [ mod1 ] and [ mod3 ] shows that the saturated spiral model gives a better fit to the double - peaked profiles . in order to quantify the goodness of the fits ,
we have obtained the root mean square ( rms ) deviation between the observed profiles and the models . in almost all observations the saturated spiral model provided a smaller rms when compared with that for the single spiral arm model , confirming what is seen in the fits .
finally , we point out that , although the models can in general reproduce the h@xmath2 profiles , there are several epochs in which we observe an excess flux in the blue wings of the profiles relative to the models . this excess flux is observed from 2010 august to february 2011 ( figures [ mod1 ] and [ mod3 ] ) . in addition , from 2011 january to march we observe a systematic blueshift of the center of the line , from @xmath39230 to @xmath39550 km s@xmath1 which could be due to an accretion disk wind .
therefore , it will be worthwhile in future work to compare the observed line profiles with models that include an outflowing wind ( e.g. , @xcite ) the physical value of the disk timescales , such as the light - travel timescale ( @xmath93 ) , as well as the inner and outer radii of the accretion disk , are strongly dependent on the mass of the smbh .
thus , uncertainties in its determination must be taken into account .
we have adopted here the smbh mass value of @xcite : @xmath124 @xmath104 .
the range of values for the inner and outer radii of the accretion disk ( @xmath125 and @xmath126 , respectively ) resulting from the uncertainties in the smbh mass quoted by @xcite are @xmath127 light - days and @xmath128 light - days .
but @xcite obtain the value for the smbh mass using the @xmath46@xmath62 relation of @xcite , pointing out that they do not take into account the scatter in this relation .
they consider only the uncertainty in the coefficients of the relation that results from the scatter ( effectively they assume that the observed scatter is not intrinsic ) .
if we now consider that the uncertainty in the smbh mass is also affected by the scatter in the @xmath46@xmath62 relation from @xcite , which can introduce an uncertainty in the mass of @xmath3 0.3dex , the resulting minimum and maximum values for the inner and outer radii become : ( 1 ) for @xmath129 @xmath104 , the minimum radii are @xmath130 light - days and @xmath131 light - days ; ( 2 ) for @xmath132 @xmath104 , the maximum radii are @xmath133 light - days and @xmath134 light - days . thus , even considering the maximum uncertainties which could affect the determination of the smbh mass , the resulting values for the inner and outer radii of the disk are still of the order of light - days .
the above results support the interpretation that the shortest variation timescale of 7days is the light - travel time between the ionizing source at the center of the disk and the mean radius of the disk .
in fact , this observed timescale can actually constrain the values of @xmath40 to @xmath135light - days and of @xmath136 to @xmath137light - days .
in addition , if we assume that 7light - days is the mean distance between the broad - line - emitting clouds and the smbh , we can estimate the mass of the smbh independently from the model , as follows . if this system of clouds is flattened , as suggested by their double - peaked profile , and we adopt as the corresponding keplerian velocity around the smbh the mean velocity of the peaks@xmath138kms@xmath1we obtain @xmath46=1.1@xmath139@xmath123 , for an inclination of 34@xmath140 relative to the plane of the sky ( as obtained from the disk model ) .
this value is in good agreement with the one obtained by @xcite .
we report the discovery of short timescale variations in the broad double - peaked h@xmath2 emission line from the liner nucleus of ngc1097 , observed in 9 epochs between 2010 march and 2011 march .
they comprise : ( 1 ) variations in the integrated flux and width of the line over a timescale of @xmath1417 days , which we identified as the light - travel time between the ionizing source and the disk ; it is the first time that such short timescale variations have been seen in a double - peaked profile from the nucleus of a galaxy ; ( 2 ) variations in the relative intensity of the blue and red peaks on a timescale of 56 months , which is compatible with the disk dynamical timescale . 1 . short timescale changes in the disk emissivity namely in the contrast @xmath50 and index @xmath69 in the _ spiral arm model _ and @xmath50 and @xmath89 in the _ saturated spiral model _ ; 2 . the rotation of the spiral arm via the change of the orientation @xmath53 of the arm in both models resulting in a estimate for the rotation period of the arm of @xmath142 months .
the values of the inner and outer radii , obtained from previous modeling , were kept in the present modeling and are , respectively , @xmath143 and @xmath144 ( corresponding to @xmath145 and 11 light - days , respectively ) .
these values are in good agreement with the short timescale variations seen in our observations .
this agreement supports the identification of the shortest timescale variations with the light travel time between the ionizing source and the accretion disk , in one hand , and gives further support to the model and the derived parameters , on the other hand .
our recent observations support the scenario in which an accretion disk whose emissivity changes due to reverberation of a variable ionizing continuum ( presumably from an riaf ) , has a non - axisymmetric feature rotating in the dynamical timescale .
gilbert , a. m. , eracleous , m. , filippenko , a. v. , & halpern j. p. 1999
, in asp conf .
175 , structure and kinematics of quasar broad line regions , ed . c. m. gaskell , w. n. brandt , m. dietrich , d. dultzin - hacyan , & m. eracleous ( san francisco , ca : asp ) , 189 | the broad ( fwhm @xmath0 10,000 km s@xmath1 ) double - peaked h@xmath2 profile from the liner / seyfert 1 nucleus of ngc1097 was discovered in 1991 , and monitored for the following 11 years .
the profile showed variations attributed to the rotation of gas in a non - axisymmetric keplerian accretion disk , ionized by a varying radiatively inefficient accretion flow ( riaf ) located in the inner parts of the disk .
we present and model 11 new spectroscopic observations of the double - peaked profile taken between 2010 march and 2011 march . this series of observations was motivated by the finding that in 2010 march the flux in the double - peaked line was again strong , becoming , in 2010 december , even stronger than in the observations of a decade ago .
we also discovered shorter timescale variations than in the previous observations : ( 1 ) the first , of @xmath07 days , is interpreted as due to reverberation of the variation of the ionizing source luminosity , and the timescale of 7 days as the light crossing time between the source and the accretion disk ; this new timescale and its interpretation provides a distance between the emitting gas and the supermassive black hole and as such introduces a new constraint on its mass ; ( 2 ) the second , of @xmath3 5 months , was attributed to the rotation of a spiral arm in the disk , which was found to occur on the dynamical timescale .
we use two accretion disk models to fit theoretical profiles to the new data , both having non - axisymmetric emissivities produced by the presence of an one - armed spiral .
our modeling constrains the rotation period for the spiral to be @xmath318 months .
this work supports our previous conclusion that the broad double - peaked balmer emission lines in ngc1097 , and probably also in other low - luminosity active nuclei , originate from an accretion disk ionized by a central riaf . |
higher - order flow coefficients @xmath1 are useful observables to constrain the properties of the quark - gluon plasma , such as a shear viscosity over entropy density ratio , in a heavy ion collision @xcite .
the @xmath0 is mainly caused by the almond shape of the nuclear overlap region , but higher - order flow , @xmath2 , @xmath3 , @xmath4 , especially their odd components are originating from the initial spatial fluctuations of participant nucleons .
recently atlas experiment has presented results of event - by - event @xmath1 @xcite , where @xmath1 values show large variations even in the fixed centrality bin due to initial fluctuations .
to control such an initial fluctuation , the event shape engineering was suggested @xcite .
this could be a useful tool to study the response of initial state to the system evolution and connect the initial and final states . in these proceedings , we present results on hbt measurements using charged pions and applying the event shape engineering technique for au+au collisions at @xmath5=200 gev recorded with the phenix experiment .
the event shape engineering focuses on the fluctuations in the transverse plane , but the presence of fluctuations in longitudinal direction could cause a twisted source along that direction @xcite .
the number of participants going to the forward and backward directions is not necessarily the same , and also participant eccentricities and participant planes might be different at both angles , which leads to different event plane angles between forward and backward rapidities .
thus the initial twist may survive as a twisted flow in the final state @xcite . in these proceedings
, we examine the possibility of a spatially twisted source in the final state using hbt interferometry in ampt model .
the event shape selection was performed by selecting the magnitude of @xmath6-order flow vectors , @xmath7 , which were measured by the reaction plane detector ( rxn , @xmath8 ) .
the @xmath7 is defined as @xmath9 , where @xmath10 , @xmath11 , and @xmath12 reflects the multiplicity within the element @xmath13 of the rxn .
figure [ fig1](left ) shows @xmath7 distributions for two centrality bins , where the higher 20% and the lower 30% @xmath7 events are shown filled .
we have tested the effect of the @xmath7 selection on charged hadron @xmath0 as shown in fig .
[ fig1](right ) , where the @xmath0 is measured at mid - rapidity ( @xmath14 ) and the systematic uncertainty from different event planes of south , north , and both combined rxn is included .
it is confirmed that the @xmath7 selection enhances or decreases the strength of @xmath0 .
( left ) @xmath7 distributions measured with the rxn in au+au 200 gev collisions , indicating 20%(30% ) higher(lower ) @xmath7 events .
( right ) charged hadron @xmath0 as a function of @xmath15 with and without event shape selection . ]
then we have applied this technique to the hbt measurement using charged pion pairs .
charged pions were identified by the electromagnetic calorimeter ( emcal , @xmath14 ) and the pair selection cuts at the drift chamber and emcal were applied to remove the effects of mis - reconstructed tracks and detector inefficiency . for the hbt analysis ,
pion pairs were analyzed with the out - side - long parameterization @xcite in the longitudinally co - moving system .
the effect of the event plane resolution was also corrected for both cases with and without @xmath7 selection .
figure [ fig2](left ) shows the extracted pion hbt radii , @xmath16 and @xmath17 , as a function of azimuthal pair angle @xmath18 relative to the second - order event plane @xmath19 .
results show that the higher @xmath7 selection increases the oscillation strength compared to the case without @xmath7 selection .
these oscillations of hbt radii are supposed to be sensitive to the final source eccentricity at freeze - out , @xmath20 .
blast - wave studies suggest that the quantity of @xmath21 would be a good probe to @xmath20 in the limit of @xmath22 , where @xmath23 denotes a mean pair transverse momentum .
the oscillation amplitudes of @xmath24 and @xmath25 in a form of the final eccentricity are plotted as a function of the number of participants calculated by glauber model in fig .
[ fig2](right ) .
the higher @xmath7 selection enhances the measured @xmath20 as well as @xmath0 .
it could be originating from a larger initial eccentricity , although there should be a contribution from the selected flow itself because the radii modulations also depend on the anisotropy in the momentum space @xcite .
( left and center ) azimuthal angle dependence of @xmath26 and @xmath27 relative to @xmath19 .
( right ) 2@xmath28-order azimuthal oscillation on @xmath29 and @xmath30 . in both panels ,
results with and without higher @xmath7 selection are shown . ]
to study the twisted source in the final state , we used the data of pb+pb 2.76 tev collisions simulated using ampt model ( v2.25 with string melting ) , where the impact parameter was fixed to @xmath31 fm . for the hbt study ,
the interference effect between two identical particles , @xmath32 , was calculated and weighted to the relative pair momentum distributions .
then the correlation functions were reconstructed by taking a ratio of the distributions with and without the weight .
also , all charged pions were allowed to make a pair with each other including @xmath33 to increase the statistics ( the consistency between results for positive and negative pairs was checked ) .
the event plane was determined using particles in @xmath34 , where particles were divided into two sub - groups .
a set of forward and backward event planes ( @xmath35 ) were used for the event cut which requires finite difference between @xmath36 and @xmath37 as a event twist selection , whereas the other set of events was used for a reference angle of azimuthal hbt measurement .
figure [ fig3](left ) shows @xmath24 , @xmath25 , and @xmath38 as a function of azimuthal pair angle relative to the @xmath37 , @xmath39 , for four @xmath40 regions , where @xmath41 was required .
the oscillations of three hbt radii measured in positive @xmath40 regions are shifted to negative direction , which is the direction of @xmath36 in the current event cut .
this phase shift can be understood to be a possible twist effect in the final source distribution .
azimuthal angle dependence of @xmath26 , @xmath27 , and @xmath42 relative to the backward @xmath19 with the event cut of @xmath41 , where the dashed lines show @xmath43.,scaledwidth=99.0% ] these oscillations were fitted with the following functions : @xmath44 to extract the magnitude of the phase shift , which is taken into account with @xmath45 .
the phase shift parameter @xmath45 obtained from the results with respect to @xmath36 and @xmath37 is plotted as a function of @xmath40 in fig .
the @xmath45 increases with going from backward to forward angle in all cases .
the variation of @xmath45 in the @xmath40 dependence is comparable to the difference between results relative to @xmath36 and @xmath37 at the same @xmath40 .
these results indicate that the source in the final state is also twisted due to longitudinal fluctuations in the initial state , as well as the twisted event plane and flow as discussed in ref
. @xcite .
phase shift parameters @xmath45 obtained from the @xmath24 , @xmath25 , and @xmath46 as a function of @xmath40 .
results measured with respect to forward and backward event planes ( @xmath36 , @xmath37 ) are shown , with event cut of @xmath41.,scaledwidth=99.0% ]
we presented the results of hbt measurements using event shape engineering for collisions recorded with the phenix experiment .
we found that the higher @xmath7 selection enhances the measured final source eccentricity as well as @xmath0 .
although the model comparison is needed to disentangle both spatial and dynamical effects on the hbt radii , this study clarifies the relation between initial and final eccentricity and constrains better the system dynamics .
we have also studied the event twist effect using the ampt model . when selecting events with finite difference between forward and backward event plane angles , the oscillations of hbt radii are shifted in the phase and the phase shift increases with @xmath40 .
the results indicate a possible twisted source in the final state preserving the initial twist due to the longitudinal fluctuations . this effect could be measured in experiments at rhic and the lhc
. both techniques could be useful to probe and control initial fluctuations in transverse plane and longitudinal directions , as well as to study the response of the system to the space - time evolution .
9 a. adare _ et al . _
( phenix collaboration ) , phys .
rev . lett . * 107 * , 252301 ( 2011 ) . c. gale , s. jeon , b. schenke , p. tribedy , and r. venugopalan , phys .
lett . * 100 * , 012302 ( 2013 ) .
et al . _ ( atlas collaboration ) , jhep * 11 * ( 2013 ) 183 . j. schukraft , a. timmins , and s. a. voloshin , phys .
b * 719 * , 394 ( 2013 ) .
p. boek , w. broniowski , and j. moreira , phys .
c * 83 * , 034911 ( 2011 ) .
f. retire and m. a. lisa , phys .
c * 70 * , 044907 ( 2004 ) . c. j. plumberg , c. shen , and u. heinz , phys .
c * 88 * , 044914 ( 2013 ) .
j. jia and p. huo , phys .
c * 90 * , 034915 ( 2014 ) .
j. jia and p. huo , phys .
c * 90 * , 034905 ( 2014 ) .
s. pratt , phys .
d * 33 * , 72 ( 1986 ) .
g. bertsch , m. gong , and m. tohyama , phys .
c * 37 * , 1896 ( 1988 ) . | azimuthal angle dependence of the pion source radii has been measured applying the event shape engineering technique at the phenix experiment . when events with higher magnitude of second - order flow vector are selected , the oscillation of the source radii is enhanced as well as @xmath0 which leads to the enhancement of the measured final source eccentricity .
the event twist effect in the spatial source distribution in the final state has been also explored with ampt model .
results indicate a possible twisted source due to the initial longitudinal fluctuations . |
due to the large electric dipole moments , there are strong and long - range interactions between rydberg atoms and these interactions can be controlled and enhanced by external electromagnetic fields .
besides , the huge polarizability of rydberg states gives rise to giant kerr coefficients @xcite , allowing nonlinear optical effects for only a few photons @xcite .
that is why the atoms excited to the high - lying rydberg states are of great interest in the recent researches of quantum many - body physics @xcite , quantum information processing @xcite and quantum nonlinear optics @xcite . the high - lying rydberg states can be a bit more easily excited from atomic ground states by laser light via absorption of more than one photon @xcite , while the spontaneous emission of the intermediary states will introduce the dissipative mechanism into the rydberg excitation schemes in spite of the long life of the rydberg states @xcite .
in general , these environment - induced dissipations inevitably lead to decoherence and noise in the quantum system and thus are undesirable in the above quantum physics researches . in recent years , however , the dissipation has been altered its role into a useful resource and tool for a lot of quantum applications , such as dissipative state engineering in trapped ions @xcite and cold atoms @xcite , dissipative quantum computation @xcite , dissipative quantum optics @xcite , bound state formation in molecules @xcite and entangled steady state production @xcite .
the key for these applications is the occurrence of `` non - equilibrium stationary state '' which can be achieved when the driving and dissipative processes arrive at a dynamical equilibrium @xcite . by appropriately arranging the system - environment couplings
, these steady states can deviate far from the classical thermal equilibrium states and keep unique quantum features resisting decoherences @xcite .
moreover , combining with the strong dipole - dipole interaction and blockade mechanics of the rydberg atoms , the dissipation could give rise to quite a few exotic phases of ultracold many - body atomic system , such as uniform phase , antiferromagetic phase , oscillatory phase and even the bistability between these phases @xcite .
in this work we investigate the effect of dissipation on the steady states and dynamics of multi - photon rydberg excitation schemes . to focus on the dissipation induced by the spontaneous emission of the intermediary states , we choose two typical four - level schemes : one level structure is @xmath0-type and the other is cascade . by appropriately arranging the laser fields , these two schemes have exactly the same coherent dynamics .
however , if including the effect of dissipation their realistic dynamics are of great difference , since a dissipative channel due to sponaneous decay running in the opposite direction in the two schemes .
we compare their non - equilibrium steady states by solving the stationary master equations under the mean - field treatment , and find that the number and distribution of their steady state solutions are entirely different when the rydberg nonlinear interaction is strong .
the bistable steady states are present in the the @xmath0-type scheme while the autler - townes splitting ( ats ) occurs in the cascade scheme resulting in the tri - stable steady states . as a consequence , for a strong nonlinear system such as rydberg atom systems
, dissipative force should be considered very carefully since the final steady states of the system will be very sensitive to it . furthermore , we study the dynamical evolution of the two schemes in the bistable region . due to
the difference in the dissipative channel , we observe two schemes may evolute into different branches under the same initial conditions , which makes sense for dissipative preparation of quantum state in strong nonlinear interacting systems .
-type and cascade atomic multi - level systems .
states @xmath1 , @xmath2 , @xmath3 and @xmath4 respectively show the ground state , two intermediate states and rydberg state , which are successively coupled by laser rabi frequencies @xmath5 , @xmath6 and @xmath7 . the three - photon laser detuning @xmath8 can be controlled in experiment .
other parameters are described in the text . ] in what follows , we will discuss two different multi - step cw - excitation schemes for the rydberg state , as typically used in experiments .
one is an @xmath0-type scheme with a raman transition before a rydberg - excited transition , as shown in fig .
[ model_scheme](a ) .
the atom is first excited from its ground state @xmath9 to an intermediate state @xmath10 with a transition strength given by the rabi frequency @xmath11 , and then transits to a metastable state @xmath12 with rabi frequency @xmath13 .
finally , another laser drives the transition between the metastable level and the desired rydberg state @xmath14 with rabi frequency @xmath15 .
the photon for the raman transition is typically provided by the mot lasers , which are tuned on resonance with the levels during the time of rydberg excitation .
so the spontaneous emissions from the intermediate state @xmath10 to state @xmath9 with rate @xmath16 and to state @xmath12 with rate @xmath17 , respectively , have to be taken into account . on the other hand ,
we allow for a detuning @xmath18 from resonant rydberg excitation and the lifetime of the rydberg state is much longer ( typically as large as tens of @xmath19s with the principle quantum number @xmath20 ) so that its decay can be safely neglected in our discussion . the second scheme shown in fig .
[ model_scheme](b ) has a cascade level structure , where @xmath12 represents the higher intermediate state , so that @xmath17 is the decay rate from state @xmath12 to state @xmath10 , opposite to the case in the first scheme .
the physical meanings of the other quantities remain unchanged .
besides , we note that in this scheme decay from state @xmath3 to state @xmath1 is physically forbidden according to the transition selection rule . except the atom - light interaction
, we also consider the strong interaction between rydberg atoms ( here we only consider a homogeneous van - der - waals interaction ) , which under the mean field approximation will cause a population - dependent energy shift to the rydberg state
. mean field theory is a classical approximation to the quantum model in which quantum correlations between atoms are ignored .
note that it works well only with large atom number @xcite . for systems of several rydberg atoms , the quantum effect of the strong rydberg blockade can no longer be effectively described by the mean field approximation @xcite . in our consideration , we assume that the number of atom is large enough so that the mean field approach can be safely used . in frames rotating at appropriate frequencies respectively , these two different rydberg - excitation schemes can be described by a same mean - filed hamiltonian ( @xmath21 ) , @xmath22 with @xmath23 being the corresponding transition operator between two internal atomic states , @xmath24 the rydberg interaction strength , and @xmath25 the rydberg state population .
the only difference is the expression of the three - photon detuning , given by @xmath26 with frequencies @xmath27 , @xmath28 , @xmath29 , @xmath30 denoting the energy of state @xmath31 , the carrier frequencies of the fields @xmath5 , @xmath6 , @xmath7 , respectively .
the upper sign is for the @xmath0-type scheme and the lower one for the cascade scheme .
that means the coherent dynamics of these two schemes can be same even they have different structures of energy levels , provided the three - photon detunings are equal .
however , we will show below in a realistic experiment where the dissipation caused by the spontaneous decay should be considered , these schemes will display distinct non - equilibrium steady states and dynamics , which is attributed to the opposite direction of decay route between level @xmath3 and level @xmath2 in two schemes . in order to investigate the influences of the spontaneous emissions from the intermediate excited states , we use the lindblad master equation @xcite @xmath32 + \mathcal{l}\left [ \hat{\rho}\right ] ,
\label{master_eq}\]]where @xmath33 is the single - atom density matrix operator which can well capture the evolution of the rydberg atoms according to the mean field approximation , and @xmath34 $ ] is the lindblad operator which is introduced phenomenologically to depict the atomic spontaneous emissions due to optical excitations @xcite:@xmath35
= \frac{\gamma } { 2}\left ( 2\sigma _ { ge}\hat{\rho}\sigma _ { ge}^{\dagger } -\left\ { \sigma _ { ee},\hat{\rho}\right\ }
\right ) + \frac{\gamma } { 2}\left ( 2d\hat{\rho}d^{\dagger } -\left\ { d^{\dagger } d,\hat{\rho}\right\ } \right)\]]where @xmath36 , and @xmath37 for the @xmath0-type system and @xmath38 for the cascade system . from
that we can obtain two different groups of differential equations of the density matrix elements @xmath39 .
for the @xmath0-type system we have , @xmath40 while for the cascade system , @xmath41 where we have assumed that the off - diagonal coherence elements @xmath42 and the conservation of the total population that the diagonal elements @xmath43 .
the effective detuning @xmath44 , including the population - dependent frequency shift of the rydberg state , which makes the differential equations nonlinear .
we can find these two groups of equations become identical when the decay @xmath45 , which means the dynamics of the two schemes are indistinguishable .
instead , a non - vanishing @xmath17 will lead to entirely different excitation dynamics , especially in the regions with high nonlinearity induced by rydberg - rydberg interaction . to show these differences in the following we will keep the common decay @xmath16 constant and focus on the influence of the decay @xmath17 which has opposite direction in the two schemes .
then it is also convenient to renormalize the time by @xmath46 and all frequencies by @xmath16 in the equations .
we first investigate the steady states of the schemes , where the competition between the oscillation caused by coherent optical driving and the dissipation due to the spontaneous emissions reaches a dynamical equilibrium so that the populations and the coherence for each level become stationary .
the steady occupancy of the rydberg state , @xmath47 , can then be readily obtained by setting all time differential terms in eq .
( [ ntypef ] ) and ( [ cascadetyf ] ) equal zero and then solving the resulting algebraic equations . for the @xmath0-type scheme
, we obtain a cubic equation about @xmath47 which may have one or three solutions corresponding to the uniform and the bistable phase of the rydberg excitation scheme , respectively ( we should check the stabilities of these solutions to ensure they are achievable in the dissipation environment ) . here the bistable phase , as a typical nonlinear effect , is caused by the rydberg - rydberg interaction .
it has been observed in recent experiment of a dilute rydberg atomic ensemble @xcite . on the other side ,
the equation becomes quintic in the case of the cascade scheme , indicating the existence of multi - stable phase of the scheme , which hasnt been found in the rydberg system .
the analytical forms of these equations and their solutions are too cumbersome to present here , we instead show the phase diagram of the rydberg atoms in the parameter space of @xmath7 and @xmath8 , which is decided by the number of the stable solutions . as shown in fig .
[ phase_diagram ] , the left column is for the @xmath0-type scheme and the right column for the cascade scheme , with the regions labeled by numbers `` 0''-3 " respectively corresponding to the oscillatory phase ( none stable solution ) , the uniform phase ( only one stable solution ) , bistable phase ( two stable solutions ) and tri - stable phase ( three stable solutions ) . space , indicating the influence of decay and nonlinearity on the phase of the rydberg atoms , with the numbers of the stable solutions marked in different gray - level . the left column is for the @xmath0-type system and the right column is for the cascade one . from top to bottom
, we fix the rabi frequencies @xmath48 , @xmath49 , and change the strength of decay and rydberg interaction : ( a ) and ( d ) @xmath50 and @xmath51 ; ( b ) and ( e ) @xmath50 and @xmath52 ; ( c ) and ( f ) @xmath53 and @xmath52 .
( all parameters are scaled by @xmath16.),width=316,height=336 ] from the figure we can find the two schemes have similar phase diagrams at small decay strength ( see the first row , @xmath54 and @xmath55 ) .
the parameter space is dominated by the uniform phase , the bistable phase presents for small rabi frequency @xmath15 , and the oscillatory phase only survives in a very narrow region where detuning @xmath18 is small . for a large value of @xmath17 ( the second row , @xmath56 and @xmath55 ) ,
the phase diagrams become distinct from each other . in the case of the @xmath0-type scheme
the oscillatory phase occupies a large area with larger value of @xmath15 , that is because the non - lossy states , @xmath3 and @xmath31 , sustain a rabi - type population oscillation between them .
instead , in the case of the cascade scheme the decay of state @xmath3 prevents such oscillation .
moreover , ats of the cascade system causes a unique tri - stable phase ( we will show it clearly below ) which is absent in the former case @xcite .
we emphasize that the effect of dissipation - sensitive steady states is more remarkable under the condition of strong nonlinearity .
when we decrease the rydberg - rydberg interaction strength , the phase diagrams of the two schemes trends to be analogous even for large value of decay ( the third row , @xmath56 and @xmath57 ) .
( first row ) , state @xmath58 ( second row ) , state @xmath59 ( third row ) , and @xmath60 ( last row ) are plotted as a function of three - photon detuning @xmath8 , with the weak rydberg - state coupling case ( @xmath61 ) in thick black curves and the strong coupling case ( @xmath62 ) in thin blue curves .
the dashed curves corresponds to the values in the unstable phases .
the left and right columns are for the @xmath0-type system and the cascade system , respectively .
other parameters are same as fig .
[ phase_diagram](b ) and ( e ) ] before focusing on the different quantum multi - stable phases of the two schemes in the parameter space of strong nonlinearity and large decay , we first investigate their steady states without nonlinearity , which will show the cause of these differences are the direction of the decay @xmath17 . by solving the stationary obes in the case of @xmath63 and other parameters same as fig .
[ phase_diagram](b ) and ( e ) , we observe that the rydberg excitation spectrum in the cascade scheme shows a double - peak structure separated by @xmath64 which is same as the feature of the ats presenting in the three - level rydberg - excitation scheme @xcite . in our @xmath0-type scheme ,
the same direction of decay @xmath17 and @xmath16 makes the atoms populate mostly on state @xmath65 and @xmath66 , resulting in the splitting of the rydberg excitation spectrum .
moreover , different from in the three - level scheme where ats is just an effect of transient stability @xcite , the existence of the intermediate state @xmath67 here has stabilized this splitting and enabled it to appear in the final steady state of the scheme . in the case of the @xmath0-type scheme
, the opposite directions of decays @xmath17 and @xmath16 , as well as small rabi frequency @xmath13 , may accumulate the population on state @xmath3 only , giving rise to a single peak instead of the ats in the rydberg excitation spectrum .
these different peak structures still remain in the presence of rydberg - rydberg interactions . as we shown in fig .
[ rydberg_popu](a)-(h ) , the nonlinearity caused by the interaction distorts the original spectrum , resulting in the different multi - stable steady states of the cascade and the @xmath0-type schemes .
again the right column of fig .
[ rydberg_popu ] is for the @xmath0-type scheme , where we display the exact values of the steady population of each states at small @xmath7 case ( thick black line ) and large @xmath7 case ( thin blue line ) , respectively , corresponding to the two bistable regions marked in phase diagram fig .
[ phase_diagram](b ) .
we can find different features of the bistable behaviors in these two cases .
when @xmath68 is small it exhibits intrinsic bistability in the region of @xmath69 with its inclined direction of hysteresis window depending on the sign of rydberg interaction @xmath24 .
when @xmath8 is negative or large positive ( i.e. @xmath70 is off - resonance ) , the ground states @xmath3 is occupied with a dominant number of atoms ( @xmath71 ) .
however , as increasing @xmath7 to @xmath72 we find that the dominant population at @xmath73 occupies the rydberg state @xmath31 with its value @xmath74 due to a quite strong rydberg excitation . meanwhile , a small region of bistable phase emerges near @xmath75 where the rydberg interaction induced energy shift has been compensated by a large detuning i.e. @xmath76 . besides ,
when @xmath15 is large , there is a large unstable region ( corresponding to the oscillatory phase ) denoted by the dashed lines , which is due to the strong coupling between the non - lossy states @xmath3 and @xmath31 . turning to the case of the cascade system ( the right column of fig .
[ rydberg_popu ] ) , a dramatic difference is the presence of the inclined double - peak configuration at small @xmath15 case , which gives rise to bistable and even tri - stable phases when @xmath8 and @xmath77 are comparable . in addition , state @xmath1 and state @xmath2 share the most atomic population ( @xmath78 ) and the case does not change much even for larger value of @xmath7 .
however , the double - peak structure degenerates a lot so that there is no multi - stable phase in large @xmath15 case .
this is due to the comparable values of @xmath11 and @xmath15 prevent the occurrence of the ats .
summarizing , for the @xmath0-type scheme , decay @xmath17 leads to a large population on state @xmath3 which , combining with rydberg state @xmath31 , makes the scheme work like a two - level system , so that the sustained rabi - typed oscillation occurs with an intensive @xmath7 . for small @xmath7
the behavior of the scheme is more close to the cooperative optical excitation @xcite . for the cascade scheme
, the existence of @xmath79 leads to the ats phenomenon , inducing multi - stable steady state phases under the condition of strong nonlinear interaction .
the above phase diagrams reveal the properties of the steady states in the parameter space of detuning @xmath18 and rabi frequency @xmath15 , which are controlled by the external lasers . specially , owing to the strong nonlinear interactions the bistable phases present , which suggests there are two different steady states , i.e. , the low- and the high - rydberg - occupied steady state , under the same laser parameters . in reality , the system would eventually evolute into one of them , which depends on the initial states and decay routes . in order to study this dependence , we perform a numerical simulation for the population dynamics by directly solving the obes ( [ ntypef ] ) and ( [ cascadetyf ] ) . ) for ( a ) the @xmath0-type system and ( b ) the cascade system , as a function of initial population on @xmath80 state ( @xmath81 ) and @xmath82 state ( @xmath83 ) .
dark and light colors represent the low and high rydberg - state occupancy , respectively .
the colorless regions are unstable and without stationary solutions .
( c ) and ( d ) show population evolution in the bistable region for the @xmath0-type system and the cascade system , respectively , where @xmath84 , and @xmath85 ( dashed blue line ) and @xmath86 ( solid black line ) .
see text for other parameters . ] in the calculation , we choose @xmath87 , @xmath88 , other parameters are same as in fig . [ phase_diagram](b ) and ( e ) where the bistable steady state gives two stationary rydberg populations @xmath89 ( low ) and @xmath90 ( high ) for the @xmath0-type scheme , and @xmath91 ( low ) and @xmath92 ( high ) for the cascade scheme , respectively .
since the rydberg state @xmath31 is not subjected to the dissipation in our model , we fix its initial population by @xmath93 for the @xmath0-type scheme and @xmath94 for the cascade scheme , and then focus on the dependence of its final population on the initial population of state @xmath1 and @xmath3 as depicted in fig .
[ dynamics](a ) and ( b ) .
note that the total particle is always conserved , @xmath95 . from fig .
[ dynamics ] we can find difference between the two schemes in the side of small @xmath96 , which means state @xmath2 and state @xmath97 are largely occupied at the initial time .
for the cascade system whose @xmath79 is large and @xmath96 is small , it is hard to pump atoms into the highly excited state @xmath31 .
however , in the @xmath0-type system dissipations from @xmath2 may create strong atomic correlations between @xmath1 , @xmath98 and @xmath3 ( e.g. a raman scattering process @xcite ) , giving rise to an enhancement for population in state @xmath3 .
it is therefore possible to realize a high rydberg excitation .
the final populations @xmath99 for the @xmath0-type scheme and @xmath100 for the cascade scheme . on the other side when @xmath96 is large , since state @xmath3
is directly coupled with state @xmath31 by laser @xmath7 , the rydberg excitation is easy .
the final populations @xmath101 for the @xmath0-type scheme , where the population in @xmath3 and @xmath31 are dominant .
for the cascade scheme , due to the effect of decay @xmath79 and @xmath102 the situation is different .
we have @xmath103 , where the populations of state @xmath1 and @xmath98 are not small . except for the above two sides , in the middle region , both of the schemes tend to a low rydberg occupancy with @xmath104 for the @xmath0-type scheme , corresponding to a dominant final occupancy of state @xmath3 , and with @xmath105 for the cascade scheme , indicating a population sharing between states @xmath1 and @xmath98
. finally , we emphasis that even the final values of the rydberg population are close , the dynamical evolution process and therefore the time spent on achieving the final stationary states are much different for the two schemes . in fig .
[ dynamics](c ) and ( d ) , we show their different population dynamics for the rydberg state in the bistable region under the same experimental parameters .
in contrast to the case of the cascade scheme , the @xmath0-type scheme is subjected to a strong population oscillation while tending towards the high branch of rydberg occupancy and it requires more time to reach steady state .
this result agrees with our former analysis of bistability as in fig .
[ phase_diagram](b ) that two non - lossy states @xmath3 and @xmath31 may lead to unstable oscillations .
before the conclusion , we discuss the experimental realization of our proposed schemes .
the three - photon cascade scheme can not only offer an easier route for the rydberg excitation but also limit the excited atoms in a narrow velocity distribution , so that it has been adopted in recent experiments to explore the nonequilibrium phase transition in a thermal rydberg atomic gas @xcite .
especially , in this experiment the existence of intrinsic optical bistable effect has been identified as we analyzed in the above theory model . however , according to our theory model we find in this scheme the tri - stable phase only exists under quite rigorous parameters as implied in fig .
[ phase_diagram](e ) , making it a challenge to verify in experiments . as for the @xmath0-type scheme ,
it is realizable by current experimental technology and has been recently advanced in several theoretical proposals to realize rydberg quantum simulator @xcite and prepare rydberg - dressed atoms @xcite .
in an open quantum system where driving and dissipative processes compete with each other and settle the system on a nonequilibrium steady state , a number of novel phenomena absent in the equilibrium system will appear @xcite . in this work
, we have investigated the effect of dissipation on the steady state properties and dynamics for the two different excitation schemes of strong interacting rydberg atoms , one is of the @xmath0-type level structure and the other of the cascade structure .
we study the phase diagram of these schemes and find bistable and tri - stable phases in the region of strong nonlinearity caused by the rydberg - rydberg interaction .
we investigate the population dynamics of the rydberg state in bistable region and display its dependence of the initial population distributions . by comparing the steady state and the dynamics of these two schemes
, we show their sensitive dependence on the path and the amount of the dissipation caused by the spontaneous emission @xcite .
future work will focus on the quantum state preparation @xcite and multi - stable switching @xcite in the rydberg atomic schemes by controlling and arranging the dissipation .
j. q. thanks p. meystre and k. zhang for useful discussions .
this work was supported by the national basic research program of china ( 973 program ) under grant no .
2011cb921604 , the nsfc under grants no .
11104076 and no . 11234003 , the specialized research fund for the doctoral program of higher education no
. 20110076120004 . | we theoretically investigate the effect of dissipation on multi - photon excitation of rydberg atoms . the steady states and the dynamics
are compared via two types of four - level excitation schemes with different dissipative paths of spontaneous emission .
we find that in the case of strong rydberg - rydberg interaction , the schemes will settle in several different non - equilibrium steady states .
the interesting aspect is that there exists the multi - stable steady states , which reveals the competition between interaction - induced nonlinearity and dissipation caused by spontaneous emission .
a numerical simulation on the rydberg population dynamics in the bistable region exhibits different features existing in the two schemes even with the same initial conditions , which accounts for the influence of the dissipation on the dynamics . |
intercalated compounds built by insertion of transition metal ( tm ) atoms into layered titanium dichalcogenides are interesting as natural nanostructures formed by alternation of magnetic tm layers and non - magnetic layers of the host lattice .
the investigations performed for @xmath8tis@xmath1 ( @xmath4 = mn , fe , co , ni ) @xcite show that the effect of intercalation on the chemical bonding is not consistent with the rigid band model @xcite often referred to in relation to these compounds .
instead , the charge carriers localization due to the formation of covalent bonds with intercalant atoms seems to become important . from this point of view , the tise@xmath1-based intercalated materials are obviously even more interesting , because they demonstrate a higher degree of two - dimensionality in comparison with tis@xmath1 , as follows from their larger @xmath9 ratio ( 1.697 in pure tise@xmath1 vs. 1.672 in tis@xmath1 , see , e.g. , ref . ) .
@xcite discussed magnetic properties of @xmath8tise@xmath1 ( @xmath4 = v , cr , mn , fe , co , ni , cu ) referring to the hybridization between the intercalant @xmath3 states and the band states of tise@xmath1 , the information on the latter being based on the band structure calculation @xcite . according to ref .
@xcite , the hybridized states with lighter tm atoms have smaller contribution of se@xmath10 states .
the aim of the present work is to deepen the understanding of the chemical bonding and related ( magnetic and transport ) properties of two tise@xmath1-based systems , intercalated with `` light '' ( cr ) and `` heavy '' ( co ) tm atoms .
monocrystalline samples over a range of intercalant concentration were prepared and used in the study of magnetic susceptibility , conductivity , x - ray emission and photoelectron spectroscopy . while for the system co@xmath0tise@xmath1 the literature data about influence of intercalation on crystal structure @xcite , electric and magnetic properties @xcite were reported earlier , for cr@xmath0tise@xmath1 system we failed to find any literature data except those of ref . @xcite
polycrystalline co@xmath0tise@xmath1 and cr@xmath0tise@xmath1 samples ( @xmath11 ) were prepared by an ampoule synthesis method from the constituent elements .
the methods of synthesis and characterization are described in detail in ref .
the initial diffraction data were obtained with a dron 1um diffractometer ( cu @xmath12 , ni filtered radiation . )
the unit cell parameters have been then determined by a least square refinement with 16 reflexes with the accuracy @xmath13 , @xmath14 .
the hexagonal unit cell parameters of intercalated tise@xmath1 are shown in fig .
[ fig : a+c ] . the crystal structure ,
electric and magnetic properties of co@xmath0tise@xmath1 are described in refs .
the information about the type of intercalant ordering in co@xmath0tise@xmath1 is controversial @xcite , which may be traced back to different thermal treatments of the samples . in the present work we used the samples quenched from 850 @xmath15c in ice water . for the samples
so prepared , no ordering of intercalant atoms was observed in both systems . at temperatures above that of liquid nitrogen ,
both materials are paramagnetic , with a temperature dependence of their magnetic susceptibility following a curie - weiss law .
the values of effective magnetic moments are close to nominal spin moments of co@xmath16 and cr@xmath17 ions , respectively .
the single crystals for the x - ray spectra measurements were grown by the gas transport reaction method in evacuated quartz ampoules with i@xmath1 as gas - carrier @xcite .
powdered co@xmath18tise@xmath1 and cr@xmath18tise@xmath1 material was used as origin materials . in the case of co@xmath18tise@xmath1
there was a flow of substance from the hot ( 1000 @xmath15c ) to the cold ( 600 @xmath15c ) edge of the ampoule .
the single crystals had a shape of thin plates @xmath19 2 - 3 mm large and @xmath19 0,1 mm thick .
the final intercalant content ( determined from the lattice parameter values ) was found to be close to the initial one : co@xmath20tise@xmath1 . in the case of cr@xmath18tise@xmath1
the substance was transferred from the middle part ( at 900 @xmath15c ) where the powder was initially placed to both hot ( 1100 @xmath15c ) and cold ( 700 @xmath15c ) edges of the ampoule .
the single crystals grown at the hot edge of the ampoule had a shape of slim needles with the size close to @xmath21 mm , whereas those grown at the cold edge were thin plates 24 mm large and @xmath190.1 mm thick .
from the unit cell parameters comparison we found that the crystals formed at the cold edge have the composition cr@xmath22tise@xmath1 , those at the hot edge cr@xmath23tise@xmath1 .
the x - ray photoelectron spectroscopy ( xps ) measurements were performed at a perkin elmer spectrometer with an energy resolution of 0.3 ev , x - ray emission spectroscopy ( xes ) measurements at a soft x - ray radiation spectrometer rsm-500 with electron excitation and an energy resolution of 0.7 ev .
the single crystals of co@xmath20tise@xmath1 and cr@xmath22tise@xmath1 were cut by a razor blade in the spectrometer chamber directly before the measurements .
in this section , we present the discussion of the electronic characteristics as they might be deduced from the analysis of the measured bulk properties .
the microscopic aspects of hybridization will be discussed in sec .
[ sec : calcul ] in relation with _ ab initio _ calculations
. a decrease of the lattice parameter @xmath24 in the course of intercalation is typical for @xmath8tise@xmath1 systems as has been mentioned in ref .
@xcite and could be seen in fig .
[ fig : a+c ] . this might be understood as a result of covalent bonds being formed between intercalant atoms and the host lattice @xcite . while free co@xmath16 and cr@xmath17 ions have the same spin value of 3/2 , in intercalated systems the
experimentally obtained effective magnetic moment at the cr atom is roughly 1.5 times larger than that per co atom ( see table [ tab : mom ] ) .
this fact may be related to a stronger hybridization of @xmath25 orbitals ( those participating in the magnetic moment formation in the @xmath26 configuration of co@xmath16 ) with the se@xmath10 orbitals .
in contrast , the @xmath25 orbitals are essentially empty in the @xmath27 configuration of cr@xmath17 .
the deviation of the effective moment from its nominal free - ion value correlates well with the lattice distortion ( which is the measure of the chemical bonding between an intercalant and the host @xcite ) at least for concentrations @xmath28 as is shown in fig .
[ fig : dmdc ] .
the systems with @xmath29 are actually beyond the percolation limit on the ordered lattice of @xmath3 `` impurities '' , so that their clustering and direct exchange interaction apparently results in the drop of magnetic moments and breakdown of the abovementioned correlation .
the lattice deformation and change in the degree of hybridization affects the conductivity @xmath30 of the @xmath8tise@xmath1 systems . while in co@xmath0tise@xmath1
@xmath30 is only weakly concentration - dependent , in cr@xmath0tise@xmath1 it exhibits an interesting behavior
being on the average higher , @xmath31 includes several irregularities .
[ fig : conduc ] shows the conductivity at room temperature , but the irregularities persist over a broad temperature range .
two different types of @xmath32 are shown in the inset of fig .
[ fig : conduc ] for two cr concentrations .
an overall decrease of @xmath30 as the unit cell parameter @xmath24 decreases ( see fig . [
fig : conduc ] ) is apparently due to the drop in the free charge carriers concentration as electrons are trapped in covalent bonds this seems to be well the case at least in all co - containing systems . due to a different charge configuration of cr@xmath17
as compared to co@xmath16 , one electron provided by cr remains free and contributes to the metallic type of conductivity see the calculated density of states ( dos ) distribution at the impurity site , sec .
[ sec : calcul ] .
a drop in the conductivity near @xmath33=0.05 of cr probably indicates the compensation of highly mobile intrinsic ( hole - type ) carriers in tise@xmath1 , the concentration of which is 0.05 per formula unit@xcite .
this is consistent with a perfectly metallic type of temperature dependency of conductivity for this concentration , as shown in the inset of fig .
[ fig : conduc ] .
another drop in @xmath31 at @xmath33=0.25 may be related to the percolation limit on a lattice of those ti atoms which have no cr neighbors .
two ti atoms ( to both sides of the intercalated cr atom ) participate in the formation of localized bonds , and the percolation limit for two - dimensional hexagonal lattices equals 0.5 @xcite . for @xmath33=0.33 , differently from lower cr concentrations , the activation region is seen ( inset in fig .
[ fig : conduc ] ) in the @xmath32 , indicating an opening of the band gap at the fermi level .
it is noteworthy that at @xmath33=0.33 , exactly one extra electron is provided by cr per titanium atom .
one can speculate that the gap formation is the consequence of spin splitting of the cr - ti hybridized band , since this band could exactly accommodate one electron per ti atom .
no such gap is seemingly formed in co@xmath0tise@xmath1 , presumably because more efficient deformation of the host lattice results in more octahedral coordination of ti atoms by chalcogen , preventing the splitting between @xmath34 and @xmath35 , @xmath36 orbitals of ti @xcite .
we refer to the xps spectra with three main purposes : to control the quality of monocrystalline samples , to get information about the electronic structure and to trace the effects of on - site magnetic interactions on the spectra .
[ fig : ti2p ] and [ fig : xpscore ] depict the @xmath37 core levels of titanium and of intercalated elements , correspondingly .
ti @xmath37 spectra are known to exhibit relatively large chemical shifts already at early oxidation states ( see , e.g. , @xcite ) .
the actual spectra indicate a good ( oxygen - free ) quality of samples .
the chemical shift still present ( 455.0 ev for the ti@xmath5 binding energy in tise@xmath1 ; 455.5 ev in co@xmath23tise@xmath1 ; 455.8 ev in cr@xmath22tise@xmath1 ) exhibits the increase of ionicity from the host to co - based to cr - based systems @xcite .
this is consistent with above discussed ( fig .
[ fig : a+c ] ) stronger @xmath38-contraction of the crystal lattice in more covalent co@xmath0tise@xmath1 , as compared to cr@xmath0tise@xmath1 . of interest
are the core @xmath37 photoemission spectra of intercalant atoms , which may exhibit an exchange splitting due to an intra - atomic @xmath37@xmath3 interaction ( similarly to what has been demonstrated for the mn - based heusler alloys@xcite ) . as is seen in fig .
[ fig : xpscore ] , the cr@xmath5 line in cr@xmath22tise@xmath1 clearly shows a splitting of about 1 ev , differently from the co@xmath5 spectrum of co@xmath20tise@xmath1 where no such splitting is observed . the splitting is qualitatively related to the difference between local magnetic moments in these compounds ( see table [ tab : mom ] ) . the valence - band ( vb ) photoelectron spectra ( fig .
[ fig : xps+xes ] ) have a relatively simple structure , where the se@xmath39 contribution ( in the region of binding energies 1216 ev ) is clearly separated from the se@xmath10ti@xmath3@xmath4@xmath3 hybridized band spanning the binding energies 06 ev . in the latter band ,
the contribution from the se@xmath10 states is the most pronounced , since the photoionization cross - section of selenium is much larger than for all other constituents @xcite .
however , the contribution of cr@xmath3 and , in particular , co@xmath3 states is noticeable just below the fermi energy , similarly to the behavior detected for the fe@xmath3 states in fe@xmath0tite@xmath1 @xcite .
an analysis of the calculated dos ( see next section ) shows that a narrow peak at @xmath191.5 ev below the fermi level , well separated from the bulk of se@xmath10 states , indeed dominates the local dos at the co site and must be visible in the valence - band xps .
the trends in the `` local '' electronic structure of intercalant atoms depending on concentration are much better visible from x - ray emission spectra which are element - selective and hence not masked by the contributions from selenium states .
the cr and co @xmath40 spectra ( @xmath41 transitions ) have been measured for several concentrations and are shown in fig .
[ fig : xes ] . based on the arguments of ref .
, the concentration - related differences in the spectra can be traced to the percolation threshold on a two - dimensional hexagonal lattice at @xmath42 . above this threshold , the intercalant atoms ( under an uniform distribution )
may have a neighbor of the same type at the distance of hexagonal lattice constant .
it is noteworthy how the width and the relative intensity of @xmath43 and @xmath44 bands vary with the concentration .
the cr @xmath40 spectra of intercalated compounds are quite different from those of pure metallic cr , whereas the corresponding differences for the co - based systems are less pronounced .
a striking difference in the xes of co- and cr - intercalated systems is a much broader emission band in the latter , in spite of the fact that the widths of the occupied valence bands in both systems are presumably not too much different .
the shift of the x - ray emission spectra by the @xmath5 core level binding energy onto the common energy scale with the xps ( fig .
[ fig : xps+xes ] ) makes obvious that the states above the nominal fermi level effectively participate in the emission process in cr@xmath22tise@xmath1 , but not in co - intercalated systems .
another peculiarity of the spectra shown in fig .
[ fig : xes ] is the increase of the @xmath45 relative intensity in more diluted ( especially in cr - based ) systems .
the broadening of @xmath43 spectra of @xmath3 components has been detected and discussed before in @xmath3 oxides ( of mn @xcite and cu @xcite ) where charge - transfer excitations were argued to be an important mechanism responsible for a rich satellite structure , as well as in pure @xmath3 metals ( cu , @xcite ) where an initial - state @xmath3 vacancy satellite was shown to contribute on the high - energy side of the @xmath43 line for excitation energies beyond the @xmath46 threshold . for our systems in question , none of these mechanisms seem to work : charger - transfer excitations can not play an important role because of a much high covalency than in oxides ; as for shake - up / shake - off processes it is difficult to argue why their intensity would be enhanced at low cr ( but not co ) concentrations . moreover , the increase of the high - energy satellite in cu @xmath40 in ref .
@xcite is accompanied by the decrease of the @xmath45 intensity ratio ( due to the coster - kronig process ) the trend opposite to that observed in the present case .
what makes the difference between low - concentrartion cr - intercalated system and other systems studied is relatively high localization of the cr@xmath3 state and high magnetic moment associated with it .
it is noteworthy that in recently studied heusler alloys nimnsb and co@xmath1mnsb where the high magnetic moment at the mn site is similarly localized , the @xmath43 emission from the states nominally above the fermi level was observed @xcite . similarly to how it was argued in ref .
@xcite for heusler alloys , we tend to qualitatively attribute the anomalous x - ray emission in the diluted cr@xmath0tise@xmath1 system to a relaxation of the intermediate state with electronic configuration @xmath47 , which involves the contribution from the states of @xmath48 symmetry nominally vacant in the ground - state configuration @xmath49 .
below , we shall refer to this effect for brevity as re - emission .
strictly speaking , the excitation and subsequent emission in a resonant process must be treated as a single event ; an extensive literature exists on this subject ( see , e.g. , refs . ) .
the effect is the more pronounced the stronger the localization of involved states is .
the peculiarity of our present spectra is that such typically resonance - like phenomenon as re - emission is observed in spite of high - energy ( i.e. non - resonance ) electronic excitation , with the excitation energy of nearly 4 kev , apparently due to sufficiently high lifetime of an electron trapped in a highly localized vacant cr @xmath3 state . for the co - intercalated systems ,
the effect is practically not observable .
assuming phenomenologically the participation of the lowest vacant cr @xmath3 states in the x - ray emission , we leave the ( otherwise single - particle ) quantitative treatment of the spectrum till next section where the microscopic information about the electronic structure becomes available .
another important observation concerning the x - ray emission spectra is a substantial increase especially at low intercalant concentrations of the @xmath45 intensity ratio as compared to that in the pure metal . to our opinion , this indicates a localization of the @xmath3 electron density and hence a magnetic moment .
@xmath43 and @xmath44 bands originate from the dipole transitions @xmath50 and @xmath51 , correspondingly .
the change in the @xmath45 relative intensity may be influenced by the selection rule on the total moment projection , @xmath52 , which in the present case favors a substantial decrease of the statistical weight of the @xmath50 transition , since the exchange interaction energy of @xmath3 electrons ( 22.5 ev ) by more than an order of magnitude exceeds their spin - orbit interaction energy . as already mentioned above ,
the changes in the observed @xmath45 weights distribution for systems with lower cr content is opposite to what one could expect from the coster - kronig transitions , which redistribute intensities from @xmath53 to the @xmath5 channel .
this indicates a relative unrelevance of the coster - kronig effect for the systems in question .
in ref . we reported the results of a first - principles optimization of lattice parameters in pure tise@xmath1 ( as well as in tite@xmath1 ) and the electronic structure of @xmath3-intercalated systems , with the use of @xmath4(tise@xmath1)@xmath54 supercells in the latter case .
the calculation has been done with the full - potential linearized augmented plane wave code wien97@xcite , using the generalized gradient approximation ( gga ) in the formulation by perdew , burke and ernzerhof@xcite .
the hexagonal lattice parameters as found in the course of total energy minimization were @xmath55=3.519 , @xmath24=6.280 , whereas the experimental measurements give the values @xmath55=3.535 , @xmath24=6.004 ( as cited in ref .
the internal coordinate @xmath38 ( the relative distance between ti and se layers ) is 0.247 in the calculation , compared to the measured value of 0.25 . at a first sight ,
the gga overestimates the @xmath9 ratio by as much as 5% .
however , considering the reduced value of @xmath38 in the calculation , one can notice that in reality just the van der waals gap between adjacent chalcogen planes is overestimated , whereas the lenghts of ti - ti and ti - se bonds remain well within 1% of the experimental estimate .
the perdew burke ernzerhof formula for gradient approximation was optimally tuned for a certain degree of localization of the exchange - correlation hole , typical for bulk solids but not for the `` empty '' van der waals gap
. some discussion on this subject can be found , e.g. , in ref .
@xcite . in the present study
, we consider systems with relatively high concentration ( @xmath56 per tise@xmath1 unit ) of intercalant atoms in the van der waals gap , therefore the abovementioned shortcoming of the gga treatment must be less serious than for pure dichalcogenides . as in ref .
@xcite , we simulated intercalated systems using the rhombohedral @xmath4(tise@xmath1)@xmath54 supercell , shown in fig .
[ fig : scell ] ( in the hexagonal setting ) .
similar supercells were earlier used in non - magnetic calculations by suzuki _
et al._@xcite . with only one type of supercell being considered in _ ab initio _ calculations
, it had little sense to address the systematical trends in the lattice parameters under doping .
instead , we concentrated in our analysis on the properties of two intercalated systems as _ impurity _ systems .
the calculations have been performed therefore in a fixed ( @xmath24 , @xmath55 ) frame , with a local atomic relaxation around the @xmath4 atoms taken into account .
this allows us to make a more clear comparison of cr- and co - intercalated systems .
from the following it becomes obvious that the noticeably perturbed region around an intercalant atom is indeed much smaller than the region defined by the translation vectors in the supercell of our choice .
in the hexagonal setting , the supercell with the lattice constants @xmath57 , @xmath58 ( fig .
[ fig : scell ] ) hosts three @xmath4(tise@xmath1)@xmath54 units with the following atomic positions in one of them : @xmath4 at ( 0,0,0 ) ; ti at @xmath59 and at @xmath60 ; six se at @xmath61 ; @xmath62 ; @xmath63 .
the optimized values of internal coordinates @xmath64 to @xmath65 , obtained from the condition that the forces on all atoms disappear , are listed in table [ tab : relax ] .
the local structural distortion , as compared to the equilibrium structure of tise@xmath1 , is quite small .
the negative value of @xmath65 corresponds to the displacement of the chalcogen planes towards the ti layers by 0.42% of the @xmath24 parameter of tise@xmath1 ; this corresponds roughly to @xmath66 rather than 0.25 found for pure tise@xmath1 .
a qualitative difference between cr@xmath67tise@xmath1 and co@xmath67tise@xmath1 is that the nearest ti atoms approach the co atom ( @xmath64 slightly negative ) but are shifted away from the cr atom .
this is consistent with the evidence of larger co ti bonding and , contrarily , higher localization of the cr@xmath3 states as addressed below . the spatial distribution of the charge density in a plane cut across the supercell is shown in fig .
[ fig : dens_line ] .
the vertical size of the plot covers 18.84 .
one can see that in the absence of an @xmath4 atom the charge density between the se layers is indeed rather low , about 0.05 @xmath68/@xmath69 .
an intercalant atom embedded in this region mediates the chemical bonding between chalcogen atoms .
an effectively larger ( see below ) co atom provides a slightly higher charge density along the se@xmath4 line .
moreover , the distrubution of the charge density along [ 001 ] ( i.e. , towards the nearest ti atoms ) is somehow different in cr- and co - intercalated crystals ( fig .
[ fig : dens_cont ] ) . on the co site ,
the occupation of @xmath3 states is substantially higher as compared to cr , with the effect that the effective size of an intercalant atom is larger , in spite of a slightly higher contraction of the co@xmath3 states . at the same time , the occupation of the ti@xmath3 shell is lower in the vicinity of co next to a cr atom . the local dos distribution on the atoms in question ( fig .
[ fig : dos ] ) helps to understand the origins and energetics of this different charge distribution . the center of gravity of the cr@xmath3 states lies closer to that of ti@xmath3 . in spite of large exchange splitting at the cr site ,
a noticeable contribution from both atoms is present near the fermi level .
specifically , mostly the @xmath70 orbitals at both cr and ti sites are responsible for this direct hybridization and the enhancement of the ti dos ( of the atoms neighboring to cr in the [ 001 ] direction ) near the fermi energy .
this factor is not important in the co - intercalated system because the co@xmath3 states are lower in energy and do not mix so efficiently with ti@xmath3 in fact , the majority - spin band at the co site is fully occupied .
apart from this enhancement of the direct cr
ti bonding , the distribution of electronic density ( spatial as well as energy - resolved ) within the tise@xmath1 remains largely unaffected by the presence of intercalant atoms and very much close to that of pure tise@xmath1 ( see ref .
we notice however a difference of our electronic structure from that calculated for tise@xmath1 in ref .
@xcite , with the use of the muffin - tin approximation ) .
the local dos at the @xmath4 site contains spin - splitted and broadened levels , typical for a magnetic impurity embedded in the vb ( the latter being formed , in this case , by the ti@xmath3se@xmath10 hybridization ) .
an anomalously narrow minority - spin peak appears at the co site because its energy position is between the separated subbands due to a bonding and antibonding ti
se interaction ( see more detailed discussion in ref .
@xcite ) , i.e. the scattering of valence - band electrons at this impurity level is suppressed .
a more detailed analysis of the crystal - field splitting and spatial distribution of the @xmath2 orbitals will be given elsewhere @xcite .
it should be noted that the present calculation for the ordered cr@xmath67tise@xmath1 structure failed to find an energy gap at the fermi level ( implied by the conductivity type as shown in fig .
[ fig : conduc ] for @xmath33=0.33 ) .
the reason may be either an effect of the lattice parameters change , important for the cr - intercalated systems but ignored in the present calculation , or the importance of correlation effects due to high cr@xmath3 localization , that would justify the use of a different ( orbital - dependent ) form of the total energy functional .
this subject yet needs a further clarification .
the values of local magnetic moments ( attributed to muffin - tin sphere sizes cited above ) are 2.81 @xmath71 for cr and 1.46 @xmath71 for co , that agrees well with the experimental values of table [ tab : mom ] for @xmath33=0.33 . in spite of a slight variation of the magnetic density over the crystal ( small antiparallel polarization of ti@xmath3 shells and parallel of the se@xmath10 shells )
, the magnetic moment remains quite localized at the intercalant atom .
for the values of the @xmath37 exchange splitting ( taken simply as energy differences between spin - up and spin - down components of self - consistently calculated relativistic core levels ) , we obtained 0.85 ev at the cr site and 0.59 ev at the co site .
the former is in good agreement with the experimental estimate , and the latter is probably already too small to be resolved with the experimental resolution ( see fig . [
fig : xpscore ] ) . with the information about the local state density distribution at intercalant sites available , it becomes possible to address the problem of x - ray re - emission in the cr@xmath0tise@xmath1 systems with low @xmath33 content .
an assumption of a strongly localized character of cr@xmath3 states is supported by the calculations of both spatial distribution ( fig .
[ fig : dens_line ] ) and the local dos ( fig .
[ fig : dos ] ) .
localized states in the vacant region ( which may host an absorbed electron in an intermediate state of a x - ray emission process ) must be necessarily those in the minority - spin channel . in order to imitate their possible participation in the x - ray re - emission , we calculated the cr@xmath40 x - ray emission spectrum , taking into account dipole transition matrix elements and the cutoff energy of 2 ev above the nominal fermi level , to allow the lowest minority - spin band localized at the cr site to contribute in full .
this band incorporates more than two extra electrons , rather than one as would be the case in a simple intermediate - state model . however , there is no simple criterium to select only part of these low - lying vacant states for their possible contribution to the re - emission process .
the result of a calculation is shown in fig .
[ fig : xes ] , top curve . without
the fermi energy artificially shifted , the higher peak in both @xmath43 and @xmath44 bands is missing , so that the calculated spectrum is rather close to that of pure chromium .
no such artificial adjustment was necessary for the co@xmath0tise@xmath1 system .
as a result of this combined study incorporating measurements of the crystal structure , kinetic properties ( conductivity ) and the analysis of the electronic structure by different spectroscopic techniques ( xps and xes ) , interpreted on the basis of electronic structrure calculations , we establish the following . in spite of nominally identical magnetic moments of a free ion for cr and co , the saturation magnetization of intercalated tise@xmath1 reveals an almost two times larger effective moment on cr than that on co sites . a large exchange splitting in the cr@xmath3 shell polarizes
the cr@xmath37 shell and leads to the exchange splitting of cr@xmath37 core levels large enough ( 0.9 ev ) to be detected by photoelectron spectroscopy .
no such splitting was resloved for co@xmath37 levels .
so different magnetic behavior can be traced to different strength of chemical bonding in cr- and co - intercalated systems .
the bonding is stronger in co@xmath0tise@xmath1 , as is evidenced by larger distortion of the host lattice .
the cr@xmath3 states , on the contrary , remain more localized , that favors a higher magnetic moment .
another indication of higher localization of cr@xmath3 states in cr@xmath0tise@xmath1 is a strong contribution in the x - ray emission @xmath40 spectrum from above the nominal fermi level , observed even at excitation energies much higher than the threshold and tentatively attributed to re - emission .
we emphasize that the exchange splitting of @xmath37 core levels and a strong re - emission due to high - energy electronic excitations are experimental findings quite rarely observed in @xmath3-based systems so far .
the reported first - principles calculations provide an adequate description of the structural and magnetic properties and simplify at least a qualitative understanding of the underlying physics
. however , an attempt to take into account correlation effects more systematically could be interesting for the treatment of highly localized cr@xmath3 states and of the excitation processes involving the latter .
this work was supported by the russian science foundation for fundamental research ( projects 96 - 15 - 96598 and 99 - 03 - 32503 ) , a nato linkage grant ( htech.lg 971222 ) and a dfg - rffi project . financial support from the deutsche forschungsgemeinschaft is greatly acknowledged . | the results of investigations of intercalated compounds cr@xmath0tise@xmath1 and co@xmath0tise@xmath1 by x - ray photoelectron spectroscopy ( xps ) and x - ray emission spectroscopy ( xes ) are presented .
the data obtained are compared with theoretical results of spin - polarized band structure calculations . a good agreement between theoretical and experimental data for the electronic structure of the investigated materials has been observed .
the interplay between the @xmath2ti@xmath3 hybridization ( @xmath4=cr , co ) and the magnetic moment at the @xmath4 site is discussed . a 0.9 ev large splitting of the core cr@xmath5 level
was observed , which reveals a strong exchange magnetic interaction of 3d-2p electrons of cr . in the case of a strong localization of the cr@xmath3 electrons ( for @xmath6 ) ,
the broadening of the cr@xmath7 spectra into the region of the states above the nominal fermi level was observed and attributed to x - ray re - emission .
the measured kinetic properties are in good accordance with spectral investigations and band calculation results . |
this paper was published in soft computing ( springer - verlag ) on 23 july 2016 .
the final publication is available at http://dx.doi.org/10.1007/s00500-016-2275-y
in this paper we study the expansion of the substructural logic fl@xmath0 , i.e. full lambek calculus with exchange and weakening , with a unary connective @xmath1 whose intended algebraic semantics is as follows : given a bounded integral commutative residuated lattice ( or residuated lattice for short ) * a * , @xmath2 is the maximum , if it exists , of the boolean elements of the universe @xmath3 below @xmath4 , which we call _ the greatest boolean below a _ , that is , @xmath5 in fact , this operator is similar to the so - called baaz - monteiro @xmath6 operator , very often used in the context of mathematical fuzzy logic systems that are semilinear expansions of mtl .
baaz @xcite studied it in connection with gdel logic while hjek @xcite investigated @xmath6 in bl logics in general , see also ( * ? ? ?
* chapter 2 ) for a more general perspective . indeed , in such a context of semilinear logics , i.e. logics that are complete with respect to a class of linearly ordered algebras , the semantics of @xmath6 is exactly the above one for @xmath1 : in a linearly - ordered mtl - algebra , @xmath7 if @xmath8 , and @xmath9 otherwise , since the only boolean elements in a chain are 1 and 0 ; moreover , from a logical point of view , @xmath10 represents the weakest boolean proposition implying @xmath11 .
the operator @xmath1 can be also related to the join - complement operation @xmath12 , also known as dual intuitionistic negation , already considered by skolem @xcite in the context of lattices with relative meet - complement , and later independently studied by e.g. moisil @xcite and rauszer @xcite as well , the latter in the context of expansions of heyting algebras .
it turns out that the operation @xmath13 and its iterations , where @xmath14 is the residual negation , has also very similar properties to @xmath1 , and in some classes of residuated lattices they even coincide . in this paper
we study the operator @xmath1 in the context of fl@xmath0 and axiomatize it .
we show that the usual axiomatics of the @xmath6 operator is actually too strong to capture the above intended semantics .
in fact , the axiom @xmath15 is not sound for @xmath1 over fl@xmath0 any longer .
thus , @xmath1 is a weaker operator than @xmath6 .
however , as we will see , @xmath1 keeps most of the properties of @xmath6 .
in particular , the expansion of fl@xmath0 with @xmath1 is conservative , its corresponding class of algebras is an equational class , and has the same kind of deduction theorem as @xmath6 .
also , @xmath1 may also be interesting as @xmath16 has a paraconsistent behaviour .
on the negative side , the expansion of a semilinear extension of fl@xmath0 with @xmath1 needs not to remain semilinear .
the paper is structured as follows . in section 2
we overview well - known facts about residuated lattices and its boolean elements , as well as basic facts about the logic fl@xmath0 .
sections 3 and 4 contain an algebraic study of the operator @xmath1 .
in particular , in section 3 we study basic properties and show , among other things , that the class @xmath17 of residuated lattices expanded with @xmath1 is an equational class and state the modalities , while in section 4 we compare @xmath1 with the mentioned @xmath6 and with an operation using the join - complement @xmath12 .
finally , in section 5 we focus on logical aspects , introducing the logic fl@xmath18 , i.e. the expansion of fl@xmath0 with the operator @xmath1 , and show that is a conservative expansion and has the finite model property , and hence it is decidable .
we conclude with some remarks and open problems .
we give appropriate references .
however , the paper is self - contained .
in this section we recall some properties of residuated lattices as well as of their boolean elements that we will use in the following sections .
following @xcite , a bounded , integral , commutative residuated lattice , or residuated lattice for short , is an algebra * a*@xmath19 of type @xmath20 such that : * @xmath21 is a bounded lattice with @xmath22 , for all @xmath23 , * @xmath24 is a commutative monoid ( i.e. @xmath25 is commutative , associative , with unit @xmath26 ) , and * @xmath27 is the residuum of @xmath25 , i.e. , + @xmath28 iff @xmath29 , for all @xmath30 , where @xmath31 is the order given by the lattice structure .
a negation operator is defined as @xmath32 .
the class of residuated lattices will be denoted by @xmath33 .
it is well known that @xmath33 is an equational class and that it constitutes the algebraic semantics of the substructural logic fl@xmath0 ( see section [ flew ] ) . [ 5 ] in what follows we will have occasion to refer several times to the residuated lattice structure defined on the five - element lattice of figure [ five - godel ] by taking @xmath34 and @xmath35 its residuum . with these operations , it actually becomes a five element gdel algebra , that is , a residuated lattice with @xmath25 being idempotent and satisfying the pre - linearity law @xmath36 .
[ ht ] = [ draw , circle , fill = white , minimum size=3pt , inner sep=0pt , label distance=1 mm ] ( 0,0 ) node ( 0 ) [ label = right:@xmath37 + + ( 90:1 cm ) node ( r ) [ label = right:@xmath38 + + ( 135:1 cm ) node ( s ) [ label = left:@xmath39 + + ( 45:1 cm ) node ( 1 ) [ label = right:@xmath26 ]
+ + ( 315:1 cm ) node ( t ) [ label = right:@xmath41 + +
( r ) ; end we omit the proof of the following well - known facts , see e.g. @xcite .
[ grl ] let @xmath42 . for any @xmath43 ,
the following properties hold : _ ( i ) _ if @xmath44 , @xmath45 , and @xmath46 , then @xmath47 , _ ( ii ) _ if @xmath44 , @xmath48 , and @xmath49 , then @xmath50 , _ ( iii ) _ if @xmath51 , then @xmath52 , _ ( iv ) _
@xmath53 , _ ( v ) _
@xmath54 , _ ( vi ) _ if @xmath44 , then @xmath55 , _ ( vii ) _ @xmath56 .
special elements in a residuated lattice are those that behave as elements in a boolean algebra .
let @xmath57 . an element @xmath4 of its universe @xmath3
is called _ boolean _ or _ complemented _ iff there is an element @xmath58 such that @xmath59 and @xmath60 . in the rest of this section we state several properties of boolean elements that will be useful in what follows .
even if most of them are folklore , we include proofs for all of them for the sake of being self - contained .
an equivalent and simpler condition for an element to be boolean is the following .
[ bl ] an element @xmath4 in the universe of a residuated lattice is boolean iff @xmath61 .
@xmath62 suppose there is an element @xmath63 such that @xmath59 and @xmath60 .
first , using that @xmath59 and @xmath64 , we have that @xmath65 .
so , @xmath66 , i.e. @xmath67 . secondly , we have that @xmath68 .
so , @xmath69 .
as @xmath70 , it follows that @xmath55 .
so , @xmath71 .
as we have that @xmath60 , it follows that @xmath72 .
@xmath73 by hypothesis , we have ( i ) @xmath74 . it is enough to see that @xmath75 . as @xmath76
, it is enough to prove that @xmath77 .
we have that @xmath78 .
so , by monotonicity of @xmath25 , we have ( ii ) @xmath79 .
we also have that @xmath80 .
so , again by monotonicity of @xmath25 , we have @xmath81 .
so , it follows ( iii ) @xmath82 .
now , using lemma [ grl](ii ) with ( i ) , ( ii ) , and ( iii ) , it follows that @xmath83 .
[ infmon ] let @xmath57 and let @xmath4 be a boolean element of its universe @xmath3 .
then , for all @xmath84 the following properties hold : _ ( i ) _
@xmath85 , _ ( ii ) _
@xmath86 , _
@xmath75 , _ ( iv ) _
@xmath87 , _ ( v ) _
@xmath88 , _ ( vi ) _
@xmath89 , _ ( vii ) _
@xmath90 , _ ( viii ) _ if @xmath91 , @xmath92 , @xmath93 , then @xmath94 , _ ( ix ) _ if @xmath95 , @xmath92 , @xmath93 , then @xmath94 , _ ( x ) _ if @xmath96 , then @xmath97 , _ ( xi ) _ if @xmath98 , then @xmath99 .
\(i ) suppose that ( i ) @xmath100 .
it is enough to see that @xmath101 .
we have that @xmath102 .
so , by monotonicity of @xmath25 , we have ( ii ) @xmath103 .
we also have that @xmath104 .
so , again by monotonicity of @xmath25 , we have @xmath105 .
so , it follows ( iii ) @xmath106 .
now , using lemma [ grl](ii ) with ( i ) , ( ii ) , and ( iii ) , it follows that @xmath101 .
\(ii ) using part ( i ) , we have @xmath107 .
also , @xmath108 .
so , @xmath109 .
\(iii ) using part(i ) , we have @xmath110 .
also , @xmath76 .
so , @xmath75 .
\(iv ) in a residuated lattice it holds that @xmath111
. let @xmath4 be boolean . then , using part ( i ) three times , it follows that @xmath112 .
\(v ) suppose that ( i ) @xmath61 .
it is enough to see that @xmath113 .
we have that ( ii ) @xmath114 . also , ( iii ) @xmath115 , as @xmath116 .
so , using lemma [ grl](ii ) with ( i ) , ( ii ) , and ( iii ) , @xmath113 .
\(vi ) it is enough to prove ( i ) @xmath117 and ( ii ) if @xmath118 , then @xmath119 . to see ( i ) , note that @xmath120 and @xmath121 , whence @xmath122 .
so , using distributivity of @xmath25 relative to @xmath123 , we get ( i ) . to see ( ii ) , suppose @xmath118 . then , @xmath124 . in order to get @xmath119 ,
it is enough to derive ( iii ) @xmath125 and use transitivity of @xmath31 . to get ( iii ) , let us use lemma [ grl](ii ) . as @xmath4 is boolean
, we have @xmath74 .
now , @xmath126 . also , @xmath127 .
so , using lemma [ grl](ii ) , we get ( iii ) .
\(vii ) as we have @xmath128 , for any @xmath129 , then , using lemma [ grl](iii ) and part ( iv ) , we get @xmath130 .
\(viii ) as @xmath91 , we have @xmath131 . now , using part ( iv ) , it follows that @xmath132 .
now , as @xmath92 and @xmath93 , we have , by a basic property of @xmath123 , that @xmath133 .
so , by transitivity of @xmath31 , @xmath94 .
\(ix ) as @xmath95 , reason as in part ( viii ) .
\(x ) suppose @xmath96
. then @xmath28 . by monotonicity of @xmath25
, it follows that @xmath134 .
then , as @xmath25 is both associative and commutative , @xmath135 .
so , @xmath136 . finally , using that @xmath4 is boolean , we get @xmath97 .
\(xi ) let @xmath98 .
then , @xmath137 . by distributivity of @xmath25
relative to @xmath123 , it follows that @xmath138 . as @xmath139
, we have that @xmath140 .
so , as @xmath4 is boolean , @xmath141 , i.e. @xmath99 . [ inflem ] let @xmath57 and let @xmath4 and @xmath63 be boolean elements of @xmath3 . then , _
@xmath142 , _ ( ii ) _ @xmath143 .
\(i ) firstly , we have that @xmath144 .
so , using lemma [ grl](iii ) , it follows that @xmath145 .
now , using proposition [ infmon](v ) and @xmath31-transitivity , we have that @xmath146 .
analogously , we get @xmath147 . secondly ,
suppose @xmath148 and @xmath149 , for @xmath129 .
then , using lemma [ grl](iii ) again , it follows that @xmath150 and @xmath151 .
so , @xmath152 .
then , using lemma [ grl](iii ) once again , @xmath153 .
now , using proposition [ infmon](v ) and @xmath31-transitivity , we get @xmath154 .
\(ii ) suppose that @xmath155 . then , @xmath156 . [ infmon2 ]
let @xmath57 and let @xmath4 and @xmath63 be boolean elements of @xmath3 .
then , ( i ) @xmath157 , ( ii ) @xmath158 , ( iii ) @xmath159 , ( iv ) @xmath160 , ( v ) @xmath161 , and ( vi ) @xmath26 are boolean .
\(i ) suppose @xmath74 .
then , as @xmath56 , we get @xmath162 .
\(ii ) use lemma [ inflem ] ( ii ) and see that @xmath163 , @xmath164 , @xmath165 , and @xmath166 .
so , @xmath167 is boolean .
\(iii ) use parts ( i ) and ( ii ) , and lemma [ inflem](i ) .
\(iv ) use parts ( i ) and ( ii ) , and proposition [ infmon](vi ) .
\(v ) use parts ( i ) and ( ii ) , and proposition [ infmon](vii ) .
\(vi ) use the definition of boolean element and the fact that @xmath168 . from proposition [ infmon2 ]
it easily follows that , in any residuated lattice @xmath169 , the set of its boolean elements @xmath170 is boolean@xmath171 is the domain of a subalgebra of @xmath169 , which is in fact a boolean algebra .
indeed , @xmath172 is the greatest boolean algebra contained in * a*. * b(a ) * is called the boolean skeleton or the _ center _ of * a*. the logics we are interested in are extensions or expansions of the logic fl@xmath0 described below .
the language of fl@xmath0 has four binary connectives , @xmath173 , @xmath123 , @xmath25 , and @xmath174 , and two constants , @xmath161 and @xmath26 .
the axioms of fl@xmath0 are : [ cols= " < , < " , ] \(ii ) follows easily using ( i ) .
clearly , fl@xmath175 is a _
rasiowa implicative _ logic ( cf .
then , it follows that it is algebraizable in the sense of blok and pigozzi @xcite .
it is also straightforward to check that the variety @xmath17 is its equivalent algebraic semantics .
algebraizability immediately implies strong completeness of fl@xmath175 with respect to @xmath17 .
for every set
@xmath176 of formulas , @xmath177 iff for every @xmath178 and every @xmath179-evaluation @xmath180 , @xmath181 , whenever @xmath182 \subseteq \{1\}$ ] . in fl@xmath175 the usual form of the deduction theorem
does not hold .
indeed , @xmath183 , but @xmath184 , which fails in the three - element gdel algebra @xmath185_g$ ] , where @xmath186 and @xmath187 : for any evaluation @xmath180 in this algebra , if @xmath181 , then @xmath188 , but for @xmath189 we have @xmath190 , and thus @xmath191 .
actually , fl@xmath175 enjoys the same form of deduction theorem holding for logics with the @xmath6 operator ( cf .
* proposition 2.2 ) ) . [ deduction ] @xmath192 iff @xmath193 .
@xmath194 ) we prove by induction on every formula @xmath195 ( @xmath196 ) of the given derivation of @xmath197 from @xmath198 that @xmath199 .
if @xmath200 , then the result follows due to axiom schema * ( b1)*. if @xmath195 belongs to @xmath201 or is an instance of an axiom , then the result follows using _ modus ponens _ and the derivability of the schema @xmath202 .
if @xmath195 comes by application of _ modus ponens _ on previous formulas in the derivation , then the result follows , because from @xmath203 and @xmath204 we may derive @xmath205 and then also @xmath206 , using transitivity of @xmath35 applied to the derivable formulas @xmath207 and @xmath208 .
finally , if @xmath209 comes using rule * ( b ) * from formula @xmath210 , then from @xmath211 we may derive @xmath212 using lemma [ lflb](ii ) .
@xmath213 ) to the derivation given by the hypothesis add a step with @xmath11 . in the next step put @xmath214 , which follows from the previous formula using rule *
( b)*. finally , derive @xmath197 using _
modus ponens_. thanks to this @xmath1-deduction theorem , the logic fl@xmath175 has the following property : if we expand fl@xmath175 with any further rule @xmath215 , , @xmath216 , then it is possible to dispose of the rule just adding the axiom @xmath217 .
this property also holds for the logics fl@xmath218 and fl@xmath219 .
fl@xmath175 is a conservative expansion of fl@xmath0 .
use proposition [ finb ] and the finite model property of fl@xmath0 ( see @xcite ) .
one could analogously define the expansion of mtl ( which is in turn the extension of fl@xmath0 with the pre - linearity axiom @xmath220 ) with @xmath1 , with the same additional axioms and rule , yielding the logic mtl@xmath221 , which is algebraizable and strongly complete with respect to the variety @xmath222 of mtl@xmath221-algebras .
however , unlike the case of expansion with @xmath6 , mtl@xmath221 is not a semilinear logic , that is , it is not complete with respect to the class of mtl@xmath221-chains .
the reason is that the @xmath223-form of rule * ( b ) * , `` from @xmath224 derive @xmath225 '' , is not derivable in mtl@xmath221 . indeed , taking the coatoms @xmath226 and @xmath227 in the gdel algebra of example [ 5 ] , it is clear that @xmath228 , while @xmath229 . as a final result
, we can show that fl@xmath175 inherits from fl@xmath0 the finite model property ( fmp ) . before proving this
, we introduce some preliminary notation . for a logic @xmath230 fl@xmath0 or fl@xmath231 , let us denote by @xmath232 the set of @xmath233-formulas built from a set @xmath234 of propositional variables .
now let us define the enlarged set of propositional variables @xmath235fl@xmath236 , where @xmath237 is intended to denote a fresh propositional variable , one for each formula @xmath238fl@xmath239 .
then , we can define a one - to - one translation of every formula @xmath240fl@xmath239 into a formula @xmath241fl@xmath242 , by just inductively defining : - @xmath243 , + - if @xmath244 , then @xmath245 , + - if @xmath246 , then @xmath247 , + - if @xmath248 , then @xmath249 , for @xmath250 .
if @xmath201 is a set of formulas , we write @xmath251 . note that for any @xmath252fl@xmath242 , there is a formula @xmath240fl@xmath239 such that @xmath253 .
moreover , we need the following result that will allow us to reduce proofs in fl@xmath18 to proofs in fl@xmath0 .
[ reduce ] let @xmath254 be the set of all instances of axioms of fl@xmath18 .
for each set @xmath255fl@xmath239 , it holds that @xmath256 iff @xmath257 , where @xmath258 .
the proof is quite straightforward and analogous to those of similar results that can be found in the literature in slightly different contexts .
fl@xmath175 enjoys the fmp , that is , if @xmath259 , then there is a finite @xmath260 and an @xmath261-evaluation @xmath180 such that @xmath262 and @xmath263 .
if @xmath259 , by lemma [ reduce ] , it holds that @xmath264 , and by strong completeness and fmp of fl@xmath0 , there is a finite algebra @xmath265 and * c*-evaluation @xmath266 such that @xmath267 and @xmath268
. then , the result will follow from the following facts : + * _ claim 1 _ * : @xmath269fl@xmath270 is a set of boolean elements of @xmath271 .
`` you _ were _ a little grave , '' it is enough to check that @xmath272 , where the latter holds because @xmath273 is the axiom * ( b2 ) * of fl@xmath18 .
* _ claim 2 _ * : let @xmath169 be the @xmath33-algebra generated by the set @xmath274fl@xmath275 , which is finite since @xmath261 is a subalgebra of @xmath276 . then , @xmath1 exists in @xmath169 and @xmath277 .
therefore , @xmath169 is indeed an @xmath17-algebra . that @xmath169 is finite is obvious , and thus , by proposition 5 , @xmath1 exists . on the other hand , the elements of @xmath278
keep being boolean in @xmath169 .
hence , the only missing thing to check is that any boolean element of @xmath169 already belongs to @xmath278 .
this is also clear since boolean elements are closed by propositional combinations with connectives . * _ claim 3 _ * : let us define the @xmath169-evaluation ( taking @xmath169 as @xmath279-algebra ) @xmath280 defined by @xmath281 . then , for any @xmath11 , @xmath282 , in particular , @xmath283 .
we prove that @xmath282 by structural induction . *
if @xmath11 is a propositional variable , it holds by construction * if @xmath284 for @xmath250 , by induction hypothesis we have @xmath285 and @xmath286 , and hence @xmath287 . *
if @xmath288 , then we have to prove that @xmath289 , the latter being equal to @xmath290 by definition .
therefore , we have to prove in turn that the three defining conditions * ( be1 ) * , * ( be2 ) * , and * ( bi ) * are satisfied by @xmath291 to be the greatest boolean below @xmath292 , assuming by induction that @xmath293 . * * since @xmath294 is axiom * ( be1 ) * of @xmath295 , we have that @xmath296 .
hence , @xmath297 . * * is clear from claim 1 . *
* we have to check that if @xmath298 is such that @xmath299 , then @xmath300 . if @xmath298 , by construction of @xmath169 , then there exists a formula @xmath301 such that @xmath302 . on the other hand , by ( ii ) of lemma 12 , we know that @xmath303 .
thus , we also know that if @xmath304 and @xmath305 , then @xmath306 .
now , the two conditions are satisfied , hence we have @xmath307 .
this closes the proof of claim 3 .
finally , from these claims it readily follows that @xmath308 and @xmath309 , as required .
in this paper we have considered the expansion of fl@xmath0 with the operator @xmath1 , that in algebraic terms provides the greatest boolean below a given element of a residuated lattice . among other things ,
we have axiomatized it and shown that the resulting logic is a conservative expansion enjoying the finite model property .
the axioms for @xmath1 turn out to be very close to those of the monteiro - baaz @xmath6 operator , in fact only one axiom is a weaker version of the one for @xmath6
. even if the properties are very similar , that small difference causes , e.g. that in the context of mtl , the expansion with @xmath1 is not any longer a semilinear logic , in contrast to the expansion with @xmath6 . as a matter of fact
, we have chosen this topic for our humble contribution to honour the memory of our beloved and late friend franco montagna , because it was suggested by franco to the first author during the preparation of their joint manuscript @xcite , together with amidei , where they study the expansion of fl@xmath0 and other substructural logics with @xmath6 .
the authors are thankful to an anonymous reviewer for his / her comments that have helped to improve the final layout of this paper .
the authors have been funded by the eu h2020-msca - rise-2015 project 689176sysmics .
esteva and godo have been also funded by the feder / mineco spanish project tin2015 - 71799-c2 - 1-p .
the authors declare that they have no conflict of interest .
this article does not contain any studies with human participants or animals performed by any of the authors .
amidei , jacopo , ertola - biraben , rodolfo c. , and montagna , franco .
conservative expansions of substructural logics .
( submitted ) .
preprint available as cle e - prints , vol .
16(2 ) , 2016 . ( http://www.cle.unicamp.br/e-prints/vol_16,n_2,2016.html )
baaz , matthias .
infinite - valued gdel logics with 0 - 1-projections and relativizations . in _
gdel96 - logical foundations of mathematics , computer science and physics ; lecture notes in logic 6 _ , 1996 , p. hjek , ed . , springer - verlag , pp .
23 - 33 .
galatos , nikolaos , jipsen , peter , kowalski , tomasz , and ono , hiroakira .
_ residuated lattices : an algebraic glimpse at substructural logics _
, studies in logic and the foundations of mathematics , vol .
151 , elsevier , new york , 2007 .
skolem , thoralf .
untersuchungen ber die axiome des klassenkalkls und ber produktations- und summationsprobleme , welche gewisse klassen von aussagen betreffen , _ skrifter uitgit av videnskapsselskapet
i kristiania , i _ , matematisk - naturvidenskabelig klasse , 3 , 1919 , pp . 1 - 37 | we expand fl@xmath0 with a unary connective whose algebraic counterpart is the operation that gives the greatest complemented element below a given argument .
we prove that the expanded logic is conservative and has the finite model property .
we also prove that the corresponding expansion of the class of residuated lattices is an equational class . |
the experiments at relativistic heavy ion collider ( rhic ) have shown evidence for the hot and dense matter , also known as quark gluon plasma ( qgp ) , created during the early stages of the collisions @xcite .
the high temperature and baryon density of the produced matter makes it most suitable environment for the production of light nuclei ( p , d , @xmath4 , @xmath5 ) , hypertriton and dibaryons ( @xmath6 , p@xmath7 , @xmath8 , @xmath9 ) as well as their antiparticles . for long time
the study of light ( anti)nuclei and ( anti)hypernuclei production has remained of interest for physicists @xcite .
these studies are important to understand the matter - antimatter symmetry , dark matter and structure of neutron star @xcite .
antihypertriton ( @xmath10 ) and antihelium-4 ( @xmath11 ) have already been observed at rhic @xcite and large hadron collider @xcite .
very recently , interaction between antiproton pairs has been also measured by the star experiment @xcite .
the production of light ( anti)nuclei and ( anti)hypernuclei in heavy ion collisions is fairly described by the thermal model @xcite and the coalescence model based on multiphase transport model as well as other transport models @xcite .
the production of light ( anti)nuclei and ( anti)hypernuclei in the most central au+au collisions at @xmath0 200 gev has been studied using coalescence model and hydrodynamic blast - wave model in @xcite .
using the same model of ref .
@xcite , the production of light ( anti)nuclei and ( anti)hypernuclei in the most central au+au collisions at @xmath0 11.5 gev are discussed in this article .
different quantum chromodynamics ( qcd ) based models have proposed existence of dibaryons as exotic form of matter . the h dibaryon was first predicted by jaffe @xcite and then later many other dibaryon states were predicted , e.g. p@xmath7 @xcite , @xmath8 @xcite and @xmath9 @xcite .
recently experiments at rhic @xcite and lhc @xcite have searched for h dibaryon . with the advancement in computation power
reasonable theoretical progress has been made to understand dibaryon structure @xcite .
however information about the invariant yield of dibaryons from heavy ion collisions remains scarce and more efforts are required in this direction .
the invariant yield of dibaryons @xmath6 , p@xmath7 , @xmath8 and @xmath9 are presented for central au+au collisions at @xmath0 11.5 and 200 gev .
the baryon - strangeness correlation coefficient @xmath12 is proposed as a diagnostic tool to understand the nature of matter formed in heavy ion collisions @xcite . for qgp state
the @xmath12 is expected to be unity , however a significant dependence of @xmath12 on hadronic environment is observed by v. koch , a. majumder and j. randrup @xcite .
measurement of @xmath12 in experiments is a technical challenge as one needs to measure baryon number and strangeness on event - by - event basis .
therefore the strangeness population factor @xmath13 was introduced by t. a. armstrong _
et al . _
@xcite , which fairly depicts the local correlation between baryon number and strangeness @xcite .
further we introduce @xmath14 , which represents the local strangeness - strangeness correlations . keeping in mind the technical challenges to measure @xmath12 , in this letter , we concentrate on the strangeness population factor @xmath13 , @xmath14 for central au+au collisions at @xmath0 11.5 and 200 gev .
= 4.8 in = 3.5 in
a naive coalescence model is used to study the production of multistrange hadrons , light nuclei and hypertriton in central au+au collisions at @xmath0 11.5 and 200 gev .
it is assumed that the production of these particles occur at the kinetic freeze - out stage . in this case
the particle production probability is proportional to the primordial hadron density and can be described by following equation @xcite : @xmath15 where e@xmath16 are the invariant yield of particles ( a , b and c ) under consideration , @xmath17 and @xmath18 are their momenta , b is the coalescence parameter and @xmath19 .
the phase space information from the hydrodynamic blast - wave model is used as an input to the equation ( [ ea ] ) to calculate the invariant yields of @xmath6 , p@xmath7 , @xmath20 , @xmath9 , light nuclei and hypertriton . in hydrodynamic blast - wave model @xcite ,
the system is characterized by these parameters : the kinetic freeze - out temperature @xmath21 , the radial flow parameter @xmath22 and elliptic flow parameter @xmath23 , the spatial anisotropy a , the average transverse radius r , and the particle emission duration @xmath24 .
it is assumed that the fireball created in heavy ion collision is in local thermal equilibrium and moves outward with velocity @xmath25 .
the phase - space emission points for hadrons are defined as a wigner function : @xmath26 where y is the rapidity , @xmath27 is transverse mass , @xmath28 is four momentum , and ( 2s+1 ) is the degeneracy due to spin of hadrons .
@xmath29 is given by @xmath30 where @xmath31 is the transverse position of the hadrons in coordinate space .
then we can define the azimuthally integrated @xmath1 spectrum as @xmath32 results obtained for the invariant yields of multistrange hadrons , nuclei and hypertriton using equation [ ea ] and [ dnptdpt ] are discussed in next section .
to study dibaryons , light nuclei and hypertriton production in the central au+au collisions at the rhic energies 200 gev and 11.5 gev , we use following parameters derived from the star experiment at the rhic as input to the hydrodynamic blast - wave model : kinetic freeze - out temperature = 89 ( 120 ) mev , baryo - chemical potential = 21.9 ( 315 ) mev , strangeness chemical potential = 6.5 ( 68 ) mev and radial flow parameter @xmath22 = 0.91 ( 0.46 ) for the @xmath33 ( 11.5 ) gev @xcite . the elliptic flow parameter @xmath23 = 0 , spatial anisotropy a = 1 , average transverse radius r = 10 fm and finite longitudinal proper time = 6.2 fm / c are set same for both @xmath34 and 11.5 gev @xcite .
similar calculations were done by k .- j .
sun and l .- w .
chen in @xcite , where the freeze - out parameters are higher than the parameters used in our calculations .
the proton spectra used from the phenix collaboration to derive the freeze - out parameter in @xcite are not corrected for the feed - down from @xmath35 and @xmath36 baryons.the coalescence of hadron occurs when @xmath37 and @xmath38 mev / c , where @xmath39 and @xmath40 are the phase - space position of the two constituent hadrons , and @xmath41 is the nuclear force radius . for deuteron and multistrange dibaryon @xmath41 = 1.57 fm
is used and for the other nuclei @xmath41 = 1.5 fm is used .
the first two panels in figure [ fig : spectra_11gev ] show differential yields of p , @xmath35 , @xmath42 , @xmath7 , @xmath6 , p@xmath7 , @xmath20 and @xmath9 produced in central au+au collisions at @xmath43 and 11.5 gev respectively .
our calculation can reproduce the data for proton , @xmath35 , @xmath42 and @xmath7 from the star experiment at both energies @xcite .
the light ( anti)nuclei and hypertriton spectra for @xmath44 gev are shown in third panel of figure [ fig : spectra_11gev ] . for @xmath43 gev the light ( anti)nuclei spectra
are taken from the article @xcite , where same coalescence model was used . the @xmath1 integrated yield for light nuclei and dibaryons in the central rapidity
are given in table [ tab : int_dndy ] .
we observe that expected yields of all the particles at @xmath44 gev are significantly higher than @xmath43 except for @xmath9 , may be because of competition between strangeness production mechanism at this energy .
figure [ fig : rapidity ] shows the rapidity distribution of p , @xmath35 , @xmath42 , @xmath7 , @xmath6 , p@xmath7 , @xmath20 , @xmath9 , d , @xmath4 and hypertriton in central au+au collision at @xmath44 gev from the hydrodynamical blast - wave model plus a coalescence model .
since uniform rapidity distribution is used for @xmath43 gev , we have not shown those rapidity distributions here .
= 3.3 in = 2.5 in the p@xmath45 integrated yields _ dn / dy _ of multistrange hadrons as a function of strangeness @xmath46 for central au+au collisions at @xmath44 gev ( left ) and 200 gev ( right ) are shown in figure [ fig : yieldvss ] , where filled symbols are data from the star experiment @xcite and different lines represent our calculations from the hydrodynamical blast - wave model plus a coalescence model .
the @xmath6 and @xmath9 dibaryon production yields at top rhic energy were estimated by the exhic collaboration based on a realistic coalescence model and statistical model @xcite .
those yields are compared with our calculations in the figure [ fig : yieldvss ] .
we observe an exponential behavior of the invariant yield of multistrange hadrons similar to light nuclei @xcite .
the yield for baryon and dibaryon systems are fitted with function @xmath47 , where @xmath48 is number of initial strange hadrons , @xmath49 is penalty factor and @xmath50 is the strangeness .
the penalty factor quantitatively tells us how hard it is to produce a hadron with strangeness ( @xmath46 + 1 ) compared to a hadron with strangeness ( @xmath46 ) .
we obtain @xmath49 = 9.86 for baryons and @xmath49 = 4.62 for dibaryon system from the model for central au+au collisions at @xmath51 gev .
similarly we obtain @xmath49 = 6.46 for baryons and @xmath49 = 4.21 for dibaryon system from the model for central au+au collisions at @xmath33 gev . by fitting the data from the star experiment for baryons , we get @xmath49 = 12.92 @xmath52 1.04 ( 5.71 @xmath52 0.34 ) for the central au+au collisions at @xmath51 ( 200 ) gev .
= 3.5 in = 2.50 in figure [ fig : yieldvsb ] shows the production rate of nuclei as a function of baryon number for the central au+au collisions at @xmath44 gev and 200 gev , where solid points are our calculations using the coalescence model and open symbols are data from the star experiment @xcite . at @xmath43 gev ,
our results are consistent with the star measurement within the uncertainties .
the production rates exhibit a decreasing exponential behavior with the increase in baryon number .
further we obtain the reduction factor by fitting the data with exponential function @xmath53 .
obtained fit values for reduction factor are 1.2@xmath54 ( 1.5@xmath54 ) and 0.33@xmath54 ( 1.95@xmath55)for adding one more nucleon ( antinucleon ) to the system for @xmath43 and 11.5 gev respectively .
the reduction factor obtained from our calculation at @xmath43 gev is comparable with @xmath56 ( @xmath57 ) obtained by the star experiment @xcite .
the production rate for nuclei at @xmath44 gev are significantly higher than @xmath43 gev where the rates decrease sharply for antinuclei at same energy compared to @xmath43 gev .
the difference in reduction factors between matter and antimatter shows a significant energy ( or temperature ) dependence , which illustrate an increasing matter - antimatter asymmetry of the yields at lower energies ( temperatures ) . if we make a rough extension to current universe at room temperature , we can hardly observe the antimatter existence , which is consistent with the current observation of the cosmic rays from which neither antideutron nor antihelium are observed @xcite .
= 3.5 in = 2.50 in the strangeness population factor @xmath58 contains the local baryon - strangeness correlation in the numerator and the baryon - baryon correlation in the denominator @xcite . therefore @xmath13 is quantitatively a good representation of @xmath59 , where @xmath60 is the second derivative of the free energy with respect to the chemical potential , from lattice qcd @xcite .
the ratio @xmath13 as a function transverse momentum is shown in figure [ fig : s3s2 ] ( left ) .
similarly we define @xmath61 for strangeness=-2 dibaryon which contains the local strangeness - strangeness correlation in the numerator and the baryon - baryon correlation in the denominator .
an increase in ratios @xmath14 and @xmath13 is observed for au+au collisions at @xmath3 = 200 gev compared to @xmath3 = 11.5 gev as shown in figure [ fig : s3s2 ] ( right ) .
the ratios @xmath62 for @xmath44 gev and 200 gev are also shown in figure [ fig : s3s2 ] ( left ) .
we observe that the ratio @xmath62 at @xmath44 gev is lower than unity , where the isospin effects become important compared to @xmath63 200 gev .
= 3.5 in = 2.50 in
we presented an interesting calculation for the production of dibaryons , light ( anti)nuclei and hypertriton , based on a naive coalescence model for au+au collisions at @xmath3 = 11.5 and 200 gev .
the exponential behavior of the invariant yields versus strangeness is studied for the multistrange hadrons and penalty factor for the baryon and dibaryon are derived .
the ratios @xmath14 and @xmath13 are discussed for au+au collisions at @xmath3 = 11.5 and 200 gev .
we observe an increase in @xmath14 and @xmath13 at @xmath3 = 200 gev compared to @xmath3 = 11.5 gev .
furthermore our study indicates that the suppression factor for nuclei production at @xmath3 = 11.5 gev is roughly four times smaller than suppression factor at @xmath3 = 200 gev ; leading to higher probability for observation of light nuclei candidates at lower energy .
our calculation will provide the motivation to carry out measurement of @xmath13 , light nuclei and dibaryons during the phase - ii of beam energy scan program at star experiment at rhic @xcite .
this work is supported in part by the major state basic research development program in china under contract no .
2014cb845401 , the national natural science foundation of china under contract nos .
11421505 , 11520101004 , 11275250 , 11322547 .
author n. shah is supported by chinese academy of sciences ( cas ) president s international fellowship initiative no . 2015pm029 .
50 i. aresene _ et al . _ ( the brahms collaboration ) , nucl .
phys . a * 757 * ( 2005 ) 1 .
b. b. back _ et al . _
( the phobos collaboration ) , nucl .
phys . a * 757 * ( 2005 ) 28 .
j. adams _ et al . _ ( the star collaboration ) , nucl . phys . a * 757 * ( 2005 ) 102 .
et al . _ ( the phinex collaboration ) , nucl .
phys . a * 757 * ( 2005 ) 184 .
y. g. ma , j. h. chen , l. xue , front .
* 7 * ( 2012 ) 637 .
y. g. ma , j. h. chen , l. xue , a. tang and z. xu , nucl .
news * 23 * ( 2013 ) 10 .
m. hori , j. walz , prog . in part . and
nucl . phys .
* 72 * ( 2013 ) 206 . c. m. ko _ et al .
sci . and tech .
* 24 * ( 2013 ) 050525 .
b. i. abelev _
et al . _ ( the star collaboration ) , science * 328 * ( 2010 ) 58 ; j. h. chen , nucl .
phys . a * 835 * ( 2010 ) 117 .
h. agakishiev _
et al . _ ( the star collaboration ) , nature * 473 * ( 2011 ) 353 ; l. xue , j. phys .
g * 38 * ( 2011 ) 124072 .
the alice collaboration , arxiv:1506.08453 .
l. adamczyk _
et al . _ ( the star collaboration ) , nature * 527 * ( 2015 ) 345 .
a. andronic , p. braun - munzinger , j. stachel , and h. stcker , phys .
b * 697 * ( 2011 ) 203 .
j. cleymans , s. kabana , i. kraus , h. oeschler , k. redlich , and n. sharma , phys .
c * 84 * ( 2011 ) 054916 .
s. zhang , j. h. chen , h. crawford , d. keane , y. g. ma , and z. b. xu , phys .
b * 684 * ( 2010 ) 224 .
g. chen , h. chen , j. wu , d. s. li and m. j. wang , phys .
c * 88 * ( 2013 ) 034908 .
a. s. botvina , j. steinheimer , e. bratkovskaya , m. bleicher , j. pochodzalla , phys .
b * 747 * ( 2013 ) 7
. l. zhu , c. m. ko and x. yin , arxiv:1510.03568 .
l. xue , y. g. ma , j. h. chen and s. zhang , phys .
c * 85 * ( 2012 ) 064912 ; phys .
c * 92 * ( 2015 ) 059901 ( e ) .
k. j. sun and l. w. chen , phys .
b * 751 * ( 2015 ) 272 .
r. jaffe , phys .
lett * 38 * ( 1977 ) 195 .
t. golman , k. maltman and f. wang , phys .
* 59 * ( 1987 ) 627 .
g. a. miller , chin .
* 51 * ( 2013 ) 466 .
z. y. zhang , y. w. yu , c. r. ching , t. h. ho and z. d. lu , phys . rev .
c * 61 * ( 2000 ) 065204 .
l. adamczyk _
et al . _ ( the star collaboration ) , phys .
* 115 * ( 2015 ) 022301 ; n. shah ( the star collaboration ) , nucl .
phys . a * 914 * ( 2013 ) 410 .
the alice collaboration , arxiv:1506.07499 [ nucl - ex ] .
_ ( the exhic collaboration ) , phys . rev
* 106 * ( 2011 ) 212001 .
et al . _ ( the exhic collaboration ) , phys .
c * 84 * ( 2011 ) 064910 .
s. r. beane _ et al . _
( the nplqcd collaboration ) , phys . rev .
lett * 106 * ( 2011 ) 162001 .
et al . _ ( the hal qcd collaboration ) , phys .
* 106 * ( 2011 ) 162002 .
p. e. shanahan , a.w .
thomas , and r. d. young , phys .
107 * ( 2011 ) 092004 . v. koch , a. majumder , j. randrup , phys .
* 95 * ( 2005 ) 182301 .
s. haussler , s. scherer , m. bleicher , phys .
b * 660 * ( 2008 ) 197 .
t. a. armstrong _
( e864 collaboration ) , phys .
c * 70 * ( 2004 ) 024902 .
h. sato , k. yazaki , phys .
b * 98 * ( 1981 ) 153 .
f. retire and m. a. lisa , phys .
c * 70 * ( 2004 ) 044907 .
l. kumar ( the star collaboration ) , nucl .
phys . a * 931 * ( 2014 ) 1114 .
k. geiger , phys .
* 258 * ( 1995 ) 237 .
j. w. harris and b. mller , annu .
nucl . part .
* 46 * ( 1996 ) 71 .
b. mller , nucl .
phys . a * 630 * ( 1998 ) 461c .
b. i. abelev _
( the star collaboration ) , phys .
lett . b*655 * ( 2007 ) 104 .
g. agakishiev _
et al . _ ( the star collaboration ) , phys .
rev . lett.*108 * ( 2012 ) 072301 .
j. adams _ et al .
_ ( the star collaboration ) , phys . rev
* 98 * ( 2007 ) 062301 .
x. zhu ( for the star collaboration ) , acta phys .
. supp . * 5 * ( 2012 ) 213 ; f. zhao , thesis ( the star collaboration ) , university of california , los angeles , 2014 . h. fuke _
* 95 * ( 2005 ) 081101 .
* 108 * ( 2012 ) 131301 .
et al . _ , phys .
d * 79 * ( 2009 ) 074505 . studying the phase diagram of qcd matter at rhic ( the star collaboration ) , sn0598 , 2014 . | the production of dibaryons , light nuclei and hypertriton in the most central au+au collisions at @xmath0 11.5 and 200 gev are investigated by using a naive coalescence model .
the production of light nuclei is studied and found that the production rate reduces by a factor of 330 ( 1200 ) for each extra nucleon added to nuclei at @xmath0 11.5 ( 200 ) gev .
the @xmath1 integrated yield of multistrange hadrons falls exponentially as strangeness quantum number increases .
we further investigate strangeness population factor @xmath2 as a function of transverse momentum as well as @xmath3 .
the calculations for @xmath0 11.5 gev presented here will stimulate interest to carry out these measurements during the phase - ii of beam energy scan program at star experiment
. strangeness , nuclei , hypertriton , dibaryon |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.